The failing heart stimulates tumor growth by circulating factorsMeijers - The failing heart stimulates tumor growth

Wouter C. Meijers MD (1), Manuel Maglione MD PhD (2), Stephan J.L. Bakker MD PhD (3), Rupert Oberhuber MD PhD (2), Lyanne M. Kieneker Msc (3), Steven de Jong PhD (4), Bernhard J. Haubner MD PhD (5), Wouter B. Nagengast MD PhD (6), Alexander R. Lyon MD PhD(7), Bert van der Vegt MD PhD (8), Dirk J. van Veldhuisen MD PhD (1), B. Daan Westenbrink MD PhD (1), Peter van der Meer MD PhD (1), Herman H.W. Silljé PhD (1), Rudolf A. de Boer MD PhD (1).

Affiliations

1. University of Groningen, University Medical Center Groningen, Department of Cardiology, PO Box 30.001, 9700 RB Groningen, the Netherlands

2. Centre of Operative Medicine, Department of Visceral, Transplant and Thoracic Surgery, Medical University Innsbruck, Innsbruck, Austria.

3. University Medical Center Groningen and University of Groningen, Department of Internal Medicine, Division of Nephrology, Groningen, The Netherlands.

4. University of Groningen, University Medical Center Groningen, Department of Medical Oncology, Groningen, The Netherlands.

5. Department of Internal Medicine III (Cardiology and Angiology), Innsbruck Medical University, Innsbruck, Austria.

6. University of Groningen, University Medical Center, Department of Gastroenterology and Hepatology, Groningen, The Netherlands.

7. National Heart and Lung Institute, Imperial College London and Royal Brompton Hospital, Sydney Street, London, SW3 6HP, UK.

8. University Medical Center Groningen, University of Groningen, Department of Pathology, Groningen, the Netherlands.

Word count: 5000 (Text; only) 8669 (Including title page, abstract, text, references, tables and figure

legends)

Corresponding author:

R.A. de Boer MD PhD FESC FHFA

University of Groningen, University Medical Center Groningen, AB31

Department of Cardiology

PO Box 30.001, 9700 RB, Groningen, The Netherlands.

Phone: +31 50 361 2355 / Fax: +31 50 361 4391

E-mail r.a.de.boer@umcg.nl

Abstract

Background: Heart failure (HF) survival has improved and nowadays many patients with HF die from

non-cardiac causes, including cancer. Our aim was to investigate whether a causal relationship exists

between HF and the development of cancer.

Methods: HF was induced by inflicting large anterior myocardial infarction (MI) in APCmin mice, which

are prone to develop precancerous intestinal tumors, and tumor growth was measured. In addition,

to rule out hemodynamic impairment, a heterotopic heart transplantation model was employed,

where an infarcted or sham-operated heart was transplanted into a recipient mouse, while the

native heart was left in situ. After 6 weeks, tumor number, volume, and proliferation were

quantified. Candidate secreted proteins were selected because they were previously associated both

with (colon) tumor growth and with myocardial production in post-MI proteomic studies. Myocardial

gene expression levels of these selected candidates were analyzed, as well as their proliferative

effects on HT-29 (colon cancer) cells. We validated these candidates by measuring them in plasma of

healthy subjects and HF patients. Finally, we associated the relation between cardiac specific and

inflammatory biomarkers and new-onset cancer in a large prospective general population cohort.

Results: The presence of failing hearts, both native and heterotopically transplanted, resulted in

significantly increased intestinal tumor load of 2.4fold in APCmin mice (all P<0.0001). The severity of

left ventricular (LV) dysfunction and fibrotic scar strongly correlated with tumor growth (P=0.002 and

P=0.016, respectively). We identified several proteins (including serpinA3 and A1, fibronectin,

ceruloplasmin, and PON1) that were elevated in human patients with chronic HF (N=101) compared

to healthy subjects (N=180, P<0.001). Functionally, serpinA3 resulted in marked proliferation effects

in human colon cancer (HT-29) cells, associated with Akt-S6 phosphorylation. Finally, elevated cardiac

and inflammation biomarkers in apparently healthy humans (N=8319), were predictive for new-onset

cancer (N=1124), independent from risk factors for cancer (age, smoking status and body mass

index).

1

Conclusion: We demonstrate that the presence of HF is associated with enhanced tumor growth and

this is independent from hemodynamic impairment and could be due cardiac excreted factors. A

diagnosis of HF may therefore be considered a risk factor for incident cancer.

Key-words: Heart failure, cancer, proteomics, myocardial infarction, biomarkers.

Clinical Perspective:

What is new?

New onset cancer is prevalent in patients with heart failure, but it remains unclear if the

failing heart itself contributes to tumor growth.

The current translational data provide for the first time direct proof that the presence of

heart failure itself stimulates tumor formation, possibly via secreted factors.

What are the clinical implications?

Heart failure as a consequence of cancer treatment is an accepted phenomenon and has

been widely studied.

In current daily practise, there is nearly no awareness that new onset cancer can develop in

patients with heart failure and the results of this study will spark interest in this hitherto

unexplored disease combination.

We speculate that it may be recommended to consider differential surveillance programmes

to screen heart failure patients that are at risk for cancer development.

Further studies of cardiac secreted factors and their potential utility as cancer biomarkers

could help to risk stratify heart failure patient regarding their cancer risk.

2

INTRODUCTION

Heart failure (HF) is associated with substantial morbidity and high mortality1. In the last decade, it

has become increasingly apparent that mortality in HF is not only due to cardiac complications and

events. It is well known that HF is characterized by the presence of multi-morbidity, and the extent of

co-morbid diseases strongly affects mortality2. Indeed, recent registries and clinical trials observed

that, in comparison to earlier trials, currently a larger percentage of HF patients die from non-cardiac

causes3,4. However, most attention has been on renal disease, diabetes, and atrial fibrillation while

little attention has been paid to cancer, despite it appears a common disease and important cause of

death, also in patients with HF.

HF as a consequence of cancer treatment has received considerable attention, and the ‘cardio-

oncology’ field is gaining interest at a rapid pace. Further, cancer might affect the heart directly, and

cardiac atrophy and apoptosis have been described in the setting of prevalent cancer 5. But the

converse, i.e. cancer development as a consequence of HF has not been studied at all. Although a

substantial proportion of patients with HF develops and dies from cancer, in the contemporary

treatment and work-up for HF, surveillance for cancer has no role whatsoever.

Although commonly thought of as two separate disease entities, cardiovascular disease (CVD,

including HF) and cancer possess various similarities and possible interactions, including a number of

shared risk factors, suggesting common trigger mechanisms6. Inflammation, obesity, oxidative stress,

diabetes, hypertension, smoking, diet, and physical inactivity are all contributors to both the

development of CVD and cancer.

In line with this hypothesis, recent epidemiological and case-control studies showed that patients

with prevalent HF were more prone to develop incident cancer7–9. In another community-based

cohort, subjects with HF had an increased risk to develop cancer, independent of age and sex 10. But

these studies are associative and cannot prove causality of the observed relation between prevalent

3

HF and new onset cancer. We therefore aimed to explore this possible causal relationship between

HF and cancer development.

