Abstract Top

Background

Sub-Saharan Africa is facing rapidly increasing prevalences of cardiovascular disease, obesity, diabetes and hypertension. Previous and ongoing undernutrition among pregnant women may contribute to this development as suggested by epidemiological studies from high income countries linking undernutrition in fetal life with increased burden of non-communicable diseases in later life. We undertook to study the risks for hypertension, glucose intolerance and overweight forty years after fetal exposure to famine afflicted Biafra during the Nigerian civil war (1967–1970).

Methods and Findings

Cohort study performed in June 27–July 31, 2009 in Enugu, Nigeria. Adults (n = 1,339) born before (1965–67), during (1968–January 1970), or after (1971–73) the years of famine were included. Blood pressure (BP), random plasma glucose (p-glucose) and anthropometrics, as well as prevalence of hypertension (BP>140/90 mmHg), impaired glucose tolerance (IGT; p-glucose 7.8–11.0 mmol/l), diabetes (DM; p-glucose ≥11.1 mmol/l), or overweight (BMI>25 kg/m2) were compared between the three groups. Fetal-infant exposure to famine was associated with elevated systolic (+7 mmHg; p<0.001) and diastolic (+5 mmHg; p<0.001) BP, increased p-glucose (+0.3 mmol/L; p<0.05) and waist circumference (+3cm, p<0.001), increased risk of systolic hypertension (adjusted OR 2.87; 95% CI 1.90–4.34), IGT (OR 1.65; 95% CI 1.02–2.69) and overweight (OR 1.41; 95% CI 1.03–1.93) as compared to people born after the famine. Limitations of this study include the lack of birth weight data and the inability to separate effects of fetal and infant famine.

Conclusions

Fetal and infant undernutrition is associated with significantly increased risk of hypertension and impaired glucose tolerance in 40-year-old Nigerians. Prevention of undernutrition during pregnancy and in infancy should therefore be given high priority in health, education, and economic agendas.

Citation: Hult M, Tornhammar P, Ueda P, Chima C, Edstedt Bonamy A-K, et al. (2010) Hypertension, Diabetes and Overweight: Looming Legacies of the Biafran Famine. PLoS ONE 5(10): e13582. doi:10.1371/journal.pone.0013582

Editor: J. Jaime Miranda, Universidad Peruana Cayetano Heredia, Peru

Received: July 5, 2010; Accepted: September 19, 2010; Published: October 22, 2010

Copyright: © 2010 Hult et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: Funding was received from the Karolinska Institute, Swedish International Development Cooperation Agency and Carl-Eriks Levins Foundation. Funding was

unrestricted and the funders had no role in the study design, data collection, analysis, decision to publish or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

* E-mail: mikael.norman@ki.se

# These authors contributed equally to this work.

Introduction Top

Sub-Saharan African countries are in a process of a rapid epidemiological transition, away from infectious diseases, towards non-communicable diseases as leading causes of death [1]. Such transition in disease pattern is generally attributed to rural-to-urban shifts in adult lifestyle and typically involves changes in diet, cigarette smoking and lack of exercise [2]. There is a growing body of evidence however, suggesting that increased susceptibility to chronic diseases in adulthood has a developmental basis, originating in fetal life [3], [4]. The rapid development and shaping of the phenotype that occurs in utero is highly sensitive to environmental - in particular, nutritional – perturbations, leading to reduced functional capacity, altered metabolism and hormone production [3], [5]. This is considered an important underlying mechanism [3], [4] explaining why adults born small – a proxy for fetal starvation - are at increased risks for cardiovascular diseases and diabetes [6], [7].

The role of undernutrition during gestation for future health in children and adults has previously been evaluated in different countries [8]–[24]. However, besides data from European birth cohorts from the second world war [9]–[12], [19], [21]–[24], there is very limited information on the effects of fetal famine from other parts of the world, and with a follow-up time reaching beyond young adulthood. In particular, there is a lack of long-term follow-up data from sub-Saharan Africa, a continent with an unresolved situation with ongoing maternal-infant undernutrition, a recent and quick introduction of obesity-enhancing lifestyle factors and a rapidly increasing prevalence of cardiovascular disease, diabetes and hypertension [1], [25]–[27]. Therefore, the strength of a link between poor fetal nutrition and later adult diseases, would be of great interest to clarify in this part of the world also.

The study aim was to determine the risks for hypertension, diabetes and overweight forty years after fetal and infant exposure to the famine hitting Biafra during the Nigerian civil war (1967–1970) [28].

Methods Top

This observational study is based on the findings from a cohort of 1,339 subjects, belonging to the ethnic group Igbo, who were born between 1965 and 1973, i.e., before, during and after the Biafran famine.

Ethics statement

Ethical approval for the study was given by the Ethics Committee of the University of Nigeria Teaching Hospital, Enugu, and the Regional Ethical Vetting Board, Stockholm, Sweden. Informed consent (verbal and written in English, translated to Igbo if necessary) was obtained from all participants.

The Nigerian civil war and the Biafran famine

The Nigerian civil war broke out on 6 July 1967, after the Igbo people in the south-eastern provinces had declared independence as the Republic of Biafra. The war was a culmination of ethnic, economic, and religious tensions among the various peoples of Nigeria. Disapproving of the secession, Nigerian forces rapidly pushed the Biafrans back into a small enclave. Inflow of food to this enclave was cut off. The result was extensive famine among the Igbos, regarded as one of the great nutritional disasters of modern times [28]. The war ended on 15 January 1970.

Of the 1 to 3 million Igbos that are estimated to have lost their lives, only a small fraction (10%) died of military violence. The majority succumbed to starvation. The nutritional emergency was most critical in the Biafran enclave, in which approximately 7 million people - mostly refugees - resided. In August 1968 the first international relief operations were launched but the amount of food provided was clearly not sufficient and the great majority of the Igbos did not get access to this relief food [29].

Setting and participants

This follow-up study took place between June 27 and July 31, 2009. All shops at the six major market places of Enugu, the former Biafran capital, were systematically covered. People at the markets were actively contacted at their work place. The selection of participants was restricted to men and non-pregnant women who knew they were born in the southeast of Nigeria between 1965 and 1973. All subjects reported year of birth and 73% reported at least month and year. To confirm the birth year specified by the subject, we asked each person for a short review of his or her family history during the Nigerian civil war. The vast majority (more than 90%) of eligible subjects accepted participation in the study. It was not possible to do a systematic categorization of those who declined.

Data collection was conducted by MH, PT, PU, as well as doctors and medical students from the University of Nigeria Teaching Hospital (please see acknowledgements). Recruitment was performed by all team members, whereas measurements of blood pressure and p-glucose were obtained by a smaller group (n = 7) of specially trained investigators. Subjects were asked about level of education, current smoking, previously diagnosed hypertension or diabetes mellitus, and treatment for these conditions. Level of education was categorized as none, primary, secondary or higher education. There was no available information on birth weight or infant nutrition.

All measurements were performed according to predefined standard operating procedures. Participants were instructed to rest seated for at least five minutes before blood pressure (BP) and heart rate were measured. Two readings were taken three minutes apart in the left arm using a validated [30] automated oscillometric device (Omron M6 HEM-7001-E, Omron Corporation, Kyoto, Japan). A third reading was taken if the first and second BP differed more than five mmHg and mean systolic and diastolic values were calculated and defined as the subject's BP. Subjects having a systolic BP (SBP)≥140 mmHg, or a diastolic BP (DBP)≥90mmHg or who had a normal blood pressure but were pharmacologically treated for hypertension (n = 41 of 1,339), were categorized as hypertensive. Subjects having a SBP≥160 mmHg, and a DBP≥100 mmHg, were categorized as severely hypertensive. Random plasma glucose (p-glucose) was measured with a minimally invasive sampling technique (Abbott Freestyle Lite, Abbott Diabetes Care Ltd, Oxon, UK). A random p-glucose of 11.1 mmol/l or

higher was classified as diabetes mellitus and impaired glucose tolerance (IGT) was defined as random p-glucose between 7.8 and 11.0 mmol/l [31]. A subgroup (n = 75) reported no food intake for at least 8 hours prior to investigation. In this group IGT was defined as p-glucose between 6.1 and 6.9 mmol/l and diabetes mellitus as a p-glucose ≥7.0 mmol/l [31]. Height and weight were obtained from all subjects using height sticks and digital scales, respectively, and body mass index (BMI) was calculated. Overweight was defined as a BMI over 25 kg/m2 and obesity as BMI over 30 kg/m2. Waist circumference was measured with a measuring tape placed around the abdomen at the level of the upper hip bone, and considered as an indicator of central obesity.

All participants were briefly counselled about their current health status. Hypertensive or glucose intolerant subjects were referred to the University of Nigeria Teaching Hospital for further investigation and treatment.

Categorization according to exposure to famine

Subjects born between 1965 and 1967 were categorized as being exposed to famine in early childhood. Subjects born between 1968 and January 1970 (end of war) were categorized as being exposed to famine in fetal life and in infancy. Subjects born immediately after the surrender of Biafra were left uncategorized because of uncertainties in their exposure to famine and local variations in nutritional relief. Accordingly, subjects born between 1971 and 1973 and after the transitional period were categorized as being unexposed to famine.

Efforts to limit any potential bias included investigator-driven recruitment of participants (in contrast to subjects actively seeking participation in the study), direct assessments of outcomes with non-operator dependent methods and predefined standardized operational procedures applied by all data collectors.

Statistical methods

We aimed to include 630 individuals in each group, based on an assumption of a standard deviation of 19 mmHg and a group difference in systolic blood pressure of 3 mmHg or more, given 80% power and a statistical significance level of 0.05. Because of exhausted recruitment at the six study markets, data collection was ended when 1,339 subjects had been included in the study. All subjects were included in logistic regression analyses according to year of birth whereas in group wise logistic regression analyses, the 173 subjects born during the transition from famine to nutritional relief, i.e., from February to December 1970, were excluded.

Data are presented as means (SD), proportions (%) and odds ratios (95% CI). Group differences were tested using ANOVA and chi-squared test. Logistic and linear regression analyses were used to calculate odds ratios (OR) and for evaluation of relations between exposure (fetal-infant famine), potential confounders (sex, smoking, educational level) and outcomes (blood pressure, p-glucose, and BMI). As several previous reports have shown that the associations between early life exposures and adult outcomes vary in relation to sex and adult body size [9], [10], [14], [22], [32]–[34], we also stratified our analyses on sex and BMI. A p-value <0.05 was considered as statistically significant. Analyses were performed using STATA/IC 11.0 (Stata Corp LP, TX, US). To evaluate gender differences analyses stratified according to sex were also performed.

Results Top

Age differed among the three groups due to inclusion criteria. The proportion of men varied between 66 and 74% in the three groups and was highest in the group born after the famine. Smoking occurred predominantly in men and the proportion of smokers differed in the three groups, Table 1. Educational levels were found to be distributed equally between groups (p = 0.38) and overall, no education was found in 1%, primary in 34%, secondary in 52% and higher education in 13%.

Table 1. Subject characteristics.

doi:10.1371/journal.pone.0013582.t001

Famine and risk for hypertension

SBP and DBP were higher after fetal-infant exposure to famine, as compared to the other two groups, i.e., those exposed in early childhood as well as those that were born after the famine, i.e., were unexposed, Table 1. In addition, the OR's for a SBP and DBP in the hypertensive range were significantly higher for the group exposed to fetal-infant famine, Table 2.

Table 2. Odds ratios for hypertension, impaired glucose tolerance, diabetes, overweight (BMI>25 kg/m2) and obesity (BMI>30kg/m2) in 40-year-old men and women exposed to famine in early childhood, fetal-infant life, or unexposed.

doi:10.1371/journal.pone.0013582.t002

In linear regression analyses, SBP and DBP were associated with BMI (β = 0.49, r = 0.13, p<0.001 and β = 0.54, r = 0.21, p<0.001, respectively) and male sex (β = 3.3, r = 0.08, p = 0.003 and β = −1.46, r = 0.06, p = 0.05, respectively), but not with smoking. After adjusting for BMI odds ratios for SBP and DBP in the hypertensive range were of the same magnitude as before adjustment and still significantly higher for the group exposed to fetal-infant famine. In addition, analyses stratified on sex showed that exposure to famine in fetal life and infancy was a risk factor for SBP and DBP in the hypertensive range in both women and men. The OR (95% CI) for severe hypertension in the group exposed to fetal-infant famine was 2.82 (1.18–6.73) in men and 2.30 (0.65–8.10) in women, Table 2.

Odds ratios for high SBP (≥140 mmHg) according to each year of birth and with 1972 as reference year (OR = 1), are presented in Figure 1. The figure shows that famine in fetal-infant life (birth-year 1968 and 1969) resulted in a three to four fold increase in risk for an adult SBP in the hypertensive range, both in women and men.

Figure 1. Odds ratios for high systolic (≥140 mmHg) blood pressure in men and women at follow-up in 2009 according to year of birth and with 1972 as reference.

doi:10.1371/journal.pone.0013582.g001

Odds ratios for high SBP (≥140 mmHg) in relation to exposure group (childhood famine, fetal-infant famine and unexposed) and stratified on current overweight (BMI≥25), are presented in Table 3. Formal statistical testing for an interaction between early exposure to famine by sex and adult overweight was not significant, suggesting that the effects of early famine and current overweight on SBP were additive to each other.

Table 3. Odds ratios (OR) for systolic blood pressure (SBP)≥140 mm Hg and impaired glucose tolerance stratified on exposure to early famine and current BMI.

doi:10.1371/journal.pone.0013582.t003

Famine and risk for impaired glucose tolerance/diabetes

Random p-glucose was higher in adults exposed to famine in fetal-infant life, Table 1. The crude odds ratio for both IGT and diabetes was significantly higher for the group exposed to fetal-infant famine in comparison to the subjects born after the famine. Exposure to famine in early childhood did not increase the risk for IGT and diabetes in later life, Table 2.

In linear regression analyses, random p-glucose was associated with BMI (β = 0.04, r = 0.12, p<0.001) and male sex (β = −0.41, r = 0.10, p = 0.001). In analyses adjusted for BMI, fetal-infant exposure to famine remained as a risk factor for IGT and diabetes in Nigerian subjects in their forties. Stratified analyses on sex showed that the OR (95% CI) for diabetes after exposure to fetal-infant famine was 3.84 (0.95–15.5) in men and 2.06 (0.48–8.86) in women, Table 2.

Famine and risk for overweight

The group born after the famine was significantly taller than the other two groups, whereas there were no group differences in weight, Table 1. Waist circumference and BMI were higher in adults exposed to famine in fetal-infant life, Table 1. The crude odds ratio for overweight (BMI>25) was significantly higher for the group exposed to fetal-infant famine in comparison to the subjects born after the famine. Stratified analyses on sex showed that this effect was confined to women Table 2.

Discussion Top

This study showed higher BP, higher p-glucose and higher weight in middle-aged Nigerian people exposed to severe undernutrition in utero and in infancy. Comparing unexposed offspring with that of starving pregnant women, fetal-infant undernutrition was associated with significant increases in the prevalence of hypertension (from 9.5 to 24%, defined as SBP≥140 mmHg) and impaired glucose tolerance or diabetes (from 8.0 to 13%). Famine in early childhood was also associated with an increased prevalence of adult blood pressure in the hypertensive range (from 9.5 to 16%). Given the additive effects of early famine and adult overweight (Table 3), early undernutrition followed by later overnutrition seem to provide two fundaments for the adverse metabolic and cardiovascular outcomes seen in today's Nigeria, and as also previously described in an Indian cohort (14).

Our results are in line with previous epidemiological and experimental studies suggesting that fetal undernutrition contributes significantly to cardiovascular disease risk in adult life [3], [5], [20], [35]. In contrast to observations in European birth cohorts [9]–[12], [19], [20]–[24], the effects of fetal undernutrition seem to be more pronounced and emerge at an earlier age in this sub-Saharan cohort. These differences could reflect variations in exposure and population differences. Such an explanation may not necessarily be confined to susceptibility for and degree of fetal-infant undernutrition alone. Accelerated growth in later childhood [22], [32] and a high BMI in adult life have previously been found to have a stronger adverse effect on hypertension [33], [34] and diabetes [9], [10] in people who were small at birth. The mismatch between the environment that people in urban Nigeria now live in, characterized by a high-calorie-high-carbohydrate diet, and the one within which they evolved during the Biafran famine, may therefore be largely responsible for the present increase in disease risks [36]. Our finding of a higher current weight and waist circumference, and a higher BMI in adults exposed to fetal-infant famine (Table 1) imply that accelerated growth later in childhood may be in the pathway between early under-nutrition and later metabolic syndrome. Such an explanation may also be valid for the finding that glucose tolerance and blood pressure in 35-year-old Gambians (predominantly women) who remained fit and lean, showed no association with moderate to severe malnutrition in early life [37]. However, as we found no attenuation of the excess odds for adult hypertension and impaired glucose tolerance after adjusting for current BMI (Table 2), the association between fetal-infant famine and adult metabolic syndrome cannot be attributed to accelerated childhood growth alone [38].

Gender differences in developmental programming have previously been described [32]. In our study, both men and women exposed to fetal-infant undernutrition exhibited higher odds ratios for elevated blood pressure at follow-up. An excess risk for IGT was also found in men whereas power limitations unable conclusions regarding the risk for IGT in women exposed to early famine. Conversely, a significant excess risk for overweight was only seen in women exposed to fetal-infant undernutrition. Although there are reports of increased susceptibility for fetal programming of BP in males [23], [32], sex differences in the association between birth weight and BP have been questioned [39]. Experimental data suggest possible sex specific mechanisms in fetal programming of insulin secretion and insulin resistance [40], mechanisms that may be relevant in explaining our findings of gender differences in glucose tolerance.

The strengths of this study include the design with prospectively set inclusion criteria and active enrolment of a large cohort of customers and traders at markets, i.e., places where many people in the urban areas gather and work, as other sectors of employment and sites for purchasing of everyday goods are not very developed in this part of the world. Thus we believe that the study cohort is representative for urban settings in sub-Saharan Africa and at highest risk for the present epidemic in non-communicable diseases. Categorization was based on date of birth – i.e., exposure to famine in early life. In addition, since subjects born in the transitional period were excluded, there was no late gestational overlap with famine in the unexposed group. The follow-up time was sufficiently long to establish relations to outcomes that are directly related to adult cardiovascular disease. Finally, we addressed the possibility that smoking confounded our results [41].

Although heritability for birth weight has been estimated to range from 25% to 40% [42], and although some experimental data suggest that birth weight may fail to reflect intrauterine factors associated with later disease risk [43], birth weight is the most commonly used proxy

for fetal undernutrition. Therefore, a limitation of our study, shared with other famine studies [19], [23], is the lack of anthropometric data at birth and in infancy. There are no records from which these data can be retrieved. In addition, we have no data on mother and infant nutrition. Given that inflow of food to Biafra was cut off, it can be assumed that access to infant formula was extremely limited and that most infants were exclusively breastfed.

It is likely that survival of the healthiest pregnant women occurred during the Biafran famine. The most severe cases of fetal and infant undernutrition are also likely to have died prior to follow-up, either in early childhood or from the increasing cardiovascular morbidity reported from adults residing in the area [25]. Some misclassification may have occurred as the nutritional situation of Biafra gradually deteriorated, making it difficult to pinpoint the start of famine. If anything, these limitations may have introduced a conservative bias into our study, underestimating the true long-term effects of undernutrition in fetal life and in infancy.

Before the war, Biafra differed from most developing nations because it had a good supply of food and water, and sound public health policies with many physicians, nurses, hospitals, and clinics [44]. Already months after the war, there are reports of vast improvement in the nutritional situation [28]. Our results are therefore most likely reflecting differences in adult health outcomes after severe early famine as compared to a significantly better nutritional situation in early life.

The external validity of data from a convenience sample from market places can also be discussed. The method of recruiting participants will have missed subsistence farmers and others not attending the markets, whose nutritional status and other factors are likely to differ from market people. We note that the prevalences of hypertension, diabetes and obesity reported herein are similar to those reported from comparable urban and rural Nigerian cohorts [45], [46].

The study design, i.e., comparing long-term outcomes in people born before, during or after famine, has previously been used [9]–[12]. By inclusion, the subjects in the unexposed group were youngest. As the prevalence of hypertension and glucose intolerance increases with age, some associations with birth year might be expected. However, given the size of the effect and that the OR's for all outcomes after fetal-infant famine were significantly lower not only in younger, but also in older people, trends in disease-risks over time and cohort effects cannot be the only explanation for our findings.

Previous studies indicate that a nutritional insult - during gestation or the first few months of postnatal life - may be important for later outcome and disease risk [9], [11]. Although the resolution and exposure data of this study do not allow for a detailed analysis of the timing of the insult, the striking dose-response effect found between birth during years of famine and over-risk for hypertension in adult life (Figure 1) suggests a causal relationship. The Biafran famine was characterized by a severe scarcity of proteins, manifested in the vast number of infants and children suffering from kwashiorkor [28]. Experimental models suggest that protein deficit in utero may programme abnormal glucose homeostasis and vascular endothelial dysfunction, whereas results are less consistent with regard to programming of high blood pressure [35], [43]. Besides the nutritional insult, pregnant women in former Biafra were living under conditions of war. Such stress for mothers and infants could also contribute to higher blood pressure in later life [47].

The implications of our findings are important. On a population level, a 3.3 mmHg increase in mean SBP and 2 mmHg in DBP can be translated into an estimated increase in cardiovascular deaths by 25% and stroke by 32% [48]. Given the combination of large blood pressure effects and increased rates of glucose intolerance resting on a basis of prevalent obesity before middle age (Table 3), which is characteristic for Nigeria today, it is not surprising that disability and deaths from stroke and coronary heart disease are rapidly increasing. To adequately address this trend in non-communicable diseases on a community level requires a developed health care infrastructure providing life-long treatment and follow-up. The increasing burden of chronic diseases therefore poses a massive challenge to the already crippled health care systems of sub-Saharan Africa.

In summary, our study demonstrates a fetal-infant contribution to adult hypertension and glucose intolerance in an African cohort. It can be assumed that fetal and infant undernutrition is a significant contributor to the increasing prevalence of hypertension and glucose intolerance also in other parts of sub-Saharan Africa. Therefore, prevention of fetal and infant undernutrition should be given high priority in national health, education, and economic agendas to limit the increase of non-communicable diseases in many African countries. Given that the highest risk for hypertension was found in those undernourished in early life and then growing overweight, it is appropriate to consider that preventing excess growth in later childhood may be as important for reducing adult ill-health as supporting fetal-infant growth.

Acknowledgments Top

This study was a collaboration between the Karolinska Institute, Stockholm, Sweden and University of Nigeria Teaching Hospital, Enugu, Nigeria. Field work in Enugu could not have been completed in time without the undermentioned members of Medix Frontiers whose invaluable contribution ensured a successful collection of data for the research: Igwedimma Michael, Ezekekwu Chinedu, Besongngem Akotanchi, Moeteke Nnamdi S, Dr. Ilomuanya Uchenna S., Alozie Izima C., Osi Sixtus S., Odumegwu Chidi A, Odiakosa Otito R., Ezenduka Johnbaptist, Awuzie Nonye P.,Ugwoke Kenneth, Ijemba Obiora C. E., Adugo Christian C., Orogwu Ifedichimma, Obumneme Chinweuba, Okechukwu Valentine, Nworgu Chinemerem, Eze Maduabuchi N., Bassey Joshua Ihemelandu, Aneke Ifeoma O., Unachukwu Chinonyerem, Elosiuba Emeka, Mba Martins, Iwuchukwu Izuchukwu, Akande Ifeoluwa, Ijeoma Austin, Ohanu Uzoamaka, Awoh Benita, Agwu Uchechukwu, Onah Chiamaka, Njeze Nneze, Onwumere Nnenna M., Enwereji Amarachi, Okpara Valentine, Agu Chinenye, Ebere Ikenna, Nwafor Nkemjika Peace, Nwosu Nneka Ifunanya, Ifeanyichi Martilord,Ihediora Amarachi, Okutepa Ocholi G., Nwose Sunday Onyemaechi, Umeh Arinze, Nwankwo kosiso O., Akalawu Martins, Nwatuzor Christopher and Ewelukwa Chukwudum.

Author Contributions Top

Conceived and designed the experiments: MH PT PU CC AKEB BO MN. Performed the experiments: MH PT PU CC BO. Analyzed the data: MH PT PU AKEB MN. Wrote the paper: MH PT PU CC AKEB BO MN.

