Familial Melanoma: A Meta-analysis and Estimates of Attributable Fraction

A familial condition or disease is one that tends to occur more often in family members than is expected by chance alone. Familial clustering of a disease is an indicator of possible heritable causes, although such clustering does not preclude shared environmental factors, and thus the relative contribution of genetic and environmental factors can be difficult to disentangle. It is well established that melanoma commonly clusters in families, and in the past two decades, there has been an explosion in research directed at elucidating the genetic basis for melanoma. The recent discovery of several common, albeit low-risk, genotypes has led to the widespread perception that melanoma is a predominantly genetic disease (1).

This perception has been affirmed by several studies that have reported statistically significantly increased relative risks (RR) of melanoma for people with a positive family history of the disease. Ford and colleagues (2) in a pooled analysis of eight case-control studies conducted in the 1980s to early 1990s reported a pooled odds ratio (OR) of 2.24 [95% confidence interval (CI), 1.76-2.86] for risk of melanoma associated with having at least one affected first-degree relative. A more recent systematic review by Gandini and colleagues (3) that included 14 studies conducted up to 2002 reported a more modest pooled OR of 1.71 (95% CI, 1.41-2.14). All of the original studies included in the pooled analysis (2) and meta-analysis (3) relied on self-reports of positive family history, which can lead to exaggerated risk estimates because cases are more likely to over-report family history and controls under-report (4, 5). In the interval since these reports were Published, there has been several large, population-based studies investigating this association, some of which have used superior methods for ascertaining family history of melanoma.

The RRs derived from epidemiologic research alone, however, are insufficient for quantifying the contribution of family history to the burden of melanoma because RRs give no weight to the population prevalence of any given factor. The population attributable fraction (PAF) is widely used to quantify the public health effect of a putative causal factor because it considers both the strength of association between risk factor and disease, as well as the prevalence of the factor in the community. Thus, in this context, the PAF estimates the proportion of all cases of melanoma attributable to having a family history of melanoma.

The aim of this work therefore was to evaluate systematically the most recent epidemiologic evidence describing the relationship between melanoma and a positive family history of the disease, and then to use these data to estimate the fraction of melanomas attributable to a positive family history.

A systematic review and meta-analysis was done in accordance with the Meta-analysis of Observational Studies in Epidemiology guidelines for reviews of observational studies (6).

We included observational studies of all designs in the meta-analysis provided that they permitted quantitative assessment of the association between histologically confirmed melanoma and family history of the disease. We included studies reporting various measures of RR because melanoma is a rare disease, and in such instances ORs and standardized incidence ratios provide a valid estimate of the RR.

Eligible studies up to June 2008 were identified by searching the following databases and by hand searching the reference lists of the retrieved articles.

• Medline 1950 (U.S. National Library of Medicine, Bethesda, MD) using PubMed software as the search interface

• Embase 1966 (Elsevier Science, Amsterdam, Holland) using the Embase search interface

• Conference Papers Index 1982 (CSA, Bethesda, MD) using the CSA Illumina search interface

• ISI Science Citation Index using the ISI Web of Science search interface

For computer searches, we used the following medical subject heading terms or text words (using both U.K. and U.S. spellings): “melanoma,” “familial,” “family history,” “proband,” “risk,” “etiological,” “etiology,” “cohort studies,” and “case-control studies.” Studies that had been commonly cited in the literature were also included as citation search terms in the ISI Science Citation Index (1990 to present) to identify subsequent studies that had referenced them. Only studies of adult populations (>18 y) were included. The search was not limited to studies published in English. We read the abstracts of all identified studies to exclude those that were clearly not relevant. The full texts of the remaining articles were read to determine if they met the study inclusion criteria. Where multiple reports from one study were found, the most recent or most complete publication was used.

An abstraction form summarizing study design, study population, and relevant raw and adjusted data was completed for each article by two independent reviewers (CO, HC); inconsistencies were resolved by consensus. The following information was recorded for each study: study design, location, calendar years of data collection (case-control studies), number of cases and controls or person-years duration of follow-up (cohort studies), age range of study population, variables for which statistical adjustment was done, point estimates (RR, OR, or standardized incidence ratio), and 95% CIs. Definitions of family history were also recorded (one or more affected first-degree relatives versus wider definitions of family history; self-reported versus verified). Where several risk estimates for family history were presented, we abstracted those adjusted for the greatest number of potential confounders. Studies that reported results separately by gender, body site, or histologic subtype with no combined data were treated as independent data sets in the meta-analysis.

We did not assess the methodologic quality of the primary studies and hence did not exclude studies on the basis of quality score (6), but instead performed subgroup and sensitivity analyses according to study features that could potentially affect the strength of the association.