4

METHODS

The data, analytic methods, and study materials will be made available to other researchers for

purposes of reproducing the results or replicating the procedure, regarding the in vitro and in vivo

study. A request for the PREVEND data can be made by www.prevend.org.

Mouse model

All animal experiments have been conducted using the C57BL/6J-ApcMin/J (APCmin) mouse strain,

purchased from The Jackson Laboratory (ME, USA). This strain is highly susceptible to spontaneous

intestinal adenoma formation11. Further detail about the strain can be found in the supplemental

methods. All experimental procedures were performed in accordance with the European Union

guidelines (DEC 6844 (Netherlands)) and with the protocol established by the Austrian Federal

Ministry of Science, Research and Economy (Austria) for the care and use of animals. Experiments

were approved by the Austrian Ministry of Education, Science, and Culture (BMWF-66011_0063-

II_10b_2010).

In vivo study design

In this study, we used two different mouse models. 1) Mice in which a myocardial infarction (MI) was

induced (N=22) and the corresponding control sham operated mice (N=10). Cardiac Magnetic

Resonance Imaging (MRI) was performed as reported 12 one week after operation to assess cardiac

function. 2) Mice receiving either an MI or sham donor heart implanted into the cervical (neck) area.

Donor mice (providing the donor heart) were subjected to an MI (N=17) or sham (N=7) operation one

week before transplantation. Just prior to transplantation, cardiac function of the to be transplanted

heart was assessed (in the donor mice) using echocardiography 13. During a follow-up period of 6

5

weeks after HTx surgery no mice were sacrificed, excluding bias. Detailed methods are in the

supplemental methods.

In vitro study design

The human colorectal carcinoma cell line HT29 was grown and tested under various conditions.

Detailed methods are in the supplemental methods.

General population (PREVEND study)

The Prevention of REnal and Vascular ENd-stage Disease (PREVEND) study is a prospective,

observational cohort study, derived from the general population and comprises 8592 participants.

This study was designed to monitor long-term development of cardiac, renal and peripheral vascular

end-stage disease. More details have been described previously 14. The study protocol conforms to

the ethical guidelines of the 1975 Declaration of Helsinki, and was approved by the institutional

review board (IRB) of the University of Groningen. Informed consent was obtained from all PREVEND

participants. In this study, we measured the 5 candidate genes in 180 subjects and cardiac markers

N-terminal pro B-type natriuretic peptide (NT-proBNP), an established biomarker of HF, and high

sensitivity troponin T (hs-TnT), and MR-pro-atrial natriuretic peptide (MR-pro ANP), inflammation

markers C-terminal proendothelin-1 (CT-proET-1), MR-pro-adrenomedullin (MR-proADM), high-

sensitivity C-reactive Protein (hs-CRP) and procalcitonin (PCT) and neuro-endocrine markers

aldosterone, renin and galectin-3.

Chronic Heart Failure (VitD-CHF trial)

We used banked plasma from 101 chronic HF patients that were enrolled in previously published

study (Clinical Trial identifier: NCT01092130) 15,16, who were ≥18 years of age, had a left ventricular

ejection fraction (LVEF) <45%, and were treated with optimal HF medication. The study was

6

approved by the IRB and all patients provided written informed consent. Further details are in the

supplemental methods section.

Statistical analyses

Mouse studies: Normally distributed variables are presented as means±SD. Non-normally distributed

variables are expressed as medians (IQR). Experimental, clinical and biochemical characteristics were

compared across two groups using two-sample t-test for continuous, normally distributed variables,

and Wilcoxon rank-sum test for continuous, non-normally distributed variables. We performed linear

regression analyses to demonstrate the association between tumor load and either the amount of

fibrosis or left ventricular function.

Analyses in PREVEND: The PREVEND study was set up to overselect participants with increased

urinary albumin excretion (UAE) (>10 mg/L). A design-based statistical weighting was used to adjust

for this overselection, allowing conclusions to be made for the general population. As a first

assessment, we performed cumulative hazard plots with log-rank test after stratifying the population

in tertiles based upon NT-proBNP level. To study the association more deeply and including all the

markers measured in the PREVEND study related to cardiac tissue, neuro-endocrine and

inflammation, we used cox proportional hazards regression models, first unadjusted and then

adjusting for age, smoking and BMI. These covariates are selected because shared risk factors might

bias our findings. Furthermore, we calculated the risk for developing incident cancer for those who

developed HF or not before and after the age of 55 years. P<0.05 (two-sided) was considered as level

of statistical significance.

All analyses were performed using Stata/MP College station, TX: StataCorp LP and GraphPad, La Jolla

California USA.

7

RESULTS

Assessment of cardiac function and remodeling in mice with HF after myocardial infarction

To study the interaction between HF and tumor growth we induced a large anterior MI to provoke

HF. We studied APCmin mice, that are prone to develop intestinal polyps and tumors11. One week after

surgery, we performed cardiac MRI (Fig. 1a) to assess cardiac function. Left ventricular Ejection

Fraction (LVEF) was markedly decreased after MI compared to sham-operated animals: 61% vs. 32%;

P<0.001 (Fig. 1b). Functional cardiac parameters in both groups are presented in supplemental tables

1 and 2: MI-induced HF was associated with lower systolic blood pressure, elevated LV end-diastolic

pressure, accompanied with increases in atrial and liver weights. Cardiac fibrosis was determined by

Masson's Trichrome staining on cardiac tissue slides and quantified (Fig. 1a), and was increased 3.2-

fold in mice with HF (P<0.0001; Fig. 1c). Related to the latter, genes associated with inflammation

and fibrosis were significantly up-regulated in HF on mRNA level, protein and plasma level (Fig. 1e-1h

and supplemental figure 1a-k); with the most prominent increase for IL-6 (both on mRNA and plasma

level). NPPA, the gene encoding Atrial Natriuretic Peptide (ANP), an established surrogate HF marker

was more than 20-fold increased in HF compared to sham (P<0.0001; Fig. 1d).

Assessment of tumor growth in MI-induced HF

After six weeks, mice were euthanized and intestinal tissue were harvested. The intestines were

pinned down directly after sacrifice and polyps were counted and measured. Significantly more

(N=57 vs. N=34; Fig. 2a; P<0.001) and larger polyps (1.69mm vs. 1.44mm; Fig. 2b; P<0.001) were

observed in APCmin mice with HF compared to sham operated animals. We calculated overall tumor

load, assuming a spherical morphology of the polyps, and demonstrate a 2.4-fold increase in HF mice

(Fig. 2c; P<0.0001). To further study the increased growth, we performed KI-67 staining, and found

increased numbers of KI-67 positive cells (54% in HF vs. 44% in sham; P<0.05, Fig. 2e). A well-

8

described feature of this model is intestinal blood loss due to polyp bleeding, resulting in anemia and

splenomegaly resulting in a significant higher spleen weight was significantly higher in HF mice

compared to sham (P=0.038) (Fig. 2d).

Association between cardiac remodeling and tumor growth

To relate the changes in tumor growth to parameters of severity of HF, we performed linear

regression analyses between tumor load and indices of cardiac remodeling: fibrosis and LVEF. Both

indices were associated with tumor load (fibrosis: β=0.51, P=0.016; Fig. 2g; and LVEF: β=-0.63,

P=0.002; Fig. 2h).