References Top

1. Unwin N (2001) Non-communicable disease and priorities for health policy in sub-Saharan Africa. Health Policy Plan 16: 351–352. Find this article online

2. Yach D, Hawkes C, Gould CL, Hofman KJ (2004) The global burden of chronic diseases: overcoming impediments to prevention and control. JAMA 291: 2616–2622. Find this article online

3. Barker DJP (1998) Mothers, babies and health in later life. Edinburgh: Churchill Livingstone.

4. Gluckman PD, Hanson MA, Bateson P, Beedle AS, Law CM, et al. (2009) Towards a new developmental synthesis: adaptive developmental plasticity and human disease. Lancet 373: 1654–1657. Find this article online

5. Barker DJ, Gluckman PD, Godfrey KM, Harding JE, Owens JA, et al. (1993) Fetal nutrition and cardiovascular disease in adult life. Lancet 341: 938–941. Find this article online

6. Bergvall N, Iliadou A, Johansson S, de Faire U, Kramer MS, et al. (2007) Genetic and shared environmental factors do not confound the association between birth weight and hypertension: a study among Swedish twins. Circulation 115: 2931–2938. Find this article online

7. Kaijser M, Bonamy AK, Akre O, Cnattingius S, Granath F, et al. (2009) Perinatal risk factors for diabetes in later life. Diabetes 58: 523–526. Find this article online

8. [No authors listed] (1991) Underlying and proximate determinants of child health: the Cebu Longitudinal Health and Nutrition Study. Am J Epidemiol 133: 185–201. Find this article online

9. Ravelli AC, van der Meulen JH, Michels RP, Osmond C, Barker DJ, et al. (1998) Glucose tolerance in adults after prenatal exposure to famine. Lancet 351: 173–177. Find this article online

10. Ravelli AC, van Der Meulen JH, Osmond C, Barker DJ, Bleker OP (1999) Obesity at the age of 50 y in men and women exposed to famine prenatally. Am J Clin Nutr 70: 811–816. Find this article online

11. Roseboom TJ, van der Meulen JH, Ravelli AC, van Montfrans GA, Osmond C, et al. (1999) Blood pressure in adults after prenatal exposure to famine. J Hypertens 17: 325–330. Find this article online

12. Roseboom TJ, van der Meulen JH, van Montfrans GA, Ravelli AC, Osmond C, et al. (2001) Maternal nutrition during gestation and blood pressure in later life. J Hypertens 19: 29–34. Find this article online

13. Victora CG, Adair L, Fall C, Hallal PC, Martorell R, et al. (2008) Maternal and child undernutrition: consequences for adult health and human capital. Lancet 371: 340–357. Find this article online

14. Bhargava SK, Sachdev HS, Fall CH, Osmond C, Lakshmy R, et al. (2004) Relation of serial changes in childhood body-mass index to impaired glucose tolerance in young adulthood. N Engl J Med 350: 865–875. Find this article online

15. Richter LM, Norris SA, De Wet T (2004) Transition from Birth to Ten to Birth to Twenty: the South African cohort reaches 13 years of age. Paediatr Perinat Epidemiol 18: 290–301. Find this article online

16. Sachdev HS, Fall CH, Osmond C, Lakshmy R, Dey Biswas SK, et al. (2005) Anthropometric indicators of body composition in young adults: relation to size at birth and serial measurements of body mass index in childhood in the New Delhi birth cohort. Am J Clin Nutr 82: 456–466. Find this article online

17. Victora CG, Barros FC, Lima RC, Behague DP, Gon alves H, et al. (2003) The Pelotas birth cohort study, Rio Grande do Sul, Brazil, 1982–2001. Cad Saude Publica 19: 1241–1256. Find this article online

18. Grajeda R, Behrman JR, Flores R, Maluccio JA, Martorell R, et al. (2005) The human capital study 2002–04: tracking, data collection, coverage, and attrition. Food Nutr Bull 26: S15–24. Find this article online

19. Stanner SA, Bulmer K, Andres C, Lantseva OE, Borodina V, et al. (1997) Does malnutrition in utero determine diabetes and coronary heart disease in adulthood? Results from the Leningrad siege study, a cross sectional study. BMJ 315: 1342–1348. Find this article online

20. Barker DJ, Eriksson JG, Forsen T, Osmond C (2002) Fetal origins of adult disease: strength of effects and biological basis. Int J Epidemiol 31: 1235–1239. Find this article online

21. Hales CN, Barker DJ, Clark PM, Cox LJ, Fall C, et al. (1991) Fetal and infant growth and impaired glucose tolerance at age 64. BMJ 303: 1019–1022. Find this article online

22. Eriksson J, Forsen T, Tuomilehto J, Osmond C, Barker D (2000) Fetal and childhood growth and hypertension in adult life. Hypertension 36: 790–794. Find this article online

23. Koupil I, Shestov DB, Sparen P, Plavinskaja S, Parfenova N, et al. (2007) Blood pressure, hypertension and mortality from circulatory disease in men and women who survived the siege of Leningrad. Eur J Epidemiol 22: 223–234. Find this article online

24. Stein AD, Zybert PA, van der Pal-de Bruin K, Lumey LH (2006) Exposure to famine during gestation, size at birth, and blood pressure at age 59 y: evidence from the Dutch Famine. Eur J Epidemiol 21: 759–765. Find this article online

25. Ike SO (2009) Prevalence of hypertension and its complications among medical admissions at the University of Nigeria Teaching Hospital, Enugu (Study 2). Niger J Med 18: 68–72. Find this article online

26. Holmes MD, Dalal S, Volmink J, Adebamowo CA, Njelekela M, et al. (2010) Non-communicable diseases in sub-Saharan Africa: the case for cohort studies. PLoS Med 7(5): e1000244. Find this article online

27. Black RE, Allen LH, Bhutta ZA, Caulfield LE, de Onis M, et al. (2008) Maternal and child undernutrition: global and regional exposures and health consequences. Lancet 371: 243–260. Find this article online

28. Miller JP (1970) Medical relief in the Nigerian civil war. Lancet 760: 1330–1334. Find this article online

29. Aall C (1970) Relief, nutrition and health problems in the Nigerian-Biafran war. J Trop Pediatr 16: 70–90. Find this article online

30. Topouchian JA, El Assaad MA, Orobinskaia LV, El Feghali RN, Asmar RG (2006) Validation of two automatic devices for self-measurement of blood pressure according to the International Protocol of the European Society of Hypertension: the Omron M6 (HEM-7001-E) and the Omron R7 (HEM 637-IT). Blood Press Monit 11: 165–171. Find this article online

31. Report of a WHO/International Diabetes Federation consultation (2006) Definition and diagnosis of diabetes mellitus and intermediate hyperglycemia. ISBN 92 4 159493 4.

32. Adair LS, Cole TJ (2003) Rapid child growth raises blood pressure in adolescent boys who were thin at birth. Hypertension 41: 451–456. Find this article online

33. Huxley RR, Shiell AW, Law CM (2000) The role of size at birth and postnatal catch-up growth in determining systolic blood pressure: a systematic review of the literature. J Hypertens 18: 815–831. Find this article online

34. Leon DA, Koupilova I, Lithell HO, Berglund L, Mohsen R, Vagero D, Lithell UB, McKeigue PM (1996) Failure to realise growth potential in utero and adult obesity in

relation to blood pressure in 50 year old Swedish men. BMJ 312: 401–6. Find this article online

35. Armitage JA, Khan IY, Taylor PD, Nathanielsz PW, Poston L (2004) Developmental programming of the metabolic syndrome by maternal nutritional imbalance: how strong is the evidence from experimental models in mammals? J Physiol 561: 355–377. Find this article online

36. Gluckman PD, Hanson MA (2004) Living with the past: evolution, development, and patterns of disease. Science 305: 1733–1736. Find this article online

37. Moore SE, Halsall I, Howarth D, Poskitt EM, Prentice AM (2001) Glucose, insulin and lipid metabolism in rural Gambians exposed to early malnutrition. Diabet Med 18: 646–653. Find this article online

38. Lucas A, Fewtrell MS, Cole TJ (1999) Fetal origins of adult disease - the hypothesis revisited. BMJ 319: 245–9. Find this article online

39. Lawlor DA, Ebrahim S, Davey Smith G (2002) Is there a sex difference in the association between birth weight and systolic blood pressure in later life? Findings from a meta-regression analysis. Am J Epidemiol 156: 1100–1104. Find this article online

40. Shepherd PR, Crowther NJ, Desai M, Hales CN, Ozanne SE (1997) Altered adipocyte properties in the offspring of protein malnourished rats. Br J Nutr 78: 121–129. Find this article online

41. Huxley R, Neil A, Collins R (2002) Unravelling the fetal origins hypothesis: is there really an inverse association between birthweight and subsequent blood pressure? Lancet 360: 659–665. Find this article online

42. Clausson B, Lichtenstein P, Cnattingius S (2000) Genetic influence on birthweight and gestational length determined by studies in offspring of twins. BJOG 107: 375–381. Find this article online

43. Nwagwu MO, Cook A, Langley-Evans SC (2000) Evidence of progressive deterioration of renal function in rats exposed to a maternal low-protein diet in utero. Br J Nutr 83: 79–85. Find this article online

44. Blum JD (1969) Who cares about Biafra anyway?. The Harvard Crimson. February 25.

45. Sani MU, Wahab KW, Yusuf BO, Gbadamosi M, Johnson OV, et al. (2010) Modifiable cardiovascular risk factors among apparently healthy adult Nigerian population - a cross sectional study. BMC Res Notes 3: 11. Find this article online

46. Oladapo OO, Salako L, Sodiq O, Shoyinka K, Adedapo K, et al. (2010) A prevalence of cardiometabolic risk factors among a rural Yoruba south-western Nigerian population: a population-based survey. Cardiovasc J Afr 21: 26–31. Find this article online

47. Johansson S, Norman M, Legnevall L, Dalmaz Y, Lagercrantz H, et al. (2007) Increased catecholamines and heart rate in children with low birth weight: perinatal contributions to sympathoadrenal overactivity. J Intern Med 261: 480–487. Find this article online

48. Yusuf S, Sleight P, Pogue J, Bosch J, Davies R, et al. (2000) Effects of an angiotensin-converting-enzyme inhibitor, ramipril, on cardiovascular events in high-risk patients. The Heart Outcomes Prevention Evaluation Study Investigators. N Engl J Med 342: 145–153. Find this article online

Abstract Top

Objectives

Poor adherence is one of the biggest obstacles in therapeutic control of high blood pressure. The objectives of this study were (i) to measure adherence to antihypertensive therapy in a representative sample of the hypertensive Pakistani population and (ii) to investigate the factors associated with adherence in the studied population.

Methods and Results

A cross-sectional study was conducted on a simple random sample of 460 patients at the Aga Khan University Hospital (AKUH) and National Institute of Cardiovascular Diseases, Karachi, from September 2005–May 2006. Adherence was assessed using the Morisky Medication Adherence Scale (MMAS), with scores ranging from 0 (non-adherent) to 4 (adherent). In addition to MMAS, patient self-reports about the number of pills taken over a prescribed period were used to estimate adherence as a percentage. AKU Anxiety and Depression Scale (AKU-ADS) was incorporated to find any association between depression and adherence. At a cut-off value of 80%, 77% of the cases were adherent. Upon univariate analyses, increasing age, better awareness and increasing number of pills prescribed significantly improved adherence, while depression showed no association. Significant associations, upon multivariate analyses, included number of drugs that a patient was taking (P<0.02) and whether he/she was taking medication regularly or only for symptomatic relief (P<0.00001).

Conclusions

Similar to what has been reported worldwide, younger age, poor awareness, and symptomatic treatment adversely affected adherence to antihypertensive medication in our population. In contrast, monotherapy reduced adherence, whereas psychosocial factors such as depression showed no association. These findings may be used to identify the subset of population at risk of low adherence who should be targeted for interventions to achieve better blood pressure control and hence prevent complications.

Citation: Hashmi SK, Afridi MB, Abbas K, Sajwani RA, Saleheen D, et al. (2007) Factors Associated with Adherence to Anti-Hypertensive Treatment in Pakistan. PLoS ONE 2(3): e280. doi:10.1371/journal.pone.0000280

Academic Editor: Bernhard Baune, James Cook University, Australia

Received: December 4, 2006; Accepted: February 13, 2007; Published: March 14, 2007

Copyright: © 2007 Hashmi et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This research was partially supported by research funds from the Department of Biological and Biomedical Sciences, Aga Khan University.

Competing interests: The authors have declared that no competing interests exist.

* To whom correspondence should be addressed. E-mail: philippe.frossard@aku.edu

¤a Current address: Department of Public Health and Primary Care, University of Cambridge, Cambridge, United Kindgom

¤b Current address: Department of Surgery, Yale University, New Haven, Connecticut, United States of America

Introduction Top

Hypertension is an overwhelming global challenge which ranks third as a cause of disability-adjusted life-years [1]. According to the Seventh Report of the Joint National Committee on Hypertension, there are approximately 50 million hypertensive individuals in the United States only and 1 billion worldwide [2]. Even though the burden of hypertension is currently centered in economically developed countries (37.3%), developing countries will feel a greater impact due to their larger population proportion. Indeed estimates indicate that up to three-quarters of the world's hypertensive population will be in economically developing countries by the year 2025 [1].

The National Health Survey of Pakistan (NHSP), conducted from 1990 to 1994, showed that hypertension affects 18% of adolescents above 15 years of age and 33% of adults above 45 years of age; less than 3% hypertensive patients, however, have their BP controlled to 140/90 mm Hg or below and more than 70% of all hypertensive patients (85% in rural areas) in Pakistan are not even aware of their disease [3].

Studies worldwide indicate that despite the availability of effective medical therapy, over half of all hypertensives do not take any treatment [4] and more than half of those on treatment have blood pressures over the 140/90 mmHg threshold [5]. The World Health Organization (WHO) describes poor adherence as the most important cause of uncontrolled blood pressure and estimates that 50–70% of people do not take their antihypertensive medication as prescribed [6].

It has been well documented that uncontrolled blood pressure increases the risk of ischemic heart disease 3-to 4-fold [7] and the overall cardiovascular risk by 2-to 3-fold [8]. The incidence of stroke increases approximately 3-fold in patients with borderline hypertension and approximately 8-fold in those with definite hypertension [9]. Moreover, a recent case control study has shown that non adherence to therapy is associated with an increased risk of stroke in patients with hypertension [10], [11].

Adherence is defined by WHO as “the extent to which a person's behavior–taking medication, following a diet, and/or executing lifestyle changes-corresponds with agreed recommendations from a health care provider” [12]. Adherence is dependent on numerous factors and has been shown to vary from 0 to 100% in different populations studied [12], [13], [14], [15]. Factors such as age [16], [17], gender [18], low socioeconomic status and severity of disease [17], class of drug prescribed [19], number of pills per day [16], [18], side effects of medication [16], [17], patient's inadequate understanding of the disease and importance of the treatment [16], [17], co-morbid medical conditions [17], lack of social support [20], poor patient-provider relationship [21], cost, forgetfulness [22], and presence of

psychological problems, especially depression [17], [21], have all been shown to affect adherence in various populations.

We undertook this research with the objectives of (i) measuring adherence through the use of validated tools and (ii) investigating the demographic, social and personal factors affecting patient adherence to antihypertensive therapy in the Pakistani population.

Methods Top

Study design and sample recruitment

This descriptive study was a questionnaire-based cross sectional analysis. A simple random sample of 460 patients was selected between September 2005 and May 2006 from two tertiary care hospitals in Karachi, Pakistan, namely Aga Khan University Hospital (AKUH) and National Institute of Cardiovascular Diseases (NICVD).

AKUH is one of the most advanced private tertiary care centers of Pakistan, and caters to a large number of people from all over the country. NICVD is a government-run, tertiary care hospital and as such, is approached by a population more indicative of the country's health status. Pakistan, as a developing country, is home to a vast majority of people living at or below poverty line. Low-cost services at this institute attract a greater number of patients. NICVD was thus chosen for sample selection to increase the reliability of our results and for better generalization of our data to the Pakistani population.

The inclusion criteria were 1) patients of age 18 years and above, 2) those who had been diagnosed with ‘essential’ hypertension and 3) those who were on prescribed antihypertensive medications since at least the previous one month. Patients with co-existing medical conditions were also included. All people who fulfilled the inclusion criteria were then assessed for familiarity with Urdu, which is the national language of Pakistan and is understood and spoken by most people throughout the country, irrespective of ethnicity. Patients who could not converse in Urdu were excluded, due to the lack of appropriate translators.

Ethical approval, informed consent and patient privacy

Patients who agreed to participate were explained the nature and the objectives of the study, and informed consent was formally obtained. The study was approved by the Ethical Review Committees of AKUH and NICVD. The information about patient's identity was not included with the other data and only the principal investigator had access to this information. No reference to the patient's identity was made at any stage during data analysis or in the paper.

Data collection

The data collection tool was a questionnaire, designed-based on an extensive literature review of similar studies [23], [24]. The questionnaire was administered by trained interviewers in the Urdu language for ease of comprehension by the patients. The tool was pilot-tested on 50 patients, who were not included in the final study sample. In view of the responses generated in the pilot study, we further modified our survey instrument to include more common responses or to modify the questions. The final survey questionnaire required approximately 20–30 minutes to administer. The questionnaire extracted information regarding patient

demographics and clinical characteristics, including co-morbidities such as diabetes, ischemic heart disease, neurological diseases and others, characteristics of hypertension and anti-hypertensive treatment, awareness about hypertension and anti-hypertensive treatment, and factors that, in the patient's views, encouraged or discouraged the patient's drug taking behavior.

Blood pressure was measured twice by the hospital health physicians and trained investigators using aneroid sphygmomanometers and stethoscopes. Blood pressure was measured in all subjects after they were in the resting state for 10 minutes and in sitting position in the right arm place at the level of the heart. An appropriate-sized cuff (cuff bladder encircling at least 80% of the arm) was used to ensure accuracy [2].

Patients were asked about details of their prescribed medication regimen. The information obtained was tabulated. All information was based on self-reporting and included number of prescribed antihypertensive drugs, trade names of all prescribed drugs along with the drug class, doses in milligrams (mg), frequency per day, duration of intake in months, time of intake of the drug (morning, afternoon, evening), and any side effects associated with the drug. A standard drug manual was used to find out the generic class of the drug. Prescriptions, whenever they were available at the time of the interview, were used in getting reliable data, particularly from illiterate patients.

Depression assessment tool

The Aga Khan University Anxiety and Depression Scale (AKU-ADS) [25] was used to determine the severity of depressive symptoms in the study population. AKU-ADS is an indigenous screening instrument that has been developed in Urdu, for assessment of anxiety-depression syndromes. The questionnaire is based on 25 items, 13 psychological and 12 somatic. At a cut-off score of 20, the sensitivity of the scale is 66%, specificity 79%, positive predictive value 83% and negative predictive value 60%. It covers most of the clinical features specified by DSM-IV criteria, including somatic complaints. Responses to the questions were recorded and scored as never (0), sometimes (1), mostly (2), and always (3). As the scale is designed in Urdu, the chances of differences in interpretation by both the interviewer and the patient were minimal.

Measurement of adherence

Adherence was defined as ‘the extent to which patients followed their medication schedules as prescribed by their health care providers’ [17]. To measure adherence, patients' self-reports were used. Patients were asked non-judgmentally how often they missed their doses [17]. They were asked the total number of tablets they had been prescribed per week and how many pills they took and missed in the last 3, 5 and 7 days, respectively. Previous studies have adopted assessment over a longer duration of time, sometimes as much as months, to obtain this data. We, however, employed a shorter time period to elicit more accurate responses by minimizing recall bias. Adherence rates were calculated as ‘pills taken over a specific period of time, divided by pills prescribed for that specific period of time’ [17].

To further increase the strength and consistency of our results, we included an adherence assessment through the Morisky Medication Adherence Scale (MMAS) [26], a 4-item questionnaire with a high reliability and validity [14], [26], which has been particularly useful in chronic conditions such as hypertension. It measures both intentional and unintentional

adherence based on forgetfulness, carelessness, stopping medication when feeling better, and stopping medication when feeling worse. The scale is scored 1 point for each ‘no’ and 0 points for each ‘yes’. The total score ranges from 0 (non adherent) to 4 (adherent) [27].

Statistical analyses

All statistical analyses were performed using SPSS 13.0 for Windows. The questionnaire was pre-coded and all data was entered and checked twice by two different investigators. Based on a recent meta-analysis [15] of 129 studies that showed a mean adherence rate of 76.6%, a sample size of 416 participants was calculated for a power of 80% at 5% alpha and with a 6% margin of error. Mean±one standard deviation were computed for all continuous data. Frequencies were calculated for categorical variables. Adherence was represented in percentage and treated as a continuous variable. As described in previous studies, for analysis of adherence a cut-off value of 80% was used for labeling patients as adherent or non-adherent [4], [16], [28], [29]. In univariate analyses, means were compared using student's t-test and ANOVA. Categorical variables were compared using Chi-squared and Fisher's exact tests, as applicable. Odds ratios (with 95% confidence intervals, CI) were calculated from the 2×2 tables. Univariate and multivariate logistic regression were done using adherent vs. non-adherent status as the outcome variable, while various study variables were used as independent variables. Associations of study variables with the MMAS score were checked with the help of linear regression using MMAS score as the dependant variable and study variables as independent variables. A p-value of less than 0.05 was considered to be statistically significant for all analyses.

Results Top

After pilot-testing and refining the questionnaire with 50 individuals, we interviewed 460 patients at both study centers. After exclusion of cases in which adherence could not be calculated, we included data on 438 cases. In this sample, 71% of the cases were recruited and interviewed at the NICVD, while 29% were recruited from AKUH. There were no statistically significant differences in the adherence and the study variables between the cases recruited from the two hospitals hence combined analysis of the data was done (data not shown). Due to significant difference in prevalence of hypertension in the various ethnic groups in Pakistan, we also recorded the ethnicity as provided by the patient. Ethnicity, however, was not associated with adherence or other study variables and therefore was not analyzed as a risk factor. According to the 80% cutoff level, 77% of the cases were adherent (n = 336, adherence≥80%, mean adherence = 98±5%) and 23% were non-adherent (n = 102, adherence≤80%, mean adherence = 39±29%).

Demographic and clinical characteristics

There were 199 males (mean age 54±10 years) and 239 females (mean age 50±11 years); 20% of the total cases were younger than 40 and 19% older than 60 years. Although we found adherence to increase with increasing age (P<0.02), age remained only marginally different after division of cases into adherent and non-adherent groups. Subjects who were less than 40 years old were less adherent than those older than 70. The highest mean adherence rate was observed in the age group 70–80 years (mean adherence = 91±14%). Although mean adherence showed an overall increase with increasing monthly income and increasing level of education, no significant difference was observed. Table 1 shows the distribution of study variables among the adherent and non-adherent groups. Most of the

cases (89%) were married. The mean AKU-ADS score was 19±11 and 190 (43%) cases were depressed according to their AKU-ADS score. A significant proportion of the depressed patients were females 113 (73%, P<0.001). Although the presence of a single co-morbid condition slightly increased the adherence, the presence of two or more co-morbids led to a gradual decrease in adherence (P>0.05). The most frequently reported co-morbids were diabetes (23.1%) and ischemic heart disease (25.8%). Associations between co-morbidities and adherence, however, were not statistically significant.

Table 1. Demographic and clinical characteristics of the study population

doi:10.1371/journal.pone.0000280.t001

Characteristics of hypertension and anti-hypertensive treatment

Table 2 shows the prevalence of various characteristics of hypertension and anti-hypertensive treatment in the adherent and non-adherent groups. Most patients (70.8%) discovered their disease during medical checkup for symptoms related to hypertension and/or its complications.

Table 2. Characteristics of hypertension and anti-hypertensive treatment

doi:10.1371/journal.pone.0000280.t002

A greater proportion of the cases suffering from hypertension-related complications were adherent (P>0.05). A large proportion of our study sample had never been hospitalized (45%). The patients who had been hospitalized in the previous two years had significantly higher adherence (P<0.05). Surprisingly adherence increased with increasing number of anti-hypertensive drugs that a person was taking (P<0.05).

Awareness about hypertension and anti-hypertensive treatment

The overall level of awareness about hypertension and its treatment was very low. As shown in Table 3, 24% of the study sample took their medication only when they thought they had symptoms of high blood pressure. This patient group had very low adherence. The patients who considered every dose to affect blood pressure had significantly higher adherence (P<0.001) and lower systolic and diastolic blood pressures. A very small proportion of patients were aware of the risk factors for hypertension and an even smaller proportion knew about the complications. Greater awareness was associated with higher adherence.

Table 3. Awareness about hypertension and anti-hypertensive treatment

doi:10.1371/journal.pone.0000280.t003

Factors associated with low MMAS

The mean MMAS score in the overall sample was 2.5±1.3. MMAS score was significantly higher (P<0.001) in the adherent group (2.7±1.2) compared to the non-adherent group (1.7±1). Table 4 lists the patient variables significantly associated with the MMAS. MMAS scores of ≤2 were associated with a mean adherence <73%. The patients with MMAS scores ≤2 were relatively younger (51 vs. 53 years), were illiterate or at a lower level of education (P<0.05), belonged to a lower income subgroup (P<0.05), and had higher systolic (140±26 vs. 136±19 mmHg, P<0.05) and diastolic (86±18 vs. 82±13, P<0.01) pressures. A significantly greater proportion of patients with MMAS score ≤2 were depressed (50% vs. 38%, P<0.05), according to AKU-ADS.

Table 4. Patient factors associated with MMAS

doi:10.1371/journal.pone.0000280.t004

Table 5 shows the prevalence of various personal, social and behavioral characteristics that in the patient's views affected their drug adherence. The prevalence of most of the factors was significantly different in the adherent and non-adherent groups. While the common encouraging factors, such as understanding the need and effectiveness of the prescribed medication and availability of support system, were significantly associated with better adherence, the most common discouraging factors cited in literature such as forgetfulness [22], side effects [16], [17], cost of medication [16] and lack of access to medication [14] did not show statistically significant associations with non-adherence.

Table 5. Factors that, in the patient’s views, encouraged or discouraged the patient’s drug taking behavior

doi:10.1371/journal.pone.0000280.t005

Upon multivariate analysis, the only factors associated with adherence were the number of drugs that a patient was taking (P<0.02) and whether he/she was taking medication regularly or only for symptomatic relief (P<0.00001)

Discussion Top

Adherence to antihypertensive therapy as measured by our study was 77% in the studied Pakistani population, when defined by the ≥80% cut off. The factors showing significant associations with adherence were age, number of drugs prescribed and patients' knowledge of the disease and treatment, including their beliefs and practices.

Our study reports a higher adherence in the Pakistani population than what has been reported previously in a local study (57%) [22]. This could be due to measurement of adherence based on different criteria in the two studies, along with variation in the subset of population which served as the study sample.

Among the studies conducted on various populations of the world, using a similar cut-off, the adherence we observe is higher than what has been reported in a similar study in Malaysia

(44.2%) [16], comparable to a study in Egypt (74.1%) [23] and lower than what a study in the Western population (Scotland) reports (91%) [4]. Hence, we found that people of a developing country, like Pakistan, are generally more adherent to their medication than what might be assumed. However, population studies with larger samples are needed to support our claim.

