fCANCER RESEARCH54, 2378-2385, May 1, 1994)

ABSTRACT

The Utah Population Database allows examination of the genetic relationships among the 35.7% of all cancer cases in the state that havegenealogical records. Familial clustering of cancer Is measured by theGenealogical Index of Familiality and is examined by site, and within siteby age of onset, histology, and gender. Most cancer sites examined showexcess familiality for aN cases considered together. Subsets of individualswith certain characteristics showed unusually high levels of familial clustering, specifically lymphocytic leukemlas and especially chronic lymphocytic leukemia, lobular breast cancer, early lip cancer, early melanoma,and female lung cancers of alveolar/adenoma hlstology. These may represent characteristics of the most penetrant forms of inherited susceptibilities, those which are enhanced by environmental factors, chance aggregations, rare inherited syndromes, or a combination of these factors.

INTRODUCTION

Efforts to construct the UPDB3 began in 1974 (1—4).The UPDB isa unique combination of 3 Utah data sources including a genealogy,the Utah Cancer Registry, and the Utah death certificates. The core ofthis database is a genealogy of the Utah pioneers and their descendants with approximately one million individual records and 180,000family records. The genealogical data have been linked to cancerrecords reported by the state, and to state death certificates by uniqueconcordance of name and birth date. The death certificate files werenot utilized for this analysis. The UPDB has made possible theinvestigation of the genetic relationships between individuals withgiven cancer characteristics within a defined population.

We have previously analyzed the database to estimate the degree offamilial coaggregation of pairs of cancer sites using Utah CancerRegistry records from 1952 through 1982 (5—9).The record linking ofthe 10 most recent years of cancer data, through 1992, has nearlydoubled the available records. This provides a view of 25 years ofcancer in this population, allowing identification of cancer in morethan one generation in many cases. We present the most recentanalysis of the UPDB. Sample sizes are significantly increased, allowing comparison of more sites than have been described previously.

Much of the Utah population are descendants of the Utah pioneerswho were mostly members of the LDS (or Mormons). Most membersof the LDS abide by its proscriptions against consumption of coffee,tea, tobacco, and alcohol. LDS teachings encourage large, close-knitfamilies and strict sexual mores. The average number of children percouple during the second half of the 19th century was 7.7 children foronce-married women (3), and Mormons continue to have large families in this century. Polygamy was practiced by between 10 and 20%

Received 10/19/93; accepted 3/1/94.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1This research was supported by NIH Grants CA 48711, CN 05222, CA 42014, andCA 54936, and a Center of Excellence Award from the state of Utah.

2 To whom requests for reprints should be addressed, at Genetic Epidemiology, 420

ChipetaWay,Suite180,SaltLakeCity,UT84108.The abbreviations used are: UPDB, Utah Population Database; LDS, Church of Jesus

Christ of Latter Day Saints; GIF, Genealogical Index of Famiiality; MRG, median of thedistribution of the ratio of the case Genealogical Index of Familiality to the controlGenealogical Index of Familiality; CLL, chronic lymphocytic leukemia.

of male pioneers prior to 1890, which further increased the size ofkindreds. There were approximately 10,000 founding pioneers, whowere largely unrelated. Genetic studies of the population have shownthat it is genetically representive of a Northern European population(10), and due to a continued influx of immigrants has normal levels ofinbreeding (1 1).

Cancer rates in Utah for the years 1966 through 1990 are low whencompared to the United States Third National Cancer Survey (12).Table 1 summarizes United States and Utah comparisons for datafrom 1966 to 1973. Utah males have a lifetime risk of developingcancer of approximately 20%, and Utah females have a lifetime riskof approximately 24%. Utah has the lowest overall cancer incidencein the Surveillance Epidemiology and End Results program of theNational Cancer Institute. This is largely due to the much lower Utahrates for smoking-associated cancers; Utah has the lowest smokingrates of any state in national surveys. Cancer incidence comparisonsbetween Mormons and non-Mormons have been described (13).

SUBJECTS AND METhODS

The Utah Population Database. Genealogical work is a key aspect ofmembership in the LDS. Members are encouraged to keep family group sheetsfor their immediate family and ancestors, and to submit them to the Genealogical Society of Utah. The society was established in 1894, and maintaineda library of the 8 million-plus group sheets compiled, the source of the UPDB.All family group sheets which contained at least one birth, death, or marriagedate in Utah or along the pioneer trail were obtained from the Society (3). Dataon the individuals and marriages contained in these family group sheets havebeen computerized, and all the individuals have been linked into a genealogyrepresenting the Utah descendants of the Mormon pioneers (1, 2). Each groupsheet contains information on a full sibship and their parents, including foreach individual: name, place and date of birth, marriage date, date of death, andspouse name. In addition, place of death and spouse death date are alsoavailable for the parents on each group sheet. The names of the grandparentsare also included. Adopted individuals are noted, as are all births.

State-of-the-art methods for record linking have been implemented in the

UPDB. A probabilistic record linking program, LNX, developed by R. Kerber,has been used to link cancer registry records through 1992 to the genealogy.The linking algorithm used by LNX is very similar to that pioneered byNewcombe and Kennedy (14), and implemented in the generalized iterativerecord linking system program (15).