Heterotopic heart transplant to rule out hemodynamic causes of HF-induced tumor growth

To rule out if hemodynamic impairment, such as hypoperfusion (due to decreased systolic blood

pressure and forward failure) or congestion (backward failure) in response to elevated filling

pressures (with liver and/or gut congestion) would explain our initial observations, we conducted a

second experiment. Herein, we again inflicted MI in APCmin mice (donors), but now, after 1 week,

performed heterotopic heart transplantation (HTx), transplanting the (extra) infarcted heart (or

sham-operated heart) into the cervical region of the recipient APCmin mice. This resulted in APCmin

mice (receivers) with a normal native heart in situ, ensuring normal circulation, but with an additional

heart, either with or without HF (according to whether the donor was subjected to MI or to sham

operation). In order to assess the severity of MI, prior to transplantation, cardiac function was

measured using echocardiography (SFig. 2a); cardiac parameters are presented in supplemental table

3. Donor mice with HF exhibited severely impaired cardiac function prior transplantation: LVEF 61%

vs. 27%; P<0.0001; SFig. 2b), which was comparable to our first experiments.

9

Six weeks post-transplantation, we performed cardiac phenotyping of the transplanted (failing)

heart, as described in the initial experiment. We again observed significant differences regarding

fibrosis and inflammation (SFig. 2c & Fig. 2e-h). Also, an increased expression of NPPA (2.4-fold

increase P<0.05; comparing sham-operated transplanted hearts vs HF transplanted hearts).

Interestingly, although the changes between the initial and HTx experiments were both directionally

comparable and significant for all tested markers, the overall changes were clearly less strong in the

HTx model as compared to the initial model. For instance, the changes in ANP were ~10 fold less. This

led us to assume that the transplanted hearts must be considered as unloaded hearts, and as a

result, produce and secrete less remodeling factors. Importantly, the endogenous hearts of the

recipient mice demonstrated no loss of function, as shown in supplemental table 3, proving that our

experimental design ensured normal systemic circulation.

Assessment of tumor growth in mice with transplanted hearts with HF

A similar sacrifice protocol as described before was performed and intestinal polyps were counted

and measured. Significantly more (55 vs 39; Fig. 3a; P<0.01) and larger polyps (1.97mm vs. 1.51mm;

Fig. 3b; P<0.01) were observed in mice receiving an HF heart transplant compared to mice receiving a

sham-operated heart transplant. Again, the tumor load was calculated and in this model we also

observed a significant 2.4-fold increase (Fig. 3c; P<0.0001), comparable to the initial experiment. To

further validate our findings we again showed increased numbers of KI-67 positive cells (54% vs. 44%;

P<0.05, Fig. 3e). Finally, we also observed a significantly larger spleen (1.4 fold increase; Fig. 3d;

P<0.05) in mice that received an HF heart transplant, compared to sham. The other organs did not

differ in weight between groups (supplemental table 4).

10

Association between cardiac remodeling and tumor growth

Also in this experimental model, we performed linear regression analyses to determine the

association between tumor load and indices of cardiac remodeling. Herein, we found comparable

associations between tumor load and fibrosis (β=0.89, P<0.01; Fig. 3g), and between tumor load and

LVEF (β=-0.60, P=0.002; Fig. 3h).

Exploration strategy to identify HF-specific secreted proteins capable of enhancing tumor growth

Our findings from the initial and HTx experiments suggest that the enhanced tumor formation in

mice with HF is independent from hemodynamic factors, and may be explained by secreted factors

from the failing hearts. We present our hypothesis based upon the in vivo findings in figure 4,

proposing that proteins may be excreted from failing hearts into the bloodstream and affect

peripheral organs, here in particularly the intestines, resulting in enhanced tumor development and

growth.

We made use of the abundant biomedical literature that is available and set out to identify proteins

secreted into the bloodstream in response to MI, focusing on articles reporting ‘shotgun proteomic’

approaches of plasma samples post-MI. Secondly, we ascertained which epitopes exist on intestinal

tissue, that theoretically can be bound or activated by these ‘cardiac’ proteins. A detailed description

of this methodology is provided in the supplemental methods. Supplemental figure 3 summarizes our

literature findings: we identified 5 proteins with potential importance: alpha-1-antitrypsin

(SerpinA1), alpha-1-antichymotrypsin (SerpinA3), fibronectin (FN), cerulopasmin (CP) and

paraoxonase 1 (PON1).

11

Validation of myocardial gene expression of identified secreted factors

First, to validate whether the identified candidates are up-regulated in the left ventricular tissue of

mice with HF (compared to sham), we performed quantitative PCR (qPCR), both for our initial and for

the HTx studies. In our discovery study, we demonstrate a significantly increased LV expression for all

five genes (sham vs. HF; all P<0.05). Then, we assayed the same genes in the HTx study, validating 3

out of the 5 genes: SerpinA3, FN, and PON1 (SFig. 4).

Proliferation of HT-29 cells due to addition of the secreted proteins

To investigate whether the identified proteins actually exert proliferative effects on colorectal cells,

we performed in vitro experiments with HT-29 cells, which is commonly used cell type of colorectal

cancer. HT29 cells were grown and seeded in DMEM medium, and prior to the experiment starved

and co-treated with Suramin, to halt cell proliferation. After 48 hours, cells were treated with either

0.1% FCS (negative control), 10% FCS (positive control) or the identified candidate proteins. To assess

proliferation, we used three assays: 1) We measured PNCA, a gene indicative of cell proliferation and

expressed the results as ratio to the positive and negative controls and we performed staining with 2)

EDU and 3) KI-67+. SerpinA3 and SerpinA1 resulted in increased proliferation, 33% and 21%

respectively. The other proteins provoked no significant differences in proliferation rate compared to

low FCS concentrations (SFig. 5a). We confirmed the proliferative effects of SerpinA3 by showing

increased PCNA expression, and EdU+ and Ki67 staining after addition of different SerpinA3

concentration to HT-29 cells - Serpina3 consistently demonstrated proliferation evidenced by these

different assays (SFig.5b-d).

12

Activation of growth pathways in colon cells in response to SerpinA3

To further elucidate potential cellular mechanisms by which SerpinA3 results in proliferation we

stimulated HT-29 cells with SerpinA3 as described above and performed Western Blot analyses for

the Erk1/2 and Akt signal transduction pathways that have been linked to SerpinA3 stimulation17. Akt

phosphorylation and a downstream target, ribosomal protein S6 (rpS6), that marks cell growth, were

both significantly phosphorylated upon SerpinA3 stimulation, whereas Erk1/2 was not targeted

(western blots and a simplified graphical scheme are displayed in Figure 5), suggesting that SerpinA3

provokes tumor growth via the Akt pathway.

Cardiac secreted proteins are elevated in plasma of human patients with HF

To explore whether we could translate our in vivo and in vitro findings to the human situation, we

measured all five candidate proteins in 180 healthy subjects, enrolled in the Prevention of Renal and

Vascular ENd-stage Disease (PREVEND) study14 and in 101 chronic heart failure (CHF) patients15. We

used human ELISA assays that have been validated (supplemental methods section). Baseline

characteristics are presented in table 1. Indeed, we demonstrated that plasma levels of all 5 proteins

were 30-100% upregulated (all P<0.001) in patients with HF, which underscores that these proteins

indeed are related to the presence of HF (Fig. 6).