Age was found to be significantly and independently associated with adherence in our study, with better adherence observed in older people. This finding is consistent with a number of other studies [30], [31], [32] including the regional study in Malaysia [16], although there are studies which show either no association [4], [23] or decreasing adherence with increasing age [33]. Increasing self-reliance in old age has been shown to decrease adherence [34]. In the Pakistani population, a better social support structure ensured by the common extended family system, reduces self-reliance and could be the reason for better adherence in this age group. It is usual for other family members to take full responsibility of the medication routine of the families' patients.

An inverse relationship was observed between adherence and number of pills prescribed. Patients on monotherapy had a mean adherence of 79% compared to 90% for those on three drugs or more (OR; 95% CI, 0.3; 0.1–0.6). This is in contrast to what has been frequently reported so far. A recent meta-analysis of eight studies reports that the average adherence for once-daily dosing was significantly higher than for multiple daily dosing (91.4% vs. 83.2%, respectively, P<0.001) [35]. Some latest studies, however, have identified no relation between increasing number of drugs and poor adherence [4], including one such study in an Asian population [16]. One reason for our finding could be that patients on multiple pills feel that the severity of their disease is significant and hence become more cautious with their treatment, compared to those on monotherapy, who may take treatment lightly. Another reason may perhaps be that when patients have to take multiple medications, they are less likely to forget to take them, compared to having to take only one pill.

Baune et al, showed a significant correlation between education and QOL among patients with hypertension in Gaza Strip and hypothesized that educational interventions would be essential in preventing high blood pressures and consequent mortality [36]. Knowledge of hypertension significantly affected adherence in our study sample. Patients who were aware of the association between certain risk factors for hypertension, such as high salt intake, stress and a positive family history, had better adherence compared to those who with poorer knowledge. Studies from the developed world, however, indicate no association between patients' knowledge and adherence [32], [37].

Patients' beliefs and attitudes have been explored in studies worldwide to explain not taking medication as prescribed [17]. Egan et al. found forgetfulness, adverse effects and not liking to take medication among the reasons for poor adherence in a nationally representative sample in the United States [38]. Commonly encouraging factors, such as understanding the need and effectiveness of medication, a good support system and employing methods to reduce forgetfulness such as keeping medication in sight, were all significantly associated with better adherence in our population. Similarly, among the discouraging factors cited in literature, most commonly reported in our population were forgetfulness (48%) as has been reported by an earlier local study [22], followed by cost (40%) and fear of getting used to medication (27%). These were, however, factors that reduced adherence among the adherent (>80% adherence) population. This was different from the major factors reducing adherence in the non-adherent (<80% adherence) patients, whose main issues were lack of

understanding of need of medication (70%) and lack of understanding of effectiveness of medication (59%).

Depression has recently been added to the list of factors associated with non-adherence to anti-hypertensive medication [32]. Wang et al. [21] demonstrated a significant association between depression, as a multivariate factor, and non-adherence. In our study, however, depression was not found to independently correlate with non-adherence.

Our study design was limited in several aspects. Self-reporting was used as the only method of measuring adherence. Although this method has the disadvantages of recall bias, of eliciting only socially acceptable responses and hence, may overestimate adherence [14], it is simple, economically feasible and the most useful method in clinical settings [17]. Response rate of the participants was not recorded and thus in this study it is assumed that responders and non-responders are similar in distribution of the recorded variables. Apart from harboring the known limitations of the cross-sectional design, our study involved patients recruited from tertiary care hospitals only, and hence the results cannot be generalized. As we included patients with co-existing illnesses, some of our results may not be purely indicative of the characteristics of hypertensive patients. We did not include patients who could not converse in Urdu and this further restricts the generalization of our findings. Most adherence studies based on self-reporting ask patients to give information about long durations (usually ranging from 1 month to 1 year) to avoid any bias introduced by the brevity of the duration. Increasing the duration of time period could give a more generalized view of the patient's adherence over a longer period of time but at the same time increases the chance of introducing recall bias. Self-reported adherence based on a short duration of time has equal chances of being under-reported as over-reported, depending on the patient's behavior in the recent past, but minimizes the chance of recall bias and hence is more accurate. The published literature suggests that people are more accurate in reporting non-adherence when asked simple questions about recent behavior [27], [39].

The adherence goal of 80% of prescribed dose is used conventionally in clinical trials of safety and efficacy [4]. Hence, we used this value as a cut point for labeling patients as adherent and non-adherent. All the analyses, however, were repeated using a higher cut-off of 90% in order to dichotomize adherence and the results reported here were found to be consistent.

The use of validated tools in our study further strengthens the reliability of our results. MMAS had a significant linear relation with adherence measured as a continuum. Most of the factors associated with non-adherence discussed above were associated with low MMAS scores of ≤2, including those which did not show an association with adherence at the cut-off of 80%, such as depression. More studies, however, are needed to demonstrate the validity of MMAS in the Pakistani population. Hence, we have reported only those factors as significantly affecting adherence that were common to both our criteria for measurement of adherence.

In conclusion, we found younger age, monotherapy, poor awareness and symptomatic treatment to be the strongest factors affecting adherence to anti-hypertensive medication amongst Pakistani patients. Future studies are recommended to confirm our findings, as adherence to medication predicts better outcomes and indicators of poor adherence to a medication regimen are a useful resource for physicians to help identify patients who are most in need of interventions to improve adherence.

We recommend the implementation of education campaigns to increase awareness about the risk factors, natural history, complications and treatment of hypertension. Global events, such as World Hypertension Day, could be used as a forum to highlight these issues. Patient support groups can be employed to help the non-adherent. Patients who have suffered complications due to non-adherence could be requested to voluntarily share their experiences. Print and audiovisual media would be very helpful in dissemination of information. Most importantly, though, physicians have to pay special attention to patient education and counseling when treating hypertensive patients.

Acknowledgments Top

The authors wish to thank Dr. Pashtoon M. Kasi for his help in designing the questionnaire, and Dr. Mohammad Shehzad, Dr. Mahmood H. Jafri, Dr. Altaf Hussain (Research Medical Officers), Ms. Tuba Ali and Ms. Zarmeneh Aly (Medical students, Aga Khan University) for help with data collection. We acknowledge the work of Mrs. Zehra Siddiqui (Manager Administration), Mrs. Gulafroze Fazli (Senior Administrative Assistant) and Mr. Shahzad Noorali (Senior Administrative Assistant) for their help with data entry. The continuous administrative support of Mr. Hilary Fernandes (senior administrative assistant) is specially acknowledged. Finally, we wish to express our gratitude to the anonymous reviewer of the original version of our manuscript for his/her keen insight in providing us with a list of comments that allowed us to greatly improve the text.

Author Contributions Top

Conceived and designed the experiments: UA. Analyzed the data: UA SH. Wrote the paper: UA SH. Other: Designed and conducted the study and had equal participation: MA KA RS SH. Overall supervisor of the project, helped at all steps: PF. Was involved in study design, data analysis and review of the manuscript: MI AA DS.

References Top

1. Kearney PM, Whelton M, Reynolds K, Muntner P, Whelton PK, et al. (2005) Global burden of hypertension: analysis of worldwide data. Lancet 365: 217–23. Find this article online

2. Chobanian AV, Bakris GL, Black HR, et al. (2003) The seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: the JNC 7 report. JAMA 289: 2560–2572. Find this article online

3. National Health Survey of Pakistan 1990–1994 (1998) Karachi, Pakistan: Pakistan Medical Research Council.

4. Inkster ME, Donnan PT, MacDonald TM, Sullivan FM, Fahey T (2006) Adherence to antihypertensive medication and association with patient and practice factors. J Hum Hypertens 20: 295–297. Find this article online

5. Falaschetti E (2004) Blood pressure. In: Sproston K, Primatesta P, editors. Health Survey for England 2003. TSO. London: pp. pp 181–220.

6. Mant J, McManus RJ (2006) Does it matter whether patients take their antihypertensive medication as prescribed? The complex relationship between adherence and blood pressure control. J Hum Hypertens 20: 551–553. Find this article online

7. World Health Organization. Chapter III Hypertension in Adherence to Long-Term Therapies-Evidence for Action. 2003. p. 27. Available:

http://www.who.int/chronic_conditions/ad herencereport/en/print.html . Accessed August 2005.

8. Berenson GS, et al. (1998) Association between multiple cardiovascular risk factors and atherosclerosis in children and young adults. The Bogalusa Heart Study. N Engl J Med 338: 1650–1656. Find this article online

9. Thompson DW, Furlan AJ (1996) Clinical epidemiology of stroke. Neurol Clin 14: 309–315. Find this article online

10. Baune BT, Aljeesh YI, Bender R (2004) The impact of non-compliance with the therapeutic regimen on the development of stroke among hypertensive men and women in Gaza, Palestine. Saudi Med J 25(11): 1683–8. Find this article online

11. Baune BT, Aljeesh Y, Bender R (2005) Factors of non-compliance with the therapeutic regimen among hypertensive men and women: a case-control study to investigate risk factors of stroke. Eur J Epidemiol 20(5): 411–9. Find this article online

12. World Health Organization. Chapter XIII Hypertension in Adherence to Long-Term Therapies-Evidence for Action. 2003. pp. 129–136. Available: http://www.emro.who.int/ncd/Publications /adherence_report.pdf. Accessed February 2007.

13. Haynes RB, McDonald HP, Garg AX (2002) Helping patients follow prescribed treatment: clinical applications. JAMA 288: 2880–2883. Find this article online

14. Krousel-Wood M, Hyre A, Muntner P, Morisky D (2005) Methods to improve medication adherence in patients with hypertension: current status and future directions. Curr Opin Cardiol 20: 296–300. Find this article online

15. DiMatteo MR (2004) Variations in patients' adherence to medical recommendations: a quantitative review of 50 years of research. Med Care 42: 200–209. Find this article online

16. Hassan NB, Hasanah CI, Foong K, Naing L, Awang R, et al. (2006) Identification of psychosocial factors of noncompliance in hypertensive patients. J Hum Hypertens 20: 23–29. Find this article online

17. Osterberg L, Blaschke T (2005) Drug therapy: Adherence to Medication. N Engl J Med 353: 487–97. Find this article online

18. Bloom BS (2001) Daily regimen and compliance with treatment. BMJ 323: 647. Find this article online

19. Payne KA, Esmond-White S (2000) Observational studies of antihypertensive medication use and compliance: is drug choice a factor in treatment adherence? Curr Hypertens Rep 2: 515–524. Find this article online

20. Lennon C, Hughes CM, Johnston GD, McElnay JC (2001) Identification of psychosocial factors which influence patient adherence with antihypertensive medication. Int J Pharm Pract 9: R8. Find this article online

21. Wang PS, Bohn RL, Knight E, Glynn RJ, Mogun H, et al. (2002) Noncompliance with antihypertensive medications: the impact of depressive symptoms and psychosocial factors. J Gen Intern Med 17: 504–511. Find this article online

22. Almas A, Hameed A, Ahmed B, Islam M (2006) Compliance to antihypertensive therapy JCPSP 16: 23–26. Find this article online

23. Youssef RM, Moubarak II (2002) Patterns and determinants of treatment compliance among hypertensive patients. East Mediterr Health J 8: 4–5. Find this article online

24. Cheng JWM, Kalis MM, Feifer S (2001) Patient-Reported Adherence to Guidelines of the Sixth Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Pharmacotherapy 21(7): 828–841. Find this article online

25. Ali BS, Reza H, Khan MM, Jehan I (1998) Development of an indigenous screening instrument in Pakistan: the Aga Khan University Anxiety and Depression Scale. J Pak Med Assoc. 48: 261–5. Find this article online

26. Morisky DE, Green LW, Levine DM (1986) Concurrent and predictive validity of a self-reported measure of medication adherence. Med Care 24: 67–74. Find this article online

27. Pratt RJ, Robinson N, Loveday HP, Pellowe CM, Franks PJ, et al. (2001) Adherence to Antiretroviral Therapy: Appropriate use of self-reporting in clinical practice, HIV Clin Trials 2: 146–159. Find this article online

28. Garfield FB, Caro JJ (1999) Compliance and hypertension. Curr Hypertens Rep 1: 502–506. Find this article online

29. Rudd P, Ahmad S, Zachary V, Barton C, Bonduelle D (1991) Antihypertensive drug trials, contributions from medication monitors. In: Cramer JA, Spilker B, editors. Patient Compliance in Medical Practice and Clinical Trials. Raven Press Ltd. New York: pp. pp 283–299.

30. Caro JJ, Salas M, Speckman JL, et al. (1999) Persistence with treatment for hypertension in actual practice. CMAJ 160: 31–37. Find this article online

31. Monane M, Bohn RL, Gurwitz JH, et al. (1996) A population-based study of compliance with antihypertensive therapy: role of age, gender and race. Am J Public Health 86: 1805–1809. Find this article online

32. Krousel-Wood M, Thomasa S, Muntner P, Morisky D (2004) Medication adherence: a key factor in achieving blood pressure control and good clinical outcomes in hypertensive patients. Curr Opin Cardiol 19: 357–362. Find this article online

33. Gryglewska B (2005) How can we improve the effectiveness of treatment in elderly hypertensives? Blood Press 2: 46–9. Find this article online

34. Yiannakopoulou ECh, Papadopulos JS, Cokkinos DV, Mountokalakis TD (2005) Adherence to antihypertensive treatment: a critical factor for blood pressure control. Eur J Cardiovasc Prev Rehabil 12: 243–9. Find this article online

35. Iskedjian M, Einarson TR, MacKeigan LD, et al. (2002) Relationship between daily dose frequency and adherence to antihypertensive pharmacotherapy: evidence from a meta-analysis. Clin Ther 24: 302–316. Find this article online

36. Baune BT, Aljeesh Y (2006) The association of psychological stress and health related quality of life among patients with stroke and hypertension in Gaza Strip. Ann Gen Psychiatry 5(1): 6. Find this article online

37. Wang PS, Bohn RL, Knight E, et al. (2002) Non-compliance with antihypertensive medications. J Gen Intern Med 17: 504–511. Find this article online

38. Egan BH, Lackland DT, Cutler NE (2003) Awareness, knowledge and attitudes of older Americans about high blood pressure. Arch Intern Med 163: 681–687. Find this article online

39. Dunbar-Jacob J, Burke LE, Puczynski S (1995) Clinical assessment and management of adherence to medical regimens In: Nicassio M, Smith T, editors. Managing Chronic Illness: A Biophysical Perspective. American Psychological Association. Washington, DC: pp. 313–350.

Abstract Top

Background

The loss of diurnal rhythm in blood pressure (BP) is an important predictor of end-organ damage in hypertensive and diabetic patients. Recent evidence has suggested that two major physiological circadian rhythms, the metabolic and cardiovascular rhythms, are subject to regulation by overlapping molecular pathways, indicating that dysregulation of metabolic cycles could desynchronize the normal diurnal rhythm of BP with the daily light/dark cycle. However, little is known about the impact of changes in metabolic cycles on BP diurnal rhythm.

Methodology/Principal Findings

To test the hypothesis that feeding-fasting cycles could affect the diurnal pattern of BP, we used spontaneously hypertensive rats (SHR) which develop essential hypertension with disrupted diurnal BP rhythms and examined whether abnormal BP rhythms in SHR were caused by alteration in the daily feeding rhythm. We found that SHR exhibit attenuated feeding rhythm which accompanies disrupted rhythms in metabolic gene expression not only in metabolic tissues but also in cardiovascular tissues. More importantly, the correction of abnormal feeding rhythms in SHR restored the daily BP rhythm and was accompanied by changes in the timing of expression of key circadian and metabolic genes in cardiovascular tissues.

Conclusions/Significance

These results indicate that the metabolic cycle is an important determinant of the cardiovascular diurnal rhythm and that disrupted BP rhythms in hypertensive patients can be normalized by manipulating feeding cycles.

Citation: Cui H, Kohsaka A, Waki H, Bhuiyan MER, Gouraud SS, et al. (2011) Metabolic Cycles Are Linked to the Cardiovascular Diurnal Rhythm in Rats with Essential Hypertension. PLoS ONE 6(2): e17339. doi:10.1371/journal.pone.0017339

Editor: Paul Bartell, Pennsylvania State University, United States of America

Received: November 18, 2010; Accepted: January 27, 2011; Published: February 22, 2011

Copyright: © 2011 Cui et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This work was supported by Grants-in-Aid for Scientific Research (22590226) and for Young Scientists (20890198) from the Japan Society for the Promotion of Science. Additional support came from the Naito Foundation, the Mochida Memorial Foundation for Medical and Pharmaceutical Research, the Mishima Kaiun Memorial Foundation, and the Suzuken Memorial Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

* E-mail: kohsaka@wakayama-med.ac.jp

# These authors contributed equally to this work.

Introduction Top

In humans, diurnal variation in cardiovascular function, including the regulation of blood pressure (BP) and heart rate (HR), is among the best recognized physiological rhythms [1], [2]. One benefit of the cardiovascular diurnal rhythm is that it enables timely modulation of blood flow to organs according to their daily demands across the sleep-wake cycle. Dysregulation of the diurnal rhythm of BP may cause pressure overload within organs during the sleep cycle, facilitating the malfunctioning of major organs, including the heart, kidneys, brain, and eyes, in which the blood vessels are highly susceptible to damage caused by excessive BP. Consistent with this notion are clinical observations that an insufficient reduction in BP during sleep is associated with an increased risk of heart failure, stroke, and glomerular dysfunction in hypertensive and diabetic patients [3]–[5]. Furthermore, recent evidence in rodents suggests that disrupted BP rhythm impairs vascular endothelial function, further implicating the cardiovascular diurnal rhythm in the maintenance of human health [6].

The cardiovascular diurnal rhythm, like many other physiological and biological rhythms, is programmed by the circadian system. In this system, the master pacemaker in the suprachiasmatic nucleus (SCN) of the anterior hypothalamus is entrained to the daily light/dark (LD) cycle and subsequently transmits synchronizing signals to local clocks in peripheral tissues [7]. On the molecular level, it has become clear that cardiovascular tissues, including the heart and aorta, possess local clocks whose core molecular structures are similar to that of the master pacemaker in the SCN [8], [9]. The molecular clock is composed of interconnected feedback loops of gene transcription and translation in which the heterodimeric CLOCK/BMAL1 transcription factor complex activates the transcription of repressor clock genes such as the Period genes (Per1 and Per2) and Cryptochrome genes (Cry1 and Cry2). These protein products in turn inhibit CLOCK/BMAL1 transcriptional activity, thereby reducing their own transcription [7]. Recent studies showing that the diurnal variation in BP is disrupted in Clock mutant, Bmal1 knockout, and Cry1/Cry2 double knockout animals indicate that genetic components of the circadian system are required to generate the diurnal BP rhythm [10], [11].

While the molecular clocks in both the SCN and the peripheral tissues are self-sustained and cell autonomous [12], they are entrained to external cues to generate the appropriate 24 hr rhythms. Among these external cues, light and food are two strong stimuli that can entrain the molecular clock [13], [14]. Light entrainment is the dominant cue for the SCN clock, but the timing of food intake also appears to be important in resetting the phase of clock gene expression in peripheral tissues, including the heart [15]–[18], indicating that metabolic signaling from ingested food might be a driving force for the generation of the cardiovascular diurnal rhythm. Very little is known about how metabolic signals communicate with local clocks in cardiovascular tissue; however, studies employing whole-transcriptome profiling and gene targeting in cardiovascular tissue have revealed a possible molecular link between the metabolic cycle and the cardiovascular rhythm. Microarray studies have shown that the expression of metabolic genes involved in carbohydrate and fat metabolism displays diurnal variation not only in metabolic tissues but also in cardiovascular tissues [19]–[21].

Furthermore, recent studies have shown that mice with a blood vessel-specific deletion of peroxisome proliferator-activated receptor-γ (PPARγ), a nuclear receptor transcription factor that both acts as a lipid sensor and interacts with the molecular clock, have disrupted diurnal BP rhythms. This indicates that the regulation of the cardiovascular diurnal rhythm is tightly coupled with metabolic cycles at the molecular level [22].

In the current study, we hypothesized that dysregulation of the daily BP rhythm in mammals may be caused by altered metabolic cycles. We used spontaneously hypertensive rats (SHR), which develop essential hypertension with disrupted BP rhythm, to demonstrate that changes in metabolic cycles can account for the dysregulation of the cardiovascular diurnal rhythm.

Results Top

Impaired Diurnal Control of Central Feeding Circuits in SHR

To demonstrate that SHR have abnormal metabolic cycles, we first characterized the diurnal rhythms of feeding behavior, hypothalamic neuropeptide expression, and circulating hormones and nutrients in these animals (Figure 1). Both SHR and control (normotensive) Wistar-Kyoto rats (WKY) were acclimated to 12:12 light:dark (LD) conditions for a week; food intake was then measured during the light and dark periods. WKY showed a clear diurnal rhythm of feeding behavior, with only 11% of food intake occurring during the rest (light) period, whereas 20% of food intake occurred during the rest period in SHR (Figure 1A). The attenuation in the diurnal variation of feeding behavior in SHR was caused by significant increase in food intake during the light period and by a mild decrease during the dark period (Table S1). Interestingly, hyperphagia during the light period resulted in a significant increase in total food intake over 24 hrs in SHR, although the SHR were leaner than the WKY (Table S1).

Figure 1. Diurnal control of central feeding circuits is impaired in SHR.

(A) The diurnal rhythm of food intake in WKY (black bars, n = 6) and SHR (red bars, n = 6). Food intake was measured during 12 hr light (L) and dark (D) periods. Average food intake during each period is expressed as a percentage of

total food intake. (B and C) Diurnal variation in the mRNA expression levels of neuropeptides (NPY, AgRP, and POMC) in the mediobasal hypothalamus (B) and in blood metabolic parameters (glucose, insulin, leptin, and FFA) (C). Tissues and blood were collected every 4 hr from WKY (black lines, n = 6 per time point) and SHR (red lines, n = 6 per time point). Values for mRNA expression are shown as relative expression levels normalized to β-actin. The 12:12 LD cycle is indicated by the bars at the bottom of the figure. All data are expressed as means ± SEM. *p<0.05.

doi:10.1371/journal.pone.0017339.g001

We next examined diurnal rhythms in the levels of transcripts encoding hypothalamic neuropeptides that affect feeding behavior and energy balance (Figure 1B). These included selected orexigenic [neuropeptide Y (NPY) and agouti-related protein (AgRP)] as well as anorexigenic [pro-opiomelanocortin (POMC)] neuropeptides expressed in the mediobasal hypothalamus (MBH). As expected, the expression patterns of neuropeptides across the LD

cycle corresponded to the feeding behavior pattern in control animals (Figure 1B). In particular, the peak expression levels of NPY and AgRP were observed just before the onset of feeding behavior in WKY (Figure 1B). The expression of POMC in WKY was sustained at a higher level during the light period, when appetite is normally suppressed (Figure 1B). In contrast, SHR exhibited no temporal change in the expression levels of NPY and POMC (Figure 1B). An altered diurnal rhythm of AgRP levels was also found in SHR, with the peak expression shifted toward the dark period (Figure 1B). Notably, SHR exhibited significantly higher levels of the orexigenic neuropeptides NPY and AgRP at several time points of the LD cycle as well as markedly lower levels of the anorexigenic neuropeptide POMC at all time points during the light period (Figure 1B).

Since circulating nutrients and hormones secreted from metabolic tissues can directly affect the expression levels of hypothalamic neuropeptides [23]–[25], we also analyzed 24 hr profiles of glucose, insulin, leptin, and free fatty acids (FFA) in serum (Figure 1C). All four of these metabolic parameters displayed diurnal variation in both WKY and SHR (Figure 1C). However, the absolute levels and/or specific profile patterns differed between these two strains (Figure 1C). Serum glucose levels in SHR were significantly increased at the end of the dark period compared to those observed in WKY (Figure 1C). The overall levels of insulin also tended to be higher at virtually all time points of the LD cycle in SHR, although this difference was not statistically significant (Figure 1C and Table S2). In contrast to glucose and insulin, the overall levels of leptin were significantly reduced in SHR, particularly during the dark-to-light transition period (Figure 1C and Table S2). Interestingly, while the diurnal patterns of glucose, insulin, and leptin levels were not different between WKY and SHR, an altered diurnal rhythm of FFA levels was observed in SHR (Figure 1C). In particular, while the peak FFA level in WKY was observed at the end of the light period, the peak in SHR occurred during the dark period (Figure 1C). We also observed a significant increase in overall FFA levels in SHR (Table S2). Collectively, these findings reveal that SHR have disordered diurnal control of feeding circuits at the molecular and behavioral levels.

Disrupted Diurnal Regulation of Energy Metabolism in Liver and Fat in SHR

We next investigated whether the diurnal regulation of energy metabolism in peripheral tissues was also altered in SHR (Figure 2). Since the disrupted FFA rhythm observed in SHR could be induced by changes in the expression of genes controlling lipid metabolism, we first examined the 24 hr expression profiles of lipogenic genes in liver and fat tissues. We investigated the sterol regulatory element-binding protein 1c (SREBP-1c) transcript because SREBP-1c is a key transcription factor that regulates fatty acids and triglyceride synthesis depending on whether the animal is in a fed or fasted state. We also analyzed the diurnal expression of acetyl-CoA carboxylase (ACC), fatty acid synthase (FAS), and stearoyl-CoA desaturase-1 (SCD1) transcripts because the expression levels of these genes are directly regulated by SREBP-1c. In WKY, levels of SREBP-1c displayed profound diurnal variation in the liver, peaking during the dark-to-light transition period (Figure 2A, left panel). Similar time-dependent variation was observed for the expression levels of ACC, FAS, and SCD1 in the liver of WKY (Figure 2A, left panel). Interestingly, however, the diurnal variation of the hepatic expression of SREBP-1c, ACC, FAS, and SCD1 was attenuated in SHR compared to WKY. In addition, the levels of all four of these lipogenic genes were significantly decreased in liver tissue from SHR (Figure 2A, left panel).