The Utah Cancer Registry was initiated in 1958, made statewide in 1966,and includes cancer records dating back to 1952. All cancers except basal andsquamous carcinomas of the skin must be reported to the Utah Cancer Registryby order of the Utah State Board of Health. In 1973, the registry became oneof 11 population-based registries of the Surveillance, Epidemiology, and EndResults program of the National Cancer Institute. The registry maintainsabstracts of clinical records and follow-up information on all cases. There arecurrently 125,904 entries in the registry files (through August 1992) representing 117,407 individuals, 41,940 of whom linked to the UPDB genealogy(35.7%). The percentage of linked records is lower for females than males, asname changes reduce the probability of successfully linking a woman's record.The percentage of linked records reflects the fact that not all cancer cases areindividuals who would be represented in the genealogy of the Utah pioneers.This is a result of 2 factors; a high percentage of cancer cases are non-Mormon,and the genealogy file is missing individual records for some Mormon incHviduals who have had cancer. The records are coded by site according to theInternational Classification of Diseases for Oncology (Ninth Revision) and

2378

Familiality of Cancer in Utah'

Lisa A. Cannon-Albright,2Alun Thomas, David E. Goidgar, Khosrow Gholami, Kerry Rowe, Matt Jacobsen,William P. McWhorter, and Mark H. SkolnickDivision of Hematology/Oncology, Deportment of Internal Medicine IL. A. C-A., W. P. M.J, and Deportment of Medical Informatics [D. E. G., K C., K It, M. J., M. H. S.],University of Utah School ofMedicine, Salt Lake City, Utah 84108, and School ofMathematical Sciences, University ofBath, Bath, England [A. T.]

on March 10, 2021. © 1994 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from

Table 1 Site-specific age-adjusted incidence ratios of Utah and the Unitedfemales and males (1.0: Utah rates = United States rates)aStates

forFemales

MalesSite

RatioSiteRatioLip

4.00Lip3.00Melanoma

1.59Melanoma1.40Thyroid1.20Testis1.20Connective

tissue 1.20Prostate1.20Acuteleukemia 1.10 Connectivetissue1.10Uterine

corpus 1.00 Non-Hodglrinslymphoma0.99Non-Hodgkinslymphoma 1.00Thyroid0.95Myeloma

0.93 HOdgkinSlymphoma0.93Breast0.91 Acuteleukemia0.92Ovary0.89 Multiplemyeloma0.88Kidney0.88Bladder0.82Gallbladder0.88Kidney0.80Bladder0.83Gallbladder0.78Hodgkins

lymphoma 0.81Stomach0.76Rectum/anus0.81 Chronicleukemia0.73Stomach0.79Colon0.70Colon0.74Pancreas0.70Lung0.74Rectum0.66Pancreas0.73Lung0.53Cervix

0.65Liver0.63a

Cancer in Utah 1967—1977.

FAMILIALITY OF CANCER IN IffAH

of the ratio rather than the mean because the ratio of case GIF over control GIFis infinite in a few cases where small control sets contain no related individuals

(i.e., control GIF 0.0). Using the median instead of the mean makes the

analysis robust to these cases.

The mean GIF for the controls is comparatively stable over all the cancersites, with less variation than that noted between cancer site cases. This is notunreasonable because for each cancer site the cases have similar year-of-birth

distributions, and have approximately the same percentage of individuals born

outside the state of Utah (15—20%);also kinship did not vary greatly in Utahfor different time periods (5).

The incompleteness of genealogical records or the lack of appropriategenealogical record links will result in an underestimation of the mean coefficient of kinship among a set of individuals. It is therefore important to choosecontrols so that there is not an inherent difference in the standard of information between cases and controls. In order for a cancer case to be considered inthis analysis, it must first be ascertained by the Utah Cancer Registry. This

registry has essentially complete ascertainment (13) so that there should not bebias introduced simply by appearance in the registry. The cancer cases considered here must also be record-linked to the genealogy. This requires that aunique individual with exactly the same name and birth date as the cancer casemust be found in the genealogy. For this reason, we also require the controlsto have birth dates recorded in the genealogy. We have previously examined

3 different methods of control selection. The first method chooses age-, sex-,

and birth place-matched controls from the entire genealogy. The secondchooses matched controls who do not have a death date before 1966 (when theUtah Cancer Registry became statewide). The third method required that adeath certificate be available for the control and that death occurred after 1966.These 3 methods gave similar results consistent with the random variationinherent in the selection process (8). The first method of control selection wasused in these analyses. Since a difference in the number of ancestors betweencases and controls might affect the measures of familiality, we compared the

average number of ancestors for controls to the average number of ancestorsfor cases and found no significant difference.

Our previous analyses using this method (5—9)considered only the shortestconnecting path between any pair. In this analysis, we found all paths connecting all pairs of individuals and calculated both the total kinship and thebreakdown by exponent. We have improved the algorithm for calculation ofGIFs, and we present 100 sets of controls for each GIF, rather than thepreviously published 5 sets. We present an additional measure of familiality,the MRG, as discussed previously.

RESULTS

Familial Clustering of Cancer by Site. Excess risk for cancer formany anatomical sites has been shown in relatives of cancer cases (seereview, Ref. 19). However, familiality of many of the less commoncancer sites remains unexamined. All cancer sites represented in theUtah Cancer Registry are examined here. The results for the majorcancer sites are shown in Table 2, in which sites are ranked by the GIFof cases. Tables 3—8summarize the results of age-, gender-, andhistology-specific subgroups by general site categories.

Each Table shows the cancer site, the number of cases from thestate of Utah which linked to the genealogy, the GIF of the cases, themean GIF of the 100 randomly selected control groups, the MRG, asdefined previously and the significance value for the comparison ofthe case and control kinship. For those sites with at least 300 cases,subsets by age of onset and gender are also shown in Tables 3—8.Forsites with a sufficient number of early onset cases, early onset isdescribed as 50 years or younger. For those sites with under 100 cases

diagnosed before age 50, the early onset group is comprised of theyoungest tercile of all cases, and is termed “youngest.―The age of theoldest individual in the set is shown for these “youngest―groups.