Cardiac and inflammatory markers are associated with new-onset cancer

In 8319 subjects of the PREVEND, a community-based cohort study, with middle-aged participants,

we evaluated the predictive value of cardiac markers including NT-proBNP, an established biomarker

of HF, and hs-TnT, and MR-pro ANP. We previously have shown that NT-proBNP in this cohort

strongly predicts new onset HF18. In addition to these markers we investigated inflammation related

proteins including CT-proET-1, MR-proADM, hs-CRP and PCT. Further neuro-endocrine markers as

13

aldosterone, renin and galectin-3 were analyzed. New-onset cancer cases were provided through the

nationwide network and registry of histo- and cytopathology in the Netherlands (PALGA) 19, and were

reviewed and adjudicated by an independent committee. During a median follow-up of 11.5 years,

1132 (13.1%) subjects were diagnosed with cancer, of which 132 (11.7%) with colorectal cancer. We

performed Kaplan Meier analysis for both all-cause cancer and colorectal cancer according to tertiles

of NT-proBNP levels. Compared to subjects in tertile 1 (low NT-proBNP), subjects in tertiles 2 and 3

(higher and highest NT-proBNP) had an increased risk of developing all-cause cancer, and also

colorectal cancer (both P<0.0001) (Fig. 7). Next, we performed cox-proportional hazard regression

analyses to adjust for common risk factors of new onset cancer. These data show that the association

of cardiac and inflammatory biomarkers with all-cause cancer remains, also after adjustment for age,

sex, smoking status and body mass index (BMI) (Table 2), while the significant association for

colorectal cancer was lost after correction, possibly due to limited power (STable 5). The neuro-

endocrine biomarkers were not associated with all-cause cancer or colon cancer, except for galectin-

3 (only unadjusted). To extrapolate these findings to other types of cancer, we repeated these

analyses regarding breast, lung, skin, urological, hematological, male and female reproductive system

cancers (STable 6-12). Both cardiac and inflammation biomarkers were (unadjusted and adjusted)

associated with new onset lung and male reproductive system cancer, and cardiac biomarkers only

for female reproductive system cancers. Clearly, these associations are exploratory, but provide

additional suggestions for the relation between HF and incident cancer.

14

DISCUSSION

We demonstrate for the first time a causal relationship between HF and tumor growth. First, we

explored this novel paradigm by creating MI-induced HF in a murine model of precancerous polyps,

and experimental HF resulted in increased tumor formation and accelerated tumor growth. Second,

we corroborated our results in an independent model in an independent laboratory, using a

heterotopic murine heart transplantation model, and validated our initial results, while ruling out

hemodynamic impairment as a cause. To probe our hypothesis that cardiac secreted factors of the

failing heart could be responsible, we conducted a literature search from databases from myocardial

secreted proteins, and connected the candidates to databases of proteins previously associated with

new onset colorectal cancer. We then validated the candidate secreted proteins both in vitro, in vivo,

and in human studies, and identify SerpinA3 as the most robust and promising culprit. We explored

the potentially contributory involvement of inflammatory factors. Finally, we provide human

validation by showing that cardiac markers including NT-proBNP and inflammatory markers including

CRP were associated with prediction of new onset cancer.

In the past decade, significant progress has been made regarding the development of HF in cancer

patients. Recent position statements of the AHA 20 and the ESC 21 have described this field in large

detail. But the reverse, i.e. cancer in the setting of heart failure received far less attention. Lately,

epidemiological data have emerged that HF patients are at higher risk to be diagnosed with and/or to

die from cancer, compared to age-matched subjects without HF8. This fits the modern face of HF, in

which mortality is no longer solely due to CV death3, but is largely due to other causes. In a study

with nearly 800 HF patients who were prospectively followed for 5 years, cancer (next to stroke) was

the second most important predictor of mortality, with a 2.5-fold increased risk in HFpEF patients 4.

Comparable results were observed in another study comprising 2,843 patients with HFpEF and 6,599

with HFrEF, with 2-year follow-up7. In a case-control study, pairing 961 patients with incident HF to

subjects without HF, HF patients had a 68% higher risk for incident cancer, which appeared to

15

progressively increase over time8. Supporting evidence comes from another study that described that

circulating levels of the NT-proBNP and hs-TnT, markers of cardiac stretch and injury, were both

elevated in patients with cancer, already prior to induction of any cardiotoxic anticancer therapy.

These markers were related to all-cause mortality, suggesting that the presence of subclinical

myocardial damage may be present in patients with cancer22. Prior to this report, one small

prospective study reported that NT-proBNP predicts future cancer development in patients with

coronary artery disease23. Independent, circumstantial evidence, which strengthens our hypothesis,

was provided by large randomized clinical trials with HF medication, the Studies of Left Ventricular

Dysfunction (SOLVD)24 trial (enalapril vs. placebo) and the Candesartan in Heart failure - Assessment

of moRtality and Morbidity (CHARM)25 trial (Candesartan vs. placebo), where a signal towards more

cancer was observed in patients on active treatment. However, the excess incidence of cancer in this

setting was suggested to be the potential consequence of competing risk, i.e. if one reduces HF

related mortality, it has been assumed cancer will have risen and that cancer related mortality “takes

over” from HF26.

Our data suggest something different: i.e. that the presence of a failing heart per se might contribute

to tumor progression and formation. Clearly, tumorigenesis is a complex, multistage process,

characterized by several features, including resistance to growth inhibitors, autonomous

proliferation independent from normal growth factor control, replication without limit, evasion of

apoptosis, and tissue invasion, formation of metastases, with supportive growth of matrix and

enhanced angiogenesis27. These different biological processes are influenced by numerous different

signal transduction pathways. Our identified proteins are involved in many of these processes as

discussed below, but given the complexity of this process, additional factors associated with HF likely

play a role in this process as well. We also speculate that indirect effects, e.g. via changes in the

immune system, may also play a role.

16

Of our panel of candidate proteins, SerpinA3 emerged consistently as a factor that is increased in HF,

with proliferative effects in vitro. SerpinA3, known to be associated with cardiac remodeling and

matrix turnover, has also been investigated in relation with cardiac disease in a study consisting of

twenty healthy individuals and 224 CHF patients. SerpinA3 levels were comparable to our study and

also significantly elevated in HF patients compared with controls, and was associated with long-term

mortality28. In another study, using explanted human hearts comparing HF patients and donor

humans, SerpinA3 was also identified as one of strongest regulated genes 29. Further, several studies

link mineralocorticoid receptor antagonists (MRAs) to SerpinA3. Early treatment with the MRA

spironolactone improved skeletal muscle pathology in mouse models. When comparing the MR

signaling with global gene expression analyses in these skeletal muscles from mice with or without

MRA, serpinA3 was one of the genes that was more than 2-fold downregulated with MRA

treatment30. Interestingly, in a murine model of MR cardiac overexpression SerpinA3 was up-

regulated, and in in vitro studies were H9C2 cells were treated for 24 hours with aldosterone

SerpinA3 also increased31. SerpinA3 has been identified as a marker of colorectal metastasis32.

