Customize your JAMA Network experience by selecting one or more topics from the list below.
Gaudy-Marqueste C, Madjlessi N, Guillot B, Avril M, Grob J. Risk Factors in Elderly People for Lentigo Maligna Compared With Other Melanomas: A Double Case-Control Study. Arch Dermatol. 2009;145(4):418–423. doi:10.1001/archdermatol.2009.1
To assess lentigo maligna (LM) as an epidemiological entity separate from other melanomas (OMs) in elderly people.
Double age- and sex-matched case-control study to compare the risk factors for LMs and OMs.
A total of 76 patients with LM were paired by age and sex with 76 patients with OMs and 152 controls.
Main Outcome Measures
The association of melanoma risk with the following potential risk factors: sun exposure history by 10-year periods, frequency of sunburns, phenotypic traits, density of freckles and sun sensitivity at age 20 years, counts of nevi larger than 2 mm in diameter on the face and forearm, skin aging features (as assessed using a photographic scale), and history of basal and/or squamous cell carcinomas.
Risk of LMs and OMs were similarly associated with history of sunburns, light skin type, and freckling. Cumulative chronic outdoor and occupational sun exposures were not risk factors in any of the 2 groups of melanomas. Lentigo maligna differed from OMs by the absence of a detectable association with the number of nevi and a greater association with nonmelanoma skin cancers.
Although chronically sun-exposed skin is a prerequisite for LM, risk of LM does not increase with the cumulative dose of sun exposure, but LM is associated with sunburn history, like all other types of melanomas. The main epidemiological characteristic of LM is the absence of an apparent relation with the genetic propensity to develop nevi. This epidemiological profile is in accordance with recent molecular findings and may also account for the histoclinical and evolutive characteristics of LM.
Melanoma in elderly people has become a public health issue because of increasing incidence,1 long diagnostic delays,2 and poor prognosis.3 Lentigo maligna (LM) is generally considered a separate epidemiological subtype of melanoma owing to its specificity for elderly people, slow growth, and apparent link with cumulative sun exposure. However, this epidemiological profile may only be a tautology resulting from the LM histopathological criteria themselves, ie, lentiginous melanocytic hyperplasia on a marked UV-induced dermal damage.4 Melanomas with lentiginous hyperplasia of melanocytes4 are likely to be those with a prolonged horizontal growth phase and thus progress slowly. Melanomas with a marked UV-induced dermal damage are likely to be located on the head and neck in elderly people. This location has led to the hypothesis of a particular link of LMs with cumulative sun exposure, while case-control studies suggested that LM risk may mostly be related to sunburns.5,6
Whether LM really represents a separate epidemiological subtype of melanoma or just an histopathological variant among melanomas in elderly people remains questionable. This question has become more crucial since molecular specificities have been found in LM.7,8 We thus designed a double case–control study to compare the risk factors for LMs and for other melanomas (OMs) in elderly people.
A double age- and sex-matched case-control study was conducted in patients older than 65 years to assess risk factors for LM compared with risk factors for OMs. Because this study involved no intervention, it did not require approval of ethical committee, according to French regulation.
The files of patients with a first melanoma diagnosed after the age of 65 years and seen in the last 3 years were systematically collected in 3 French hospitals (Centre Hospitalier Universitaire [CHU] de Marseille, Marseille; Institut Gustave-Roussy, Paris; and CHU de Montpellier, Montpellier). Any melanoma considered to be LM by the histopathologists in the 3 institutions was defined as a case. Pathological reports were all reviewed by one of us (N.M.) and classified into the following 3 categories: (1) typical description of LM in situ or invasive, (2) typical description of OMs (superficial spreading [SSM] and nodular [NM] subtypes), and (3) apparently ambiguous cases including LM diagnoses without characteristic description and melanomas classified as SSM or NM but with some lentiginous component. Patients with typical cases of LM and OMs from the 3 hospitals were grouped by age (groups of 5-year age intervals) and sex. All of those patients were contacted by telephone. Following this telephone interview, all of those who were considered to be able to answer questions and who volunteered for either a visit to the hospital or for a consultation by the physician at home were enrolled in the study.