Figure 2. Diurnal control of peripheral energy metabolism is disrupted in SHR.

Animals were maintained on a 12:12 LD cycle (indicated by the bars at the bottom) and fed ad libitum. Liver and visceral fat tissues were collected every 4 hr from WKY (black lines, n = 6 per time point) and SHR (red lines, n = 6 per time point). (A) Diurnal variation in the mRNA expression levels of lipogenic genes in the liver and visceral fat. (B) Diurnal variation in the transcript levels of gluconeogenic genes in the liver. (C) Diurnal expression patterns of PPARα in the liver and PPARγ in visceral fat. Transcript levels were determined by real-time PCR. Values are shown as relative expression levels normalized to β-actin. Results are expressed as means ± SEM. *p<0.05.

doi:10.1371/journal.pone.0017339.g002

In fat tissue, WKY also exhibited time-dependent variation in SREBP-1c, ACC, FAS, and SCD1 levels, although the patterns of expression were different from those observed in the liver (Figure 2A). While the overall levels of SREBP1c in fat tissue did not differ between WKY and SHR, the levels of ACC, FAS, and SCD1 in fat tissue were decreased in SHR compared to WKY, resulting in attenuated diurnal variation in the levels of each of these transcripts (Figure 2A, right panel).

Since blood glucose and insulin levels were increased in SHR, we also analyzed the temporal regulation of genes involved in hepatic gluconeogenesis (Figure 2B). We studied phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (G6P) mRNA expression in the liver because an increase in the expression of these genes is associated with hepatic insulin resistance. In both WKY and SHR, the levels of PEPCK and G6P varied across the 24 hr LD cycle, with peaks during the light period (Figure 2B). Although the diurnal patterns of PEPCK and G6P expression did not differ between the two strains, the levels of these gluconeogenic genes were increased at the beginning and end of the 12 hr light phase in SHR (Figure 2B). These increases were consistent with the mild insulin resistance observed in SHR (Figure 1C and Figure 2B).

We also examined the mRNA levels of PPARα in the liver and PPARγ in fat because these proteins have been shown to directly regulate both clock gene function and lipid metabolism (Figure 2C) [22], [26]. Both WKY and SHR displayed diurnal variation in the levels of PPARα in the liver and PPARγ in fat (Figure 2C). In particular, we observed a peak in PPARα expression at the end of the light period and a peak in PPARγ expression at the beginning of the dark period in both strains (Figure 2C). While the diurnal patterns of expression of PPARα and PPARγ were similar in WKY and SHR, increased expression levels were observed in tissues from SHR (Figure 2C).

Altered Clock Gene Expression in Metabolic Tissues of SHR

To further explore whether the disrupted rhythms of energy metabolism observed in SHR are due to changes in clock gene expression in metabolic tissues, we analyzed the diurnal patterns of the mRNA expression of core clock genes, including Clock, Bmal1, Per2 and Rev-erbα, within the MBH, liver, and fat (Figure 3). In the MBH of WKY, all four of these clock genes displayed small but clear diurnal variations in expression levels (Figure 3A). In SHR, however, the expression of Clock and Bmal1 within the MBH was increased at the beginning of the light phase and decreased at the end of light phase, resulting in a phase shift of the diurnal expression of these clock genes (Figure 3A). We also observed a shift in the phase of

Per2 expression levels in the MBH of SHR (Figure 3A). In contrast to Clock, Bmal1, and Per2, the timing of Rev-erbα expression did not differ between WKY and SHR, although the expression levels of Rev-erbα were significantly higher in SHR from the middle to the end of the dark period (Figure 3A).

Figure 3. Clock gene expression is altered in metabolic tissues of SHR.

[Note that the data in this figure were derived from the same animals represented in Figure 1 (the mediobasal hypothalamus, MBH) and Figure 2 (liver and visceral fat).] Transcript levels of core clock genes (Clock, Bmal1, Per2, and Rev-erbα) in (A) the MBH, (B) the liver, and (C) visceral fat were determined by real-time PCR in WKY (black lines, n = 6 per time point) and SHR (red lines, n = 6 per time point). Values are displayed as relative expression levels normalized to β-actin. White and black bars below the figures represent the 12:12 LD cycle. All data represent means ± SEM. *p<0.05.

doi:10.1371/journal.pone.0017339.g003

In contrast to the MBH, the diurnal expression patterns of Clock, Bmal1, Per2, and Rev-erbα in both liver and fat tissues were similar in WKY and SHR (Figures 3B and 3C), but the expression levels of all four of these clock genes differed. In particular, increased levels of Clock transcripts in the liver of SHR were observed during the light-to-dark transition period, whereas Clock levels in fat at the beginning of the light period were decreased in SHR compared to WKY (Figures 3B and 3C). SHR also displayed increased levels of Bmal1 in fat (Figure 3C). In addition, SHR showed generally increased levels of Per2 in fat and Rev-erbα in the liver during the light period (Figures 3B and 3C). Overall, we observed broad changes in the diurnal regulation of both metabolic and clock genes in tissues involved in energy metabolism in SHR.

Restricted Feeding Restores Diurnal Rhythms of BP and HR in SHR

To test our hypothesis that alteration of metabolic cycles affects the diurnal cardiovascular rhythm, we first characterized the cardiovascular phenotypes of WKY and SHR using radiotelemetry. We examined average levels of BP and HR during light and dark periods in WKY and SHR (Table S3). Consistent with previous findings, SHR displayed significantly higher levels of systolic and diastolic BP during both the light and the dark periods (Table S3). In contrast, HR was significantly reduced throughout the 12:12 LD cycles in SHR compared to WKY (Table S3).

We also performed a detailed analysis of BP and HR profiles over 24 hours in WKY and SHR (Figure 4). These are shown as percent changes in BP and HR to visually clarify differences in the diurnal variation of these cardiovascular parameters between experimental groups (Figures 4, top panels) as well as in the absolute levels of these cardiovascular parameters (Figure 4, bottom panels). WKY exhibited a robust diurnal rhythm of both systolic and diastolic BP, with higher levels during the dark period and persistently lower levels throughout the light period (Figures 4A and 4B). Interestingly, although SHR displayed diurnal rhythms of systolic and diastolic BP, the patterns of diurnal variation were different from those observed in WKY (Figures 4A and 4B, top panels). In particular, the distinct BP patterns in SHR were characterized by (1) a more gradual reduction after the

beginning of the light period, (2) earlier increase prior to the active (dark) period, and (3) sustained, increased levels during the mid-dark period, while WKY exhibited relatively lower levels in the dark period (Figures 4A and 4B). We also observed a clear diurnal HR pattern in WKY (Figure 4C). In parallel with the changes in the BP rhythm, SHR exhibited a similar alteration in the diurnal rhythm of HR, although the change was not statistically significant (Figure 4C).

Figure 4. Effects of restricted feeding on diurnal rhythms and levels of BP and HR in SHR.

Blood pressure (BP) and heart rate (HR) were telemetrically monitored in freely moving animals maintained on a 12:12 LD cycle (indicated by the bars at the bottom). Data were collected for the first 5 min of every 60 min; the value at each time point represents an average of values taken over 2 consecutive days. (A) Systolic and (B) diastolic BP and (C) HR data are expressed as the mean (± SEM) of the percent change from the highest value in each individual animal (top panels) and as absolute levels (bottom panels). After BP and HR were recorded in WKY (black lines, n = 6) and SHR (red lines, n = 6) fed ad libitum, the same animals were exposed to a restricted feeding regimen in which food was provided only during the active (dark) period. BP and HR were monitored again during the fourth and fifth days of the RF regimen. Only RF data from SHR (blue lines, n = 6) are shown for visual clarity. The value shown at each time point represents the mean ± SEM. #, SHR versus WKY; *, SHR ad libitum versus RF; p<0.05.

doi:10.1371/journal.pone.0017339.g004

To further explore whether the changes in cardiovascular diurnal rhythm observed in SHR are due to changes in feeding cycles, we exposed SHR to a restricted feeding (RF) regimen in which food was available only during the dark period and monitored BP and HR (Figure 4). Interestingly, the increased percent changes in systolic and diastolic BP at the beginning of the light period were significantly reduced in SHR exposed to this regimen (Figures 4A and 4B, top panels). Similar hypotensive effects of RF on percent changes in systolic and diastolic BP at the end of the light period were also observed in SHR, although these were not statistically significant (Figures 4A and 4B, top panels). Furthermore, increased percent changes in the levels of systolic and diastolic BP during the mid-dark period in SHR were also markedly suppressed by RF (Figures 4A and 4B). In addition, we were surprised to find that the overall levels of systolic BP were decreased in SHR exposed to RF, although we did not observe such effects of RF on diastolic BP levels (Figures 4A and 4B, bottom panels). We also analyzed the effects of RF on diurnal variation in the HR of SHR. Although the difference was not statistically significant, increased levels of HR at the beginning of the light period were somewhat suppressed by RF (Figure 4C, top panel). In addition, SHR exposed to RF displayed lower overall HR during both the light and the dark periods (Figure 4C, bottom panel).

To ensure that the RF regimen did not induce a negative energy balance, which might on its own affect the regulation of cardiovascular function, we examined food intake and body weight in SHR fed ad libitum and SHR under RF conditions (Table S4). Neither total food intake over 24 hrs nor body weight differed between feeding paradigms (ad libitum versus

RF) in SHR (Table S4). Taken together, these results demonstrate that feeding cycles affect the cardiovascular diurnal rhythm in SHR.

Restricted Feeding Restores the Diurnal Rhythms of Clock and Metabolic Gene Expression in Cardiovascular Tissues in SHR

To examine the molecular mechanisms that link feeding cycles with the diurnal cardiovascular rhythm, we analyzed the expression of clock and metabolic genes within the heart and aorta (Figure 5). Consistent with previous findings, the levels of Clock, Bmal1, Per2, and Rev-erbα expression displayed robust diurnal variation in both the heart and the aorta in WKY (Figure 5A). In SHR, all four of these clock genes exhibited expression rhythms similar to those observed in WKY; however, Bmal1 and Rev-erbα levels were altered in SHR (Figure 5A). In particular, Bmal1 expression in the heart was significantly increased during the dark period, and the peak levels of Rev-erbα expression in both heart and aortic tissue were significantly reduced (Figure 5A). Surprisingly, however, the expression of these genes was restored to the levels observed in WKY upon exposure to an RF regimen (Figure 5A, left panel). We also investigated the effects of RF on the expression of Clock and Per2 in the heart of SHR (Figure 5A, left panel). In particular, the overall levels of Clock expression were increased in SHR exposed to RF (Figure 5A, left panel). RF also caused an increase in Per2 expression at the end of the light period in SHR (Figure 5A, left panel). In contrast to the broad effects of RF on clock genes in the heart, this feeding regimen affected a decrease in Bmal1 only at the beginning of the light period in the aorta (Figure 5A, right panel).

Figure 5. Effects of restricted feeding on clock and metabolic gene expression in cardiovascular tissues of SHR.

The levels of transcripts from (A) clock and (B) metabolic genes in the heart and aorta were determined by real time PCR. Animals were either fed ad libitum or exposed to an RF regimen, in which food was provided exclusively during the dark period, for 5 consecutive days. Heart and aorta tissues were harvested every 4 hr on the fifth day under these feeding conditions. For visual clarity, only data from the ad libitum-fed WKY (black lines, n = 4 per time point) and SHR (red lines, n = 4 per time point) and RF-fed SHR (blue lines, n = 4 per time point) are shown. Values for mRNA expression are displayed as relative expression levels normalized to β-actin. The 12:12 LD cycle is indicated by the bars at the bottom of the figure. All data are expressed as means ± SEM. #, SHR versus WKY; *, SHR fed ad libitum versus RF; p<0.05.

doi:10.1371/journal.pone.0017339.g005

Because genes encoding CRY1 and CRY2 have recently been shown to play an important role in the development of salt-induced hypertension [27], we examined the expression levels of these clock genes (Figure S1). The levels of Cry1 displayed clear diurnal variation in both the heart and the aorta in WKY (Figure S1). SHR exhibited similar diurnal rhythms for Cry1 in both tissues, but levels in the aorta were decreased at the beginning of the light phase regardless of feeding condition (Figure S1). In heart and aortic tissues, Cry2 also showed minor diurnal changes in WKY (Figure S1). However, Cry2 expression in SHR heart tissue displayed an augmented diurnal rhythm that could be partially restored by RF (Figure S1).

Interestingly, in parallel with the changes in Cry1 expression in the aorta, levels of Cry2 in the aorta were also reduced at the beginning of the light period in SHR (Figure S1).

We also studied PPARα and PPARγ transcription in cardiovascular tissues because these PPARs are abundantly expressed not only in metabolic tissues, but also in cardiovascular tissues (Figure 5B). In WKY, levels of PPARα displayed diurnal variation in both the heart and the aorta, although the patterns of expression were distinct between the two tissues (Figure 5B). SHR also exhibited time-dependent variation in PPARα levels in the heart and aorta; however, in contrast to the liver, the phase of peak PPARα expression differed between the two strains (Figure 2C and Figure 5B). Similarly, we observed strain-dependent expression patterns in PPARγ levels in both heart and aortic tissues (Figure 5B). Both the diurnal patterns and the expression levels of the PPARs differed between WKY and SHR. Elevated PPARα and PPARγ levels in the heart were observed during the dark period, although the expression of both PPARs in the aorta was reduced during the early light phase and the mid-dark period in SHR (Figure 5B). In heart tissue, while a RF regimen did not alter PPARα expression, it was able to completely restore normal PPARγ expression levels during the dark period (Figure 5B). In addition, the overall PPARα and PPARγ levels in the aorta were lower in SHR exposed to RF than in those fed ad libitum, although this difference was not significant (Figure 5B).

Since cardiovascular tissues constantly require energy throughout the sleep/wake cycle, we examined the expression levels of genes that can sense changes in energy balance. We studied SCD1 because recent studies suggest that this gene plays an important role in glucose and lipid metabolism in the heart (Figure 5B) [28]. We also analyzed PEPCK, which is known to cycle in the aorta (Figure 5B) [21]. In heart tissue, neither SCD1 nor PEPCK displayed diurnal variation in expression in WKY (Figure 5B), while increased levels of both SCD1 and PEPCK during the dark period were observed in SHR (Figure 5B). Interestingly, however, a RF regimen completely abolished these changes such that gene expression was reduced to the levels of WKY (Figure 5B).

Although we did not observe diurnal variation in SCD1 expression in the aorta of WKY, PEPCK RNA levels displayed a diurnal rhythm of expression, peaking during the first part of the light period (Figure 5B). However, the peak expression levels of PEPCK were significantly attenuated in the aorta of SHR (Figure 5B). In contrast to PEPCK, SCD1 expression in the aorta was significantly higher in SHR than in WKY; however, the increases occurred during the light period and were only partially restored by RF, in contrast to its expression in heart tissue (Figure 5B). In addition, when compared to SHR fed ad libitum, SHR on a RF regimen displayed elevated levels of SCD1 in the aorta during the dark period (Figure 5B).

Circulating FFA directly affects the expression of metabolic genes in cardiovascular tissues [29]. We therefore analyzed the effects of RF on the diurnal pattern of FFA levels in SHR (Figure S2). When SHR were placed on a RF schedule, these animals displayed restored rhythms as well as decreased levels of FFA levels across the 24 hr LD cycle (Figure S2). Collectively, these findings suggest that metabolic cycles originating from feeding behavior determine temporal changes in the expression of both clock and metabolic genes in cardiovascular tissues.

Effects of Restricted Feeding on the Diurnal Rhythms of Serum Corticosterone and Urine Norepinephrine in SHR

We also evaluated diurnal variation in serum corticosterone levels because the dysregulation of rhythms in both metabolic and cardiovascular function observed in SHR could be explained by changes in glucocorticoid regulation (Figure 6A). Consistent with previous findings, WKY animals displayed a clear diurnal rhythm in corticosterone levels with a peak just before the active (dark) period (Figure 6A). In SHR, we observed a similar time-dependent cycling of corticosterone, but the levels of the hormone differed between WKY and SHR (Figure 6A). In particular, increased levels of corticosterone during the light period were observed in SHR compared to WKY, although this difference did not reach statistical significance (Figure 6A). In contrast to the broad effects of RF on serum FFA and gene expression levels, neither diurnal variation nor the levels of corticosterone were altered in SHR exposed to RF (Figure 6A).

Figure 6. Effects of restricted feeding on diurnal rhythms in glucocorticoid and catecholamine levels in SHR.

Animals maintained under a 12:12 LD cycle were fed ad libitum or maintained under restricted feeding (RF) conditions for 5 consecutive days. Under the RF

condition, food was available only during the active (dark) period. (A) Serum corticosterone levels were analyzed by ELISA. On the fifth day of the feeding regimen, blood was collected every 4 hr from the free-fed WKY (black lines, n = 4 per time point) and SHR (red lines, n = 4 per time point) and the RF-fed SHR (blues lines, n = 4 per time point). (B) Urine norepinephrine levels were determined for the light (L) and dark (D) periods. On the fourth and fifth days of the RF regimen, urine was collected from the free-fed WKY (black bars, n = 4) and the SHR (red bars, n = 4) as well as the RF-fed SHR (blue bars, n = 4) in vials containing 100 µl 6 M HCl. The results shown represent averages over 2 consecutive days. All values are expressed as means ± SEM. *p<0.05.

doi:10.1371/journal.pone.0017339.g006

Since sympathetic nerve activity also affects cardiovascular and metabolic processes, we analyzed urine norepinephrine levels; norepinephrine is the predominant catecholamine synthesized in the peripheral sympathetic nerves (Figure 6B). While not statistically significant, WKY exhibited a slight diurnal variation in urine norepinephrine levels (Figure 6B). In SHR, we observed attenuated diurnal variation in urine norepinephrine levels (Figure 6B). Interestingly, RF provoked a clear diurnal rhythm in the norepinephrine levels in SHR (Figure 6B).

Discussion Top

The Daily Cycle of Energy Metabolism Drives the Diurnal Cardiovascular Rhythm

When clock genes were identified in mammals, it became clear that regulation of these genes is tightly coupled to energy metabolism as well as to the photoperiod. Perhaps most striking was the demonstration that feeding rodents exclusively during the rest (light) period, when they normally sleep, inverts the phase of diurnal variation in the clock gene's expression in metabolic tissues, including the liver and fat [30], [31]. A similar phase resetting of clock genes has also been observed in the heart [15], [17], [18], suggesting that, even in cardiovascular tissues, clock genes sense metabolic cycles. Our observation that changes in

feeding cycles affect diurnal variation in BP and HR indicates that the response to changes in metabolic cycles is not limited to the level of gene expression (i.e., clock and metabolic genes) but reaches the level of physiological functions (e.g., BP and HR). Although the difference between the feeding rhythms of SHR exposed to RF and ad libitum feeding was subtle (i.e., 100% versus 80% food intake during the active period), this small alteration in feeding cycles was sufficient to affect the cardiovascular diurnal rhythm. Recent studies showing that the db/db mouse, an animal model of obesity and diabetes, exhibits a disrupted diurnal rhythm of BP also suggest that the cardiovascular diurnal rhythm is sensitive to changes in energy metabolism [32], [33].

We also examined the 24 hr profile of locomotor activity as measured by infrared beam crossing (data not shown) because voluntary locomotor activity is closely associated with BP and HR. However, we observed no correlation between activity change and BP (or HR) change. For example, SHR exhibited increases in BP levels at the beginning and the end of the light period, but we did not observe corresponding increases in locomotor activity at identical time periods. In addition, despite the significant effects of the RF regimen on BP and HR rhythms in SHR, RF did not influence the diurnal variation of locomotor activity in these animals. These results suggest that the alterations of diurnal BP and HR rhythms observed in SHR were not caused by a change in locomotor activity.

Complex Interplay between Circadian and Metabolic Genes in the Cardiovascular System

At the molecular level, we observed that metabolic cycles directly or indirectly affect the temporal patterns and levels of expression of both circadian and metabolic genes in cardiovascular tissues. These results indicate that the diurnal rhythms in BP and HR are coordinated by complex temporal networks of circadian and metabolic genes in cardiovascular tissues and raise several interesting possibilities. First, it is possible that feeding cycles affect diurnal variation primarily via clock gene expression, which then determines metabolic gene expression. In fact, the CLOCK/BMAL1 heterodimer can directly activate the transcription of PPARα through the E-box enhancer on its promoter [34], [35]. Interestingly, we found that SHR displayed increased levels of Bmal1 expression in the heart in parallel with elevated PPARα expression during the dark period; changes in feeding cycles affected Bmal1 expression but did not change PPARα expression.

Another possibility is that feeding cycles initially regulate the expression of metabolic genes, which then affect clock gene expression. Among the metabolic genes examined in this study, the PPARs have been shown to directly regulate Bmal1 expression [22], [26]. Similar to PPARα, the expression of PPARγ in the heart was elevated during the dark period in SHR. However, in contrast to PPARα, the increased levels of PPARγ in the heart tissue of SHR were restored to the levels observed in WKY when exposed to an RF regimen. Importantly, we also observed similar effects of RF not only on Bmal1 but also on Rev-erbα, which is another clock gene target of PPARγ [36], suggesting that PPARγ may play an important role in linking metabolic cycles arising from food ingestion with clock gene function in cardiovascular tissues. Indeed, recent work by Wang and coworkers [22] demonstrated that the diurnal BP rhythm is disrupted in mice with a blood vessel-specific deletion of PPARγ. Additional evidence that PPARγ coordinates the diurnal rhythm of BP comes from human studies showing that thiazolidinediones, which are PPARγ-selective ligands, are able to normalize the diurnal BP rhythm in diabetic subjects [37].

Our findings that even subtle changes to the feeding cycle can affect diurnal variation in BP and HR raise several important questions. First, what major metabolic signal(s) affect the daily cycle of gene expression in cardiovascular tissues? Among the circulating nutrients and hormones examined in this study, FFA is a strong candidate because only FFA exhibited a disrupted diurnal rhythm in SHR and because this disruption was normalized when SHR were exposed to an RF regimen. Our observation that the diurnal rhythm of lipid metabolism, but not glucose metabolism, was predominantly impaired in SHR supports this idea. For example, SHR displayed attenuated rhythms in lipogenic gene expression in the liver and in fat tissues, whereas the diurnal rhythms of gluconeogenic gene expression were nearly intact in the livers of these animals. Notably, the expression of a large number of lipid-sensing nuclear receptors in the liver, fat, and muscle follows a diurnal rhythm [38]. In addition, Durgan and coworkers [29] demonstrated that the molecular clock in cardiomyocytes is able to sense temporal changes in FFA levels. Combined with these previous reports, our results indicate that the diurnal rhythm of lipid metabolism may play an important role in the generation of circadian rhythms of cardiovascular function.

Metabolic Cycles and Autonomic Function

Our observation that attenuation of the diurnal rhythm of urine norepinephrine levels in SHR is restored by the correction of feeding cycles indicates a close association between diurnal rhythms of feeding behavior and sympathetic nerve activity. It has been shown that the hypothalamic nuclei involved in feeding behavior have direct neuronal projections to spinal sympathetic preganglionic neurons [39]. Remarkably, in SHR, we observed impaired diurnal rhythms of the expression of neuropeptides in the MBH, suggesting that the attenuation of the norepinephrine rhythm could be due to the dysregulation of feeding circuits in the MBH. On the other hand, strong evidence has established a tight connection between the circadian and sympathetic nerve systems [40], and therefore, peripheral clock gene function can be altered if sympathetic nerve activity is changed. Importantly, the effects of sympathetic nerve activity on the peripheral clock might be tissue-specific; studies using parabiosis between intact and SCN-ablated animals demonstrated that the molecular clock requires neuronal signals to maintain circadian function in the heart, but not in the liver [41]. Indeed, in SHR, we observed that diurnal patterns of Bmal1 and Rev-erbα expression were impaired more severely in the heart than in the liver. Together with previous studies, our results suggest that, in SHR, disrupted rhythms in BP and clock gene expression in the heart could be in part due to altered daily rhythms in sympathetic nerve activity. However, the role of sympathetic input in maintaining the cardiovascular diurnal rhythm remains ambiguous. At the molecular level, although catecholamines affect the phase of clock gene expression in the ex vivo heart [42], the rhythms of clock gene expression in the heart can be entrained to daytime feeding even in catecholamine-deficient mice, which display attenuated BP rhythms [43], [44]. This finding suggests that the sympathetic signal is one of important but not the sole factor that modulates clock gene function in cardiovascular tissues [45]. Further studies should clarify the role of the sympathetic nerve system in mediating the circadian, metabolic, and cardiovascular systems.

Conclusions

We have shown here that diurnal rhythms of BP and HR are tightly coupled with the daily cycle of feeding behavior in rats with essential hypertension. Furthermore, changes in feeding cycles affect both metabolic and clock gene regulation in cardiovascular tissues. Recent studies have revealed the complex nature of interactions between the circadian and

cardiovascular systems [46], as well as between the circadian and metabolic systems [47]. Our observations provide additional evidence that these three systems cooperate to generate diurnal cardiovascular rhythms and suggest that monitoring and/or maintaining the daily cycle of feeding behavior could be beneficial for predicting and improving the prognosis of hypertension, diabetes, and obesity in humans.

Materials and Methods Top

Animals

Male WKY/Izm and SHR/Izm rats were purchased from Japan SLC, Inc. (Hamamatsu, Japan) and maintained in the Laboratory Animal Center at Wakayama Medical University. All experiments were carried out in rats between 8 and 10 weeks of age. All animal care and use procedures were conducted in accordance with guidelines approved by the Wakayama Medical University Institutional Animal Care and Use Committee (Permit Number: 332).