Further, elevated SerpinA3 is associated with a worse prognosis and increased migration and

invasion in melanoma33. Finally, both SerpinA3 and A1 have been proposed as markers of tumor

progression of adenoma into carcinoma 34 are also linked with breast, prostate and liver cancer.

Collectively, SerpinA3 could directly link HF and cancer via its pleiotropic effects, as it acts as an acute

phase protein, and is related to systemic inflammation35,36. We now extend these previous findings by

showing that SerpinA3 is produced in murine HF, is elevated in plasma of human HF patients, and

stimulates proliferation of colon tumor cells via an Akt-dependent pathway.

FN is an established central player in the cardiac extracellular matrix (ECM), during cardiac

remodelling and other physiological circumstances, and is strictly regulated, e.g. by the RAAS system.

Like in HF, ECM is important for tumor integrity and growth. FN has been implicated as a pro-

angiogenic factor. FN acts on tumor-associated fibroblasts and enhances tumor growth and

vasculature, stimulating expression of pro-angiogenic factors37. Indeed, FN has been proposed as a

17

possible target to prevent angiogenesis due to its interaction with integrin receptors38. In patients,

circulating FN levels have been significantly elevated in patients with colorectal cancer than in

controls, and serum FN levels rose further with cancer progression 39. PON1 expression and secretion

in lung cancer has been shown pro-oncogenic and supported metastatic progression by decreasing

G1/S ratio and cell senescence40. Serum PON1 activity is increased in patients with colon cancer

compared to control subjects41. In a large Finnish registry of nearly 40.000 subjects the overall

incidence of cancer was positively associated with serum CP levels. The strongest association was

observed in lung cancer and in males42. In breast cancer studies, it was hypothesized that it can be

used as a biomarker of disease progression or as a marker for response to therapy 43. Recently SARI, a

direct CP target has been studied in colon cancer and inhibits angiogenesis and tumor growth44.

In addition to the 5 proteins that were discovered using our bio-informatical approach, the

importance of inflammation as shared pathway in HF and cancer must be considered, especially since

the interleukin 1β antibody canakinumab was recently shown to decrease new onset cancer in the

CANTOS study45. In our mouse studies, we also observed clear increases in IL-6 levels, and strong

associations between the inflammatory markers CRP and MR-proADM and incident cancer in the

human (PREVEND) study.

The relation between cardiac secreted markers and the levels that can be measured systemically is

complex. We show that for cardio-specific proteins such as ANP this is straightforward: increased

cardiac production is reflected by increases in plasma levels (SFig. 1f-h). There are no well-validated

ELISA assays for mouse plasma for our candidate proteins. We have measured the HF-marker NT-

proANP, the fibrosis marker TIMP1 and the inflammatory marker IL-6 for which validated mouse

ELISA assays are available. The relations appear straightforward, although we cannot claim that this

will also be true for the other proteins.

18

We acknowledge this as a limitation of our study, and although these results do support the idea that

elevated cardiac expression can confer effects on distant organs and tumors via plasma proteins,

further research will be required to solve this complex inter-tissue interaction.

Further, in this study, we studied a post MI model of HF, which is characterized by ischemia, cell

death, and a large fibrotic scar, which likely is associated with a distinct proteomic signature. HF due

to other (non-ischemic) etiologies will likely be characterized by another set of secreted proteins,

which may cause differential effects on tumor growth. Future studies should address if etiology of HF

is important in this regard.

Another form of (acute) heart failure is stress cardiomyopathy (also called Tako-Tsubo). An increased

prevalence of malignancies has been observed by Sattler K et al. 46 in patients with Tako-Tsubo

cardiomyopathy, both at initial diagnosis and during follow-up. It has been hypothesized that this

may originate from a common pathway of the two conditions, especially the catecholamine excess in

CV disease and cancer. Yalta et al.47 proposed that malignant diseases (or yet undiagnosed

malignancy) might be the trigger for Tako-Tsubo. However, no molecular pathways have yet been

discovered to definitely link these two disease modalities to each other.

Clearly, cancer as a co-morbid condition in HF is very serious, and not surprisingly, cancer increases

the mortality in HF6. Should our data implicate that cancer surveillance might be incorporated in the

management of HF patients? In an additional exploratory analysis regarding the screening for new

onset cancer, we investigated whether subjects enrolled in the PREVEND study who developed HF

before the age of 55 (which in the Netherlands is the age where surveillance for colon cancer starts)

were in fact at higher risk for developing cancer, compared to patients who develop HF after the age

of 55. We observed that patients who developed HF before the age of 55 have a significantly higher

risk to develop cancer compared to those without HF (HR 2.43 [1.33-4.43]; P=0.004), whereas there

was no difference between patients with or without HF after the age of 55 (HR 1.05 [0.84-1.33];

P=0.636). This observation suggests (but does by no means prove) that people who are not yet

19

eligible for screening because of ‘young age’ might benefit from early screening for colorectal cancer

once they develop HF.

In summary, our data show that the presence of HF is associated with increased formation and

accelerated tumor growth in a mouse model of colon polyps. We identify several myocardial

markers, with established effects on tumor growth, that we demonstrate are chronically elevated in

human HF. One of our most promising candidate proteins might also be targeted for therapy, e.g.

using MRAs. Our preclinical and clinical data strengthen the link between HF and incident cancer. We

propose that this might have implications for screening programs for new onset cancer.

Strengths and limitations:

We conducted two independent laboratory studies with mice, using different operating techniques

and in different laboratories. All analyses were done blinded and we did not exclude animals. A such,

this study confirms to the ARRIVE criteria for stringent methodology in animal studies 48. Clearly, the

mouse APCmin model is a model of colon cancer formation, with all limitations, and does not allow

extrapolation to other tumor types. Our current proteomic approach was not based upon the in vivo

models used in this study, but rather based upon post-MI proteomic studies published by others.

Since the effects of the initial study and the heart transplantation study were comparable with

respect to tumor growth, additional proteomic analyses might be instrumental in identifying which

proteins exert the strongest effects on tumor growth.

Funding sources

This work was supported by the Netherlands Heart Foundation, Cardiovasculair Onderzoek

Nederland (CVON)-Talent program 2017 to dr. Meijers and the CVON-DOSIS, grant 2014-40, to Dr. de

Boer and the Innovational Research Incentives Scheme program of the Netherlands Organization for

Scientific Research (NWO VIDI, grant 917.13.350, to Dr. de Boer).

20

Acknowledgements

We would to acknowledge the expert technical assistance of Martin Dokter, Katharina Heinz,

Bernhard Haubner, Kees van de Kolk, Marloes Schouten, Silke U. Oberdorf-Maass, Elles Screever and

Moniek Smith.

Potential conflicts of interest

The UMCG, which employs a number of the authors, received research grants and consultancy fees

from Roche, Bristol Myers Squibb, Pfizer, Trevena, Thermofisher GmbH, and Sphingotec GmbH, for

work done by UMCG employers. Dr. de Boer received speaker fees from Novartis and is a minority

shareholder of scPharmaceuticals, Inc.

21

References

1. Ponikowski P, Anker SD, AlHabib KF, Cowie MR, Force TL, Hu S, Jaarsma T, Krum H, Rastogi V, Rohde LE, Samal UC, Shimokawa H, Budi Siswanto B, Sliwa K, Filippatos G. Heart failure: preventing disease and death worldwide. ESC Hear Fail. 2014;1:4–25.

