Abstract
Sun exposure
Epidemiological and latitude studies have consistently shown a higher incidence of melanoma in individuals with light skin types and in countries at lower latitudes. 1–4 Melanoma is more common in fair-skinned subjects and very rare in non-Caucasian populations, supporting the role of sunlight as a risk factor for this tumour. 5 In non-Caucasians, however, melanoma is a very different tumour as it affects the palms and soles and is extremely rare. 6,7 Most epidemiological studies in Caucasian populations demonstrate sunburns, and short sharp bursts of sun exposure in childhood are more significantly associated with melanoma (albeit with low relative risks) than cumulative lifetime exposure. 8 Occupational exposure, however, may be protective for melanoma especially in good tanners. The relationship between sunlight and melanoma is complex. 8 Even for recreational exposure, the association with melanoma is controversial as the magnitude of the relative risks in relation to recreational sun exposure is low and not consistent between studies. 8 Furthermore, the risks are often no longer significant once adjusted for the ease of tanning, defined as skin type (Fitzpatrick classification). 9 Sunburns especially in childhood have also been associated with melanoma but the magnitude of the relative risks is again low, around 1.5. 8 Although sun exposure may be relevant in melanoma, this does not necessarily imply that a particular dose of ultraviolet radiation (UVR) will result in melanoma. There may be a genetic susceptibility to ultraviolet radiation which varies greatly between individuals. Furthermore, genetic susceptibility for melanoma may not always correlate with skin type (ability to tan using the Fitzpatrick classification).
A rise in melanoma incidence over the last 20 or 30 years corresponds with a change in our sun-seeking behaviour over the same period with more holidays abroad and a desire to tan. This fact is also used to support the view that sun exposure is causing the observed rise in melanoma. What is not really addressed is the fact that melanoma mortality is not changing in parallel with incidence as mortality is remaining stable or even decreasing in many parts of the world. 9,10 There is therefore a widening gap between incidence and mortality curves which needs explaining. If sunlight caused a rising melanoma incidence because of our changing lifestyles as suggested, there are no obvious reasons why this increase would not include invasive tumours with greater metastatic potential which in turn would have a significant impact on mortality. This is not the case, as all incidence data in countries where the rise in incidence has been the greatest has shown that thin melanomas are responsible for the bulk of the extra tumours seen (including melanoma in situ). Thinner tumours have a better prognosis and have therefore little impact on mortality. 2,11 Some may argue that we are now picking up all melanomas much earlier so this is why the burden of melanoma is mainly for early disease. While this may be partly true, it is not conceivable biologically that the extra burden of melanomas over the last 20 years would only include thin melanomas with little metastatic potential. The more logical explanation is that the increase in incidence seen over the last 20 years has been, in part, artificially caused by increased ascertainment with surveillance and low threshold for biopsies. It is therefore likely that these lesions may have never progressed if left alone. This would also explain why in countries where public health campaigns are most active such as Australia and the USA, the surveillance and screening leads to the removal of large numbers of pigmented lesions including very thin melanomas and this, in turn, further supports the theory that melanoma incidence rises because of increased sun exposure in these parts of the world. 2,12–14 In countries where access to dermatology is difficult and where public health campaigns are not as widespread, much lower incidence rates of melanoma are reported often with thicker tumours. Mortality rates are similar or higher supporting the view that screening is partly responsible for the rapid rise in melanoma in more affluent countries. 11
The biological behaviour of melanoma is also helpful in understanding the relative contribution of genetic and environmental factors. Mean age at diagnosis, distribution of melanoma tumours according to gender and histological subtypes are all comparable in many parts of the world (within Caucasian populations) despite different patterns of sun exposure so melanoma behaviour is not significantly affected by exposure. Melanoma sites in men and women are also very different with women more likely to have their melanoma on the lower legs while men are more likely to have their melanoma on the trunk. 15 This difference according to gender is again very similar across all latitudes showing that sun exposure does not really affect body site significantly and that other gender-related factors affect melanoma behaviour. Women survive melanoma better than men and this is despite adjusting for all melanoma prognostic factors so gender effects are also important for survival. 16,17
Naevi and melanoma
Naevi are benign melanocytic tumours which start to appear in childhood, rise in number in early adulthood, stabilize in the mid-30s to decrease gradually in late adulthood. Naevi therefore senesce with age and the speed at which they disappear with age varies greatly between individuals. 18 The distribution of naevi on the body is affected by gender with girls having more naevi on the limbs while boys have more naevi on the trunk. 19 Interestingly, this reflects the distribution of melanoma in men and women and is consistent in all Caucasian populations despite different levels of ultraviolet radiation. Why gender affects distribution of naevi is not known but this is unlikely to be explained by behaviour in the sun as this difference by gender is not affected by latitude and is also seen in countries where clothing covers large parts of the body. The explanation is more likely to be genetic differences in melanocyte differentiation between boys and girls. Naevi numbers and size are strongly influenced by genetic factors with 60% of the variation explained by genes. 20,21 It is therefore clear that genetic factors are very important in naevi development and involution and that although environmental factors may play a role as well, these are likely to be minor. Sunscreens have not been shown to produce any obvious reduction in naevi numbers with age in randomized studies of sunscreen use in children so the role of UV in naevi induction is unlikely to be significant. 22
