The naevus count on the arms as a predictor of the number of melanocytic naevi on the whole body

Transcription

1 British Journal of Dermatology 1999; 140: The naevus count on the arms as a predictor of the number of melanocytic naevi on the whole body C.FARIÑAS-ÁLVAREZ, J.M.RÓDENAS,* M.T.HERRANZ AND M.DELGADO-RODRÍGUEZ Division of Preventive Medicine and Public Health, University of Cantabria School of Medicine, Av. Cardenal Herrera Oria s/n, Santander, Spain *Department of Dermatology, School of Medicine, University of Granada, Granada, Spain Service of General Internal Medicine, Hospital Virgen de las Nieves, Granada, Spain Accepted for publication 7 October 1998 Summary A study of melanocytic naevi was carried out in southern Spain to examine the relationship between numbers of naevi at different body sites as predictors of whole-body naevus count and to determine whether the naevus count on the arms is valid for identifying the risk factors for total naevi. Subjects were the control group from a case control study on risk factors for cutaneous melanoma. They were selected from visitors to the University of Granada Hospital (southern Spain) between 1989 and Of 200 people invited to participate, 146 accepted (73%). Data were collected by personal interview, and melanocytic naevi were counted over the entire body surface by clinical skin examination performed by one dermatologist. Partial correlation coefficients (R) estimated by multiple linear regression were calculated. Comparisons between whole-body naevi and naevi on the arms, and their relationship with risk factors, were assessed by analysis of variance and covariance. Arms in men (adjusted R ¼ 0 88) and thighs in women (adjusted R ¼ 0 82) were the best predictors of total naevi after adjusting for age and sun exposure. Age, occupational and leisure sun exposure, and sunburns showed significant correlations with the total number of naevi. Similar results were found for the naevus count on the arms. In conclusion, the prediction of whole-body numbers of naevi by a naevus count on specific sites differs between men and women: arms in men and thighs in women are the best predictors. Nevertheless, naevus counts on the arms allowed us to study the risk factors for total naevi as well as whole-body naevus count: age and occupational sun exposure were the strongest determinants. Keywords: anatomical naevus distribution, Mediterranean ethnicity, pigmented naevus, regression analysis, risk factors, skin pigmentation, Spain, sunlight Several case control studies have shown that naevus counts are a strong risk factor for cutaneous malignant melanoma 1 3 even in Mediterranean populations. 4 The magnitude of this risk is much greater than for any other risk factor so far established, 4 and thus the aetiology and epidemiology of naevi are currently important topics for investigation. This stronger relation between naevi and cutaneous melanoma is compatible with the hypothesis that melanocytic naevi are precursors of melanoma. 5,6 Correspondence: Dr C.Fariñas-Álvarez. farinamc@medi.unican.es Dr Ródenas is currently affiliated with the Department of Dermatology, Hospital General Universitario Morales Meseguer, Murcia, Spain British Association of Dermatologists It is difficult to study total numbers of naevi in population surveys. Several studies, almost exclusively on populations of Anglo-Saxon or north European origin, 7 9 have examined the relationship between naevus number at different sites and whole-body counts. These studies have found that naevi on the arms are good predictors of the total number of naevi. Recently, several studies on risk factors for the development of naevi have been based on naevus counts on the arms Nevertheless, it has not been shown whether the risk factors for naevus counts on the arms are the same as for total number of naevi. The main objectives of this study were to examine the relationship between naevus numbers at different sites as predictors of whole-body naevus count and to 457