Table 2 shows that for all of the 24 sites considered, the case GIFexceeded the mean control GIF, but the excess was not significant for6 sites: small intestine, gallbladder, kidney, liver, pancreas, anduterus. Cancers of the small intestine showed the highest GIF, but the

contain detailed information on diagnosis, histology, treatment, and survival ofpatients. Definitions of cancer sites are by primary site and histology andbehavior codes as described in The International Classification of Diseases forOncology (16). The lymphoma site grouping is based on histology, not primarysite, and includes alllymphoma histologies regardless ofprimary site. All othersite groups exclude lymphoma histologies. A Technical Report summarizingcancer groupings is available upon request (17).

GIF. The GIF was initially developed to measure the degree of familyclustering in the UPDB. The method measures the degree of relationshipbetween all possible pairs of individuals in a group using the Mal&ot coefficient of kinship (18) to quantify the degree of relatedness between the 2individuals. The coefficient of kinship for each pair is defined as the probability that randomly selected homologous genes from the 2 individuals areidentical by descent from a common ancestor. Calculations of kinship weremade by counting paths of common descent. Each path contributes an exponent of 1/2 to the total kinship; the value of the exponent is equal to the numberofindividuals along a path connecting a pair ofcases. For instance, full sibs areconnected by 2 routes, one through each parent; their kinship coefficient is thesum of 2 terms of exponent 3, giving 1/4. Half sibs are connected by only onepath; their kinship is one term of exponent 3, or 1/8. By examining thedistribution of all relationships by exponent, it is possible to detect whether anexcess of familiality is due to an excess of close relationships or an excess ofmore distant relationships.

The mean of the coefficients of kinship between all pairs of cases ismultiplied by@ and presented as a single measurement of familiality calledthe GIF. The GIF and the breakdown by exponents for any set of individualscan be compared to the same measurement from a set of randomly selected,matched controls from the same population for meaningful interpretation of theobserved familiality. Age-, sex-, and birthplace-matched controls are chosen at

random from the UPDB. The distribution of kinship and the mean for thecontrols varies randomly, depending upon the actual controls chosen, socontrol groups are selected a number of times and the calculations are repeated.This repetition gives an empirical distribution for the control GIF. One hundredcontrol sets were used in this analysis for each GIF measured.

This empirical method allows a significance test of excess familiality for

any group considered. The P value displayed in the Tables represents thesignificance of the standardized case GIF:

[(case GIF —mean control GIF)/control SD]

when compared with a standard normal distribution. In addition, the MRG isshown in each Table. The MRG is a measure of how extreme the observed GIFof the cases is, taking into account the variability of the control GIFs, which isan important factor for groups with smaller sample sizes. We use the median

2379

on March 10, 2021. © 1994 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from

Table 4 Familializy of cancer for urogenital and male reproductive sites withgenderandage ofonsetsubgroupsMeanCase

controlSitea GIF GIF MRG0 PvalueTestis

275 4.60 2.65 1.800.0003Prostate

8060 3.70 2.76 1.340.000Youngest(69 yes) 2688 3.89 2.80 1.390.000Oldest

5372 3.76 2.77 1.350.000Kidney

781 3.13 2.67 1.170.067Youngest(60 yes) 260 4.30 2.78 1.590.026Oldest

521 3.46 2.65 1.300.011Female292 4.19 2.59 1.610.005Male489 2.84 2.73 1.050.410Bladder

1837 2.99 2.75 1.090.031Youngest(66 yes) 610 3.20 2.77 1.170.094Oldest

1227 2.90 2.75 1.050.227Female385 3.61 2.81 1.260.098Male

1452 2.93 2.75 1.070.129aMedian of the distribution of the ratio of the case GIF to the 100 control GIFs.

FAMILIALITYOF CANCERIN UTAH

subgroups, but in all others. For granulocytic leukemia, the excessfamiiality is exclusively seen in the late onset and in the malesubgroups. Since leukemias may be histologically subdivided to agreater degree, it would be of interest to see precisely which histological subgroups of CLL, CGL, and AOL are most responsible forthis excess famiiality. Among the lymphoma cases, it is interesting tonote that the non-Hodgkins subgroup shows excess famiiality, but notthe Hodgkins group. Among the non-Hodgkins subgroups, the excessfamiliality is observed in all except the early cases. The excess offamiliality observed for myeloma is due primarily to the more common late onset and is high for both males and females.

Table 6 summarizes the results for all gastrointestinal sites exammed. Of these sites, in addition to cancers of the colon and rectum,which have been long recognized to have a familial component, onlystomach cancers also show an excess of famiiality. All subgroups ofcolon cancer show excess familiality, the measure of familiality ishighest in the early cases.

