ReviewThe Genetic Epidemiology of Cancer: Interpreting Family and TwinStudies and Their Implications for Molecular Genetic ApproachesNeil Risch1Department of Genetics, Stanford University School of Medicine, Stanford,California 94305-5120, and Division of Research, Kaiser Permanente,Oakland, California 94611-5714AbstractThe recent completion of a rough draft of the humangenome sequence has ushered in a new era of moleculargenetics research into the inherited basis of a number ofcomplex diseases such as cancer. At the same time, recenttwin studies have suggested a limited role of geneticsusceptibility to many neoplasms. A reappraisal of familyand twin studies for many cancer sites suggests thefollowing general conclusions: (a) all cancers are familialto approximately the same degree, with only a fewexceptions (both high and low); (b) early age of diagnosisis generally associated with increased familiality; (c)familiality does not decrease with decreasing prevalenceof the tumor–in fact, the trend is toward increasingfamiliality with decreasing prevalence; (d) a multifactorial(polygenic) threshold model fits the twin data for mostcancers less well than single gene or geneticheterogeneity-type models; (e) recessive inheritance is lesslikely generally than dominant or additive models; (f)heritability decreases for rarer tumors only in the contextof the polygenic model but not in the context of single-locus or heterogeneity models; (g) although the familyand twin data do not account for gene-environmentinteractions or confounding, they are still consistent withgenes contributing high attributable risks for most cancersites. These results support continued search for geneticand environmental factors in cancer susceptibility for alltumor types. Suggestions are given for optimal studydesigns depending on the underlying architecture ofgenetic predisposition.IntroductionLast June, human genome scientists announced completion ofand more recently published a rough draft of the human genomesequence, ushering in a new era of human molecular genetics(1, 2). This accomplishment was heralded with great fanfareand with predictions of a significant impact on the understand-ing and treatment of chronic human diseases such as cancer. Itis perhaps for this reason that a recent twin study of cancer (3)received so much attention from both the scientific and laymedia, because its conclusion was that susceptibility to canceris primarily environmental and not inherited and thus seem-ingly at odds with the claims of the genomicists. The fact thatthis was by far the largest twin study in cancer yet reported(nearly 45,000 twin pairs) also lent credence to this conclusion,although an editorial appearing in the same issue (4) discussedsome of the limitations of that study. Given the increasingemphasis on molecular genetic approaches to address familialdisorders coupled with the latest evidence questioning the roleof genetics in cancer susceptibility, a reassessment of the roleof genetic factors in cancer susceptibility generally and forsite-specific cancers in particular appears warranted.Study DesignsFamilial aggregation of a trait is a necessary but not sufficientcondition to infer the importance of genetic susceptibility, be-cause environmental and cultural influences can also aggregatein families, leading to family clustering and excess familial risk.Family aggregation is usually assessed by studying relatives ofaffected subjects and contrasting their rates of illness with thoseof a suitable control group, typically the relatives of unaffectedsubjects.Several approaches for disentangling genetic from envi-ronmental influences are also possible in studies of humandisease, although practical difficulties often limit their use. Themost powerful design examines risks in biological relatives ofaffected versus control adoptees, because adoption creates aseparation between an individual’s biological and environmen-tal influences. Because it is often difficult to obtain access toinformation on biological relatives of adoptees, adoption stud-ies typically focus only on common disease or trait outcomes.Another study design often used to separate genetic andenvironmental influences involves twins. Identical (MZ2) twinsderive from the fission of a single fertilized egg and thus inheritidentical genetic material. By contrast, fraternal (DZ) twinsderive from two distinct fertilized eggs and thus have the samegenetic relationship as full siblings, although they may be more“biologically” related because of sharing the same prenatalintrauterine experience.Comparing the similarity of MZ twins with same-sex DZtwins is a common approach for gleaning the magnitude ofgenetic influence on a disease or trait and has been appliedextensively to a broad range of disorders, including cancer. Astandard measure of similarity used in twin studies is theconcordance rate. The “pairwise” concordance is calculatedsimply as the proportion of twin pairs with both twins affectedof all ascertained twin pairs with at least one affected. On theother hand, the “probandwise” concordance allows for doublecounting of doubly ascertained twin pairs and has the advantageof being interpretable as the recurrence risk in a co-twin of anaffected individual (5). Usually, the most critical assumption inReceived 2/1/00; revised 4/27/01; accepted 5/2/01.1To whom requests for reprints should be addressed, at Department of Genetics,M322, Stanford University School of Medicine, Stanford, CA 94305-5120. Fax:(650) 725-1534; E-mail: [email protected] abbreviations used are: MZ, monozygotic; DZ, dizygotic; MFT, multifac-torial threshold; FRR, family risk ratio; SIR, standardized incidence ratio; PAF,population attributable fraction; RR, relative risk.733Vol. 10, 733–741, July 2001 Cancer Epidemiology, Biomarkers & Preventiontwin studies is that MZ and DZ twins display a comparabledegree of similarity because of the sharing of environmentalfactors, so that the difference in concordance rates between MZand DZ twins is only a reflection of genetic factors.Genetic Models and the Interpretation of Family andTwin StudiesUnderstanding empirical evidence about genetic susceptibilityto cancer requires a discussion of models of genetic inheritanceand their implications. The simplest way to measure geneticeffects is through familial risk ratios, defined as the risk to agiven type of relative of an affected individual divided by thepopulation
View Full Document
Unlocking...