2 458 C.FARIÑAS-ÁLVAREZ et al. determine whether naevus counts on the arms are valid for identifying the risk factors for total number of naevi. Subjects and methods The subjects were the control group from a case control study investigating risk factors for melanoma in a Mediterranean population in East Andalusia, southern Spain. 4 Subjects were selected from visitors to the University of Granada Hospital on a random basis between 1989 and Visitors to patients in wards other than dermatology, and who were without acute disease at the time of the study, were asked to participate. Every Wednesday from October to June, the first visitor who accepted the invitation to participate was included. People were asked to participate in an investigation on skin diseases; no specific mention was made of skin cancer or the study proposal. Of 200 people invited to participate, 146 accepted (73%). The main reason for not participating was refusal to take part in the skin examination. Information was gathered by personal interview and clinical skin examination performed by one dermatologist. By means of a structured questionnaire with 203 items, data were collected on socio-economic factors (social class was classified in five levels according to the Black Report, 13 from I, highest to V, lowest level), education, employment, alcohol intake and tobacco consumption, and past medical record including personal and family history of cutaneous disorders, gynaecological history, and exposure to drugs, immunosuppressive therapy and ionizing radiation (in diagnostic or therapeutic procedures). Particular emphasis was placed on data on sunlight susceptibility and sun exposure. Individuals were asked about their skin response to an initial sun exposure (45 60 min of noontime exposure in the early summer), taking into account the reaction after 24 h (propensity to burn) and after 7 days (ability to tan). Each subject was classified from I to VI according to the working classification of sunreactive skin types of Fitzpatrick (I, always burn, never tan; II, usually burn, tan less than average with difficulty; III, sometimes mild burn, tan about average; IV, rarely burn, tan more than average with ease; V, never burn, always tan, brown skin; VI, never burn, black skin). 14 Individual sunlight exposure history was assessed in detail. We asked about places of birth and residence (altitude, rural or urban setting and number of years in each place), number of hours spent daily, weekly and monthly in occupational outdoor exposure in summer (April to September) and winter (October to March), with specific detail about every period of life (childhood, adolescence and adulthood), and clothing habits and body areas exposed (hat use, and exposure of trunk and extremities). Sunlight exposure during recreational activities and vacations was also examined using the same method for each period of life. The entire cutaneous surface of the participants was explored for pigmented lesions. Findings were noted for every subject in a body surface map with 30 anatomical areas. 15 Initially, the analyses were done with the 30 sites; then, bearing in mind the objective of the study, the analyses were repeated reducing the number of skin sites from 30 to seven: face and neck, thorax, arms, hands, thighs, legs and feet. This reduction in number of sites did not affect the results. Common acquired melanocytic naevi were counted and their diameter measured. Only naevi over 2 mm were considered, to minimize confusion with freckles and lentigines. Ephelides (freckles) on the face, forearms, shoulders and upper back were assessed by comparison with an analogue scale from 0 (absence) to 100 (maximum intensity). 15 Colour on sun-exposed skin (dorsum of hand) and an unexposed site (upper inner arm) was evaluated for each patient from direct observation, and the patient classified as fair or dark. Eye colour was recorded as black, brown, hazel, green/grey or blue. Hair colour was classified into four categories (black, brown, blonde or red); if the colour had changed with age, hair colour at age years was noted. Body weight and height were also measured. Clinical examinations were carried out in winter to minimize confounding factors such as evanescent freckles and summer tanning. The complete interview and skin examination lasted approximately h. We compared the number of melanocytic naevi on each of the seven areas mentioned above (face and neck, thorax, arms, hands, thighs, legs and feet) with wholebody naevus count. For each area, we used the wholebody naevus count minus the number of naevi on the area analysed, e.g. when we studied the ability of the naevus count on the arms to predict the total number of naevi, we used the total number of naevi minus the number on the arms, and correlated this number with the number of naevi on the arms. The distribution of the number of naevi was normalized by a natural logarithmic transformation. Partial correlation coefficients (R) were calculated by multiple linear regression analysis controlling for age, occupational sun exposure and recreational sun exposure. The R-value represents the correlation between total number of naevi and naevus counts on each area. Stepwise regression and regression