Table 7 reviews the famiiality for cancers of the skin and connective tissues. Lip cancer is the site with the highest significant excessof familiality (Table 2). Care has been taken in the registry todifferentiate between lip cancer and squamous cell skin cancer, which

Table 3 Familiality of cancer for female reproductive and related sites with age ofonset and histology subgroups

MeanCase control

n GIF GIF

966 3.38 2.72195 3.98 2.44770 3.28 2.75

Table 2 Genealogical Index ofFamiliality by cancer site

MeanSitenCase

GIFcontrol GIFMRGapvalueSmall

intestine1394.833.061.780.107Lip8254.752.791.720.000Testis2754.602.651.800.0003Thyroid5844.432.611.72<0.0001Melanoma,

skin1,1574.062.641.54<0.0001Myeloma6283.962.821.430.0008Leukemia1,4803.842.661.45<0.0001Connective

tissue3143.722.631.400.002Prostate8,0603.702.761.340.000Gallbladder3243.682.681.410.055Brain5713.582.751.320.004Colon3,5303.532.761.29<0.0001Lymphoma1,9863.382.671.27<0.0001Ovary9663.382.721.240.001Lung2,4773.332.771.19<0.0001Breast5,8113.232.731.180.000Stomach1,0343.172.671.180.006Kidney7813.132.671.170.067Cervix1,0313.122.681.170.022Rectum1,3123.052.741.110.044Bladder1,8372.992.751.090.031Liver1692.952.581.210.368Pancreas9592.902.751.040.268Uterus1,9452.902.781.040.205All

cancers41,7872.902.711.070.000

a Median of the distribution of the ratio of the case GIF to the 100 control GIFs.

excess was not statistically significant (P = 0.11). Although the meancontrol GIF for this site did not differ from other sites, it has a muchlarger variance due to the small number of cases. The 4 sites with thehighest measure of familiality are among the less common cancersites, including small intestine, lip, testis, and thyroid. Of the morecommon cancers, melanoma ranks the highest, followed by leukemia,prostate, and colon cancer. Breast cancer, usually thought to be themost familial of the common cancers, is among the cancer sites withrelatively low GIF.

Many of the cancer sites with the highest familiality are sites forwhich Utah incidence is higher than United States incidence. Table 1shows Utah incidence compared to United States Caucasian incidencefor females and males. The sites which have higher incidence in Utahthan in United States Caucasians are lip, melanoma, thyroid, connective tissue, and acute leukemia for females; and lip, melanoma, testis,prostate, and connective tissue for males. All are among those cancersites with the highest measured familiality in the UPDB.

Table 3 shows that cancer at 3 of the 4 female reproductive orgenital sites examined demonstrate excess familiality; cancer of theuterus does not. For breast, ovarian, and cervical cancer, the measureof familiality is higher for the early onset cases than the late onsetcases. The lobular breast cancer subgroup shows the highest excessfamiliality, which, although it is higher in the youngest group, appearsto be present in both groups.

Table 4 reviews the urogenital and male reproductive sites. Each ofthe sites except kidney showed significant excess familiality compared to controls. The excess familiality for prostate cancer wasobserved in both the youngest third and the oldest two-thirds. Although bladder cancer showed excess familiality as a group, nosubgroup had a significant excess.

Table 5 reviews the familiality of cancer for hematopoietic sites.Some subgroups of the leukemias showed the highest familialityobserved of all cancers considered. Both lymphocytic and granulocytic subgroups showed excess clustering; however, of the lymphocytic leukemias, the excess familiality is observed solely in the CLLcases. Both granulocytic subgroups (CGL and AOL) show excessfamiliality. In the subgroups examined, the significant excess familiality for lymphocytic leukemia is not observed in the youngest

MRGa

1.241.641.20

Site P value

0.0010.0070.040

OvaryEarlyLate

Breast

EarlyLate

Lobular breastYoungest (57 yes)Oldest

Ductal breastEarlyLate

CervixEarlyLate

Uterus

EarlyLate

5811 3.23 2.73 1.18 0.000

1051 3.96 2.72 1.45 <0.00014761 3.25 2.76 1.18 <0.0001

302 5.66 2.73 2.09 <0.0001101 13.17 2.62 5.74 <0.0001201 6.16 2.64 2.36 <0.0001

5530 3.23 2.74 1.18 <0.0001999 3.92 2.74 1.43 <0.0001

4532 3.24 2.77 1.17 <0.0001

1031 3.12 2.68 1.17 0.022588 3.27 2.71 1.21 0.020443 2.46 2.78 0.87 0.761

1945 2.90 2.78 1.04 0.205

184 239 2.73 1.01 0.5511761 2.95 2.81 1.05 0.167

a Median of the distribution of the ratio of the case GIF to the 100 control GIFs.

2380

on March 10, 2021. © 1994 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from

Table 6 Familiality ofcancer forgastrointestinalsites withgender andage ofonsetsubgroupsMeanCasecontrolSitenOfFGIFMRGap

valueSmall

intestine1394.833.061.780.107

Table 7 Familialityof cancer for skin and soft tissue sites with gender and age ofonset and histology subgroups

MeanCasecontrolSitenGIF GIF MRGa p value

1.72 0.0002.09 <0.00011.77 <0.00011.15 0.5921.60 <0.0001

2.792.922.802.912.80

FAMILIAL1TYOF CANCER IN UTAH

is reported differently. Melanoma of all sites has been considered asa single group, as well as subdivided into melanoma of the skin,ocular melanoma, and lentigo maligna. Most of the excess familialityobserved for melanomas is due to the cases of melanoma of the skinrather than ocular melanomas and lentigos, which are less common.Cancers of the connective tissues show excess familiality as a whole,and for the male subgroup.

Table 8 summarizes famiiality for the remaining cancer sites,including brain, thyroid, and lung. The late and female subgroupsappear to contribute most of the excess familiality observed for braincancer. For lung cancer, which as a whole showed excess familiality,there are 4 histological tumor types considered. All except small celltypes show familiality in excess of that expected. The squamous andsmall cell subtypes are those more strongly associated with lungcancer due to smoking.

