Sunbeds and Melanoma Risk: Many Open Questions, Not Yet Time to Close the Debate

Results and Discussion

The literature search process identified 103 articles. Depending on the content, title, abstract and/or full text were analyzed to identify relevant new findings. The updated information that these articles gave is summarized in the following paragraphs. Overall, we found no studies that demonstrate a causal relationship between moderate solarium use and melanoma risk.

Our analysis of the report of Suppa and Gandini (1) revealed severe weaknesses (Table I). The criticism previously addressed by us (8, 9) and others (7) has not been adequately addressed in the report of Suppa and Gandini (1) or by others (2, 3, 5, 6). The conclusions reported in those publications (1-3, 5, 6) are not sufficiently supported by present scientific knowledge and therefore are not in agreement with generally accepted principles of evidence-based medicine.

Randomized controlled trials are able to provide unequivocal proof of a causal relationship between moderate solarium use and melanoma risk. However, such trials are lacking [reviewed in (8, 9) for various reasons: a) they are unfeasible-they take too long, are too costly, and too demanding on compliance; and b) they would now be considered unethical by many, as voiced by Suppa and Gandini (1). The results of cohort and case–control studies do not prove causality [reviewed in (8, 9)], not even by the Hill criteria (22), contrary to what is suggested by Suppa and Gandini (1) (see below). Moreover, the overall quality of those observational studies and the resulting evidence levels are low as a result of severe limitations (including unobserved or unrecorded confounding), leading to bias [reviewed in (8, 9)].

In most studies published to date, many of the confounding factors, including sun exposure, sunburn, and skin type, have not been adequately and systematically recorded and adjusted for [reviewed in (8, 9)]. As pointed out in our previous meta-analysis (8), only a few studies published so far reported odds ratios (ORs) adjusted for the same confounding factors. As many as 35.5% (n=11) of all (n=31) studies included in that meta-analysis (8) did not account for a single confounder. The remaining studies (n=20) adjusted mainly for age (n=15), sex (n=11), and skin color (n=11). Fewer studies adjusted for hair color (n=10), sun exposure (n=8), sunburn (n=8), family history of melanoma (n=7), nevi (n=7), freckles (n=5), and education (n=5). Moreover, individual confounders were assessed differently across studies included in that meta-analysis (8) and were only partly comparable.

In this context, risk estimates (e.g. ORs) as given in meta-analyses published to date, including those by Burgard and colleagues (8), Boniol and colleagues (11) and Colantonio and colleagues (7), might well have been affected by the issues of lack of standardization in terms of confounding factors for sunbed studies [8,9] and could well be obtained through the scenario indicated before (8, 9): Moderate sunbed use has no effect on melanoma risk, but an “unhealthy lifestyle” (e.g. extensive sunbathing, alcohol, smoking) resulted in an inflated OR of 1.2 in association with sunbed use [“sun worshippers” and individuals with an unhealthy lifestyle go more often to tanning salons (8, 9)].

The gap between sunbed studies and earlier studies on risk of sun exposure remains remarkable. Confounding caused by exposure to the sun – the major UV source – is often neglected or corrected for inadequately. Most often, such work lacks a proper analysis of covariance (collinearity or other) to eliminate a possible dominance of sun over sunbed exposure due to an a priori highly plausible strong correlation between sunbed use and sunbathing; OR=2 to 7 for sunbathing among sunbed users versus nonusers (12-14). As pointed out in a recent French study (15), sunbeds were estimated to have only a minor contribution to melanoma incidence (1.5% in men and 4.6% in women) compared with the sun (83%); that is, sunbeds were not likely to be a major driver of the increase in melanoma incidence. Moreover, the authors noted that disentangling risk from use of sunbeds from that of the sun is difficult [anecdotal attribution of melanoma to sunbed use is often offset by excessive sunbathing, as exemplified by an Australian publicity campaign to regulate solaria (16)]. The importance of solar UV exposure and sunburn for melanoma development is reflected by the frequency of melanomas that occur after the diagnosis of a cutaneous melanoma, reported to range from 8.2% of previously diagnosed melanoma in European countries to 23% in countries with more intense ambient solar UV radiation (34-42). Earlier studies identified the number of sunburns as a good proxy of “at risk” sun exposure in relation to melanoma (17, 18). Virtually all studies on sunbed and melanoma fail to use that proper proxy of effective UV dosimetry, except for two studies that confirmed a strong relationship between UV burns and melanoma risk (19, 20). That finding would imply that UV burns specifically increase melanoma risk, but it should be considered that genuine sunburn is far more common than UV burns from sunbeds. The confounding effects of sun exposure and impact of UV burns would also be compelling reasons why melanoma risk in relation to sunbed use varies so strongly between studies: Meta-analyses of European studies showed no net significant melanoma risk associated with sunbed use, in contrast to U.S. and Australian studies (7, 8).