For each LM case examined and interviewed, a patient with OM within the same age and sex group was randomly selected from the list and contacted by telephone. If this patient could not be contacted, was unable to answer the telephone, or refused to visit the hospital, the next on the list was contacted.
Two age- and sex-matched controls were selected for each LM-OM pair within the consultation lists of the same hospitals as the patients with LM and those with OMs. People with the following characteristics were excluded: (1) those already consulting a physician in the Department of Dermatology of the 3 hospitals; (2) those with a melanoma before the age of 65 years; (3) those with a chronic disease in the first 50 years of their lives; (4) those with a history of chemotherapy or long-term immunosuppressive treatment; and (5) those with a history of laser resurfacing, dermabrasion, or chemical peeling.
All cases and controls were examined by the same dermatologist (N.M.). Skin aging features were assessed using a photographic scale.9 Nevi larger than 2 mm in diameter were counted on the face and left forearm.10 Respondents were interviewed using a standardized questionnaire. First, they were asked to recall their phenotypic traits when they were 20 years old: (1) their skin color and hair color were assessed using a photographic scale; (2) the density of freckles was assessed using a pictorial scale,11 and (3) their typical reaction to sun exposure was determined according to Fitzpatrick classification.12
Second, they were asked whether they previously had skin lesions removed or destroyed, apart from moles (pooling actinic keratoses, nonmelanoma skin cancers [NMSCs], and seborrheic keratoses) and/or had been treated for basal and/or squamous cell carcinomas. No research in files or laboratory reports was done to ascertain this information. Third, a brief sun exposure history was obtained by 10-year intervals and included (1) living place; (2) frequency (0-4) of sunburns (defined as episodes of painful erythema and/or scaling after sun exposure); (3) periods of outdoor occupation and mean occupational time spent outside per week; and (4) mean time spent outdoor in weekends and vacations. All these sun exposure times were transformed into a dose expressed in minimal erythemal doses, using geographical ambient sun exposure tables13 to estimate a cumulative lifetime dose received during outdoor activities, outdoor occupation, and everyday activities.2 In addition doses were estimated separately for the first 20 years of life and for the last 20 years.
Univariate analyses were performed to determinate risk factors for OMs and LM among quantitative and qualitative variables. Exposure to risk factors in patients with LMs and those with OMs, adjusted by age and sex, were compared using the McNemar test for qualitative variable and Wilcoxon test for quantitative variables. Variables with significant odds ratios in the univariate analysis were further candidates to determine independent risk factors for LMs and OMs in 2 multivariate analyses using logistic regression with a forward step-by-step approach.
A total of 76 pairs of LM and OM cases were constituted, with 152 controls. Among the 76 cases of LM, 29 were in situ (LM) and 47 invasive (LM melanoma). Among the OMs, 43 were SSMs (57%), 15 were NMs (20%), 9 were acrolentiginous (12%), and 9 unclassified (12%).
The mean age was 71.6 years for the patients with LM, 70.8 years for those with OMs, and 74. 5 years for controls; these 3 groups comprised 44% men and 56% women. Familial history of melanoma was reported for 4 LM cases and 4 OM cases (5% each).
Most of the LMs (n = 68 [89%]) vs a minority of OMs (n = 10 [13%]) were located on the face. Conversely, 2 (3%) of the LMs vs 25 (33%) of OMs were located on the trunk (shoulders included).
The OMs as well as the LMs were significantly associated with clear or pale skin and with red or blond hair at age 20 years, although this was only a trend for LM. Only LMs were significantly associated with clear eyes, with low tanning ability, and with a high density of freckles at age 20 years (Table 1).
Patients with OMs had significantly more nevi larger than 2 mm in diameter on the forearms (mean number, 1.04) compared with controls (mean number, 0.34) and patients with LMs (mean number, 0.32), whereas patients with LMs did not differ from controls. Patients with OMs had significantly more nevi larger than 2 mm in diameter on the face (mean number, 1.34) compared with patients with LMs (mean number, 0.99), but they did not significantly differ from controls (Table 1).