2. Baldi I, Azzolina D, Berchialla P, Gregori D, Scotti L, Corrao G. Comorbidity-adjusted relative survival in newly hospitalized heart failure patients: A population-based study. Int J Cardiol. 2017;243:385–388.

3. Lund LH, Donal E, Oger E, Hage C, Persson H, Haugen-Löfman I, Ennezat PV, Sportouch-Dukhan C, Drouet E, Daubert JC, Linde C. Association between cardiovascular vs. Non-cardiovascular co-morbidities and outcomes in heart failure with preserved ejection fraction. Eur J Heart Fail. 2014;16:992–1001.

4. Tribouilloy C, Rusinaru D, Mahjoub H, Soulière V, Lévy F, Peltier M, Slama M, Massy Z. Prognosis of heart failure with preserved ejection fraction: A 5 year prospective population-based study. Eur Heart J. 2008;29:339–347.

5. Murphy KT. The pathogenesis and treatment of cardiac atrophy in cancer cachexia. Am J Physiol - Hear Circ Physiol. 2016;310:H466–H477.

6. Koene RJ, Prizment AE, Blaes A, Konety SH. Shared risk factors in cardiovascular disease and cancer. Circulation. 2016;133:1104–1114

7. Ather S, Chan W, Bozkurt B, Aguilar D, Ramasubbu K, Zachariah AA, Wehrens XHT, Deswal A. Impact of noncardiac comorbidities on morbidity and mortality in a predominantly male population with heart failure and preserved versus reduced ejection fraction. J Am Coll Cardiol. 2012;59:998–1005.

8. Hasin T, Gerber Y, McNallan SM, Weston SA, Kushwaha SS, Nelson TJ, Cerhan JR, Roger VL. Patients with heart failure have an increased risk of incident cancer. J Am Coll Cardiol. 2013;62:881–886.

9. Banke A, Schou M, Videbaek L, Moller JE, Torp-Pedersen C, Gustafsson F, Dahl JS, Kober L, Hildebrandt PR, Gislason GH. Incidence of cancer in patients with chronic heart failure: A long-term follow-up study. Eur J Heart Fail. 2016;18:260–266.

10. Hasin T, Gerber Y, Weston SA, Jiang R, Killian JM, Manemann SM, Cerhan JR, Roger VL. Heart Failure After Myocardial Infarction Is Associated With Increased Risk of Cancer. J Am Coll Cardiol. 2016;68:265–271.

11. Moser AR, Luongo C, Gould KA, McNeley MK, Shoemaker AR, Dove WF. ApcMin: A mouse model for intestinal and mammary tumorigenesis. Eur J Cancer. 1995;31:1061–1064.

12. Booij HG, Yu H, De Boer RA, Van De Kolk CWA, Van De Sluis B, Van Deursen JM, Van Gilst WH, Silljé HHW, Westenbrink BD. Overexpression of A kinase interacting protein 1 attenuates myocardial ischaemia/reperfusion injury but does not influence heart failure development. Cardiovasc Res. 2016;111:217–226.

13. Cannon M V, Sillje HH, Sijbesma JW, Vreeswijk-Baudoin I, Ciapaite J, van der Sluis B, van Deursen J, Silva GJ, de Windt LJ, Gustafsson JÅ, van der Harst P, van Gilst WH, de Boer RA. Cardiac LXR protects against pathological cardiac hypertrophy and dysfunction by enhancing glucose uptake and utilization. EMBO Mol Med. 2015;7:1229–1243.

22

14. Hillege HL, Janssen WMT, Bak AAA, Diercks GFH, Grobbee DE, Crijns HJGM, Van Gilst WH, De Zeeuw D, De Jong PE. Microalbuminuria is common, also in a nondiabetic, nonhypertensive population, and an independent indicator of cardiovascular risk factors and cardiovascular morbidity. J Intern Med. 2001;249:519–526.

15. Schroten NF, Ruifrok WPT, Kleijn L, Dokter MM, Silljé HH, Heerspink JL, Bakker SJL, Kema IP, van Gilst WH, van Veldhuisen DJ, Hillege HL, de Boer RA. Short-term vitamin D3 supplementation lowers plasma renin activity in patients with stable chronic heart failure: An open-label, blinded end point, randomized prospective trial (VitD-CHF trial). Am Heart J. 2013;166:357–364.e2.

16. Meijers WC, van der Velde AR, Muller Kobold AC, Dijck-Brouwer J, Wu AH, Jaffe A, de Boer RA. Variability of biomarkers in patients with chronic heart failure and healthy controls. Eur J Heart Fail. 2017;19:357–365.

17. Yang G-D, Yang X-M, Lu H, Ren Y, Ma M-Z, Zhu L-Y, Wang J-H, Song W-W, Zhang W-M, Zhang R, Zhang Z-G. SERPINA3 promotes endometrial cancer cells growth by regulating G2/M cell cycle checkpoint and apoptosis. Int J Clin Exp Pathol. 2014;7:1348–58.

18. Brouwers FP, Van Gilst WH, Damman K, Van Den Berg MP, Gansevoort RT, Bakker SJL, Hillege HL, Van Veldhuisen DJ, Van Der Harst P, De Boer RA. Clinical risk stratification optimizes value of biomarkers to predict new-onset heart failure in a community-based cohort. Circ Heart Fail. 2014;7:723–731.

19. Casparie M, Tiebosch a TMG, Burger G, Blauwgeers H, van de Pol a, van Krieken JHJM, Meijer G a. Pathology databanking and biobanking in The Netherlands, a central role for PALGA, the nationwide histopathology and cytopathology data network and archive. Cell Oncol. 2007;29:19–24.

20. Lipshultz SE, Adams MJ, Colan SD, Constine LS, Herman EH, Hsu DT, Hudson MM, Kremer LC, Landy DC, Miller TL, Oeffinger KC, Rosenthal DN, Sable CA, Sallan SE, Singh GK, Steinberger J, Cochran TR, Wilkinson JD. Long-term cardiovascular toxicity in children, adolescents, and young adults who receive cancer therapy: Pathophysiology, course, monitoring, management, prevention, and research directions: A scientific statement from the American Heart Association. Circulation. 2013;128:1927–1955.

21. Zamorano JL, Lancellotti P, Rodriguez Muñoz D, Aboyans V, Asteggiano R, Galderisi M, Habib G, Lenihan DJ, Lip GYH, Lyon AR, Lopez Fernandez T, Mohty D, Piepoli MF, Tamargo J, Torbicki A, Suter TM, Achenbach S, Agewall S, Badimon L, Barón-Esquivias G, Baumgartner H, Bax JJ, Bueno H, Carerj S, Dean V, Erol Ç, Fitzsimons D, Gaemperli O, Kirchhof P, Kolh P, Nihoyannopoulos P, Ponikowski P, Roffi M, Vaz Carneiro A, Windecker S, Minotti G, Cardinale D, Curigliano G, De Azambuja E, Dent S, Ero C, Ewer MS, Farmakis D, Fietkau R, Kohl P, McGale P, Ringwald J, Schulz-Menger J, Stebbing J, Steiner RK, Szmit S. 2016 ESC Position Paper on cancer treatments and cardiovascular toxicity developed under the auspices of the ESC Committee for Practice Guidelines. Eur. Heart J. 2016;37:2768–2801.