Naevi are the strongest risk factor for melanoma with odds ratios in the order of 5 to 10 which is far greater than any relative risks associated with sun exposure. 23,24 Risk factors associated with an excess of naevi are consistent across the world even with variable levels of sun exposure so latitude does not appear to be relevant. 23,24 Although some studies have suggested that naevi are slightly more numerous in Australia compared to the UK, this mainly relates to small junctional naevi. 19,25 There has also been the issue of mole counting in Australia which is likely to be overestimated as many solar lentigines may be counted as melanocytic lesions in adulthood. There are no significant differences in the magnitude of relative risk for developing naevi in Australia compared to the UK, in populations with similar genetic pools and susceptibility to melanoma, i.e. mainly of British and Northern European ancestry. 23,24,26
The atypical mole syndrome is the presence of multiple both small and atypical naevi, is a powerful predictor of risk for melanoma and is consistently associated with melanoma in parts of the world that have different levels of sun exposure. 26,27 Indeed genetic research shows similar founder mutations in p16 or CDKN2A (an important tumour suppressor gene affecting the germline in melanoma families) which result in an increased melanoma risk detected in Europe, the USA and Australia. 27 Many melanoma families, with or without atypical naevi may have mutations in the tumour suppressor gene p16, on chromosome 9p21 and the prevalence of the mutation depends on the number of melanomas within a family. 27 Overall the mutation is found in 25% of all melanoma families worldwide so other genes are likely to be involved. It is likely that melanoma is a genetically heterogeneous tumour with different at-risk phenotypes. Testing for p16 mutation is not clinically indicated in melanoma patients with a positive family history as the chance of detecting a mutation is small and does not change the management of the patient. 27 The atypical mole syndrome can also be found in individuals belonging to rare family cancer syndromes such as neurofibromatosis, retinoblastoma and Li-Fraumeni syndrome but is also seen in families with an excess of cancers such as breast, pancreas, brain tumours and osteosarcoma suggesting that multiple atypical naevi may be a marker of cancer susceptibility in general and not only melanoma. 28 Individuals with BRAC2 and BRCA1 mutations also have an increased risk of melanoma of the skin but also the eye. 29
Skin pigmentation with susceptibility to sunlight
Having fair skin, fair hair and blue eyes makes an individual more prone to melanoma and this association is also used to support the role of sun exposure in melanoma causation. 5,30 However, research in skin and hair pigmentation has revealed that the link with melanoma may not be explained by pigmentation alone. Fair skin types are more likely to have several variants in the MC1R receptor (Melanocortin 1 Receptor) which at first seemed to indicate that these polymorphisms would be useful in detecting melanoma susceptibility. However, the attributable proportion of melanomas explained by these MC1R variants is very small so would not be useful in population screening. Furthermore, some of these polymorphisms are associated with melanoma risk but are not always predictive of fair skin/hair so the association may also be explained by factors other than pigmentation. 31 The research in this field is, in the meantime, revealing interesting associations between polymorphisms in MC1R and the oncogene BRAF. The BRAF oncogene is mutated in over 70% of melanoma tumours. 32 Furthermore, melanomas in individuals with MC1R variants are more likely to show BRAF mutations than those without MC1R variants which highlight potential links between the MSH/MC1R pathways and the RAS/BRAF pathways.
Another phenotype related to melanoma in patients with fair hair and skin is solar damage which causes solar elastosis, solar lentigines and solar keratosis. 25,30 Melanoma in these patients tend to occur in older age groups, affect the face or chronically sun-exposed sites and are more likely to be of the lentigo maligna or nodular type rather than superficial spreading melanoma. 25,33 These melanomas also have better prognosis. Sun exposure may therefore be more relevant for these types of melanomas as these patients are also susceptible to non-melanoma skin cancers. These tumours are also more likely to show increased p53 expression (another oncogene involved in many cancers) which is more rarely seen in melanomas affecting younger individuals without sun damage. This again supports the theory of different genetic pathways leading to melanoma.
Sun avoidance and vitamin D deficiency
There are concerns that public health campaigns advocating sun avoidance may be detrimental in the long term as this may lead to vitamin D deficiency in temperate climates. 34 If a drastic reduction in sun exposure occurs, vitamin D may become too low in winter with potential deleterious health effects. Although many argue that vitamin D supplements or diet should easily address the effects of sun avoidance campaigns, many Caucasian populations do not take vitamin D supplementation and are not eating oily fish which is the main source of vitamin D from food. The evidence that vitamin D is crucial for health is now very well documented with deleterious effects including an increased susceptibility to cancer, metabolic syndrome, autoimmune disease, bone loss and psychiatric diseases in those with low levels. 35,36 It is clear that it would be counter-productive to recommend excessive sun exposure without precautions as the burden of non-melanoma skin cancer and pre-cancerous skin lesions is significant for most health organizations and can be associated with significant morbidity. There is a need to reassess what the level of evidence is to justify the campaigns stating that melanoma is caused by sunlight as this is may not be entirely accurate and may have detrimental effects in the long term, especially in temperate climates.