Compared with controls, patients with OMs and those with LMs showed a significantly higher number of senile lentigines on the dorsum of the hands, and a significantly higher frequency of erythrosis interfollicularis coli and guttate hypomelanosis. Only OMs were associated with higher frequency of cutis rhomboidalis nuchae and higher number of senile lentigines on the forearms. No significant differences in actinic keratosis and wrinkles density were observed among the 3 groups (data not shown) (Table 1).
More patients (with OMs or LMs) than controls recalled the removal of skin lesions and NMSCs. More patients with LMs recalled undergoing an NMSC resection compared with patients with OMs (Table 1).
The frequency of sunburns represented a significant risk factor for LMs and OMs. There was no clear relation between LMs or OMs and cumulative sun exposure (including leisure, occupational, and daily ambient exposure) (Table 2).
The best predictive factors for OM retained by the logistic regression model (Table 3) were a history of NMSCs (P < .001), a history of other skin lesions removed (P = .005), the presence of a cutis rhomboidalis nuchae (P = .002), the number of nevi on the forearms (P = .003), the frequency of sunburns in the last 20 years (P < .001), and a pale skin type (P = .03). The best predictive variables for LMs (Table 4) were a history of NMSCs (P < .001), the frequency of sunburns during the first 20 years (P < .001), and the presence of guttate hypomelanosis (P = .03). The predictive accuracy of the model was satisfactory, with an area under the curve of 0.83 for OM (95% confidence interval CI, 0.78-0.89) and 0.81 for LM (95% confidence interval, 0.75-0.86).
Our data support the idea that the risks factors for LM are highly similar to those for other subtypes of melanoma (SSM and NM), whatever the age at which they occur (ie, sunburns,14 light skin type, and tendency to freckle).15 The main difference between LMs and OMs was the absence of any detectable association between LM and number of nevi and an apparently tighter association of LM with a history of NMSC.
We expected a direct link between LM and chronic outdoor or occupational exposure,16 because of the usual development of LM in the chronically sun-exposed areas in elderly people. However, our study shows that LMs do not grow more often in the elderly people who have received more cumulative sun exposure than the others at the same age. Advanced chronic sun damage may be a prerequisite for LM, but there is no dose-effect relationship with cumulative exposure. Conversely, a clear link was found between the risk of LM and history of sunburns. The multivariate model even retained sunburns in the first 20 years of life as one of the best predictors of LMs, such as with other types of melanomas.14 This association between LM and sunburns has already been suggested in previous case-control studies.4
Although the numeric differences of nevi in cases of LM and OM and the controls were small, there was a link between the number of nevi and the risk of OM but not with the risk of LM. This is in line with a previous study showing that melanomas located on the head and neck are less likely to have a higher number of nevi than SSM or NM located on the trunk.17
These results raise very important questions. Why are LMs located almost exclusively on the face, although risk increases with sunburns and not with the amount of chronic sun exposure? How can we explain the absence of a detectable link between LM risk and nevi? The “dual pathway” hypothesis18 may provide an explanation. Whiteman et al18 distinguished “nevus-prone” people who can develop a melanoma, growing from a nevus on the trunk early in life after minimal sun exposure, and “nevus-resistant” people who require a lot of sun exposure to develop a melanoma on a chronic sun-exposed area. This hypothesis is supported by the different epidemiologic profiles of melanomas associated with nevi and de novo melanomas.19
Lentigo maligna could represent a particular type of melanoma, without any link to genetic propensity to develop nevi, and could thus develop even in people with no genetic propensity at all to develop nevi and melanomas. In this situation, generation of a melanoma could require not only a history of sunburns but additional promoting factors, such as a high density of mutations, which is only encountered in regularly exposed skin after decades of sun exposure,20 the so-called cancerization field. This assumption is supported by the particular link that was observed between LM risk and NMSCs in both the univariate and multivariate analyses. This is also in line with recent studies showing that melanomas occurring on sun-exposed areas display a molecular profile different from that of OMs, with a lower incidence of BRAF mutations and higher incidence of p53 mutations.7,8 This hypothesis could explain not only the location of LMs on the face of elderly people but also, indirectly, the additional particular features that have justified the individualization of LMs among OMs. Indeed, epidermal atrophy of severely UV-damaged skin could influence the pattern of the melanocytes growth and force it to a lentiginous expansion. In such melanomas independent from nevus propensity, the stability of the melanocytic system may also contribute to the usual slowness of the LM progression.