22. Pavo N, Raderer M, Hulsmann M, Neuhold S, Adlbrecht C, Strunk G, Goliasch G, Gisslinger H, Steger GG, Hejna M, Kostler W, Zochbauer-Muller S, Marosi C, Kornek G, Auerbach L, Schneider S, Parschalk B, Scheithauer W, Pirker R, Drach J, Zielinski C, Pacher R. Cardiovascular biomarkers in patients with cancer and their association with all-cause mortality. Heart. 2015;101:1874–1880.

23. Tunon J, Higueras J, Tarin N, Cristobal C, Lorenzo O, Blanco-Colio L, Martin-Ventura JL, Huelmos A, Alonso J, Acena A, Pello A, Carda R, Asensio D, Mahillo-Fernandez I, Lopez Bescos

23

L, Egido J, Farre J. N-Terminal Pro-Brain Natriuretic Peptide Is Associated with a Future Diagnosis of Cancer in Patients with Coronary Artery Disease. PLoS One. 2015;10:e0126741.

24. The SOLVD Investigators. Effect of enalapril on mortality and the development of heart failure in asymptomatic patients with reduced left ventricular ejection fractions. The SOLVD Investigattors. N Engl J Med. 1992;327:685–91.

25. Pfeffer MA, Swedberg K, Granger CB, Held P, McMurray JJ V, Michelson EL, Olofsson B, Ostergren J, Yusuf S, Pocock SJ, Committees CI and. Effects of candesartan on mortality and morbidity in patients with chronic heart failure: the CHARM-Overall programme. Lancet. 2003;362:759–766.

26. Pfeffer MA. Cancer in cardiovascular drug trials and vice versa: A personal perspective. Eur Heart J. 2013;34:1089–1094.

27. Hanahan D, Weinberg RA. The hallmarks of cancer. Cell. 2000;100:57–70.

28. Lok SI, Lok DJ, Van Der Weide P, Winkens B, Bruggink-André De La Porte PW, Doevendans PA, De Weger RA, Van Der Meer P, De Jonge N. Plasma levels of alpha-1-antichymotrypsin are elevated in patients with chronic heart failure, but are of limited prognostic value. Netherlands Heart J. 2014;22:391–395.

29. di Salvo TG, Yang K-C, Brittain E, Absi T, Maltais S, Hemnes A. Right Ventricular Myocardial Biomarkers in Human Heart Failure. J Card Fail. 2015;21:398–411.

30. Chadwick JA, Hauck JS, Lowe J, Shaw JJ, Guttridge DC, Gomez-Sanchez CE, Gomez-Sanchez EP, Rafael-Fortney JA. Mineralocorticoid receptors are present in skeletal muscle and represent a potential therapeutic target. FASEB J. 2015;29:4544–4554.

31. Latouche C, Sainte-Marie Y, Steenman M, Chaves PC, Naray-Fejes-Toth A, Fejes-Toth G, Farman N, Jaisser F. Molecular signature of mineralocorticoid receptor signaling in cardiomyocytes: From cultured cells to mouse heart. Endocrinology. 2010;151:4467–4476.

32. Long NP, Lee WJ, Huy NT, Lee SJ, Park JH, Kwon SW. Novel biomarker candidates for colorectal cancer metastasis: A meta-analysis of in vitro studies. Cancer Inform. 2016;15:11–17.

33. Zhou J, Cheng Y, Tang L, Martinka M, Kalia S, Zhou J, Cheng Y, Tang L, Martinka M, Kalia S. Up-regulation of SERPINA3 correlates with high mortality of melanoma patients and increased migration and invasion of cancer cells. Oncotarget. 2017;8:18712-18725.

34. Peltier J, Roperch JP, Audebert S, Borg JP, Camoin L. Quantitative proteomic analysis exploring progression of colorectal cancer: Modulation of the serpin family. J Proteomics. 2016;148:139–148.

35. Bode JG, Albrecht U, Häussinger D, Heinrich PC, Schaper F. Hepatic acute phase proteins - Regulation by IL-6- and IL-1-type cytokines involving STAT3 and its crosstalk with NF-κB-dependent signaling. Eur J Cell Biol. 2012;91:496–505.

36. Surinova S, Choi M, Tao S, Schuffler PJ, Chang C-Y, Clough T, Vyslouzil K, Khoylou M, Srovnal J, Liu Y, Matondo M, Huttenhain R, Weisser H, Buhmann JM, Hajduch M, Brenner H, Vitek O, Aebersold R. Prediction of colorectal cancer diagnosis based on circulating plasma proteins. EMBO Mol Med. 2015;7:1166–1178.

37. Limoge M, Safina A, Truskinovsky AM, Aljahdali I, Zonneville J, Gruevski A, Arteaga CL BA. Tumor p38MAPK signaling enhances breast carcinoma vascularization and growth by

24

promoting expression and deposition of pro-tumorigenic factors. Oncotarget. 2017;8:61969–61981.

38. Murphy PA, Begum S, Hynes RO. Tumor angiogenesis in the absence of fibronectin or its cognate integrin receptors. PLoS One. 2015;10:e0120872.

39. Saito N, Nishimura H, Kameoka S. Clinical significance of fibronectin expression in colorectal cancer. Mol Med Rep. 2008;1:77–81.

40. Aldonza MBD, Son Y, Sung H-J, Mo Ahn J, Choi Y-J, Kim Y-I, Cho S, Cho J-Y, Aldonza MBD, Son YS, Sung H-J, Mo Ahn J, Choi Y-J, Kim Y-I, Cho S, Cho J-Y. Paraoxonase-1 (PON1) induces metastatic potential and apoptosis escape via its antioxidative function in lung cancer cells. Oncotarget. 2017;8:42817–42835.

41. Afsar CU, Gunaldi M, Okuturlar Y, Gedikbasi A, Tiken EE, Kahraman S, Karaca F, Ercolak V, Karabulut M. Paraoxonase-1 and arylesterase activities in patients with colorectal cancer. Int J Clin Exp Med. 2015;8:21599–21604.

42. Knekt P, Aromaa A, Maatela J, Rissanen A, Hakama M, Aaran RK, Nikkari T, Hakulinen T, Peto R, Teppo L. Serum ceruloplasmin and the risk of cancer in Finland. Br J Cancer. 1992;65:292–296.

43. Schapira D V., Schapira M. Use of ceruloplasmin levels to monitor response to therapy and predict recurrence of breast cancer. Breast Cancer Res Treat. 1983;3:221–224.

44. Dai L, Cui X, Zhang X, Cheng L, Liu Y, Yang Y, Fan P, Wang Q, Lin Y, Zhang J, Li C, Mao Y, Wang Q, Su X, Zhang S, Peng Y, Yang H, Hu X, Yang J, Huang M, Xiang R, Yu D, Zhou Z, Wei Y, Deng H. SARI inhibits angiogenesis and tumour growth of human colon cancer through directly targeting ceruloplasmin. Nat Commun. 2016;7:11996.