Blood and Tissue Collection

WKY and SHR (8 weeks old) were maintained under a 12:12 LD cycle with free access to food and water for 1 week. Animals were then sacrificed at 4 hr intervals across the 24 hr LD cycle. Blood was collected by cardiac puncture. The mediobasal hypothalamus (MBH) was dissected from the brain with two coronal cuts, each just posterior to the optic chiasm and the pituitary stalk. A pair of sagittal cuts was made 2.5 mm from the midline. A final horizontal cut was made 2 mm dorsal to the floor of the MBH. Peripheral tissues (i.e., liver, visceral fat, heart and aorta) were also obtained after the blood and MBH were collected. Serum was separated from whole blood by centrifugation (2,000 x g) for 20 min at 4°C and stored at −80°C. Tissues were preserved in RNA stabilization solution at −20°C for subsequent analysis. The specific procedures used to analyze blood and tissue samples are described below.

Blood Analysis

Serum insulin and leptin concentrations were determined by ELISA (Morinaga Institute of Biological Science, Inc., Yokohama, Japan). Glucose levels were measured with a glucometer (Glutest Neo; Sanwa Kagaku Kenkyusho Co., Ltd., Nagoya, Japan). Free fatty acid (FFA) levels were determined by the enzymatic colorimetric method (NEFA C-test Kit; Wako Diagnostics, Osaka, Japan). Serum corticosterone was analyzed by ELISA (AssayPro, St. Charles, MO).

RNA Extraction and Quantitative Real-Time PCR

Total RNA was extracted from tissues preserved in RNA stabilization solution with Trizol (Invitrogen, Carlsbad, CA). cDNAs were synthesized using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). Quantitative real-time PCR was performed and analyzed using a TP850 Thermal Cycler Dice Real-time System (Takara Bio, Inc., Otsu, Japan). Samples contained 1 X SYBR Premix Ex Taq II (Takara Bio, Inc.), 250 nM of each primer, and cDNA in a 10 µl volume. The PCR conditions were as follows: 3 min at 95°C, then 35 cycles of 5 s at 95°C and 20 s at 60°C. Relative expression in comparison with β-actin was calculated by the comparative CT method. The sequences of primers used in this study are shown in Table S5.

Feeding Rhythm Analysis

WKY and SHR (8 weeks old) were caged individually with free access to food and water and maintained under a 12:12 LD cycle. Food intake was measured immediately before the onset of the dark period and immediately after the onset of the light period. Averages represent intake over 2 consecutive days.

Analysis of Cardiovascular Parameters

To obtain arterial blood pressure (BP) and heart rate (HR) measurements, a battery-operated transmitter with a pressure-sensing catheter (PA-C40; Data Science International, St Paul, MN) was implanted into 7 week old WKY and SHR. Briefly, the catheter was inserted into the abdominal aorta of animals anesthetized with pentobarbital (50 mg/kg, i.p.), and the transmitter body was placed in the abdominal cavity according to the manufacturer's instructions. Following surgery, animals were given food and water ad libitum and allowed to recover for at least 1 week before data were collected through a receiver (RPC-1; Data Science International). The collected data were stored and analyzed with a computerized system (PowerLab/8s; ADInstruments Japan, Inc., Nagoya, Japan) to determine 24 hr diurnal variation in systolic and diastolic BP and HR. These cardiovascular parameters were monitored for the first 5 min of every 60 min for at least 2 days. Results at each time point represent an average over 2 consecutive days. Both BP and HR were calculated in absolute values and in percent changes from the highest value of each cardiovascular parameter in each individual animal.

Restricted Feeding Studies

All animals were maintained under a 12:12 LD cycle and provided with food ad libitum for at least 1 week. One group of animals was then exposed to a restricted feeding (RF) regimen for 5 days, during which food was available only during the dark period while the other group continued to have free access to food. Three sets of animals were used in the RF studies. In the first set of animals, we monitored BP and HR as described above on the fourth and fifth days of the RF regimen. We also measured BP and HR before starting the RF regimen; this measurement served as a baseline condition. In this set of animals, total food intake and BW were also measured on the last day of the RF regimen. In the second set of animals, heart, aorta, and blood were collected every 4 hr on the fifth day of the RF regimen for measurements of serum FFA and gene expression levels. The last set of animals was placed in metabolic cages. After allowing 1 week for acclimatization to the cage, the RF regimen was started, and urine was collected during the light and dark periods on the fourth and fifth days of the RF regimen. We also collected urine for 2 consecutive days before beginning the RF regimen, which served as the baseline condition. Urine was collected in a vial containing 100 µl of 6 M HCl and stored at −20°C for subsequent analysis.

Catecholamine Assay

Norepinephrine levels in urine were determined by ELISA (GenWay Biotech, Inc., San Diego, CA). The results are shown as average values over the 2 days of urine collection.

Statistical Analysis

All results are presented as means ± SEM. Two-way analysis of variance (ANOVA) and Bonferroni tests were used for comparisons among multiple groups. A paired or unpaired two-tailed Student's t test was also used for comparisons between two groups. Significance was assumed for p values <0.05.

Supporting Information Top

Figure S1.

Effects of restricted feeding on the Cry gene expression in cardiovascular tissues of SHR. The levels of Cry1 and Cry2 mRNA expression in the heart and aorta were determined by real time PCR. Animals were either fed ad libitum or exposed to an RF regimen, in which food was provided exclusively during the dark period, for 5 consecutive days. On the fifth day, heart and aortic tissues were harvested from animals at 4-hr intervals. For visual clarity, only data from ad libitum-fed WKY (black lines, n = 4 per time point) and SHR (red lines, n = 4 per time point) and RF-fed SHR (blue lines, n = 4 per time point) are shown. Values for mRNA expression are displayed as relative expression levels normalized to β-actin. The 12:12 LD cycle is indicated by the bars at the bottom of the figure. All data are expressed as means ± SEM. #, SHR versus WKY; *, SHR fed ad libitum versus RF; p<0.05.

(DOC)

Figure S2.

Effects of restricted feeding on diurnal rhythms of serum free fatty acids levels in SHR. Animals maintained under a 12:12 LD cycle were fed ad libitum or maintained under restricted feeding (RF) conditions for 5 consecutive days. Under the RF condition, food was available only during the active (dark) period. On the fifth day of the feeding regimen, blood was collected every 4 hr from the free-fed WKY (black lines, n = 4 per time point) and SHR (red lines, n = 4 per time point) and the RF-fed SHR (blues lines, n = 4 per time point). The levels of free fatty acids (FFA) were determined by the enzymatic colorimetric method. Note that the results for WKY (black lines) and SHR (red lines) fed ad libitum were reproduced from data used in Figure 1C to clarify the difference in diurnal patterns between groups. Values are expressed as means ± SEM.

(DOC)

Table S1.

Food intake and body weight in WKY and SHR. Food intake (grams) was determined for the 12-hr light, 12-hr dark and total 24-hr periods in the same set of WKY (n = 6) and SHR (n = 6) represented in Figure 1A. Body weight was also compared between these animals. Values are displayed as means ± SEM.

(DOC)

Table S2.

Metabolic parameters in WKY and SHR. [Note that the data in this table were derived from the same animals represented in Figure 1C.] The levels of glucose, insulin, leptin, and

free fatty acids (FFA) in the blood were determined in 8- to 9-week-old animals fed ad libitum. Blood was collected at 4-hr intervals over a 24-hr period from the hearts of WKY (n = 6 per time point) and SHR (n = 6 per time point). The data were pooled to provide an average level of each metabolic parameter over a 24-hr period. Values are presented as means ± SEM.

(DOC)

Table S3.

Cardiovascular parameters in WKY and SHR. [Note that the data in this table were derived from the same animals represented in Figure 4.] Systolic and diastolic blood pressure (BP) and heart rate (HR) were telemetrically monitored in freely moving WKY (n = 6) and SHR (n = 6). Data were collected for the first 5 min of every 60 min for 2 consecutive days and pooled to determine an overall mean (± SEM) value for the 12-hr light and 12-hr dark periods.

(DOC)

Table S4.

Food intake and body weight in SHR under free and restricted feeding conditions. SHR were maintained on a 12:12 LD cycle and exposed either to a free feeding condition or to a restricted feeding (RF) regimen in which food was provided only during the active (dark) period. On the fifth day of the feeding regimens, total food intake over a 24-hr period and body weight were determined in SHR fed ad libitum (n = 6) and SHR exposed to RF (SHR-RF, n = 6). Note that the data in this table were derived from animals 1 to 2 weeks older than those represented in table S1 because the animals in this table were exposed to feeding paradigms for nearly 1 week. All data are expressed as means ± SEM.

(DOC)

Table S5.

Primer sequences used for real-time PCR.

(DOC)

Acknowledgments Top

We thank Y. Miyazaki and other members of the Maeda Laboratory for helpful discussions and technical assistance.

Author Contributions Top

Conceived and designed the experiments: HC AK MM. Performed the experiments: HC AK HW MERB SSG. Analyzed the data: HC AK HW. Wrote the paper: HC AK MM.

References Top

1. Millar-Craig MW, Bishop CN, Raftery EB (1978) Circadian variation of blood-pressure. Lancet 1: 795–797. Find this article online

2. Pickering TG, Harshfield GA, Kleinert HD, Blank S, Laragh JH (1982) Blood pressure during normal daily activities, sleep, and exercise. Comparison of values in normal and hypertensive subjects. JAMA 247: 992–996. Find this article online

3. Lurbe E, Redon J, Kesani A, Pascual JM, Tacons J, et al. (2002) Increase in nocturnal blood pressure and progression to microalbuminuria in type 1 diabetes. N Engl J Med 347: 797–805. Find this article online

4. O'Brien E, Sheridan J, O'Malley K (1988) Dippers and non-dippers. Lancet 2: 397. Find this article online

5. Verdecchia P, Schillaci G, Guerrieri M, Gatteschi C, Benemio G, et al. (1990) Circadian blood pressure changes and left ventricular hypertrophy in essential hypertension. Circulation 81: 528–536. Find this article online

6. Anea CB, Zhang M, Stepp DW, Simkins GB, Reed G, et al. (2009) Vascular disease in mice with a dysfunctional circadian clock. Circulation 119: 1510–1517. Find this article online

7. Lowrey PL, Takahashi JS (2004) Mammalian circadian biology: elucidating genome-wide levels of temporal organization. Annu Rev Genomics Hum Genet 5: 407–441. Find this article online

8. Reilly DF, Westgate EJ, FitzGerald GA (2007) Peripheral circadian clocks in the vasculature. Arterioscler Thromb Vasc Biol 27: 1694–1705. Find this article online

9. Young ME (2006) The circadian clock within the heart: potential influence on myocardial gene expression, metabolism, and function. Am J Physiol Heart Circ Physiol 290: H1–16. Find this article online

10. Masuki S, Todo T, Nakano Y, Okamura H, Nose H (2005) Reduced alpha-adrenoceptor responsiveness and enhanced baroreflex sensitivity in Cry-deficient mice lacking a biological clock. J Physiol 566: 213–224. Find this article online

11. Curtis AM, Cheng Y, Kapoor S, Reilly D, Price TS, et al. (2007) Circadian variation of blood pressure and the vascular response to asynchronous stress. Proc Natl Acad Sci U S A 104: 3450–3455. Find this article online

12. Yoo SH, Yamazaki S, Lowrey PL, Shimomura K, Ko CH, et al. (2004) PERIOD2::LUCIFERASE real-time reporting of circadian dynamics reveals persistent circadian oscillations in mouse peripheral tissues. Proc Natl Acad Sci U S A 101: 5339–5346. Find this article online

13. Reppert SM, Weaver DR (2002) Coordination of circadian timing in mammals. Nature 418: 935–941. Find this article online

14. Rutter J, Reick M, McKnight SL (2002) Metabolism and the control of circadian rhythms. Annu Rev Biochem 71: 307–331. Find this article online

15. Goh BC, Wu X, Evans AE, Johnson ML, Hill MR, et al. (2007) Food entrainment of circadian gene expression altered in PPARalpha-/- brown fat and heart. Biochem Biophys Res Commun 360: 828–833. Find this article online

16. Wu T, Jin Y, Ni Y, Zhang D, Kato H, et al. (2008) Effects of light cues on re-entrainment of the food-dominated peripheral clocks in mammals. Gene 419: 27–34. Find this article online

17. Damiola F, Le Minh N, Preitner N, Kornmann B, Fleury-Olela F, et al. (2000) Restricted feeding uncouples circadian oscillators in peripheral tissues from the central pacemaker in the suprachiasmatic nucleus. Genes Dev 14: 2950–2961. Find this article online

18. Oishi K, Kasamatsu M, Ishida N (2004) Gene- and tissue-specific alterations of circadian clock gene expression in streptozotocin-induced diabetic mice under

restricted feeding. Biochem Biophys Res Commun 317: 330–334. Find this article online

19. Panda S, Antoch MP, Miller BH, Su AI, Schook AB, et al. (2002) Coordinated transcription of key pathways in the mouse by the circadian clock. Cell 109: 307–320. Find this article online

20. Storch KF, Lipan O, Leykin I, Viswanathan N, Davis FC, et al. (2002) Extensive and divergent circadian gene expression in liver and heart. Nature 417: 78–83. Find this article online

21. Rudic RD, McNamara P, Reilly D, Grosser T, Curtis AM, et al. (2005) Bioinformatic analysis of circadian gene oscillation in mouse aorta. Circulation 112: 2716–2724. Find this article online

22. Wang N, Yang G, Jia Z, Zhang H, Aoyagi T, et al. (2008) Vascular PPARgamma controls circadian variation in blood pressure and heart rate through Bmal1. Cell Metab 8: 482–491. Find this article online

23. Gao Q, Horvath TL (2007) Neurobiology of feeding and energy expenditure. Annu Rev Neurosci 30: 367–398. Find this article online

24. Morton GJ, Cummings DE, Baskin DG, Barsh GS, Schwartz MW (2006) Central nervous system control of food intake and body weight. Nature 443: 289–295. Find this article online

25. Rosen ED, Spiegelman BM (2006) Adipocytes as regulators of energy balance and glucose homeostasis. Nature 444: 847–853. Find this article online

26. Canaple L, Rambaud J, Dkhissi-Benyahya O, Rayet B, Tan NS, et al. (2006) Reciprocal regulation of brain and muscle Arnt-like protein 1 and peroxisome proliferator-activated receptor alpha defines a novel positive feedback loop in the rodent liver circadian clock. Mol Endocrinol 20: 1715–1727. Find this article online

27. Doi M, Takahashi Y, Komatsu R, Yamazaki F, Yamada H, et al. (2010) Salt-sensitive hypertension in circadian clock-deficient Cry-null mice involves dysregulated adrenal Hsd3b6. Nat Med 16: 67–74. Find this article online

28. Dobrzyn P, Sampath H, Dobrzyn A, Miyazaki M, Ntambi JM (2008) Loss of stearoyl-CoA desaturase 1 inhibits fatty acid oxidation and increases glucose utilization in the heart. Am J Physiol Endocrinol Metab 294: E357–364. Find this article online

29. Durgan DJ, Trexler NA, Egbejimi O, McElfresh TA, Suk HY, et al. (2006) The circadian clock within the cardiomyocyte is essential for responsiveness of the heart to fatty acids. J Biol Chem 281: 24254–24269. Find this article online

30. Stokkan KA, Yamazaki S, Tei H, Sakaki Y, Menaker M (2001) Entrainment of the circadian clock in the liver by feeding. Science 291: 490–493. Find this article online

31. Zvonic S, Ptitsyn AA, Conrad SA, Scott LK, Floyd ZE, et al. (2006) Characterization of peripheral circadian clocks in adipose tissues. Diabetes 55: 962–970. Find this article online

32. Su W, Guo Z, Randall DC, Cassis L, Brown DR, et al. (2008) Hypertension and disrupted blood pressure circadian rhythm in type 2 diabetic db/db mice. Am J Physiol Heart Circ Physiol 295: H1634–1641. Find this article online

33. Senador D, Kanakamedala K, Irigoyen MC, Morris M, Elased KM (2009) Cardiovascular and autonomic phenotype of db/db diabetic mice. Exp Physiol 94: 648–658. Find this article online

34. Oishi K, Shirai H, Ishida N (2005) CLOCK is involved in the circadian transactivation of peroxisome-proliferator-activated receptor alpha (PPARalpha) in mice. Biochem J 386: 575–581. Find this article online

35. Inoue I, Shinoda Y, Ikeda M, Hayashi K, Kanazawa K, et al. (2005) CLOCK/BMAL1 is involved in lipid metabolism via transactivation of the peroxisome proliferator-

activated receptor (PPAR) response element. J Atheroscler Thromb 12: 169–174. Find this article online

36. Fontaine C, Dubois G, Duguay Y, Helledie T, Vu-Dac N, et al. (2003) The orphan nuclear receptor Rev-Erbalpha is a peroxisome proliferator-activated receptor (PPAR) gamma target gene and promotes PPARgamma-induced adipocyte differentiation. J Biol Chem 278: 37672–37680. Find this article online

37. Anan F, Masaki T, Fukunaga N, Teshima Y, Iwao T, et al. (2007) Pioglitazone shift circadian rhythm of blood pressure from non-dipper to dipper type in type 2 diabetes mellitus. Eur J Clin Invest 37: 709–714. Find this article online

38. Yang X, Downes M, Yu RT, Bookout AL, He W, et al. (2006) Nuclear receptor expression links the circadian clock to metabolism. Cell 126: 801–810. Find this article online

39. Elmquist JK (2001) Hypothalamic pathways underlying the endocrine, autonomic, and behavioral effects of leptin. Physiol Behav 74: 703–708. Find this article online

40. Buijs RM, Scheer FA, Kreier F, Yi C, Bos N, et al. (2006) Organization of circadian functions: interaction with the body. Prog Brain Res 153: 341–360. Find this article online

41. Guo H, Brewer JM, Champhekar A, Harris RB, Bittman EL (2005) Differential control of peripheral circadian rhythms by suprachiasmatic-dependent neural signals. Proc Natl Acad Sci U S A 102: 3111–3116. Find this article online

42. Durgan DJ, Hotze MA, Tomlin TM, Egbejimi O, Graveleau C, et al. (2005) The intrinsic circadian clock within the cardiomyocyte. Am J Physiol Heart Circ Physiol 289: H1530–1541. Find this article online

43. Reilly DF, Curtis AM, Cheng Y, Westgate EJ, Rudic RD, et al. (2008) Peripheral circadian clock rhythmicity is retained in the absence of adrenergic signaling. Arterioscler Thromb Vasc Biol 28: 121–126. Find this article online

44. Swoap SJ, Weinshenker D, Palmiter RD, Garber G (2004) Dbh(-/-) mice are hypotensive, have altered circadian rhythms, and have abnormal responses to dieting and stress. Am J Physiol Regul Integr Comp Physiol 286: R108–113. Find this article online

45. Vujovic N, Davidson AJ, Menaker M (2008) Sympathetic input modulates, but does not determine, phase of peripheral circadian oscillators. Am J Physiol Regul Integr Comp Physiol 295: R355–360. Find this article online

46. Durgan DJ, Young ME (2010) The cardiomyocyte circadian clock: emerging roles in health and disease. Circ Res 106: 647–658. Find this article online

47. Maury E, Ramsey KM, Bass J (2010) Circadian rhythms and metabolic syndrome: from experimental genetics to human disease. Circ Res 106: 447–462. Find this article online

Abstract Top

Background

The validity of Doppler echocardiographic (DE) measurement of systolic pulmonary artery pressure (sPAP) has been questioned. Recent studies suggest that mean pulmonary artery pressure (mPAP) might reflect more accurately the invasive pressures.

Methodology/Principal Findings

241 patients were prospectively studied to evaluate the diagnostic accuracy of mPAP for the diagnosis of PH. Right heart catheterization (RHC) and DE were performed in 164 patients mainly for preoperative evaluation of heart valve dysfunction. The correlation between DE and RHC was better when mPAP (r = 0.93) and not sPAP (r = 0.81) was assessed. Bland-Altman analysis revealed a smaller variation of mPAP than sPAP. The following ROC analysis identified that a mPAP≥25.5 mmHg is useful for the diagnosis of PH. This value was validated in an independent cohort of patients (n = 50) with the suspicion of chronic-thromboembolic pulmonary hypertension. The calculated diagnostic accuracy was 98%, based on excellent sensitivity of 98% and specificity of 100%. The corresponding positive and negative predictive values were 100%, respectively 88%.

Conclusion

mPAP has been found to be highly accurate for the initial diagnosis of PH. A cut-off value of 25.5 mmHg might be helpful to avoid unnecessary RHC and select patients in whom RHC might be beneficial.

Citation: Er F, Ederer S, Nia AM, Caglayan E, Dahlem KM, et al. (2010) Accuracy of Doppler-Echocardiographic Mean Pulmonary Artery Pressure for Diagnosis of Pulmonary Hypertension. PLoS ONE 5(12): e15670. doi:10.1371/journal.pone.0015670

Editor: German Malaga, Universidad Peruana Cayetano Heredia, Peru

Received: September 24, 2010; Accepted: November 19, 2010; Published: December 17, 2010

Copyright: © 2010 Er et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: The authors have no support or funding to report.

Competing interests: The authors have declared that no competing interests exist.

* E-mail: Fikret.Er@uk-koeln.de

Introduction Top

Pulmonary hypertension (PH) is associated with restricted flow through the pulmonary circulation, increased pulmonary vascular resistance and right heart failure [1]. Since the last decades PH has been identified as a devastating disease with a high mortality in dependence on the clinical classifications [2], [3]. Early diagnosis is essential to identify patients at high risk, treat the PH and modify the etiologic substrate. PH has been defined as an elevation of mean pulmonary artery pressure (mPAP)≥25 mmHg in right heart catheterization (RHC) [1], [4]. When PH is suspected in patients based on the history, risk factor assessment, and physical examination, an echocardiogram has been addressed as the next appropriate study [4]. The Doppler echocardiogram (DE) can simultaneously provide an estimate of right ventricular systolic pressure (RVSP), functional and morphologic cardiac sequelae of PH, and identification of possible cardiac causes of PH or the presented clinical symptoms. The need for further invasive diagnostics is often triggered by the DE assessment of the peak systolic PAP (sPAP). But the reported accuracy of sPAP determination by DE is controversial. While initial comparisons between DE and RHC revealed an acceptable correlation [5], [6], recent studies questioned the diagnostic value of DE in PH [7]–[9]. In the present study we aimed to determine whether echocardiographic assessment of mPAP is more accurate than sPAP for initial diagnosis of PH and estimation of real pulmonary artery pressure.

Methods Top

The study was performed in accordance with the STARD criteria to improve the quality of diagnostic accuracy [10]. The calculated diagnostic accuracy was validated in an independent cohort.

Ethics Statement

The study complies with the Declaration of Helsinki. All participants gave their written informed consent and the Ethics Committee of the University of Cologne approved the conduct of this study.

Subjects

Consecutive patients referred to the Cardiology Department of the University of Cologne were included in this prospective study from December 2008 to June 2010. All patients had a clinical indication for RHC due to heart failure, aortic or mitral valve dysfunction or the suspicion of PH. All patients in the validation group had a history of pulmonary embolism and the clinical suspicion of PH.

Right Heart Catheterization

RHC was performed without sedation at rest in the cardiac catheter laboratory of the Cardiology Department of the University of Cologne. End-expiratory pressure measurements were taken from the right atrium, right ventricle, pulmonary artery and pulmonary capillary.

Transthoracic echocardiography

Comprehensive two-dimensional echocardiography was performed in all patients within 120 minutes before right heart catheterization using a Phillips iE33 ultrasound device equipped with a standard transducer operating at 1–5 MHz without using saline contrast. The

echocardiography was performed by three different cardiologists. In each patient only one cardiologist performed the examination, randomly. Multiple views were recorded to identify optimal view for analysis as recommended in actual guidelines [11]–[13]. The right atrial pressure (RAP) was estimated by evaluating the inferior vena cava (IVC) diameter (IVCd) and change with respiration [14], [15]: when IVCd was less than 20 mm and the collapsibility greater than 50% RAP was estimated to be 5 mmHg versus 10 mmHg when the collapsibility was less than 50%. When the IVCd was greater than 20 mm RAP was estimated to be 15 mmHg when the collapsibility was greater than 50% and to be 20 mmHg when the collapsibility was less than 50%.

Additionally to the mean gradient estimation, mPAP was calculated using the Chemla formula (mPAP = 0.61×sPAP +2 mmHg) and the Syyed formula (mPAP = 0.65×sPAP +0.55 mmHg) [16]–[18].

Continuous wave Doppler was used to determine the peak velocity of the tricuspid regurgitant (TR) jet at end-expiration. Patients were excluded when TR jet was not available. The highest TR velocity was measured and traced to obtain the peak and mean systolic right-ventricular-right-atrial (RV-RA) gradient. The mean gradient was calculated by tracing the TR time-velocity integral plus RAP [19]. The sPAP was calculated using the highest RV-RA gradient plus estimated RA pressure. The mPAP was calculated as mean RV-RA pressure plus estimated RA pressure. The quality of continuous Doppler envelope was graded by a blinded cardiologist from 1 (excellent visualization with full spade shaped Doppler envelope with exactly detectable peak) to 5 (poorly visualization of Doppler signal and peak velocity).

The potential confounding factors in echocardiographic right heart assessment like right heart dimensions, the presence of atrial fibrillation or severe tricuspid regurgitation were documented, but not corrected in any direction. In these cases the measurements were performed like in all other patients.

The right ventricular function was estimated by calculating the tricuspid annular plane systolic excursion (TAPSE). Left ventricular ejection fraction (LVEF) was estimated by the Simpson's rule in the four and two chamber views.