3 PREDICTIVE VALUE OF ARM NAEVUS COUNTS 459 models were constructed to examine both univariate and multivariate effects of these variables on the total naevus count and to identify confounders. 16 The number of naevi on the arms was analysed in a similar manner. All P-values used were two-tailed. Statistical analyses were performed using the BMDP statistical package (Dynamic version). 17 Results The male/female ratio was 1 : 2 (Table 1). The mean SD age was years (range 21 78). The distribution of naevi by anatomical site and sex is displayed in Table 2. The mean SEM number of naevi was in men and in women. The sites with the highest numbers of naevi were the thorax and arms, in both sexes. There were statistically significant differences between men and women in the mean number of naevi on the thorax (P < 0 05) and legs (P ¼ 0 02). The thighs showed a higher number of naevi in women than in men and this difference achieved a borderline statistical significance (P ¼ 0 05). The relationship between naevus count on different body sites and whole-body naevus count is shown in Table 3. The naevus count on the arms was the best predictor of total number of naevi for all subjects (R ¼ 0 78) and for men (R ¼ 0 87), but not for women (R ¼ 0 74). In women, thighs was the best predictor (R ¼ 0 78). These results did not change when estimates were adjusted by age, number of Table 1. Characteristics of the study population Variable n ¼ 146 (%) Sex Male 50 (34) Female 96 (66) Social class a V 75 (51) IV 31 (21) III 29 (20) I/II 11 (8) Age (years) < (7) (21) (15) (15) (26) > (16) Mean SD a According to the Black Report. 13 hours of occupational sun exposure and number of hours of recreational sun exposure, the best predictors for total numbers of naevi being arms in men (adjusted R ¼ 0 88) and thighs in women (adjusted R ¼ 0 82). Table 4 compares correlations of studied factors (constitutional, demographic and sun exposure factors) with whole-body naevus counts and naevus counts on the arms. No differences were found, except for vacational sun exposure and episodes of painful erythema, which gave lower R-values for naevi on the arms than for whole-body naevi. The effects of constitutional and demographic factors were similar on total naevus count and on naevus count on the arms. Age showed the highest correlation with both, even after controlling for occupational sun exposure and total episodes of sunburns (R ¼ 0 49 for whole-body naevi and R ¼ 0 41 for naevi on the arms); there was a negative correlation between age and number of naevi. Hair and eye colour did not show any significant effects on the number of naevi even after controlling for confounders. One measure of skin type was the self-reported relative ease of sunburn, and greater levels of susceptibility were not significantly associated with higher naevus counts. Social class did not show any significant effect on naevus count; nevertheless, income and years of education did show a significant correlation with both whole-body naevus count and naevus count on the arms. Regarding sun-exposure factors, occupational sun exposure showed a significant negative correlation with total naevi (R ¼ 0 33, P < 0 001) and with naevus count on the arms (R ¼ 0 32, P < 0 001). Vacational sun exposure also showed a positive relationship with total number of naevi (R ¼ 0 27, P ¼ 0 002), but not with naevus count on the arms (R ¼ 0 14, Table 2. Anatomical distribution of melanocytic naevus counts by sex Males Females (n ¼ 50) (n ¼ 96) Region Mean (SEM) Mean (SEM) P-value Face and neck 4 7 (0 9) 5 1 (0 5) Thorax 16 0 (2 6) 8 9 (1 2) <0.05 Arms 10 5 (2 0) 12 6 (1 5) 0.40 Hands 0 7 (0 1) 0 6 (0 1) Thighs 2 2 (0 5) 4 0 (0 6) Legs 1 4 (0 5) 2 8 (0 4) 0.02 Feet 0 3 (0 1) 0 2 (0 1) 0.33 Total naevus count 36 4 (6 2) 34 8 (3 5) 0.80