Several different graphical presentations have been made, displaying different aspects of the measures. Fig. 1 shows the breakdown ofthe GIF by kinship exponent, with the contribution from each expo

GallbladderYoungest (67 yes)OldestFemaleMale

ColonEarlyLateFemaleMale

StomachYoungest (65 yes)OldestFemaleMale

324 3.68108 5.18216 2.85195 3.33129 3.17

3530 3.531178 4.072352 3.421769 3.551761 3.91

1034 3.17344 4.09690 3.12388 338646 2.72

2.68 1.41 0.0552.79 2.29 0.0812.76 1.07 0.4652.74 1.23 0.2802.71 1.31 0.385

2.76 1.29 <0.00012.77 1.47 <0.00012.77 1.23 <0.00012.76 1.28 <0.00012.73 1.43 <0.0001

2.67 1.18 0.0062.81 1.44 0.0082.61 1.20 0.0532.74 1.35 0.0722.71 1.02 0.496

Liver 169 2.95 238 1.21 0.368Table5 Familialityofcancerfor hematopoieticsiteswithgender,age ofonse4 and

histology subgroups

MeanCase control

Site n GIF GIF MRc? p value

1480 3.84 2.66 1.45 <0.0001245 1.62 233 0.68 0.915

1224 4.28 2.77 135 0.000589 3.91 2.66 1.47 <0.0001891 3.66 2.68 1.38 <0.0001600 6.30 2.70 2.35 0.000200 3.46 2.70 1.35 0.199400 9.38 2.80 3.46 0.000220 7.84 2.68 3.07 <0.0001380 5.67 2.70 2.11 <0.0001112 2.04 234 0.92 0.661451 8.10 2.68 3.05 0.000151 339 2.87 1.37 0.294300 10.60 2.78 3.76 0.000157 12.84 237 5.56 <0.0001294 7.30 2.65 2.81 <0.0001629 3.39 2.65 1.26 0.006209 2.78 2.63 1.16 0.428420 4.27 2.80 133 0.001268 1.49 2.69 035 0.953361 3.74 2.72 1.39 0.028200 4.34 2.72 1.67 0.027394 3.90 2.69 1.48 0.005131 1.82 231 0.80 0.714263 4.81 2.76 1.74 0.002168 1.27 2.63 0.51 0.907226 4.23 2.64 1.65 0.012

1986 3.38 2.67 1.27 <0.0001412 2.99 2.67 1.14 0.210

1571 3.54 2.76 1.29 <0.0001895 3.23 2.65 1.22 0.007

1091 3.40 2.70 1.26 <0.00011607 3.62 2.73 1.32 <0.0001212 3.14 2.47 1.30 0.171

1392 338 2.76 1.32 <0.0001746 3.39 2.69 1.27 0.005861 3.95 2.70 1.47 <0.0001383 2.79 2.66 1.05 0.393200 3.59 2.80 1.39 0.202183 3.58 2.83 1.41 0.264150 3.15 2.69 1.25 0.329233 2.82 2.61 1.17 0.401

628 3.96 2.82 1.43 0.0008210 3.95 2.84 1.42 0.107418 4.36 2.86 130 0.0004267 5.05 2.78 1.84 0.0006361 3.89 2.70 1.48 0.014

Pancreas 959 2.90Youngest (67 yes) 320 2.88Oldest 639 2.86Female 417 3.18Male 542 2.64

2.75 1.04 0.2682.72 1.06 0.3982.82 1.03 0.4492.64 1.19 0.1272.77 0.96 0.618

Leukemia

EarlyLateFemaleMale

LymphocyticYoungest (63 yes)OldestFemaleMale

CLLYoungest(65 yes)OldestFemaleMale

GranulocyticEarlyLateFemaleMale

CGLICMLAGLIAML

Youngest (58 yes)OldestFemaleMale

LymphomaEarlyLateFemaleMale

Non-HodgkinsEarlyLateFemaleMale

HOdgkinSEarlyLateFemaleMale

Rectum 1312 3.05 2.74 1.11 0.044Youngest (65 yes) 437 4.09 2.75 1.49 0.0005Oldest 875 2.99 2.80 1.06 0.237Female 549 3.27 2.74 1.20 0.095Male 763 3.40 2.71 1.27 0.006

a Median of the distribution of the ratio of the case GIF to the 100 control GIFs.

LipEarlyLateFemaleMale

Melanoma, allEarlyLateFemaleMale

Melanoma, skinEarlyLateFemaleMale

LentigomalignaYoungest (63 yes)OldestFemaleMale

Ocular melanoma

825 4.75275 6.05550 4.9872 2.29

753 4.49

1657 336454 5.81

1201 3.39772 3.08885 3.87

1157 4.06406 6.74749 3.77550 3.61607 4.70384 2.79128 1.23256 3.45165 3.28219 3.10124 4.23

2.67 1.34 <0.0001239 2.31 <0.00012.75 1.23 0.00052.65 1.17 0.0562.69 1.44 <0.00012.64 134 <0.00012.65 234 0.0002.76 1.35 0.00042.65 1.38 0.0032.68 1.76 <0.00012.71 1.04 0.4242.84 0.47 0.8652.91 1.17 0.2632.77 1.28 0.3302.76 1.20 0.3652.82 1.72 0.171

Connective tissue 314 3.72 2.63 1.40 0.022Youngest (52 yes) 103 3.65 2.68 1.63 0.266Oldest 211 3.85 2.88 1.35 0.143Female 135 3.88 2.55 1.73 0.160Male 179 435 2.55 1.86 0.015

a Median of the distribution of the ratio of the case GIF to the 100 control GIFs.Myeloma