We also disagree with the conclusion of Suppa and Gandini (1) that the criteria defined by Hill (22) [or as modified by Weed and Gorelic (23)] to provide epidemiological evidence of a causal relationship between a potential cause and an observed effect are fulfilled for the inference that moderate solarium use per se increases melanoma risk (Table I). The following criteria are not fulfilled for the relationship between moderate solarium use and melanoma risk:

• Consistency (consistent findings observed by different persons in different places with different samples strengthen the likelihood of an effect)

• Specificity (causation is likely if there is a very specific population at a specific site and disease with no other likely explanation)

• Plausibility (a plausible mechanism between cause and effect is helpful in determining causality)

• Coherence (coherence between epidemiological and laboratory findings increases the likelihood of a causal effect)

• Experiment (experimental evidence is helpful in determining causality)

Therefore, based on Hill's criteria, the evidence does not support causality. Consistency and specificity are not fulfilled for many reasons, including the obvious difficulties of confounding factors. In a recent meta-analysis, subgroup analyses for studies performed in Europe, studies with low risk of bias, and studies with recruitment between 1991 and 1999 showed no association between melanoma risk and solarium use (ever vs. never) (8). The lack of association in the subgroup analysis is unlikely to be caused by a lack of power, for example, because the number of participants in studies performed in Europe is much greater than in U.S. studies. The lack of association in studies performed in Europe may be due to several factors. Firstly, as outlined above, the role of solar UV exposure represents a major confounding factor that is difficult to document or adjust for and that may well in part explain why latitude-dependent variations in melanoma risk in association with sunbed arise (e.g. due to shifts in effects from sunburns).

However, other region-specific factors, which include technical differences in solarium devices, must also be taken into account, as must skin type, also an important confounding factor. Since 2008, solarium devices in Europe and Oceania (Australia and New Zealand) are restricted in intensity to a UV index of 12 and 36 (which was 60 before 2002), respectively. In contrast, the intensity of a U.S. solarium is not restricted, but a maximum recommended exposure time is often given. In a previous meta-analysis, another observation of sensitivity and subgroup analyses was the finding that the recruitment period strongly affected the association of melanoma risk with solarium use (17). For recruitment before 1991, a higher OR was found than for recruitment from 1991 to 1999 or since 2000. This observation might be due to changes in operation and technical modifications of UV-emitting devices (about two decades ago, the solarium industry started to produce devices with higher pressure bulbs emitting larger doses of long-wave UVA). Moreover, as shown in our recent meta-analysis (8), many published studies most likely overestimated the association of melanoma risk with solarium use.

A large body of evidence from epidemiological and animal studies demonstrates no increase in melanoma risk after chronic (moderate) UV exposure (24-31). Many studies show that suberythemal chronic exposure to the sun may even be protective and that outdoor workers may have a reduced risk of melanoma (18, 27-30). Driver mutations in the B-rapidly accelerated fibrosarcoma (BRAF) gene or in other important drivers of melanomagenesis do not carry the specific UV signature [mutations in BRAF are similar to those found in guanine nucleotide-binding protein subunit alpha-11 (GNA11) and GNAQ driver genes in uveal melanomas from UV-protected parts of the inner eye (31)]. Initiating melanoma by UV exposure in mice without predisposition by an activated oncogene proved difficult [with few exceptions, e.g. by neonatal exposure of inhibitor of cyclin-dependent kinase 4-alternate open reading frame-null xeroderma pigmentosum, complementation group C-null (Ink4a-Arf^−/−Xpc^−/−) mice (32) and incidentally successful, 3/20, with repeated sunburn exposure (33)].

The results of many recent investigations clearly argue against the hypothesis that moderate solarium use may increase melanoma risk. Unfortunately, however, the review of Suppa and Gandini (1) does not adequately consider those studies. For example, the retrospective case–control study of Müller and colleagues investigated risk factors for subsequent primary melanomas after the diagnosis of a cutaneous melanoma in Austria (1,648 participants with histologically verified cutaneous melanoma, including 1,349 with single and 299 with multiple primary melanomas) (34). Suppa and Gandini mentioned the study, but the main findings were not adequately presented (1). In that study, solarium use was not associated with an increased but rather a reduced risk for developing melanoma (OR=0.71; 95% confidence interval=0.51–0.99; p=0.04) (34). Those findings indicate that solarium use may even have a protective effect.

In summary, the main conclusions drawn by Suppa and Gandini (1) and others (2-6) are scientifically not proven and are not in accordance with generally accepted principles of evidence-based medicine. They are not in line with recommendations of the Oxford Centre for Evidence-based Medicine (21). Furthermore, as outlined in this minireview (Table I) and previously (9), the conclusions do not fulfill Hill's criteria for plausibility in a biological system (22). Other researchers added the ruling out of confounding factors and bias (23). The review of Suppa and Gandini (1), as well as recent publications by others (2, 3, 5, 6) are therefore not correct in reaching their conclusions, and the debate is not closed.