Sentinel lymph node biopsy (SLNB) controversies
There is a wide spectrum of opinion regarding the role of this procedure in the management of malignant melanoma. Preoperative lymphoscintigraphy and intraoperative lymphatic mapping with vital dyes are used to track the drainage from the site of the melanoma to the first lymph node which is termed sentinel. 37 If the sentinel lymph node is positive patients will then undergo complete lymph node dissection of this drainage basin.
Interpretation of SLNB can be difficult due to the fact that melanoma can metastasize in small groups of cells; additionally benign melanocytic naevus cells can be found in sentinel nodes. The location of the cells within the node can help to differentiate naevus cells from melanoma. Several techniques are used to detect melanoma cells apart from conventional staining with haematoxylin and eosin. Immunohistochemical staining with antibodies to HMB45, melanA and S100 is carried out widely. Each special stain has differing levels of sensitivity and specificity, and a combination of stains is often used to clarify the result. MIB1 can be used as an indicator of cell proliferation. Molecular biology techniques such as reverse transcriptase polymerase chain reaction (rtPCR) are associated with better diagnostic sensitivity and have been used in the detection of melanoma in sentinel lymph nodes. There have been concerns about the false-positive rates 38 and as such the role for rtPCR remains uncertain. 39
Prognosis is obviously pivotal in the patient's mind. Prognostic information is also important for clinicians to determine appropriate follow-up, potential future adjuvant treatment and is often used to define inclusion criteria for trials. With experience SLNB has a low incidence of well-documented complications and side-effects. What is of concern is the number of sentinel node biopsies leading to complete lymphadenectomy when there is evidence that in a significant percentage of patients this procedure may not be needed. 40
The use of SLNB in staging and prognosis of melanoma is widely accepted and forms the basis of the American Joint Committee on Cancer (AJCC) staging criteria. 41 However, in Europe, and in the UK especially, the value of sentinel node biopsy is being challenged not least because of the morbidity of the procedure (especially when leading to a full lymphadenectomy) but also because of the lack of evidence so far that this technique increases melanoma survival. The interpretation and analysis of the multicentre selective lymphadenectomy trial, which is the only published randomized controlled trial comparing SLNB with nodal observation in melanoma, has provoked much of the current debate. Over 2000 patients with intermediate thickness melanoma were randomized into two groups. In the treatment arm, patients underwent SLNB and, if positive, complete lymphadenectomy. The control group were given lymphadenectomy only if they developed palpable nodal metastases. No significant difference was observed between the groups in the primary endpoint of melanoma specific survival (i.e. death due to melanoma) after follow-up five years later. 42 Differing levels of disease-free survival have been found only in the analysis of non-randomized subgroups.
The assumption that SLNB will improve survival relies on the assumption that all tumour cells passing through a lymph node will lead to eventual haematogenous spread. This may of course happen in a proportion of cells but there is no evidence to suggest that passing through a lymph node is a prerequisite for blood-borne spread. 43 Furthermore, the detection of melanoma cells in a lymph node does not always imply that these will lead to palpable lymph nodes and further spread as this all depends on the site and size of the melanoma deposit in the lymph node. 44 There are alternatives to this stepwise spread (incubator hypothesis) which proposes a lymphatic and haematogenous spread occurring simultaneously. 45 The reality may be that melanoma does not fit into any one of these models. Each melanoma may have differing dissemination potential and based on the ability of the melanoma cell to spread to lymph, blood and viscera. 43 This is the reason why it is difficult to predict the behavior of these melanoma cells in individuals cases especially when they are in small numbers and subcapsular. 44
However, 10 years of follow-up of the MSLT-1 groups demonstrates that the incidence of nodal metastases in the observation group will eventually equal the total incidence of nodal metastases detected by biopsy or by biopsy with false-negative results. 40 There is no survival advantage at present in doing sentinel node biopsies and eventually the nodes will become palpable which can lead to complete lymphadenectomy at that stage so many argue that clinical detection of enlarged lymph nodes alone should be the indication for complete clearance of the node basin.
Until more evidence is available, SLNB remains at least a reliable staging tool in melanoma. The problem is that, at present, this investigation is not used solely as a staging tool, as almost all positive sentinel node biopsies lead to lymphadenectomy with significant morbidity. This is especially relevant when this technique is offered for thin melanomas between 0.75 and 1 mm in thickness as these have a low propensity to spread and the positive sentinel may represent a false-positive result or may never lead to further spread of the disease.