To pool RR estimates for the positive family history of melanoma, a weighted average of the log RR was estimated, taking into account the random effects using the method of DerSimonian and Laird (7). Statistical heterogeneity among studies was evaluated using the Cochrane Q test and I² statistics. The Cochrane Q test, which is calculated as the weighted sum of squared differences between individual study effects and the pooled effect across studies, is widely known to have too much power to detect clinically unimportant heterogeneity if the number of studies is large (8). The I² statistic describes the percentage of variation across studies that is due to heterogeneity rather than chance (9), and does not inherently depend on the number of studies considered (I² = 100% × (Q−df)/Q). We also conducted separate analyses by study design, geographic location, year of publication (before and after 2000), whether family history was verified (yes or no), definition of family history (first-degree relative or other), and by confounders controlled for in the analyses. Finally, we conducted sensitivity analyses, omitting each study in turn to determine whether the results could have been influenced excessively by a single study. We evaluated publication bias by assessing the funnel-plot asymmetry (10, 11).

To estimate the PAF for a positive family history of melanoma using the adjusted RR derived from meta-analysis, we used the method of Bruzzi et al. (12) that accounts for possible confounding and effect modification. CIs for the PAFs were derived using the substitution method described by Daly (13). We estimated the prevalence of a positive family history of melanoma by calculating an average of the prevalences reported in study cases, which was weighted by the size of the case group. Due to heterogeneity in the prevalence of family history across studies from different locations, we decided a priori to calculate PAFs for different geographic locations (North America, South America, Northern Europe, Central Europe, Mediterranean Europe, and Australia).

All analyses were conducted using Stata 10.

The primary computerized literature search identified 54 potentially eligible studies. After a review of the study abstracts, we retrieved 24 articles for further assessment of which 21 were found to be relevant (2, 14-33); 3 studies were excluded because they were duplicate reports from the same study (34-36). One of the articles identified, by Ford and colleagues (2), represented a pooled analysis of eight case-control studies; this article presented estimates from two other eligible studies that did not publish their data as original reports (37, 38). The estimates from these two studies as published by Ford et al. (2) were included in the meta-analysis. In total, 22 studies were included: two cohort studies (14, 15), two population-based record linkage studies (16, 17), 1 nested-case-control study (18), and 17 case-control studies (Table 1; refs. 19-33, 37, 38). One study presented data stratified by sex only (19), and the data for men and women were included in the meta-analyses as two independent data sets.

Using a random effects model, the pooled RR for a positive family history of melanoma was 2.06 (95% CI, 1.72-2.45; Fig. 1), with evidence of significant heterogeneity (P = 0.01). The results of subgroup analysis to examine the heterogeneity are provided in Table 2. Eleven of the studies were population based; the pooled RR for these studies was lower than that for the nonpopulation-based studies (2.03 versus 2.51). The pooled estimate was highest for the cohort studies (OR, 3.11; 95% CI, 1.40-6.92) due to a higher estimate from the nonpopulation-based cohort (15). Only two studies assessed family history using record linkage through population-based cancer registries (16, 17); the pooled RR for these two studies was 2.52 (95% CI, 2.11-3.00) compared with 1.97 (95% CI, 1.61-2.42) for the remainder of the studies for which family history of melanoma was self-reported. As expected, the pooled RR was higher for studies that reported on family history in first-degree relatives (RR, 2.22; 95% CI, 1.81-2.72) than for studies that used wider definitions of family history (OR, 1.79; 95% CI, 1.51-2.12).

Summary statistics were not influenced by excluding one study at a time, with the pooled RR ranging from 1.98 (95% CI, 1.67-2.35), with the omission of Freedman et al. (15), to 2.13 (95% CI, 1.80-2.54), with the omission of Holly et al. (26). The funnel plot of the effect estimates for the risk of melanoma related to a positive family history was close to symmetrical and there was no evidence of publication bias using the Egger weighted regression method (P for bias = 0.89) or the Begg rank correlation method (P for bias = 0.245).

Estimates of the PAF associated with family history of melanoma, calculated using the meta-analysis–derived summary RRs and the weighted average of the prevalence estimates for all studies, and stratified by geographic location are presented in Table 3. The prevalence of a positive family history of melanoma ranged from 1.3% in Northern European studies to 15.8% in Australian studies; the corresponding PAFs ranged from 0.007 in Northern Europe to 0.064 in Australia. For all studies, the PAF estimate was 0.040 (95% CI, 0.032-0.045).

We have systematically reviewed the most recent epidemiologic data reporting the relationship between melanoma and a positive family history of the disease, and conducted a meta-analysis. Our findings suggest that only a small percentage of melanoma cases (most likely 4%, but always <7%) are attributable to having an affected family member. Variation in PAF by geographic location reflects large differences in the prevalence of reported family history, which ranged from 1.3% for studies conducted in Northern Europe to 15.8% for those conducted in Australia. The reasons for this remain to be explored, but might include over-reporting or genuinely higher penetrance of melanoma phenotypes in highly sun-exposed populations. To our knowledge, this is the first study to systematically evaluate the PAF of melanoma associated with familial risk using estimates of RR derived through systematic review and meta-analysis.