Youngest(63yes)OldestFemaleMale nent value displayed for the case GIF, the median control GIF, and theaMC&anofthedistributionoftheratioofthecaseGIFtothe100controlGIFs.5thand95thpercentilesofthecontrolGIFsshownforallcancer

b A@, acute lymphocytic leukemia; CLL, chronic lymphocytic leukemia; COL,

chronic granulocytic leukemia; CML, chronic myclocytic leukemia; AGL, acute granu- C8SC5@ Fig. 2 shows the breakdown of the GIF by kinship exponentlocytic leukemia; AML, acute myelocytic leukemia. similarly to Fig. 1 for all cancer sites considered separately. Fig. 3

2381

on March 10, 2021. © 1994 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from

Table 8 Familiality of cancer for brain, lung, and endocrine sites with gender, ageofonse4and histologysubgroupsMeanCase

controlSiten GIF GIF MRGa pvalueBrain

571 338 2.75 1.320.004Early135 1.39 2.62 0.580.859Late435 435 2.89 1.56<0.0001Female255 4.76 2.74 1.830.004Male316 3.35 2.71 1.260.133Thyroid

584 4.43 2.61 1.72<0.0001Early297 4.79 2.69 1.890.0002Late287 4.95 2.69 1.970.0009Female418 4.16 2.60 1.650.0003Male166 4.02 2.51 1.790.062Lung

2477 3.33 2.77 1.19<0.0001Youngest(64 yes) 824 3.95 2.81 1.43<0.0001Oldest

1653 3.20 2.78 1.150.007Female510 330 2.82 1.250.043Male

1967 3.37 2.78 1.21<0.0001Alveolar/adenocarcinoma477 3.67 2.72 1.360.008Youngest159 2.86 2.80 1.140.479Oldest318 4.71 2.72 1.750.001Female163 631 2.73 2.600.0007Male314 3.80 2.71 1.420.031Large

cell 635 3.34 2.75 1.230.031Youngest212 1.39 2.83 0.510.950Oldest423 3.69 2.86 1.300.074Female128 1.63 2.89 0.700.762Male507 3.30 2.75 1.210.093Squamous

cell 762 3.84 2.74 1.41<0.0001Youngest253 5.28 2.76 2.010.0004Oldest509 335 2.68 1.330.032Female

63 0.85 2.61 0.470.734Male699 3.74 2.75 1.360.0002Small

cell 296 333 2.83 1.290.137a

Median of the distribution of the ratio of the case GIF to the 100 control GIFs.

FAMILIALITY OF CANCER IN UTAH

noted for many different cancer sites (19). Since exposure tocommon environmental variables may also result in excessivefamilial risk, familial clustering of a disease is best examinedamong individuals with known genetic relationships, both near anddistant. In the more distant relationships, shared familial environment might be less, and the probability of shared genotypes can bemeasured. This suggests stronger evidence for a genetic componentfor cancer sites for which there is an excess of familiality at evendistant relationships (Fig. 2).

The UPDB is a unique database which has allowed the hypothesis of familial clustering of a genetic nature to be examined.

Previous analyses on a much smaller data set (8) showed a verysimilar ranking of cancer sites by familiality to that presented here.More sites have been examined here, and some of these lesscommon sites have taken places high in the site ranking of Table2. These include cancer of the testis, thyroid, and myeloma, which

now rank among the top 6 sites for significant excess familiality.The only significant kinship difference in sites which have beenpreviously reported occurred for ovarian cancer, which ranked asthe third most familial of sites previously and has now dropped tothe level of breast cancer. The number of ovarian cancer cases hasincreased from 435 to 966. If there are only a few clusters of closerelatives among cases, then in the presence of the new samples theeffect of these few clusters would be reduced. Table 3 shows thatthere is still significant excess familiality among the 195 ovariancases diagnosed before age 50.

It is interesting that many of the sites showing highest familialityare also those for which incidence in Utah is higher than United Statesincidence. There may be an excess of these sites in Utah due to thepresence of a susceptibility gene or genes in earlier generations whichare now manifesting themselves in many descendants. Exposure to anadditional risk factor in an individual with an inherited susceptibilitymight also be responsible, e.g., the excess sun exposure in Utah couldresult in a higher percentage of individuals with a lip cancer susceptibility gene actually expressing the cancer.

It is well recognized that breast and ovarian cancer have a familialcomponent; however, cervical cancer is thought of primarily as anongenetic disease of viral origin. It appears that only the subgroup ofearly cervical cancer is responsible for the excess famiiality observed.

shows the distributions of the ratios of the case GIF to the controlGIFs (MRG) ranked by median MRG, showing the median andinterquartile-range and range for all cancer sites.

DISCUSSION

Many diseases occur more frequently among the relatives ofdiseased individuals than in the general population. This has been

U-

C00

.0.c

Case GIF

—.—.—.- Median control GIF

5th& 95thpercentilesofcontrolGlFs

U,0

0

Fig. 1. Genealogical Index of Familiality for all41,787 cancer cases taken together is shown bycontributions from paths of descent of differentlength. The total GIF is given by the sum of thecontributions over path lengths. Also shown are themedian, 5th, and 95th percentiles of the distributionof control GIFs, again shown by path length.

0

00

2 4 6 8 10

Pathlength

2382

on March 10, 2021. © 1994 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from

FAMILIALITY OF CANCER IN UTAH

Smallintestine.139 Up.825 Testis.275 ThyroId,584

0

1@

a

00

0

a00

0

a

0a

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10

Myelorna, 628 Leukemia,1480

2 4 6 8 10

Connective tIssue. 314

0

0

0a

a

a0a

2 4 6 8 10

Brain.571 Colon.3530

0

0

0a

Melanoma,skin.1157

2 4 6 8 10

Prostate.8060 Gallbladder.324

0

,@.0

00

0

0

0a

a

a0a

0

0

0aFig. 2. Breakdowns of OW by path length are

given for the cancer sites listed in Table 2. Themedian, 5th, and 95th percentiles of the distributions of control GIFs are given for comparison. Thename of the site and the number of cases are shownabove each graph.