Statistics

All variables were tested for normal distribution with the Kolmogorov-Smirnov test. Continuous variables are expressed as means ± standard deviation (SD). Comparison of 2 means was performed with the t test for normally distributed variables and the Mann-Whitney U test for non-Gaussian variables. Chi-square test was used for nonparametric comparisons. For diagnostic utility calculations receiver operating characteristic (ROC) curves were used. Results are expressed in terms of area under the curve (AUC) and 95% CI for this area. Sensitivity and specificity were estimated with ROC curves. Accuracy, positive predictive value (PPV) and negative predictive value (NPV) were calculated accordingly. Pressure comparisons were done using analysis described by Bland-Altman with predefined accuracy as 95% limits of agreement ±2xSD [20]. All statistical tests were 2-tailed, and p<0.05 was considered statistically significant. Statistical analysis was performed using SPSS 18 (SPSS GmbH Software – IBM Company, Munich, Germany).

Results Top

Baseline characteristics

191 consecutive patients with an indication for RHC were eligible for the study. 7 patients refused to participate in the study, in 9 patients echocardiography could not be performed within the predefined time schedule and 11 in patients an analyzable TR jet was not available. Data from a total of 164 patients were available for final analysis. The baseline characteristics of the study participants are displayed in Table 1. The majority of the patients underwent RHC for invasive evaluation of the aortic (n = 74) or mitral (n = 49) valve dysfunction. In 41 patients RHC was performed for heart failure assessment, in 11 of these patients for scheduling for heart transplantation.

Table 1. Characteristics of 164 patients and indications for RHC.

doi:10.1371/journal.pone.0015670.t001

Correlation of invasive versus echocardiographic sPAP, mPAP and RAP

In all patients echocardiographic evaluation of mPAP and sPAP were performed. The comparison of invasive versus echocardiographic mPAP revealed a better correlation than the sPAP in RHC versus echocardiography (Figure 1). The correlation coefficient for DE and invasive sPAP was 0.81 (p<0.001) compared to 0.93 for mPAP in DE versus RHC (p<0.001). Both, echocardiographic sPAP and mPAP calculations were influenced by the estimated RAP. The correlation of DE and invasive RAP was weaker than seen for mPAP and sPAP (r = 0.67; p<0.001). Better Doppler signal quality improved the documented correlations (Figure 1).

Figure 1. DE versus RHC correlations.

DE mPAP (B) was better correlated with RHC than sPAP (A). Dotted lines mark virtual best correlation of 1 and solid lines mark the real correlation. r indicates the correlation coefficient, sPAP indicates systolic pulmonary artery pressure.

doi:10.1371/journal.pone.0015670.g001

Using Bland-Altman analysis, the bias for the echocardiographic estimates of sPAP was -4.1 mmHg with 95% limits of agreement ranging from +23 mmHg to −29 mmHg (Figure 2A). In contrast bias for the mPAP measurements was 0.3 mmHg with 95% limits of agreement ranging from +12 to −12 mmHg (Figure 2B). Absolute values of mPAP lower than sPAP values, which could explain smaller variations of mPAP. To rule out this fact the relative variation of mPAP and sPAP in DE vs. RHC were calculated. But even the relative differences between DE and RHC were larger for sPAP than mPAP indicating a better correlation of DE mPAP and RHC vs. DE sPAP and RHC (Figure 3A).

Figure 2. Bland-Altman plot of DE estimates of PA and RHC pressures for sPAP (A) and mPAP (B).

Smaller bias and limits of agreement present in mPAP measurements compared to sPAP measurements.

doi:10.1371/journal.pone.0015670.g002

Figure 3. Analysis of relative differences and ROC analysis of mPAP for diagnosis of PH.

A, the relative positive and negative deviation between DE and RHC were larger for sPAP than mPAP. B, ROC analysis reveal an excellent diagnostic accuracy of

mPAP for the diagnosis of PH with an area under the curve (AUC) of 0.95.

doi:10.1371/journal.pone.0015670.g003

Correlation and Bland-Altman analysis identified an excellent correlation of echocardiographic mPAP and RHC between 20 and 40 mmHg (Figure 1). We identified this range to be important for the diagnosis of PH. Therefore using the ROC analysis we tested the usefulness of echocardiographic mPAP for the diagnosis of PH. For this calculation PH was defined as invasive mPAP≥25 mmHg [21]. The ROC analysis revealed a useful sensitivity and specificity of mPAP for diagnosis of PH with DE, reflected by an area under curve (AUC) of 0.95 (95 CI; 0.914–0.983; p<0.001; Figure 3B). A cut-off value of mPAP≥25.5 mmHg could detect PH with a sensitivity of 92% and a specificity of 84%. The corresponding likelihood ratios (LR) were 5.76 for positive LR and 0.18 for negative LR (Table 2).

Table 2. Diagnostic value of different mPAP cut-offs for diagnosis of PH.

doi:10.1371/journal.pone.0015670.t002

Validation of mPAP for the Diagnosis of PH

The diagnostic accuracy of mPAP for the detection of patients with PH was tested in an independent cohort of patients. 50 consecutive patients who were referred to the Cardiology Department for invasive RHC with a history of pulmonary embolism and the clinical suspicion of chronic-thromboembolic PH were included. The demographics and invasive vs. DE pressures are displayed in table 3. There were no exclusions. Using the DE cut-off value of 25.5 mmHg 42 patients were identified to have a PH (Figure 4). RHC confirmed in all 42 patients the diagnosis of PH. In 8 patients DE suggested that a PH could be excluded. In one of these patients the result was false negative. The calculated diagnostic accuracy was 98%, based on excellent sensitivity of 98% and specificity of 100%. The corresponding positive and negative predictive values were 100% and 88%.

Figure 4. Study flowchart in accordance to the STARD criteria.

In phase 0 diagnostic criteria were evaluated in 164 patients undergoing DE and RHC. In phase 1 the calculated cut-off value for mPAP was validated in a cohort of patients with the suspicion of PH.

doi:10.1371/journal.pone.0015670.g004

Table 3. Demographics, RHC and DE measurements of the validation group (n = 50).

doi:10.1371/journal.pone.0015670.t003

Discussion Top

We examined the diagnostic accuracy of mPAP for the diagnosis of PH. We applied DE assessment in a large number of patients undergoing RHC for several reasons. Comparison analysis revealed that DE mPAP reflects more precisely the invasive pressures than DE sPAP does. Based on this observation the performed ROC analysis displayed that mPAP is useful for diagnosis of PH. The accuracy of mPAP for PH was validated in an independent high-risk cohort of patients with the suspicion of PH. An excellent sensitivity of 98%, specificity of 100% and accuracy of 98% of this diagnostic tool could be confirmed. Only in one of forty-three patients with borderline PH DE was false negative and in none of the patients false positive.

Despite of promising therapeutic options emerged in the last decades, mortality remains high among patients with PH [22], [23]. Early diagnosis of PH may change its detrimental character and its annual mortality rate of 15% [24]. But still the diagnosis of PH is challenging. The invasive assessment of the right heart and pulmonary arteries has been established as the gold standard in the diagnosis of PH [21]. It helps to differentiate pre- and postcapillary PH. Due to its invasive character RHC is not useful as a routine screening method. Echocardiography is a widely available and accepted noninvasive diagnostic instrument for assessment of right heart function and dimension [13], [25]–[29]. Initial studies reported a good correlation between DE and RHC [5], [6].

But recent DE studies questioned the diagnostic value of DE for PH assessment displaying a large variation of sPAP [7]–[9]. In a well-designed study Fisher and Colleagues assessed the usability of DE for evaluation of sPAP and found despite a good correlation a wide and inacceptable discrepancies between DE and RHC in Bland-Altman analysis [7]. Our results are in agreement with previous findings. Therefore DE sPAP may not useful for the diagnosis and estimation of pulmonary artery pressure.

It is of clinical interest to identify patients with PH early and in a feasible and economic way. We identified DE as a helpful tool for the diagnosis of PH when mPAP and not sPAP is used. Our results concerning the correlation of mPAP in DE vs. RHC are supported by recent studies [30], [31]. Aduen and coworkers identified echocardiographic mPAP as a valuable parameter for the assessment of pulmonary artery pressure and confirmed that the method used in our study of mean gradient estimation is equally applicable to the Chemla formula and the Syyed formula [16]–[18], [30], [31].

In 43 of 50 patients with the suspicion of PH the diagnosis of PH could be confirmed. Of these patients almost one third had mild, one third moderate and one third severe PH. Hence, echocardiographic mPAP may be suitable in a wide range of PH severity. Especially the detection of PH in mildly symptomatic or mildly elevated mPAP is crucial. Echocardiographic mPAP may be a helpful screening tool in these patients.

In conclusion we identified the echocardiographic mPAP measurement as accurate for diagnosis of PH. mPAP measurement may prevent unnecessary RHC and identify patients at high risk for PH.

Study limitations

This study is limited by its focus on a single center's experience and the limited sample size of the validation group. In almost 6% of the patients a TR jet was not analyzable. Saline contrast may increase the rate on available TR jet signals. The estimation of RAP is generally challenging as in our study. But despite the large variation in RAP the calculated accuracy of mPAP was excellent, suggesting that TR signal alone reflects accurately the PAP. RHC and DE were not performed simultaneously. While during Swan-Ganz catheterization simultaneous measurements may be suitable, it's a technical challenge performing echocardiography in the catheter lab during RHC.

Author Contributions Top

Conceived and designed the experiments: FE. Performed the experiments: SE AN NG EC KMD. Analyzed the data: SE AN NS FE. Contributed reagents/materials/analysis tools: SE AN NG EC NS. Wrote the paper: FE.

References Top

1. Galie N, Hoeper MM, Humbert M, Torbicki A, Vachiery JL, et al. (2009) Guidelines for the diagnosis and treatment of pulmonary hypertension: The Task Force for the Diagnosis and Treatment of Pulmonary Hypertension of the European Society of Cardiology (ESC) and the European Respiratory Society (ERS), endorsed by the International Society of Heart and Lung Transplantation (ISHLT). Eur Heart J 30: 2493–2537. Find this article online

2. Benza RL, Miller DP, Gomberg-Maitland M, Frantz RP, Foreman AJ, et al. (2010) Predicting survival in pulmonary arterial hypertension: insights from the Registry to Evaluate Early and Long-Term Pulmonary Arterial Hypertension Disease Management (REVEAL). Circulation 122: 164–172. Find this article online

3. Simonneau G, Robbins IM, Beghetti M, Channick RN, Delcroix M, et al. (2009) Updated clinical classification of pulmonary hypertension. J Am Coll Cardiol 54: S43–54. Find this article online

4. McLaughlin VV, Archer SL, Badesch DB, Barst RJ, Farber HW, et al. (2009) ACCF/AHA 2009 expert consensus document on pulmonary hypertension a report of the American College of Cardiology Foundation Task Force on Expert Consensus Documents and the American Heart Association developed in collaboration with the American College of Chest Physicians; American Thoracic Society, Inc.; and the Pulmonary Hypertension Association. J Am Coll Cardiol 53: 1573–1619. Find this article online

5. Berger M, Haimowitz A, Van Tosh A, Berdoff RL, Goldberg E (1985) Quantitative assessment of pulmonary hypertension in patients with tricuspid regurgitation using continuous wave Doppler ultrasound. J Am Coll Cardiol 6: 359–365. Find this article online

6. Currie PJ, Seward JB, Chan KL, Fyfe DA, Hagler DJ, et al. (1985) Continuous wave Doppler determination of right ventricular pressure: a simultaneous Doppler-catheterization study in 127 patients. J Am Coll Cardiol 6: 750–756. Find this article online

7. Fisher MR, Forfia PR, Chamera E, Housten-Harris T, Champion HC, et al. (2009) Accuracy of Doppler echocardiography in the hemodynamic assessment of pulmonary hypertension. Am J Respir Crit Care Med 179: 615–621. Find this article online

8. Fisher MR, Criner GJ, Fishman AP, Hassoun PM, Minai OA, et al. (2007) Estimating pulmonary artery pressures by echocardiography in patients with emphysema. Eur Respir J 30: 914–921. Find this article online

9. Arcasoy SM, Christie JD, Ferrari VA, Sutton MS, Zisman DA, et al. (2003) Echocardiographic assessment of pulmonary hypertension in patients with advanced lung disease. Am J Respir Crit Care Med 167: 735–740. Find this article online

10. Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, et al. (2003) The STARD statement for reporting studies of diagnostic accuracy: explanation and elaboration. Ann Intern Med 138: W1–12. Find this article online

11. Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, et al. (2005) Recommendations for chamber quantification: a report from the American Society of Echocardiography's Guidelines and Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr 18: 1440–1463. Find this article online

12. Quinones MA, Otto CM, Stoddard M, Waggoner A, Zoghbi WA (2002) Recommendations for quantification of Doppler echocardiography: a report from the Doppler Quantification Task Force of the Nomenclature and Standards Committee of the American Society of Echocardiography. J Am Soc Echocardiogr 15: 167–184. Find this article online

13. Rudski LG, Lai WW, Afilalo J, Hua L, Handschumacher MD, et al. (2010) Guidelines for the echocardiographic assessment of the right heart in adults: a report from the American Society of Echocardiography endorsed by the European Association of Echocardiography, a registered branch of the European Society of Cardiology, and the Canadian Society of Echocardiography. J Am Soc Echocardiogr 23: 685–713. quiz 786-688. Find this article online

14. Moreno FL, Hagan AD, Holmen JR, Pryor TA, Strickland RD, et al. (1984) Evaluation of size and dynamics of the inferior vena cava as an index of right-sided cardiac function. Am J Cardiol 53: 579–585. Find this article online

15. Brennan JM, Blair JE, Goonewardena S, Ronan A, Shah D, et al. (2007) Reappraisal of the use of inferior vena cava for estimating right atrial pressure. J Am Soc Echocardiogr 20: 857–861. Find this article online

16. Chemla D, Castelain V, Humbert M, Hebert JL, Simonneau G, et al. (2004) New formula for predicting mean pulmonary artery pressure using systolic pulmonary artery pressure. Chest 126: 1313–1317. Find this article online

17. Chemla D, Castelain V, Provencher S, Humbert M, Simonneau G, et al. (2009) Evaluation of various empirical formulas for estimating mean pulmonary artery

pressure by using systolic pulmonary artery pressure in adults. Chest 135: 760–768. Find this article online

18. Syyed R, Reeves JT, Welsh D, Raeside D, Johnson MK, et al. (2008) The relationship between the components of pulmonary artery pressure remains constant under all conditions in both health and disease. Chest 133: 633–639. Find this article online

19. Abbas AE, Fortuin FD, Schiller NB, Appleton CP, Moreno CA, et al. (2003) Echocardiographic determination of mean pulmonary artery pressure. Am J Cardiol 92: 1373–1376. Find this article online

20. Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307–310. Find this article online

21. Galie N, Hoeper MM, Humbert M, Torbicki A, Vachiery JL, et al. (2009) Guidelines for the diagnosis and treatment of pulmonary hypertension. Eur Respir J 34: 1219–1263. Find this article online

22. Archer SL, Weir EK, Wilkins MR (2010) Basic science of pulmonary arterial hypertension for clinicians: new concepts and experimental therapies. Circulation 121: 2045–2066. Find this article online

23. Thenappan T, Shah SJ, Rich S, Tian L, Archer SL, et al. (2010) Survival in pulmonary arterial hypertension: a reappraisal of the NIH risk stratification equation. Eur Respir J 35: 1079–1087. Find this article online

24. Thenappan T, Shah SJ, Rich S, Gomberg-Maitland M (2007) A USA-based registry for pulmonary arterial hypertension: 1982-2006. Eur Respir J 30: 1103–1110. Find this article online

25. Lai WW, Gauvreau K, Rivera ES, Saleeb S, Powell AJ, et al. (2008) Accuracy of guideline recommendations for two-dimensional quantification of the right ventricle by echocardiography. Int J Cardiovasc Imaging 24: 691–698. Find this article online

26. Lindqvist P, Calcutteea A, Henein M (2008) Echocardiography in the assessment of right heart function. Eur J Echocardiogr 9: 225–234. Find this article online

27. Arora H, Virani SS, Simpson L, Stainback RF (2008) Contrast echocardiography for right-sided heart conditions: case reports and literature review. Tex Heart Inst J 35: 38–41. Find this article online

28. Swanson KL, Utz JP, Krowka MJ (2008) Doppler echocardiography-right heart catheterization relationships in patients with idiopathic pulmonary fibrosis and suspected pulmonary hypertension. Med Sci Monit 14: CR177–182. Find this article online

29. D'Cruz IA (1998) The Role of Echocardiography in Right Heart Failure. Echocardiography 15: 759–760. Find this article online

30. Aduen JF, Castello R, Daniels JT, Diaz JA, Safford RE, et al. (2010) Accuracy and Precision of 3 Echocardiographic Methods for Estimating Mean Pulmonary Artery Pressure. Chest. Find this article online

31. Aduen JF, Castello R, Lozano MM, Hepler GN, Keller CA, et al. (2009) An alternative echocardiographic method to estimate mean pulmonary artery pressure: diagnostic and clinical implications. J Am Soc Echocardiogr 22: 814–819. Find this article online

Abstract Top

Background

Working mostly at night has been suggested to be associated with upset of chronobiological rhythms and high blood pressure, but the evidence from epidemiological studies is weak.

Methods

In a cross-sectional survey, we evaluated the association between shift work and blood pressure, pre-hypertension and hypertension. In total, 493 nurses, nurse technicians and assistants, were selected at random in a large general hospital setting. Hypertension was diagnosed by the mean of four automatic blood pressure readings ≥140/90 mmHg or use of blood pressure lowering agents, and pre-hypertension by systolic blood pressure ≥120–139 or diastolic blood pressure ≥80–89 mmHg. Risk factors for hypertension were evaluated by a standardized questionnaire and anthropometric measurements. The association between the shift of work and blood pressure, pre-hypertension and hypertension was explored using univariate and multivariate analyses that controlled for risk factors for hypertension by covariance analysis and modified Poisson regression.

Results

The mean age of the participants was 34.3±9.4 years and 88.2% were women. Night shift workers were older, more frequently married or divorced, and less educated. The prevalence of hypertension in the whole sample was 16%, and 28% had pre-hypertension. Blood pressure (after adjustment for confounding) was not different in day and night shift workers. The prevalence of hypertension and pre-hypertension by shift work was not different in the univariate analysis and after adjustment for confounding (all risk ratios = 1.0).

Conclusion

Night shift work did not increase blood pressure and was not associated with hypertension or pre-hypertension in nursing personnel working in a large general hospital.

Citation: Sfreddo C, Fuchs SC, Merlo ÁR, Fuchs FD (2010) Shift Work Is Not Associated with High Blood Pressure or Prevalence of Hypertension. PLoS ONE 5(12): e15250. doi:10.1371/journal.pone.0015250

Editor: James M. Wright, University of British Columbia, Canada

Received: August 24, 2010; Accepted: November 8, 2010; Published: December 14, 2010

Copyright: © 2010 Sfreddo et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This study was supported by the National Institute of Science and Technology for Health Technology Assessment (IATS) - CNPq/Brazil and Fundo de Incentivo a Pesquisa (FIPE-HCPA), Brazil. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

* E-mail: ffuchs@hcpa.ufrgs.br

Introduction Top

Working mostly at night has been suggested to be associated with deleterious consequences of general health, as a consequence of disturbance of chronobiological rhythms [1]. Disturbances of the circadian sleep rhythm could prevent the dipping pattern of blood pressure [2], [3], and increase the incidence of hypertension. Studies with short duration have found higher ambulatory blood pressure among shift workers [4]. Shorter periods of sleep, which have been described for shift workers, could also lead to higher blood pressure [5], [6].

Despite the biological plausibility and evidence from short term studies, there is very few evidence that acute changes in blood pressure regulation in shift workers lead to chronic hypertension. Epidemiological studies linking shift work to the risk for hypertension are few, old and of questionable quality. Two recent and better designed studies found opposite effects. In a Japanese cohort study, habitual shift workers of a steel company had higher systolic and diastolic blood pressure than habitual daily workers [7]. In contrast the Finnish Twin Cohort [8], showed no association between the usual period of work and incidence of hypertension.

The aim of the present study was to investigate the association between shift work and the prevalence of hypertension and pre-hypertension in the nursing staff of a general hospital.

Methods Top

Ethical statement

The Review Board of the Hospital São Vicente de Paulo from the University of Passo Fundo approved the investigation, and all participants gave their written consent to participate.

Study design

Nursing personnel (nurses, nurse technicians and nurse assistants) of a large general hospital (Hospital São Vicente de Paulo), Passo Fundo, Brazil, were selected at random during the periodic medical examination between January 2008 and August 2009. A total of 493 individuals, predominantly women (88.2% vs. 11.8%), aged from 19 to 68 years agreed to participate.

A standardized questionnaire was used to investigate aspects pertaining to shift working and other risk factors for hypertension. Certified investigators interviewed the participants at the place of work, using a manual with instructions for every section of the questionnaire and for blood pressure and anthropometric measurements. Ten percent of the interviews were repeated at random for quality assurance, and the data were entered in duplicate.

Study variables

Demographic (age, self-reported skin color), socioeconomic (years at school, previous employments), and lifestyle (physical activities, smoking and alcoholic beverages consumption) characteristics were recorded. Individuals were categorized as current smokers or non-smokers (never smokers or ex-smokers). The pattern of alcohol consumption in the previous year was investigated by a type/frequency questionnaire. The risk of alcoholic beverages consumption was defined by a binge drinking pattern, i.e., 5 or more glasses of drinking in one occasion. Skin color was categorized as white or non-white. Years at school were categorized as more or less than 8 years. Sleeping hours corresponded to the hours of sleep during 24 h in a typical week.

Shifts of work were characterized as morning, afternoon, and night and the combination of them. Since just a few worked exclusively afternoon shifts, this category was combined with the morning shift.

Blood pressure was measured four times during the interview, using an automatic device (OMRON CP-705), and the average was used in the analyses. Hypertension was defined by systolic blood pressure ≥140 or diastolic blood pressure ≥90 mmHg, or use of blood pressure-lowering drugs. Pre-hypertension was defined by systolic blood pressure ≥120–139 or diastolic blood pressure ≥80–89 mmHg.

Weight was measured within 100 g intervals and height within 0.1 cm using a scale with a stadiometer (model 31, Indústria Filizola AS, São Paulo, Brazil), with the participants wearing light clothes and without shoes. Body mass index (BMI) was calculated by dividing weight by height squared.

Sample size calculation and statistical analyses

The sample size calculation was done with the Epi-Info software, version 6.04. Assuming a prevalence of hypertension of 14% in daily workers and an increase to 25% in workers in the night shift, a power of 80% and confidence limits of 95%, the sample size was estimated as 486 to reject the null hypothesis. Proportions were compared by Chi-square test and means by the Student t-Test for independent samples. Multivariate analyses were carried out using co-variance analyses and Modified Poisson regression.

Results Top

Among 2,140 employees of the hospital, 862 (23.1%) were nurses, nurse technicians or nurse assistants. From these, 493 (57.1%) were selected at random, and none refused to participate. The prevalence of hypertension was 16%, and 28% of the participants had pre-hypertension. Table 1 shows that the participants were relatively young, mostly women, predominantly white, and with an average number of years of education corresponding to high school education. Workers in the night shift were older, less frequently single, and slept more hours per day. Diastolic blood pressure was higher in the night shift workers, but this difference disappeared after adjustment for age, BMI and years at school (P = 0.4). Figure 1 shows that the distribution of hypertension and pre-hypertension by shift work was not different.

Figure 1. Proportion of the nursing personnel with pre-hypertension and hypertension by shift of work (P = 0.8).

doi:10.1371/journal.pone.0015250.g001

Table 1. Characteristics of the nursing personnel by shift of work [N (%) or mean ±SD].

doi:10.1371/journal.pone.0015250.t001

Table 2 presents the distribution of age, BMI, systolic and diastolic blood pressure, and the classification of blood pressure by the combination of shifts of work. Age and BMI were different by shift of work, without any consistent association with the night shift. Blood pressure and the proportion of participants with hypertension and pre-hypertension were not associated with the shift of work.

Table 2. The association between MAEN (morning, afternoon, evening, night) working hours with age, BMI and blood pressure.

doi:10.1371/journal.pone.0015250.t002

Table 3 presents the adjusted risk ratios for hypertension or pre-hypertension by selected categories and by shift of work characterized as day or night or by the combination of different shifts. Age, male gender and BMI were positively associated with hypertension, while higher number of sleeping hours was associated with lower prevalence of hypertension and pre-hypertension. Shifts of work, either classified as day or night, or by the combination of shifts, were not associated with the prevalence of hypertension or pre-hypertension. The exclusion of sleeping hours from the model did not change the estimates at all. The risk ratios for hypertension alone were similar but had wider confidence limits.

Table 3. Risk factors for pre-hypertension or hypertension analyzed by Modified Poisson Regression (Risk ratio and 95% CI).

doi:10.1371/journal.pone.0015250.t003

Discussion Top

This cross-sectional survey of a representative sample of the nursing staff of a large general hospital demonstrated that the shift of work was not associated with the magnitude of blood pressure or prevalence of hypertension. The absence of association was confirmed by multivariate analyses and was not influenced by longer periods of total sleep during a typical work week for the nurses that worked the night shift. Traditional risk factors, such as age and BMI, were associated with a higher prevalence of hypertension.

Short term studies with small samples have identified a trend towards higher blood pressure among shift workers [3], [4]. There are few epidemiological surveys with large samples. Two recent cohort studies report opposing findings. Habitual shift workers of a Japanese steel company had higher systolic and diastolic blood pressure than habitual daily workers [7]. However, in the Finnish Twin Cohort, which is a large population-based study with long follow-up, there was no association between the usual period of work and incidence of hypertension [8]. The kind of night shift work and secondary post-hoc analyses of both studies may explain the different findings. Our study was prospectively designed, and had a detailed evaluation of shift of work and blood pressure. The negative findings strengthen the conclusion that work at night, at least in some jobs, such as nursing, is not associated with an increase in blood pressure, hypertension or pre-hypertension. An association between shift work and shorter period of sleeping was expected, but was not confirmed in our survey, since nurses that regularly worked at night reported a longer weekly duration of sleep. Longer periods of sleep could be an intermediate mechanism diminishing the incidence of hypertension in our study, but the absence of an association both in the bivariate and multivariate analyses is against that possible explanation of our findings.