The dual pathway theory may however be an oversimplification of epidemiological and molecular subtypes of melanomas. The first pathway linked to nevus could in fact cover a spectrum of propensity to develop nevus and melanoma. The upper extremity of the spectrum may be represented by the most melanoma- and nevus-prone patients with dysplastic nevus syndrome, who are likely to develop a melanoma early in life. Down on the spectrum are less melanoma- and nevus-prone people who may be less and less likely to develop a melanoma, and if they do will develop it later in life. The OMs of the elderly people in the present study probably correspond to the lowest part of this spectrum. The LMs may represent a second pathway totally independent from nevus propensity. Other pathways are likely for mucosal melanomas or acrolentiginous melanoma.21
One possible bias affecting the results of this study comes from the fact that it partly relied on long-past retrospective information (sun exposure), potentially altered by the memory of the elderly participants, as well as on more direct information potentially transformed by the aging process (eg, number of nevi, hair color). Efforts were made to limit these biases. Patients with memory problems were excluded. Some variables that could not be directly assessed because of their modification in old age (eg, freckling, hair color) were likely to be correctly recalled. We thus used a questionnaire with visual supports for variables (hair color panel and freckling scale) and probably obtained a good reflection of the real phenotype at age 20 years. Conversely, we considered that patients older than 65 years could not be reliable in the estimation of their nevi at age 20 years. Although the number of nevi decreases with age and is thus less and less representative of the nevus propensity, we chose to count the nevi at the inclusion of the patients, assuming that the groups with more nevi later in life were also the groups who had more nevi earlier in life. For simplicity and to ensure acceptance by all individuals, the nevi were only counted on the face and the arms, which has been shown to be representative of the whole nevus count.22,23 History of skin cancer as reported by patients was considered as trustworthy.24 The fact that the same investigator performed both the examination and the questionnaire in controls and patients was a major advantage in the consistency of the measures, but the investigator could not be really blinded for case or control status. However, if this should have resulted in a bias, it would have been in a way compatible with our initial hypotheses (ie, that there is a link between LM and cumulative sun exposure and no link between number of nevi and LM or OM in patients older than 65 years), which were not confirmed by the results. Finally, when dealing with 2 groups of cancers in elderly people, a retrospective inclusion of only surviving patients may clearly bias the results. This is the reason this study was voluntarily limited to melanomas diagnosed in the last 3 years.
In conclusion, this epidemiological study suggests that LM differs from OMs, not by an association with a different type of sun exposure, but by an absence of link to genetic propensity to develop nevi. This specificity is in line with recent molecular findings and may also account for most of the histoclinical and evolutive specificities of LM.
Correspondence: Caroline Gaudy-Marqueste, MD, Department of Dermatology, Hôpital Sainte Marguerite, 270 Blvd de Sainte Marguerite, Marseille 13009, France (email@example.com).
Accepted for Publication: November 26, 2008.
Author Contributions: Dr Gaudy-Marqueste had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Guillot, Avril, and Grob. Acquisition of data: Madjlessi. Analysis and interpretation of data: Gaudy-Marqueste, Madjlessi, Guillot, Avril, and Grob. Drafting of the manuscript: Gaudy-Marqueste, Madjlessi, and Grob. Critical revision of the manuscript for important intellectual content: Guillot, Avril, and Grob. Study supervision: Grob.
Financial Disclosure: None reported.
Additional Contributions: J. Gouvernet, MD, and R. Giorgi, MD, provided statistical support.
This article was corrected online for typographical errors on 4/20/2009.