0

‘@.

0

00

0

0

00

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10

Lymphoma. I 986 Ovary, 966

0

a0a

2 4 6 8 10

Lung. 2477 Breast, @811

0

0

0a

0

0

0a

0

0

0a

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10

Stomach, 1034 Kidney,781 Cervix.1031 Rectum.1312

a1@

0

00

a,@.

0

00

0

0

00

0

1@

0

0a

0

0

00

0

0

0a

2 4 6 8 10 2 4 6 8 10

Bladder.1837 Liver.169 Pancreas. 959 Uterus. I945

2 4 6 8 10246810 246810 246810

The distribution of GIF by exponent (Fig. 2) shows that the excess offamiliality is observed out to the level of exponent 5, indicating thatthe excess of familiality is not just due to close relationships.

Lung cancer showed excess familiality (Table 2), and 3 of the 4histological tumor types still showed excess familiality when considcrud separately (Table 8). This finding may simply represent the effectof the familiality of smoking; it may represent the interaction of asusceptibility gene for lung cancer and another environmental variablesuch as smoking; or it may be the result of an inherited lung cancersusceptibility which is entirely unrelated to smoking, but is expressedin a histologically similar manner. Individuals represented in theUPDB do not necessarily practice the LDS's teachings prohibitingtobacco use. Further studies of these families are necessary to resolvethese alternative explanations.

The familiality among the leukemia cases was second only to thatobserved for lobular breast cancer. Leukemia alone was the seventhmost familial of all cancer sites (Table 2), and when examined byhistological type, much higher familialities were observed (Table 5).Lymphocytic leukemias clustered more than all leukemias together,

and when leukemias were separated into CLL and ALL, only CLLshowed excess familiality. Granulocytic leukemias alone showed cxcess famiiality, and showed even further excess when divided intochronic and acute subgroups. This may indicate that there are uniquesusceptibility genes for each of these leukemia types.

It has been widely reported that cancer which appears at an earlyage is more likely to be due to an inherited predisposition; however,most of these studies are based on rare kindreds and not on apopulation. This hypothesis can be tested appropriately by exam

2383

on March 10, 2021. © 1994 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from

Testis Small Lip Thyroid Melanoma, Leukemia Myeloma GallbladderintestineskinConnective

Prostate Brain Colon Lymphoma Ovary Liver Lungtissue

FAMILIALITY OF CANCER IN UTAH

0c'J

0

0c'J

0

Fig. 3. Box and whisker plots show the medians,interquartile ranges, and ranges of the distributionsof the ratios of case GIFs to control GIFs for thesites given in Table 2. The sites are ordered bymedian ratio.

0c%J

0

Kidney Breast stomacn GerVix Hectum Bladder Allcancers Pancreas Uterus

ining the clustering in an entire population, as was done here with

the UPDB. Many of the sites which were examined, includingbreast, gallbladder, kidney, lip, lymphoma, melanoma, ovary, andthyroid cancers, showed excess familial clustering in both the earlyand late onset groups. This suggests that there are inherited predispositions to both early and late onset cancers; one or morepredispositions could be responsible. This finding is supported by

the fact that cancer susceptibilities which are responsible for bothearly and late cancer have been mapped. For several cancer sitesexamined, the familiality as measured by the GIF in both the earlyand the late onset groups exceeded the familiality measure whenthe 2 age groups were considered together (Table 2), suggestingindependent, age-specific susceptibilities, among other possibilities. Of the common cancers, breast cancer, ovarian cancer, andmelanoma all showed higher GIF in early than late onset cases.Although some of this effect may be due to earlier screening anddiagnosis in the relatives of young probands, this result alsosupports the observations of clusters of cancer occurring at veryearly age which are probably due to rare inherited susceptibilityloci. Leukemia and lymphoma both showed much higher familiality in the late onset group than in the early onset group. This lateage onset familiality held true for many of the subgroups analyzed.

The sites for which only one of the sexes showed a statisticallysignificant excess include kidney, rectum, lip, melanoma of all sites,connective tissue, brain, and thyroid. For all except kidney, brain, andthyroid, males showed excess familiality and females did not. Whenlung cancer was subdivided, the alveolar/adenocarcinoma typeshowed greater excess clustering for females; squamous cell lungcancer showed excess clustering for males only. If the clusters of lungcancer observed are due to an inherited susceptibility, the genderdifference may indicate a different gene/environment interaction effect which leads to higher penetrance in females.

Many cancer susceptibility genes have already been isolated, andother cancer susceptibility loci have been identified, but the geneshave not yet been isolated (for review, see Ref. 20). Major geneshave been hypothesized for other sites including lung cancer andprostate cancer, for which supportive linkage results have not yetbeen reported. The results presented here support the hypothesis ofan inherited basis to cancer of almost all sites, and supports theexistence of more than one susceptibility locus for some cancers,

since subsets of individuals with certain characteristics show cxcess clustering. These may represent characteristics of individualswith the most penetrant forms of inherited susceptibilities. Itremains unclear what fraction of cancer of any particular site is dueto an inherited component. Studies of common cancer susceptibilities in the Utah population (21—22) support the similarity to other

populations, and suggest that the results should be generally applicable to populations of similar origin.

REFERENCES

1. Skolnick, M. Prospects for population oncogenetics. In: J. J. Mulvihill, R. W. Miller,and J. F. Fraumeni, Jr. (eds.), Genetics of Human Cancer, pp. 19—25.New York:Raven Press, 1977.