45. Ridker PM, MacFadyen JG, Thuren T, Everett BM, Libby P, Glynn RJ, Ridker P, Lorenzatti A, Krum H, Varigos J, Siostrzonek P, Sinnaeve P, Fonseca F, Nicolau J, Gotcheva N, Genest J, Yong H, Urina-Triana M, Milicic D, Cifkova R, Vettus R, Koenig W, Anker SD, Manolis AJ, Wyss F, Forster T, Sigurdsson A, Pais P, Fucili A, Ogawa H, Shimokawa H, Veze I, Petrauskiene B, Salvador L, Kastelein J, Cornel JH, Klemsdal TO, Medina F, Budaj A, Vida-Simiti L, Kobalava Z, Otasevic P, Pella D, Lainscak M, Seung KB, Commerford P, Dellborg M, Donath M, Hwang JJ, Kultursay H, Flather M, Ballantyne C, Bilazarian S, Chang W, East C, Everett B, Forgosh L, Glynn R, Harris B, Libby P, Ligueros M, Thuren T, Bohula E, Charmarthi B, Cheng S, Chou S, Danik J, McMahon G, Maron B, Ning MM, Olenchock B, Pande R, Perlstein T, Pradhan A, Rost N, Singhal A, Taqueti V, Wei N, Burris H, Cioffi A, Dalseg AM, Ghosh N, Gralow J, Mayer T, Rugo H, Fowler V, Limaye AP, Cosgrove S, Levine D, Lopes R, Scott J, Thuren T, Ligueros M, Hilkert R, Tamesby G, Mickel C, Manning B, Woelcke J, et al. Effect of interleukin-1β inhibition with canakinumab on incident lung cancer in patients with atherosclerosis: exploratory results from a randomised, double-blind, placebo-controlled trial. Lancet. 2017;390:1833–1842.

46. Sattler K, El-Battrawy I, Lang S, Zhou X, Schramm K, Tülümen E, Kronbach F, Röger S, Behnes M, Kuschyk J, Borggrefe M, Akin I. Prevalence of cancer in Takotsubo cardiomyopathy: Short and long-term outcome. Int J Cardiol. 2017;238:159–165.

47. Yalta K, Yalta T. Cancer and takotsubo cardiomyopathy: More questions than answers. Int J Cardiol. 2017;242:13.

48. Kilkenny C, Browne WJ, Cuthill IC, Emerson M, Altman DG. Improving bioscience research reporting: The arrive guidelines for reporting animal research. PLoS Biol. 2010;6:e1000412.

25

26

Figure Legends:

Figure 1. Cardiac phenotype of the failing heart compared to the sham operated animals a. MRI

scan (sagittal & ventral view) of the heart, both sham and MI operated, and the Masson’s trichome

staining. b. Left ventricular (LV) ejection fraction of mice with and without HF. c. The abundance of

cardiac fibrosis present in mice with and without HF. d-h. Represent the fold change of expression

levels of different genes in the myocardial tissue d. NPPA, e. Interleukin-6, f. Collagen 3a1, g. Cluster

of Differentiation 68, h. Galectin-3. Data are presented as mean ± SEM ***, p<0.001; ****,

p<0.0001.

Figure 2. Effect of a failing heart or sham after 6 weeks regarding intestinal tumorigenesis a. Crude

number of intestinal polyps. b. The average size of the tumors (smallest diameter was measured) c.

Calculated tumor load d. spleen mass measured in milligram and corrected for tibia length e. KI-67

positive cell count f. Representative depiction of KI-67 staining g. Association between tumor and

load fibrosis h. Association between tumor load and LVEF. Data are presented as means ± SEM *,

p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001. Associations are presented as ß.

Figure 3. Effect of a failing heart (or sham-operated heart) 6 weeks after transplantation on

intestinal tumor growth a. Crude number of intestinal polyps. b. The average size of the tumors

(smallest diameter was measured) c. Calculated tumor load d. Spleen mass measured in milligram

and corrected for tibia length e. Representative depiction of KI-67 staining f. Representative

overview of the KI-67 staining g. Association between tumor load and fibrosis h. Association between

tumor load and LVEF. Data are presented as means ± SEM *, p<0.05; **, p<0.01; ***, p<0.001; ****,

p<0.0001. Associations are presented as ß.

Figure 4. Graphical depiction of the hypothesis of ‘secreted proteins by the failing heart that

enhance tumorigenesis’

27

Figure 5. Growth pathways in colon cells in response to SerpinA3 a. Western blot analysis of total

and phosphorylated Akt, rpS6 and Erk1/2. b. Quantification of Akt, rpS6 and ERK 1/2 phosphorylation

Data are presented as means ± SEM *, p<0.05; c. Scheme of Akt and rpS6 phosphorylation that marks

cell growth, due to SerpinA3 stimulation in HT-29 cells, whereas Erk1/2 was not targeted.

Figure 6. Human plasma levels of the identified candidate proteins. Plasma concentration of

candidate factors in plasma from healthy subjects and in plasma from patients with HF. Data are

presented as means ± SEM. ****, p<0.0001.

Figure 7. Cumulative incidence of cancer per NT-proBNP level (tertiles)

a. Cumulative incidence curves showing new-onset cancer in patients stratified by NT-proBNP

tertiles. b. Cumulative incidence curves showing new-onset colorectal cancer in patients stratified

by NT-proBNP tertiles. Log-rank test for both plots: p <0.0001.

28

Table 1. Baseline characteristics of the HF patients and healthy subjects (PREVEND)

Variables HF Patients (n=101) PREVEND (n=8319)

Age, years (mean, SD) 64±10 49±13

Male, n (%) 93 (93) 4127 (50)

NYHA class, II/III 89/11 NA

Diabetes mellitus, n (%) 14 (14) 131 (2)

Current smoking, n (%) 22 (22) 3686 (45)

LVEF, % (mean, SD) 35±8 NA

Systolic blood pressure, mm Hg (mean, SD) 118±18 129±20

Diastolic blood pressure, mm Hg (mean, SD) 72±12 74±10

Serum creatinine, μmol/L (mean, SD) 90±18 84±20

29

Table 2. Hazard ratio for new onset cancer per biomarker doubling.

Unadjusted Adjusted for model 1HR 95% CI P-value HR 95% CI P-value

New onset cancer

Cardiac markersNT-proBNP 1.39 1.32-1.46 <0.001 1.06 1.00-1.12 0.046hs-TnT 1.61 1.52-1.69 <0.001 1.10 1.02-1.19 0.018MR-proANP 1.80 1.66-1.95 <0.001 1.11 1.01-1.22 0.027

Neuro-endocrineAldosterone 0.94 0.85-1.05 0.277Renin 1.05 0.99-1.10 0.075Galectin-3 1.69 1.50-1.90 <0.001 1.05 0.91-1.21 0.534

InflammationCT proET-1 1.51 1.37-1.68 <0.001 1.11 1.00-1.23 0.040MR-proADM 2.47 2.17-2.81 <0.001 1.22 1.06-1.39 0.004hs-CRP 1.19 1.14-1.23 <0.001 1.08 1.04-1.13 <0.001PCT 1.40 1.29-1.53 <0.001 1.07 0.96-1.21 0.228

Model 1: adjustment for age, smoking and BMI

Cox proportional hazard analyses of doubling per biomarker level and the risk for new-onset cancer

HR = Hazard ratio; CI = Confidence interval; BMI = Body mass index.

30