4 460 C.FARIÑAS-ÁLVAREZ et al. Table 3. Partial correlation coefficients between naevus counts in different anatomical regions and whole-body naevus count Partial correlation coefficients (R) Unadjusted Adjusted a Region Total Males Females Total Males Females Face and neck Thorax Arms Hands Thighs Legs Feet a Adjusted by multiple linear regression analysis controlling for age, total number of hours of occupational sun exposure, and total number of hours of recreational sun exposure. The highest R-value for each column is in italics. P > 0 05). A similar result was found for episodes of painful erythema: a positive significant correlation with whole-body naevi (R ¼ 0 28, P ¼ 0 002), but not with naevi on the arms. No relationship was found for outdoor leisure exposure, childhood sunbathing, blistering sunburns and freckles, either with the total number of naevi or with the naevus count on the arms. Discussion There are several potential limitations of this study. One is selection bias: subjects were the control group in a case control study investigating the risk factors for melanoma in southern Spain, 4 and they were chosen among healthy people who went to visit patients Table 4. Comparison between partial correlation coefficients of related covariates of total naevi for whole-body naevus counts and naevus counts on the arms Whole body Arms Variable R a P-value b R P-value Constitutional factors Sex Age (years) Eye colour < < Hair colour Skin type c Unexposed skin Demographic factors Social class d Income Years of education Sun exposure factors Occupational sun exposure (h) 0.33 < < Outdoor leisure exposure (h) Vacational sun exposure (h) Childhood sunbathing Total episodes of painful erythema Total episodes of blistering sunburns Freckles a Pearson s correlation coefficient; b two-tailed P-value from F-test (only P 0 05 shown); c according to Fitzpatrick; 14 d according to the Black Report. 13

5 PREDICTIVE VALUE OF ARM NAEVUS COUNTS 461 admitted in wards other than dermatology in the hospital. In a case control study, controls are the people who represent the overall population, so initially, controls were planned to be obtained by random selection of municipal rolls. The lack of co-operation, mainly due to the required skin examination, forced us to use another source of controls. The absence of major differences between cases and controls regarding life-style and social class, and the identification of well-known risk factors for cutaneous malignant melanoma in our study population, 4 support the contention that the controls adequately represent the general population in whom cutaneous melanoma arises. Another factor that may limit the validity of the conclusions of this study is sample size: it is small, especially when compared with other studies, 10,18 and this may imply a lack of statistical power to detect some associations, especially when we analysed the factors associated with naevus counts on the arms. This may have obscured identification of the risk factors for naevi, particularly those showing a small influence. Most data on naevi have been drawn from Anglo- Saxon populations. The present study has been carried out in Mediterranean subjects, a population with a darker complexion and lower incidence of cutaneous malignant melanoma than Anglo-Saxon people, so the numbers of naevi in our population should be lower than in northern Europeans. 18,19 Our results show less than half the number of naevi reported in a Swedish population. 9 The site distribution of naevi in the subjects analysed was similar to that reported in previous studies of Caucasians, 7 9 where the mean number of naevi in adult men was greater than in women on the trunk, and greater in women than in men on the arms, thighs and legs. It has been shown that the number of melanocytic naevi is a very strong risk factor for developing cutaneous malignant melanoma 1 3 even in Mediterranean populations. 4 The study of naevi is hindered by practical problems relating to whole-body counts, so it is useful to count naevi on smaller areas to predict the whole-body naevus count. Recently, the area used has been the arms, based on several studies done in Anglo-Saxon people. 7,8 Our results have shown that the best area to predict total naevus count differs between men and women: arms in men, but thighs in women. This agrees with Augustsson et al. 9 who found the anterior surface of the thighs in women and the lateral surface of the arms in men to be the best areas to predict wholebody naevus counts. Furthermore, it is difficult for nonprofessionals to differentiate between naevi, freckles and lentigines on the lateral aspects on the arms. For nonprofessionals, the thighs seem the most suitable area to count. 