Several limitations must be considered when interpreting these findings. First, the studies contributing to some of the pooled RR estimates are prone to the usual biases associated with case-control studies, including selection and recall bias. Selection bias due to control recruitment from dermatology clinics or hospitals in the majority of studies would result in an attenuated effect if controls recruited in this way were more likely to have a family history of melanoma than those recruited randomly from a population-based source; this would result in an underestimate of the true PAF (39). This seems unlikely given that the pooled OR for population-based studies was lower than for clinic/hospital-based studies. It is also possible that cases were more likely to remember if a family member was diagnosed with melanoma than controls (4), although there was no evidence of such recall bias in the pooled analysis. Indeed, the pooled RR of the two record-linkage studies in which family history was verified through cancer registries was actually higher than for the remainder of studies where history was self-reported, an unlikely eventuality if the observed association was due to systematic over-reporting of family history by patients with melanoma. There are also challenges associated with the use and interpretation of PAFs (40, 41), including the interpretation of multiple competing risks, and the fact that PAFs computed separately for different risk factors are not constrained to sum to 1.0.

An earlier systematic review and meta-analysis (3) included studies conducted up to 2002; however, a large number of relevant studies have entered the literature since that review was published in 2005. Indeed, we included in our meta-analysis nine new studies, comprising 33,643 new cases (14-22) that were not published in time for the earlier analysis; two of these studies were the only studies conducted to date reporting registry-verified family history data (16, 17). Our study substantially extends previous investigations by estimating the PAFs associated with family history of melanoma.

Compared with other common chronic diseases, the familial risk for cancer is low. For example, the prevalence of familial disease is ∼25% for coronary artery disease and 20% for noninsulin-dependent diabetes (42). As reviewed by Hemminki and colleagues (43), only a small percentage of certain types of cancers are familial (<5% for most cancer sites), and the magnitude of the risk is ∼2-fold for most cancers. Data from the Swedish Family Cancer Database Cancers indicate that in that population, the highest familial attributable fractions among first-degree relatives include prostate (9.1%), colorectal (5.2%), breast (3.7%), and lung cancer (2.93%) compared with 1.4% for melanoma (based on a standardized incidence ratio of 2.54 and prevalence of 2.36%; ref. 44). Our overall estimate, using data from studies around the world, suggests a slightly higher proportion attributable to affected family members than was observed in Sweden.

It has long been acknowledged that melanoma commonly clusters in families; this increased risk in relatives may occur as a result of shared genes and/or shared environment. Technological advances and new methods for identifying susceptibility loci have directed much of the research on melanoma toward defining the genetic basis for familial melanoma. Twin studies, which allow the heritable and environmental components of risk to be estimated, suggest that much of the risk of melanoma is heritable (45). A large study of Swedish families, however, found that heritable effects accounted for only 21% of the variance in melanoma risk between all types of family members, compared with 10% shared environmental and 69% random environmental effects (46).

Family linkage studies have identified three major melanoma susceptibility genes: CDKN2A, ARF, and CDK4 (47-51). More recently, linkage studies have identified other major susceptibility loci on chromosomes 1p22 (52) and 9q21 (53). Together, these genes account for only 20% to 50% of the inherited forms of melanoma (1, 54). A combination of multiple lower penetrance alleles and/or shared sun exposure habits common to affected relatives probably account for the balance of familial cases.

Candidate gene approaches and Genome-Wide Association Studies (GWAS) have identified several low-penetrance susceptibility genes including MC1R (55, 56), OCA2 (57-59), ASIP, TYR, SLC45A2, and TRYPT1 (60, 61) that are associated with pigmentary traits, and MTAP and PLA2G6 that are associated with the development of nevi (62, 63). These higher prevalence/lower penetrance polymorphisms, which may interact with environmental exposures, are relevant to a much larger proportion of the melanoma population than the small number of familial cases caused by low-prevalence/high-penetrance mutations in CDKN2A, ARF, and CDK4.

Given the high profile of this growing body of genetic research, it is important not to lose sight of the public health goals of disease prevention and early detection, processes for which the identification of high-risk individuals is paramount. Our analyses draw attention to the overall low disease burden associated with familial risk and remind us that the bulk of cutaneous melanomas arise through other pathways. The role of environment in melanoma etiology, and the interaction between gene and environment, are underscored by the increasing incidence in recent decades (64), the large differences in incidence rates by latitude (65), and migrant studies (66) that strongly implicate sunlight as the major risk factor. In terms of disease prevention, the low burden of disease associated with familial factors must be considered against this backdrop of significant environmental contribution to melanoma etiology.

No potential conflicts of interest were disclosed.

Grant Support: Xstrata Community Partnership Program Queensland. The researchers are independent of the funding source. D.C. Whiteman is a Principal Research Fellow of the National Health and Medical Research Council of Australia.

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 with 18 U.S.C. Section 1734 solely to indicate this fact.