The absence of blood pressure measurements during sleep with an Ambulatory Blood Pressure machine and the lack of an objective evaluation of sleep disorders are possible limitations of our study. The cross-sectional design is not the strongest design to establish the risks of shift work for chronic hypertension, but the absence of any cross-sectional association makes it unlikely that we missed an increase in blood pressure that would develop over time. The fact that the study was limited to nurses makes it impossible to exclude the possibility that hypertension could be linked to shift work in other professions. The prospective planning, adequate power, careful measurement of blood pressure, and evaluation of risk factors for hypertension are strengths of our survey.

In conclusion, night shift work was not associated with an increase of blood pressure, hypertension or pre-hypertension in the nursing staff of a general hospital.

Author Contributions Top

Conceived and designed the experiments: CS SCF ÁRM FDF. Performed the experiments: CS SCF FDF. Analyzed the data: CS SCF ÁRM FDF. Contributed reagents/materials/analysis tools: SCF FDF. Wrote the paper: CS SCF ÁRM FDF.

References Top

1. Barger LK, Lockley SW, Rajaratnam SM, Landrigan CP (2009) Neurobehavioral, health, and safety consequences associated with shift work in safety-sensitive professions. Curr Neurol Neurosci Rep 9: 155–164. Find this article online

2. Yamasaki F, Schwartz JE, Gerber LM, Warren K, Pickering TG (1998) Impact of shift work and race/ethnicity on the diurnal rhythm of blood pressure and catecholamines. Hypertension 32: 417–423. Find this article online

3. Lo SH, Liau CS, Hwang JS, Wang JD (2008) Dynamic blood pressure changes and recovery under different work shifts in young women. Am J Hypertens 21: 759–764. Find this article online

4. Su TC, Lin LY, Baker D, Schnall PL, Chen MF, et al. (2008) Elevated blood pressure, decreased heart rate variability and incomplete blood pressure recovery after a 12-hour night shift work. J Occup Health 50: 380–386. Find this article online

5. Stranges S, Dorn JM, Cappuccio FP, Donahue RP, Rafalson LB, et al. (2010) A population-based study of reduced sleep duration and hypertension: the strongest association may be in premenopausal women. J Hypertens 28: 896–902. Find this article online

6. Pickering TG (2006) Could hypertension be a consequence of the 24/7 society? The effects of sleep deprivation and shift work. J Clin Hypertens 8: 819–822. Find this article online

7. Suwazono Y, Dochi M, Sakata K, Okubo Y, Oishi M, et al. (2008) Shift work is a risk factor for increased blood pressure in Japanese men: a 14-year historical cohort study. Hypertension 52: 581–586. Find this article online

8. Hublin C, Partinen M, Koskenvuo K, Silventoinen K, Koskenvuo M, et al. (2010) Shift-work and cardiovascular disease: a population-based 22-year follow-up study. Eur J Epidemiol 25: 315–323. Find this article online

Abstract Top

Background

A link between early mismatched nutritional environment and development of components of the metabolic syndrome later in life has been shown in epidemiological and animal data. The aim of this study was to investigate whether an early mismatched nutrition produced by catch-up growth after fetal protein restriction could induce the appearance of hypertension and/or atherosclerosis in adult male mice.

Methodology/Principal Findings

Wild-type C57BL6/J or LDLr−/− dams were fed a low protein (LP) or a control (C) diet during gestation. Catch-up growth was induced in LP offspring by feeding dams with a control diet and by culling the litter to 4 pups against 8 in controls. At weaning, male mice were fed either standard chow or an obesogenic diet (OB), leading to 4 experimental groups. Blood pressure (BP) and heart rate (HR) were assessed in conscious unrestrained wild-type mice by telemetry. Atherosclerosis plaque area was measured in aortic root sections of LDLr−/− mice. We found that: (1) postnatal OB diet increased significantly BP (P<0.0001) and HR (P<0.008) in 3-month old OB-C and OB-LP offspring, respectively; (2) that maternal LP diet induced a significant higher BP (P<0.009) and HR (P<0.004) and (3) an altered circadian rhythm in addition to higher plasma corticosterone concentration in 9 months-old LP offspring; (4) that, although LP offspring showed higher plasma total cholesterol than control offspring, atherosclerosis assessed in aortic roots of 6-mo old mice featured increased plaque area due to OB feeding but not due to early mismatched nutrition.

Conclusions/Significance

These results indicate a long-term effect of early mismatched nutrition on the appearance of hypertension independently of obesity, while no effect on atherosclerosis was noticed at this age.

Citation: Bol V, Desjardins F, Reusens B, Balligand J-L, Remacle C (2010) Does Early Mismatched Nutrition Predispose to Hypertension and Atherosclerosis, in Male Mice? PLoS ONE 5(9): e12656. doi:10.1371/journal.pone.0012656

Editor: Carlo Polidori, University of Camerino, Italy

Received: April 15, 2010; Accepted: July 21, 2010; Published: September 9, 2010

Copyright: © 2010 Bol et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This work was supported by the European Commission (FOOD-CT-2005-007036), the Partenon Trust (London, UK), the Belgian Fonds National de la Recherche Scientifique (FNRS), Politique Scientifique Federale (PAI-P6/30) and the Belgian Fonds pour la

Recherche dans l'Industrie et l'Agriculture (FRIA). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

* E-mail: claude.remacle@uclouvain.be

Introduction Top

Hypertension is estimated to affect 25% of the world adult population. It contributes to death from cardiovascular disease which remains a major cause of morbidity in the world [1] Hypertension coexists with other pathologies that are part of the metabolic syndrome. Although genetic and environmental factors are thought to play a role in these diseases, convincing evidence from human studies indicated that a suboptimal intrauterine environment, resulting in impaired fetal growth, is associated with an increased risk of developing insulin resistance, obesity, hypertension and cardiovascular disease in adult life. Seminal observations made by Barker and colleagues showed an increased death rate from ischemic heart disease, impaired glucose tolerance and type-2 diabetes in men with low birth weight [2], [3]. These observations gave rise to the “thrifty phenotype hypothesis” where the authors postulated that early malnutrition induces physiological and metabolic adaptations promoting early survival, which may however become deleterious for later health [4]. Since then, several animal studies have supported the concept that maternal malnutrition during pregnancy may exert long-term changes in the offspring [5], [6]. Recently, it has been suggested that the long-term effects of these adaptations to a suboptimal intrauterine environment might be more detrimental if there was a mismatch between individual's adaptations to the environment predicted to be experienced and the postnatal environment encountered. The predictive adaptive response (PAR) hypothesis was then formulated [7]. An example is provided by the comparison of the epidemiological data collected from two famines, in the North of Holland [8] and in Leningrad [9], [10]. Adult men that were submitted in utero to the Dutch famine in early pregnancy revealed more coronary heart disease, raised blood lipids, altered clotting and higher obesity compared to those who were not exposed, but such difference did not appear in the progeny analyzed in the Leningrad study. Whereas the Leningrad Siege lasted from 1941 to 1944, the Dutch famine lasted only for 6 months in a previously and subsequently well nourished population. This discrepancy between the two famines suggested that not only famine exposure, but also the rapid transition from suboptimal nutrition during pregnancy to adequate nutrition later on, could increase the risk for diseases in later life. Moreover, several studies indicated that rapid infant growth in humans is associated to an increased risk of presenting several components of the metabolic syndrome at adulthood [11] including hypertension [12].

Since then, new animal models of early mismatched environment have been established [13]–[15]. A recent publication by Cleal et al. [16] indicated that adult male sheep that had been underfed only during gestation and normally fed thereafter had altered cardiovascular function. The same was noted for sheep from the group of postnatal undernutrition alone but no alteration was observed in sheep that were constantly malnourished during gestation and lactation. Consequently, catch-up growth that was previously considered a process of recovery from detrimental effects of poor growth is now viewed as a risk factor for later health.

Boubred et al, 2009 [15] investigated in rat the effect of catch-up growth after protein restriction during gestation (LP) and found a lower nephron number in the LP offspring with or without catch-up growth that was accompanied by a transient hypertension in young age. They claimed that reduced nephron number associated with intrauterine growth retardation may not be sufficient to induce long-lasting hypertension and that a second hit such as early postnatal overfeeding is necessary. In their study, an exaggerated increase in visceral fat deposit was observed in the LP overfed group compared to the control overfed group. Previous work carried out in our laboratory also showed that adult male rats featuring a catch-up growth after fetal growth retardation induced by maternal protein restriction increased their susceptibility for obesity in adulthood [13], [14]. It is well known that obesity, together with a central pattern of fat distribution, represents a significant risk factor for hypertension and cardiovascular disease such as atherosclerosis. Indeed, adipose tissue is able to secrete a number of “adipokines” [17] some of them, like resistin, leptin, adiponectin, Monocyte Chemoattractant Protein-1, Interleukin-6, angiotensinogen, angiotensin-converting enzyme, plasminogen activator inhibitor-1 (PAI-1), being implicated in cardiovascular disease [rev. in 18, 19]. We have previously shown an increased expression of leptin, PAI-1 and angiotensinogen mRNA in offspring of dams underfed during gestation but only when catch-up growth had occurred [13], [14].

The aim of this study was to investigate whether a mismatch in early nutrition caused by catch-up growth after fetal protein restriction could induce hypertension and/or atherosclerosis in adult life and to assess if their development was obesity-dependent or not. As atherosclerosis does not develop spontaneously in rodents, several transgenic mouse models exist that allow study of atherogenesis when fed western-type diet [20]. We used LDL receptor deficient mouse (LDLr−/−), a well-characterized model.

Using these two types of experiment, we were able to show a long-term programming of early mismatched nutrition on the development of hypertension which is however independent of obesity, but we could not find an effect on atherogenesis.

Methods Top

Animals and experimental diets

To be able to reach the aim of this paper, 2 strains of mice with the same diet protocol were used. The C57BL6/J mouse strain (Janvier, Le Genest Saint Isle, France) was used for analyzing hypertension (Exp 1). Because normal mice do not spontaneously develop atherosclerosis, its programming was investigated in a strain which spontaneously develops such pathology, namely homozygous LDL-receptor knockout (LDLr−/−) mice (Jackson Laboratory, Bar Harbor, ME, USA) (Exp 2). Both strains were bred in our laboratory and maintained under controlled conditions (22°C, 12 h:12 h dark-light cycle). Non-primiparous females from both strains were housed with males overnight and mating was confirmed by the presence of a vaginal plug. The pregnant dams were housed individually during gestation and lactation. During gestation, dams of both strains were fed a control diet (C: 192 g protein/kg- 20% of energy) or an isocaloric low protein diet (LP: 81 g protein/kg- 8% of energy) purchased from Hope Farm (Woerden, Netherlands Hary block). The composition and source of diets were described elsewhere [21]. At birth, pups were weighed and litter size was recorded. In order to induce a catch-up growth, the number of pups was reduced to four in LP groups instead of eight in C litters. Moreover, during lactation, C and LP dams from both strains were fed with control diet. At weaning (4 weeks), in order to minimize the litter

effect, only 2 male mice per litter were included in the study. For each litter, one male was fed standard lab chow (Carfil Quality, Turnhout, Belgium) and the other one was fed an obesogenic (OB) diet containing 16% sucrose, 18% lard, 27% casein and 23% rice starch (w/w); energy 4.56 kcal/g (Special Diet Services, Witham, UK). For each C57BL6/J cohort, half was analyzed at 3 months of age while the other half was followed until 9 months. The 2 (LDLr−/−) cohorts were analyzed at 6 months.

Body weight and food intake were recorded weekly after weaning in the C and LP groups of the 2 strains. At the end of experiments, mice received a lethal injection of Nembutal® (Natrii pentobarbitalum, 60 mg/kg bodyweight, CEVA, Brussels, Belgium). Different organs were dissected, weighed and stored at −80°C until analysis.

All LDLr−/− offspring were genotyped by PCR technique [22]. All procedures were carried out in accordance with “Principles of laboratory animal care” (NIH publication no. 85–23, revised 1985) and with the approval of the animal ethics committee of the Université catholique de Louvain, Belgium, Approval code 052803, to the laboratory LA1220028.

(Exp 1) Circadian variation and frequency analysis of blood pressure and heart rate by implanted telemetry

Wild-type C57BL6/J male offspring were implanted with a telemetry device at 3 or 9 months of age because we aimed to verify if the early malnutrition followed by a rapid catch-up growth will precipitate the appearance of hypertension. Blood pressure catheters were surgically inserted in the carotid artery and miniaturized implants (TA11PA-C10, Data Science International Inc) placed in a subcutaneous pouch under anesthesia (ketamine/xylasine mixture). Recovery from the surgical procedure was estimated by body weight regain after 10 days. The mortality rates observed after telemetric implantation were <2%. Short- or long-term (24 hours) recordings were digitized (range 20 to 2000 Hz) and stored for further analysis. Short-term recordings were always performed between 10 to 12 a.m. In order to evaluate the blood vessels reactivity, the acute effect of L-NAME (30 mg/kg), an eNOS inhibitor, that induces an increase in BP by inhibition of NO production and activation of sympathetic activity and phenylephrine (100 µg/kg), a vasopressor alpha-1 adrenergic receptor agonist that also increases blood pressure were analyzed from short-term recordings after drug intra-peritoneal injection. Delta values of SBP and HR were calculated by subtracting the mean values of the 5 time-point after to the mean values of the 5 time-point before injection.

(Exp 2) Analysis of atherosclerosis

Atherosclerosis plaque area was determined by analyzing cross sections from the aortic root of 6-mo LDLr−/− mice male offspring. After lethal injection, hearts were dissected, frozen in tissue freezing medium (Leica Microsystems, Nussloch, Germany) and stored at – 80°C until analysis. Approximately six 9-µm frozen sections per animal were used for morphometrical analysis and 6 aortic roots per group were analyzed following this procedure. Lipids were stained with oil red O and nuclei with hematoxylin. Blinded analysis of lipid stained area was performed with Axiovision quantification image analyzer (Axiovision Software 4.6.3, Carl Zeiss, Gottinghem, Germany) after capture of images with Axioscop 2 Mot Plus microscope (Carl Zeiss).

(Exp 2) Real time RT-PCR

Total RNA was extracted from livers of LDLr−/− mice with Gen Elute™ Mammalian Total RNA miniprep kit (Sigma, St. Louis, MO, USA). The quality of RNA extracted from livers was assessed previous to reverse transcription by electrophoresis in 1% agarose gel and samples were stored at −80°C. First strand cDNA was generated from total RNA by RT using Ready-To-Go You-Prime first strand beads (GE Healthcare, Buckinghamshire, UK) following manufacturer's instructions. Quantitative real-time PCR was performed using SybrGreen master mix (Westburg Leusden The Netherlands). Sequences were designed using Primer Express software (Applied Biosystems, Foster City, CA, USA): TBP (NM_013684) 5′-TCCCAAGCGATTTGCTGC-3′ and 5′-GCAGTTGTCCGTGGCTCTCT-3′; PPARγ (NM_) 5′-TCTGGGAGATCCTCCTGTTGA-3′ and 5′-GAAGTTGGTGGGCCAGAATG-3′ and SREBP-1c (NM_011480) 5′- CGGCTGTCGTCTACCATAAGCT-3′ and 5′-CCAGTGTTGCCATGGATATAG-3′; HMG-CoASyn (NM_008256) 5′- TGCAGGAAACTTCGCTCACA-3′ and 5′- CCAGGTCAGTTTGGTCCACA-3′; HMG-CoARed (NM_ 008255) 5′- TCTTCCCGGCCTGTGTGT-3′ and 5′- CCTCACGGCTTTCACGAGAA-3′; SREBP-2 (NM_033218) 5′- GAGGTCACGAGGCTTTGCA-3′ and 5′- ACCGCTCCGCAGACGAGGA-3′. Oligonucleotides were purchased from Sigma. The level of mRNA expression was calculated using the ΔΔCt method were ΔCt of a sample represents the difference between Ct of target gene and the housekeeping gene and ΔΔCt is ΔCt of sample-ΔCt of calibrator consisting of cDNA from control mouse. The efficiency for housekeeping gene and each target gene was estimated in preliminary analyses and was assessed using standard curve method as explained previously [14]. The relative expression was calculated from 2−ΔΔCt and expressed relatively to C where values are arbitrary set to 1. RNA levels were expressed as relative abundance ± SEM.

Blood sampling and analysis

Blood was collected at sacrifice between 2 and 4 pm after 6 hours fasting by cardiac puncture in EDTA tubes and plasma was obtained by centrifugation. For glucose concentration analysis, proteins were precipitated by addition of 200 µl HClO4 (0.33N) to 20 µl blood and concentration was determined by the glucose oxidase method using Trinder's reagent (Stanbio Laboratory, Texas, USA). Triglyceride and total cholesterol concentrations were determined by using respectively TRF400CH and CTF400CH kits following the manufacturer instructions (Chema Diagnostica, Jesi, Italy). HDL-cholesterol concentration was assessed by the total cholesterol kit after precipitation of low-density and very low-density lipoproteins by “HDL precipitating reagent CD0400CH” (Chema Diagnostica, Jesi, Italy). Plasma insulin was measured using ultrasensitive insulin enzyme-linked assay (Mercodia, Uppsala, Sweden). Leptin concentration was determined by ELISA following manufacturer's instructions (Biovendor, Modrice, Czech Republic). Plasma corticosterone concentration was determined with an enzyme immunoassay kit (Assay Designs, Ann Harbor, MI, USA).

Intraperitoneal Glucose Tolerance Test

Intraperitoneal glucose tolerance test (IPGTT) was performed after an overnight fast at 3 and 9 months in male offspring of C57BL6/J dams as well as at 6 months in male offspring of LDLr−/− dams. Glucose was measured after a single intraperitoneal bolus of glucose (20% glucose solution, 2 g/kg) in samples collected by tail bleeding after 0, 30, 60 and 120 minutes. Glucose concentration was determined with a glucometer (Ascencia Contour, Bayer Health Care, Mishawak, IN, USA).

Statistical Analysis

Results are reported as mean ± S.E.M. Two-way analysis of variance was used to test for effects of maternal and post-weaning diets and interaction between these variables using the Prism Software (GraphPad Software Inc., San Diego, CA, USA). When specified, a Bonferroni post-hoc test was performed to look at statistical differences between groups. Differences with P<0.05 were considered as significant.

Results Top

Experiment 1: Programming of blood pressure and heart rate by mismatch of pre- and post-natal nutrition in C57BL6/J male offspring

Body weight and plasma parameters.

As it has been shown in models of maternal protein restriction during gestation, pups from LP litters were growth restricted at birth when compared to C (C: 1.57±0.08 vs. LP: 1.21±0.03 g; N = 12 litters/group, P<0.001). As a consequence of forced catch-up growth during lactation, LP pups presented at weaning a higher body weight when compared to C (C: 10.94±0.82 vs. LP: 13.41±0.62 g; N = 12 litters/group, P<0.05).

At 3 months of age, both C and LP animals fed the OB diet after weaning presented a significant increase in body weight when compared to usual chow fed mice but no influence of maternal diet was found (Table 1). Analysis of plasma parameters indicated no difference for triglycerides or insulin concentration. However, the post-weaning OB diet increased glucose concentration in both the C and LP group but again no effect of the maternal diet was observed (Table 1). Moreover, intraperitoneal glucose tolerance test (IPGTT) indicated an altered glucose clearance in these OB fed animals as illustrated by the higher area under the curve (Figure 1A and B) but no effect of the maternal diet was noted.

Figure 1. Blood glucose concentration and area under the curve (AUC) during intraperitoneal glucose tolerance test.

The test was performed in 3-mo (A and B), 9-mo (C and D) C57BL6 male mice and 6-mo LDLr−/− (E and F). Values are presented as mean ± SEM with n = 5 to 8 individuals per group. Statistical analysis by Two-way ANOVA indicates a significant influence of post-weaning OB diet for AUCs at 3-mo (B; P<0.0001) and 9-mo (D; P<0.0001) in C57BL6 mice (exp 1) and at 6-mo (F; P<0.0001) in LDLr−/− mice (exp 2).

doi:10.1371/journal.pone.0012656.g001

Table 1. Body weight and Plasma parameters.

doi:10.1371/journal.pone.0012656.t001

After 9 months, OB diet continued to induce a significant overweight in both C and LP groups, but no difference due to maternal diet was however observed. Plasma triglycerides

concentration was significantly raised in LP animals when compared to C, showing an influence of maternal diet. Plasma glucose concentration was significantly decreased in LP group compared to the C group (Table 1). However, feeding mice offspring with OB diet increased glucose concentration in C and LP animals. Similarly, IPGTT indicated an altered response to glucose load in both OB groups as already observed at 3 months (Figure 1C and D). In addition, induced obesity was accompanied by hypercholesterolemia, hyperinsulinemia and hyperleptinemia in both groups. Maternal LP diet increased significantly the plasma corticosterone levels in LP offspring which was amplified by the OB post-weaning diet (Table 1).

Circadian variation of blood pressure and heart rate

Using telemetry, systolic blood pressure (SBP), diastolic blood pressure (DBP) and heart rate (HR) were continuously recorded during 24 h in 3- and 9-mo C57BL6/J male mice. At 3-mo we observed a slight but significant increase DBP and HR due to OB diet in C-OB and LP-OB groups during the day period. No difference was observed for these parameters during the night period. Maternal diet had however no effect at that age (Table 2). When these parameters were analyzed at 9-mo of age, a significant higher SBP and HR due to maternal diet was observed independently of obesity development (Figure 2 and Table 2). Statistical analysis of mean day and night parameters indicated a significant difference of SBP and HR only during day phase. Indeed, HR was significantly increased in the LP group under OB diet, and SBP was higher in the LP group under chow diet. As a consequence, LP animals lost the difference in blood pressure between day and night which normally occurs, reflecting therefore a loss of circadian rhythmicity in LP offspring (Table 2). Indeed, while C and C-OB mice showed a rise of blood pressure and heart rate during night phase, in the LP and LP-OB groups, blood pressure and heart rate remained constant along the 24 h period.

Figure 2. Long-term recording blood pressures and heart rate in 9-mo male C57BL6 mice (exp1).

Recording of SBP (A), DBP (B) pressure and HR (C) were assessed during 24 h in n = 5 animals per group. Values are presented as mean values of SBP, DBP and

HR calculated for each 60-minutes sequence of recording. Shaded zone on the X-axis indicates night period (activity period for mice).

doi:10.1371/journal.pone.0012656.g002

Table 2. Day and night average values of blood pressure and heart rate in 3- or 9-mo male mice.

doi:10.1371/journal.pone.0012656.t002

Short-term recording of SBP and HR after acute treatment by L-NAME or phenylephrine on 9-mo old mice.

Intraperitoneal injection with L-NAME (30 mg/kg), an e-NOS inhibitor that induces an increase in BP, produced a physiological response with a rapid enhancement of BP and a

reduction in HR in the four experimental groups (Figure 3A). The magnitude of delta blood pressure reduction in mice from LP was not significant (Figure 3B). HR reduction tended to be lower in animals fed after weaning with OB diet (P = 0.07) (Figure 3C–D). As expected, the intraperitoneal injection of phenylephrine (100 µg/kg), an α1-adrenoceptor agonist that also increases blood pressure produced a slight increase in SBP but an important rise of HR in C animals (Figure 3E–F). When C and LP animals were fed the OB diet, they showed a minimal response to phenylephrine treatment either on SBP or HR, indicating a significant alteration of HR response due to OB diet (P = 0.01) (Figure 3G–H). Moreover, LP offspring displayed also an altered HR response to treatment indicating an influence of the maternal diet (P = 0.04) (Figure 3H).

Figure 3. Short-term recording of SBP and HR after an acute treatment with L-NAME (A–D) and phenylephrine (E–H).

Recording of cardiovascular parameters was assessed in 9-mo male C57BL6 mice (exp1) during 1 h in n = 5 animals per group. Values are presented as mean values of SBP or HR ± SEM calculated for each 5 minutes sequence of recording. Arrows indicate the time of injection. Δ values of SBP and HR were calculated by subtracting the mean values of the 5 time-point after to the mean values of the 5 time-point before injection. Statistical analysis by Two-way ANOVA indicates a significant influence of post-weaning OB diet (P = 0.0127) as well as a significant influence of maternal LP diet (P = 0.0429) for ΔHR after phenylephrine treatment.

doi:10.1371/journal.pone.0012656.g003

Experiment 2: Programming of atherosclerosis in LDLr−/− male mice by mismatch of pre- and post-natal nutrition

Body weight and plasma parameters.

LDLr−/− pups from protein-restricted dams presented at birth a slight weight reduction that was however not significant (C: 1.37±0.04 vs. LP: 1.22±0.06 g; N = 8 litters/group). During lactation LP pups caught-up weight from day 14 and remained heavier than controls until weaning (C: 7.33±0.36 vs. LP: 8.62±0.16 g; N = 8 litters/group, P<0.001). After suckling period, OB fed animals gained rapidly more weight than chow fed mice and presented obesity when analyzed at 6-mo of age. Nevertheless, no influence of maternal diet on their weight gain was observed (Table 1).

Measurements of plasma parameters indicated a significantly higher concentration of triglycerides, free fatty acids and total cholesterol due to OB feeding. Total cholesterol was also increased due to early nutritional mismatch (Table 1) whereas HDL-cholesterol was increased in a similar way in C-OB and LP-OB animals when compared to chow fed animals. Although glucose concentration did not show any difference due to early nutrition or to OB diet, obese C and LP mice featured an altered response to glucose load as indicated by IPGTT (Figure 1E, F). Insulin concentration was raised in LP and LP-OB offspring as well as in C-OB fed animals compared to C offspring. Leptin concentration was also raised in obese mice without significant influence of maternal diet.