2. Skolnick, M. The Utah genealogical database: a resource for genetic epidemiology.In: J. Cairns, J. L. Lyon, and M. Skolnick (eds.), Banbury Report No. 4: CancerIncidence in Defined Populations, pp. 285—297.New York: Cold Spring HarborLaboratory Press, 1980.

3. Skolnick, M., Bean, L., May, D., Arbon, V., de Nevers, K., and Cartwright, P.Mormon demographic history. I. Nuptiality and fertility of once-married couples.Popul. Stud., 32: 5—19,1978.

4. Skolnick, M., Bean, L. L., Dintelman, S. M., and Mineau G. A computerized familyhistory data base system. Sociol. Soc. Res., 63: 506—523,1979.

5. Hill, J. A kinship survey of cancer in the Utah mormon population. Ph.D. Thesis,University of Utah, Salt Lake City, 1980.

6. Hill, J. A survey of cancer sites by kinship in the Utah mormon population. In: J.Cairns, J. L. Lyon, and M. Skolnick (eds.), Cancer incidence in defined populations,Banbury Report 4, pp. 299—318.New York: Cold Spring Harbor Laboratory Press,1980.

7. Skolnick, M., Bishop, D. T., Carmelli, D., Gardner, E., Hadley, R., Hastedt, S., Hill,J. R., Hunt, S., Lyon, J. L., and Smart, C. R. A population-based assessment offamilial cancer risk in Utah mormon genealogies. In: F. E. Arrighi, P. N. Rao, and E.Stubblefield (eds.), Genes, Chromosomes, and Neoplasia, pp. 477—500.New York:Raven Press, 1981.

8. Cannon, L., Bishop, D. T., Skolnick, M., Hunt, S., Lyon, J. L., and Smart, C. R.Genetic epidemiology of prostate cancer in the Utah mormon genealogy. CancerSurv., 1: 48—69, 1982.

9. Bishop, D. T., and Skolnick, M. H. Genetic epidemiology of cancer in Utah genealogies: a prelude to the molecular genetics of common cancers. In: T. W. Mak and I.Tannock (eds.), Cellular and Molecular Biology of Neoplasia J. Cell Physiol. Suppl.,3: 63—77,1984.

10. McLellan, T., Jorde, L. B., and Skolnick, M. H. Genetic distances between the UtahMormons and related populations. Am. J. Hum. Genet., 36: 836—837, 1984.

11. Jorde, L. B., and Skolnick, M. H. Demographic and genetic application of computerized record linking: the Utah Mormon genealogy. Information Ct Sciences Humaines, 56—57:105—117,1981.

12. Cancer in Utah Report No. 3. 1967—1977.Salt Lake City: Utah Cancer Registry,1977.

13. Lyon, J. L., Gardner, J., and West, D. Cancer incidence in mormons and nonmormons in Utah during 1967—1975.J. NatL Cancer Inst., 65: 1055—1061,1980.

14. Newcombe, H. B., and Kennedy, J. M. Record linkage: making maximum use of

2384

on March 10, 2021. © 1994 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from

FAMILIALITY OF CANCER IN UTAH

the discriminating power of identifying information. Commun. Assoc. ComputingMachinery, 5: 563—566,1962.

15. Hill, T. Generalized iterative record linking system. GIRLS. Ottawa: Statistics Canada, 1981.

16. The International Classification of Diseases for Oncology, Field Trial Edition, Geneva,Switzerland:WorldHealthOrganization,1988.

17. Cannon-Albright, L A., Jacobsen, M., and McWhorter, W. P. Classification of cancerby site in the Utah Population Database. In: Technical Report No. 24. Salt Lake City,Utah: Department of Medical Informatics, 1993.

18. Malécot,G. Las Mathematiques de l'Heredite. Paris: Masson et Cie, 1948.19. Cannon-Albright, L. A., Bishop, D. T., Goldgar, C., and Skolnick, H. H. Genetic

predisposition to cancer. In: V. T. DeVita, Jr., S. Heilman, and S. A. Rosenberg (eds.),

Important Advances in Oncology, pp. 39—55.Philadelphia: J. B. Lippincott Co.,1991.

20. Knudson, A. G. All in the (cancer) family. Nat. Genet., 5: 103—104,1993.21. Cannon-Albright, L. A., Goldgar, D. E., Meyer, L. J., Lewis, C. M., Anderson, D. E.,

Fountain, J. W., Hegi, M. E., Wiseman, R. W., Petty, E. M., Bale, A. E., Olopade,0. I., Diaz,M.0., Kwiatkowski,D. J., Piepkorn,M.W.,Zone,J. J., andSkolnick,M. H. Assignment of a locus for familial melanoma, MLM, to chromosome9pl3-p22. Science (Washington DC), 258: 1148—1152, 1992.

22. Goldgar, D. E., Cannon-Albright, L. A., Oliphant, A., Ward, J. H., Linker, G.,Swensen, J., Tran, T. D., Fields, P., Uharriet, P., and Skolnick, M. H. Chromosome17q linkage studies of 18 Utah Breast Cancer Kindreds. Am. J. Hum. Genet., 52:743—748,1993.

2385

on March 10, 2021. © 1994 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from

1994;54:2378-2385. Cancer Res   Lisa A. Cannon-Albright, Alun Thomas, David E. Goldgar, et al.   Familiality of Cancer in Utah

  Updated version

  http://cancerres.aacrjournals.org/content/54/9/2378

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/54/9/2378To request permission to re-use all or part of this article, use this link

on March 10, 2021. © 1994 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from