A comparison of body weight and blood parameters has also been done between the two strains (Experiment 1 and Experiment 2) and only few parameters showed a significant difference by strain. Body weight was significantly lower for LDLr −/− mice than for wild type mice when submitted to OB diet and might easily be explained by the age at which mice weight were recorded (9-mo vs. 6-mo). Not surprisingly, total cholesterol values for animals receiving OB diet were significantly higher for LDLr −/− when compared to wild type strain. Finally, plasma insulin values are statistically different in C-OB group between wild-type and KO mice and triglycerides values are significantly higher in LDLr −/− mice when compared to wild type, in LP-OB group.

Atherosclerosis measurements in aortic root of LDLr−/− mice and liver mRNA analysis with real-time PCR.

Figure 4A shows representative oil red-O stained sections of the aortic root from male mice of the 4 experimental groups. Measurements of lesion area indicated an increase of plaque surface in both OB fed animals (P = 0.0002; Figure 4B). No difference according to maternal diet was observed in this experiment. Nevertheless, since plasma total cholesterol was higher in LP and LP-OB offspring, we investigated the mRNA expression of few liver genes involved in cholesterol metabolism. PPARγ and SREBP-1c mRNA levels were both significantly increased according to post-weaning OB diet (P = 0.0002; Figure 5). Although the expression of PPARγ was higher in LP groups when compared to C groups, the difference was not significant. HMG-CoA reductase mRNA abundance showed a significant interaction between the maternal diet and OB diet effect meaning that the effect of OB feeding was different according to maternal diet (P = 0.0243) while HMG-CoA synthase expression showed a significant effect of OB feeding (P = 0.004) as well as a significant effect of maternal nutrition (P = 0.0406). Finally SREBP-2 mRNA relative abundance showed an increase due to OB diet (P = 0.034). However, the increase was more important in LP-OB group than in C-OB as indicated by significant Bonferroni post-test (Figure 5).

Figure 4. Atherosclerosis plaque area assessment in aortic root of 6-mo LDLr−/− male mice.

(A) Representative images of oil red-O staining of the aortic root sections in the 4 experimental groups (scale bar = 200 µm). (B) The plaque area values are

presented as mean values ± SEM for n = 6 animals per group. Two-way ANOVA indicates a significant influence of post-weaning OB diet (P = 0.0002) on plaque surface.

doi:10.1371/journal.pone.0012656.g004

Figure 5. Relative abundance of gene mRNA assessed in livers of 6-mo LDLr−/− male mice.

Values are presented as mean relative abundance ± SEM for n = 6 animals per group. Two-way ANOVA indicated a significant influence of post weaning OB diet for SREBP-1c (P = 0.0002), HMG-CoA synthase (P = 0.004) and PPARγ (P = 0.0002) as well as a significant influence of maternal LP diet for SREBP-2 (P = 0.034) and HMG-CoA synthase (P = 0.0406). HMG-CoA reductase showed a

significant interaction effect (P = 0.0243). Bonferroni post-test has been performed to assess statistical differences in mRNA relative abundance within chow fed vs. OB fed animals within C and LP groups. * P<0.05 and ** P<0.01.

doi:10.1371/journal.pone.0012656.g005

Discussion Top

This study aimed to determine whether an in utero growth retardation followed by a rapid catch-up growth after birth was able to favour hypertension and atherosclerosis at adult age, and, if this was dependent on obesity. Two main results may be pointed out in the first experiment where hypertension was analyzed. Firstly, at 9-mo LP offspring presented higher SBP and HR than C offspring, indicating a programming of cardiovascular function by the poor prenatal diet followed by a rapid catch-up growth. In addition, this programming is independent of obesity development because hypertension was already observed in LP offspring fed a normal diet. A second important finding is the loss of circadian rhythms of the blood pressure and heart rate in the LP offspring.

Early postnatal overfeeding is known to induce cardiovascular and metabolic changes such as hypertension and central obesity with increased leptin, glucose, insulin, and glucocorticoid level [23]. In addition to intrauterine growth retardation, this may be a second hit which favors hypertension and cardiovascular disease [15]. Consistent with this, in our study, overweight resulting from post-weaning high fat diet induced an altered glucose tolerance in both OB fed groups at the age of 3-mo and these animals presented a rise in DBP and HR during the resting (day) period (Exp 1). However, contrary to what has been shown by others [15], [24], [25] no effect due to maternal diet was observed at this young age either on BP or HR. Higher SBP and HR were only revealed with age in LP offspring after 9-months. In a rat study, when the LP diet was given during gestation and maintained throughout life, catch-up growth did not occur and blood pressure as well as renal glomerular structure remained unaltered [26]. Whereas some studies using tail cuff technique depicted a rise of BP due to the maternal protein restriction during pregnancy in 3-mo rats [13], [24], [25] another group using telemetry recording showed only moderate effect of a more severe maternal protein deprivation on hypertension in young offspring [27]. This puts forwards the impact of the technique used, as data obtained by tail-cuff method are based on indirect blood pressure measurements which require restraint, a potential source of stress, and it is know that the LP offspring are more susceptible to stress at least in female offspring [28].

Many bodily functions such as temperature, eating, blood pressure and heart rate are influenced by circadian clock. Blood pressure in rodents, as in human, presents a circadian pattern and its loss correlates with a higher risk for cardiovascular complications [29], [30]. One of the factors that limit the study of cardiovascular function is the difficulty to measure blood pressure in conscious non-stressed animals. Indeed, the measurement of the basal levels and the circadian patterns is not possible using the tail cuff technique. Here we have used the gold standard technique of implanted telemetry to investigate this function. Our results highlighted a programming of alteration in the circadian clock by the early diet that may have influenced the blood pressure. In the control groups, we found a greater SBP, DBP as well as HR during the night (activity period) than during the day which represents a normal pattern [31]. This was not found in our LP groups and the HR was even higher during the day than during the night (under OB diet). Thus, LP male offspring featured a loss of circadian rhythms of both blood pressure and heart rate. Very recently, the study of

Samuelsson et al [32] revealed that maternal diet-induced-obesity was able to program hypertension in male and female offspring at 6 months but only the male offspring of the obese mother lost the diurnal blood pressure variation.

It was postulated that the functional significance of the normal increase in blood pressure and heart rate during the active period, which is during the dark phase in rodent, is to allow the animals to meet the metabolic demands of increased activity [31]. In obese Zucker rat circadian disturbance of activity and feeding occurs with enhanced food-anticipatory activity [33], [34]. Programming of dysregulation in the day-light behaviour was also observed for food intake rhythm in rats after a drastic maternal food restriction during gestation. In such condition, a decreased food intake in the dark phase and an increased food intake during the light phase compared to control group were reported [35]. Another recent study evaluated the consequence of an early or late catch up-growth after IUGR due to maternal protein restriction on the feeding behaviour during day and night phase and found that more than the quantity of food ingested, it was the duration of meals that was modified by the nutritional programming [36]. In this model, the early catch-up growth reduced the abnormal organization of hypothalamic pathways involved in energy homeostasis [37]. Our data together with those reported above open new perspectives for investigating another pathway involved in early programming because the disturbance of the circadian rhythmicity may influence food behaviour, obesity, hypertension and other diseases which have been found to have early origin.

The molecular clock in mammals is located in the suprachiasmatic nucleus (SCN). However there are molecular clocks in many peripheral tissues as in the liver, kidneys and heart [38]. Even if mechanisms linking the SCN and peripheral clocks are still poorly understood, they might involve circulating hormones [39]. Recent results suggest that the rhythmic secretion and the ability to phase shift peripheral clocks place glucocorticoids as possible candidates for the entrainment of peripheral oscillators [40] and interestingly plasma corticosterone levels were increased in the LP offspring.

On the other hand, HR and BP are under the control of different factors among which nitric oxide and several hormones. Therefore, we investigated BP and HR after an acute treatment with L-NAME, whose vasoconstrictive effects are directly modulated by the endothelium and with phenylephrine, an alpha1-adrenergic vasoconstrictor. Little change in the vascular reactivity was observed in relation to the maternal diet. Indeed, only HR showed an altered response due to maternal LP diet after phenylephrine injection. Sathishkumar et al [41] investigated the blood pressure and the vascular reactivity in 3-month old female offspring from mothers fed a low protein diet during gestation and showed an increased BP pressure and contractile response to phenylephrine, as well as, a reduced vasodilatation in response to acetylcholine. It should be noted that in this study, no postnatal catch-up growth occurred and the LP offspring remained lighter throughout life. Glucocorticoid plasma concentration which was increased in LP offspring is thought to play a main role in the early programming of adult disease. In rodents, fetal exposure to excess glucocorticoids via maternal stress (calorie restriction) or drug treatment (dexamethasone) lowered birth weight, increased glucose and insulin levels and also increased blood pressure in the adult progeny [42]. Correlatively, treatment of LP mothers during pregnancy with metyrapone, an inhibitor of corticoids synthesis, prevented the development of high blood pressure in male offspring [43]. Other studies identified increased adrenal-to-body weight ratio in parallel to an increased resting heart rate [44] and higher levels of peripheral epinephrine levels in LP male offspring, as well as perturbation of β-adrenergic response and signaling [45]. Finally, it was found that

prenatal dexamethasone decreased basal blood pressure but induced stress-hypertension in adult rat offspring [46]. In parallel to our observations, these reports indicate a link between programming of hypertension and stress-response axis. Indeed, it is thought that fetal environment could program an enhanced activity of the mediators of stress response including hypothalamic-pituitary-adrenal (HPA) axis and autonomic nervous system [47]. A suboptimal in utero environment could alter the set points of such stress axes in order to provide a biological advantage. However, if the environment encountered later on is not well matched to the fetus predictions, it could provide a disadvantage and lead to hypertension later in life. Although in our experiments, an acute treatment with phenylephrine rapidly increases SBP in C animals, only a slight increase was observed for LP offspring. We postulate that a permanent elevated level in stress hormone for these animals may reduce the response to the acute treatment with the α1-adrenoreceptor agonist.

In the second experiment, we used the well-characterized LDLr-/- mice to study the impact of a mismatched early nutrition on the development of atherosclerosis. Although we were unable to show an influence due to maternal diet, feeding animals after weaning with OB diet increased the atherosclerosis plaques extent. Only a few papers examined the influence of early diet on the development of atherosclerosis in transgenic mice models. These studies mainly focused on the impact of maternal hypercholesterolemia for the programming of atherosclerosis. In LDL receptor knock-out (LDLr−/−) mouse, as well as, in ApoE heterozygous deficient mouse, maternal hypercholesterolemia -induced either by diet or by cross-breeding wild type male mouse with ApoE deficient females- promoted atherosclerosis in aortic arch of the respective offspring [48], [49]. Another recent study has investigated the programming of atherosclerosis by fetal malnutrition in ApoE*3 Leiden mouse [50]. Such mice have an impaired clearance of lipoproteins from the plasma, a raised plasma lipid levels and a greater susceptibility to develop atherosclerosis when the mice are fed diets rich in cholesterol [51]. No change in cholesterol level and atherosclerosis was observed when the offspring were fed a normal diet. The post-weaning atherogenic diet induced however a higher level of cholesterol and a greater area of atherosclerotic lesions within the aortic arch in animals exposed to a low protein diet in utero than those exposed to the control diet, but this was noted only in females [50]. In our study, although no difference in atherosclerosis development was observed due to early mismatched nutritional environment, plasma total cholesterol concentration was however increased due to maternal diet. Moreover, the analysis of plasma HDL-cholesterol concentration showed a similar increase for animals of the C-OB and LP-OB group indicating that the higher values of total cholesterol observed in LP-OB mice were due to increased amount of LDL- and VLDL-cholesterol. Therefore we also investigated the mRNA expression of genes involved in liver cholesterol biosynthesis. The results indicated an increase due to OB feeding for almost all genes. However, the expression of HMG-CoA reductase, the rate limiting enzyme required for endogenous cholesterol biosynthesis was lower in LP-OB group when compared to C-OB group while conversely; it was increased in the LP group compared to the C group. It indicates that the higher concentration of circulating total cholesterol observed in LP-OB offspring is not due to an enhanced biosynthesis.

It has been shown that plasma total cholesterol tended to be higher in 18-mo rats that were protein restricted only during fetal life [52] while another experiment in similar conditions indicated a lower plasma cholesterol for these animals at 6-mo [53]. A possible explanation for the differences observed in these studies is the establishment of an insulin resistant phenotype that develops with aging [54]. In our study, animals from the LP-OB group were fed with an obesogenic diet after weaning which worsened the insulin resistant state as shown

by their plasma insulin levels and their IPGTT. In addition, it has been recently demonstrated that mice with diabetes and that lack LDL receptor develop severe hyperlipidemia with decreased clearance in lipoproteins [55]. Finally, statistical analysis of body weight and plasma parameters between both strains (experiment 1 and 2), demonstrated consistency in the maternal LP model since no main influence of the strain was observed for comparable plasma parameters, except for blood lipids, as expected.

In conclusion, our results showed a long-term effect of early mismatched nutritional environment on the development of hypertension independently of obesity. Moreover, 24 h recording of blood pressure and heart rate indicated impairment in circadian rhythms for animals that were exposed in utero to LP diet and forced to catch-up growth during lactation. Although no influence due to maternal diet was observed, atherogenesis was enhanced by postnatal OB diet. Finally, even if there is now convincing evidence from epidemiological and animal studies, supported by our results, that early environmental mismatch can lead in adulthood to physiological changes, the mechanisms underlying these long-term alterations are not yet understood. Research needs to be carried on further and identification of key players should be important for prevention of metabolic diseases in adulthood.

Author Contributions Top

Conceived and designed the experiments: VB BR CR. Performed the experiments: VB. Analyzed the data: VB FD BR JLB CR. Contributed reagents/materials/analysis tools: VB FD JLB. Wrote the paper: VB BR CR.

References Top

1. Staessen JA, Wang J, Bianchi G, Birkenhager WH (2003) Essential hypertension. Lancet 361: 1629–1641. Find this article online

2. Barker DJ, Winter PD, Osmond C, Margetts B, Simmonds SJ (1989) Weight in infancy and death from ischaemic heart disease. Lancet 2: 577–580. Find this article online

3. Hales CN, Barker DJ, Clark PM, Cox LJ, Fall C, et al. (1991) Fetal and infant growth and impaired glucose tolerance at age 64. BMJ 303: 1019–1022. Find this article online

4. Hales CN, Barker DJ (2005) The thrifty phenotype hypothesis. Br Med Bull 60: 5–20. Find this article online

5. McMillen IC, Robinson JS (2005) Developmental origins of the metabolic syndrome: prediction, plasticity, and programming. Physiol Rev 85: 571–633. Find this article online

6. Remacle C, Bieswal F, Reusens B (2004) Programming of obesity and cardiovascular disease. Int J Obes Relat Metab Disord 28: Suppl 3S46–53. Find this article online

7. Gluckman PD, Hanson MA, Pinal C (2005) The developmental origins of adult disease. Maternal & child nutrition 1: 130–141. Find this article online

8. Painter RC, Roseboom TJ, Bleker OP (2005) Prenatal exposure to the Dutch famine and disease in later life: an overview. Reprod Toxicol 20: 345–352. Find this article online

9. Stanner SA, Bulmer K, Andres C, Lantseva OE, Borodina V, et al. (1997) Does malnutrition in utero determine diabetes and coronary heart disease in adulthood? Result from the Leningrad siege study, a cross sectional study. BMJ 315: 1342–1348. Find this article online

10. Stanner SA, Yudkin JS (2001) Fetal programming and the Leningrad Siege study. Twin Res 4: 287–292. Find this article online

11. Cole TJ (2007) Early causes of child obesity and implications for prevention. Acta Paediatr: Suppl 962–4. Find this article online

12. Adair LS, Cole TJ (2003) Rapid child growth raises blood pressure in adolescent boys who were thin at birth. Hypertension 41: 451–456. Find this article online

13. Bieswal F, Ahn MT, Reusens B, Holvoet P, Raes M, Rees WD, Remacle C (2006) The importance of catch-up growth after early malnutrition for the programming of obesity in male rat. Obesity 14: 1330–1343. Find this article online

14. Bol V, Delattre A, Reusens B, Raes M, Remacle C (2009) Forced catch-up growth after fetal protein restriction alters the adipose tissue gene expression program leading to obesity in adult mice. Am J Physiol Regul Integr Comp Physiol 297: R291–R299. Find this article online

15. Broubred F, Daniel L, Buffat C, Feuerstein JM, Tsimaratos M, et al. (2009) Early postnatal overfeeding induces early chronic renal dysfunction in adult male rats. Am J Physiol-Renal Physiol 297: F943–F951. Find this article online

16. Cleal JK, Poore KR, Boullin JP, Khan O, Chau R, et al. (2007) Mismatched pre- and postnatal nutrition leads to cardiovascular dysfunction and altered renal function in adulthood. Proc Natl Acad Sci USA 104: 9529–9533. Find this article online

17. Trayhurn P, Wood IS (2004) Adipokines: inflammation and the pleiotropic role of white adipose tissue. Br J Nutr 92: 347–355. Find this article online

18. Poulos S, Hausman D, Hausman G (2010) The development and endocrine functions of adipose tissue. Mol Cell Endocrinol 323: 20–34. Find this article online

19. Fantuzzi G, Mazzone T (2007) Adipose tissue and atherosclerosis. Exploring the connection. Arterioscler Thromb Vasc Biol 27: 996–1003. Find this article online

20. Russell JC, Proctor SD (2006) Small animal models of cardiovascular disease: tools for the study of the roles of metabolic syndrome, dyslipidemia, and atherosclerosis. Cardiovasc Pathol 15: 318–330. Find this article online

21. Bieswal F, Hay SM, McKinnon C, Reusens B, Cuignet M, et al. (2004) Prenatal protein restriction does not affect the proliferation and differentiation of rat preadipocytes. J Nutr 134: 1493–1499. Find this article online

22. Hasty AH, Shimano H, Osuga J, Namatame I, Takahashi A, et al. (2001) Severe hypercholesterolemia, hypertriglyceridemia, and atherosclerosis in mice lacking both leptin and the low density lipoprotein receptor. J Biol Chem 276: 37402–37408. Find this article online

23. Boullu-Ciocca S, Dutour A, Guillaume V, Achard V, Oliver C, et al. (2005) Postnatal diet-induced obesity upregulates systemic and adipose tissue glucocorticoid metabolism during development and in adulthood; its relationship with the metabolic syndrome. Diabetes 54: 197–203. Find this article online

24. Langley-Evans S, Gardner D, Welham S (1994) Intrauterine programming of cardiovascular disease by maternal nutritional status. Nutrition 14: 39–47. Find this article online

25. Langley-Evans SC (2001) Fetal programming of cardiovascular function through exposure to maternal undernutrition. Proc Nutr Soc 60: 505–513. Find this article online

26. Hoppe C, Evans R, Moritz K, Cullen-McEwen L, Fitzgerald S, et al. (2007) Combined prenatal and postnatal protein restriction influences adult kidney structure, function, and arterial pressure. Am J Physiol Regul Integr Comp Physiol 292: R462–R469. Find this article online

27. Tonkiss J, Trzcińska M, Galler JR, Ruiz-Opazo N, Herrera VL (1998) Prenatal malnutrition-induced changes in blood pressure: dissociation of stress and nonstress responses using radiotelemetry. Hypertension 32: 108–114. Find this article online

28. Watkins AJ, Ursell E, Paton R, Papenbrock T, Hollis L, et al. (2008) Adaptive responses by mouse early embryos to maternal diet protect fetal growth but predispose to adult onset disease. Biol Repod 78: 299–306. Find this article online

29. Young ME (2006) The circadian clock within the heart: potential influence on myocardial gene expression, metabolism, and function. Am J Physiol Heart Circ Physiol 290: H1–H16. Find this article online

30. Rothwell PM (2010) Limitations of the usual blood-pressure hypothesis and importance of vaiability, instability, and episodic hypertension. Lancet 375: 938–948. Find this article online

31. Li P, Sur S, Mistlberger R, Morris M (1999) Circadian blood pressure and heart rate rhythms in mice? Am J Physiol 276: R500–R504. Find this article online

32. Samuelsson AM, Matthews P, Argenton M, Christie M, McConnel J, et al. (2008) Diet-induced obesity in female mice leads to offspring hyperphagia, adiposity, hypertension, and insulin resistance. A novel murine model of developmental programming. Hypertension 51: 383–392. Find this article online

33. Mistlberger RE, Lukman H, Nadeau BG (1998) Circadian rhythms in the Zucker obese rat: assessment and intervention. Appetite 30: 255–267. Find this article online

34. Mistlberger RE, Marchant EG (1999) Enhanced food-anticipatory circadian rhythms in the genetically obese Zucker rat. Physiol Behav 66: 329–335. Find this article online

35. Breton C, Lukaszewski MA, Risold PY, Enache M, Guillemot J, et al. (2009) Maternal prenatal undernutrition alters the response of POMC neurons to energy status variation in adult male rat oddspring. Am J Physiol Endocrinol Metab 296: E462–E472. Find this article online

36. Coupé B, Grit I, Darmaun D, Parnet P (2009) The timing of “catch-up growth” affects metabolism and appetite regulation in male rats born with intrauterine growth restriction Am J Physiol Regul Integr Comp Physiol 297: R813–R824. Find this article online

37. Coupé B, Amarger V, Grit I, Benani A, Parnet P (2010) Nutritional programming affects hypothalamic organization and early response to leptin. Endocrinology 151: 702–713. Find this article online

38. Schibler U, Ripperger J, Brown SA (2003) Peripheral circadian oscillators in mammals: time and food. J Biol Rhythms 18: 250–260. Find this article online

39. McNamara P, Seo SP, Rudic RD, Sehgal A, Chakravarti D, et al. (2001) Regulation of CLOCK and MOP4 by nuclear hormone receptors in the vasculature: a humoral mechanism to reset a peripheral clock. Cell 105: 877–889. Find this article online

40. Balsalobre A, Brown SA, Marcacci L, Tronche F, Kellendonk C, et al. (2000) Resetting of circadian time in peripheral tissues by glucocorticoid signaling. Science 289: 2344–2347. Find this article online

41. Sathishkumar K, Elkins R, Yallampalli U, Yallampalli C (2009) Protein restriction during pregnancy induces hypertension and impairs endothelium-dependent vascular function in adult female offspring. J Vasc Res 46: 229–239. Find this article online

42. Sugden MC, Langdown ML, Munns MJ, Holness MJ (2001) Maternal glucocorticoid treatment modulates placental leptin and leptin receptor expression and materno-fetal leptin physiology during late pregnancy, and elicits hypertension associated with hyperleptinaemia in the early-growth-retarded adult offspring. Eur J Endocrinol 145: 529–539. Find this article online

43. McMullen S, Langley-Evans SC (2005) Sex-specific effects of prenatal low-protein and carbenoxolone exposure on renal angiotensin receptor expression in rats. Hypertension 46: 1374–1380. Find this article online

44. Fernandez-Twinn DS, Ekizoglou S, Wayman A, Petry CJ, Ozanne SE (2006) Maternal low-protein diet programs cardiac beta-adrenergic response and signaling in 3-mo-old male offspring. Am J Physiol Regul Integr Comp Physiol 291: R429–R436. Find this article online

45. Petry CJ, Dorling MW, Wang CL, Pawlak DB, Ozanne SE (2000) Catecholamine levels and receptor expression in low protein rat offspring. Diabet Med 17: 848–853. Find this article online

46. O'Regan D, Kenyon CJ, Seckl JR, Holmes MC (2008) Prenatal dexamethasone ‘programmes’ hypotension, but stress-induced hypertension in adult offspring. J Endocrinol 196: 343–352. Find this article online

47. Phillips DI, Jones A, Goulden PA (2006) Birth weight, stress, and the metabolic syndrome in adult life. Ann N Y Acad Sci 1083: 28–36. Find this article online

48. Napoli C, de Nigris F, Welch JS, Calara FB, Stuart R, et al. (2002) Maternal hypercholesterolemia during pregnancy promotes early atherogenesis in LDL receptor-deficient mice and alters aortic gene expression determined by microarray. Circulation 105: 1360–1367. Find this article online

49. Goharkhay N, Sbrana E, Gamble PK, Tamayo EH, Betancourt A, et al. (2007) Characterization of a murine model of fetal programming of atherosclerosis. Am J Obstet Gynecol 197: 416.e1–416.e5. Find this article online

50. Yates Z, Tarling E, Langley-Evans S, Salter A (2009) Maternal undernutrition programmes atherosclerosis in the ApoE*3-Leiden mouse. Br J Nutr 101: 1185–94. Find this article online

51. Groot P, van Vlijmen B, Benson G, Hoker M, Schiffelers R, et al. (1996) Quantitative assessment of aortic atherosclerosis in APOE*3 Leiden transgenic mice and its relationship to serum cholesterol exposure. Arterioscl Thromb Vasc 16: 929–933. Find this article online

52. Erhuma A, Salter AM, Sculley DV, Langley-Evans SC, Bennett AJ (2007) Prenatal exposure to a low-protein diet programs disordered regulation of lipid metabolism in the aging rat. Am J Physiol Endocrinol Metab 292: E1702–1714. Find this article online

53. Lucas A, Baker BA, Desai M, Hales CN (1996) Nutrition in pregnant or lactating rats programs lipid metabolism in the offspring. Br J Nutr 76: 605–612. Find this article online

54. Petry CJ, Dorling MW, Pawlak DB, Ozanne SE, Hales CN (2001) Diabetes in old male offspring of rat dams fed a reduced protein diet. Int J Exp Diabetes Res 2: 139–143. Find this article online

55. Goldberg IJ, Hu Y, Noh HL, Wei J, Huggins LA, et al. (2008) Decreased lipoprotein clearance is responsible for increased cholesterol in LDL receptor knockout mice with streptozotocin-induced diabetes. Diabetes 57: 1674–1682. Find this article online

All site content, except where otherwise noted, is lic