9 The arms have been used to predict whole-body naevus counts in recent population studies on the epidemiology of melanocytic naevi, and the effects of counting naevi only on restricted areas have not been previously examined. Our results show no great differences regarding several risk factors and total number of naevi when we compare the analyses of naevus counts on the arms and on the whole body. The major problem found when we analysed the naevus counts on the arms was the small sample size, and therefore we have a smaller statistical power to detect some associations analysed than with whole-body naevus count. This drawback may be obviated in population surveys with a greater sample size. Regarding the risk factors associated with an increased number of naevi, our results show a significant negative correlation with age. Several studies indicate that in childhood, youth and young adulthood the number of naevi increases with age, while it decreases in adulthood. 8,11,15,20 22 This has been ascribed to a hormonal influence in puberty 7 and a greater influence of sun exposure on naevi in childhood than at older ages. 11 An alternative theory is that recent birth cohorts have suffered environmental exposure, which has increased the number of naevi in the young. 20 Studies of the prevalence of naevi have produced circumstantial evidence implicating exposure to sun, 9,15,18,23,24 and others have found evidence for a relationship with sunburns. 10,15,25 Our findings show a negative correlation between whole-body naevus count and occupational sun exposure. These results suggest that intermittent ultraviolet exposure has a potent naevogenic effect, whereas chronic ultraviolet exposure has a protective effect against naevus formation. 9 In summary, these results indicate that prediction of whole-body naevus numbers by a naevus count on specific sites differs between men and women: arms in men and thighs in women are the best predictors. However, arms may be a good predictor to investigate the risk factors associated with total naevus numbers in population surveys where it is difficult to measure whole-body naevus numbers. Moreover, naevus counts on the arms allowed us to study the risk factors for total naevi as effectively as by using whole-body naevus counts: age and occupational sun exposure were the strongest determinants of numbers of naevi even in Mediterranean populations.

6 462 C.FARIÑAS-ÁLVAREZ et al. References 1 Gallagher RP, Elwood JM, Hill GB. Risk factors for cutaneous malignant melanoma: the Western Canada Melanoma Study. Recent Results Cancer Res 1986; 102: Swerdlow J, English J, MacKie RM et al. Benign melanocytic naevi as a risk factor for malignant melanoma. Br Med J 1986; 292: Elwood JM, Whitehead SM, Davison J et al. Malignant melanoma in England: risk associated with naevi, freckles, social class, hair colour and sunburn. Int J Epidemiol 1990; 19: Ródenas JM, Delgado-Rodríguez M, Fariñas-Álvarez C et al. Melanocytic nevi and risk of cutaneous malignant melanoma in southern Spain. Am J Epidemiol 1997; 145: Black WC. Residual dysplastic and other naevi in superficial spreading melanoma. Clinical correlations and association with sun damage. Cancer 1988; 62: Skender Kalnemas TM, English DR, Heenan PJ. Benign melanocytic lesions: risk markers of precursors of cutaneous melanoma? J Am Acad Dermatol 1995; 33: MacKie RM, English J, Aitchison TC et al. The number and distribution of benign pigmented moles (melanocytic naevi) in a healthy British population. Br J Dermatol 1985; 113: English J, Swerdlow J, MacKie RM et al. Site-specific melanocytic naevus counts as predictors of whole body naevi. Br J Dermatol 1988; 118: Augustsson A, Stierner U, Rosdhal I, Suurkula M. Regional distribution of melanocytic naevi in relation to sun exposure, and sites-specific counts predicting total number of naevi. Acta Derm Venereol (Stockh) 1992; 72: Dennis LK, White E, Lee JAH et al. Constitutional factors and sun exposure in relation to nevi: a population-based cross-sectional study. Am J Epidemiol 1996; 143: Dennis LK, White E, McKnight B et al. Naevi and migration within the United States and Canada: a population-based cross-sectional study. Cancer Causes Control 1996; 7: Fritschi L, McHenry P, Green A et al. Naevi in schoolchildren in Scotland and Australia. Br J Dermatol 1994; 130: Townsend P, Davidson N. Inequalities in Health. The Black Report. Harmondsworth: Penguin Books Ltd, Fitzpatrick TB. The validity and practicability of sun-reactive skin types I through VI. Arch Dermatol 1988; 124: Gallagher RP, McLean DI, Yang P et al. Suntan, sunburn, and pigmentation factors and the frequency of acquired melanocytic naevi in children. Similarities to melanoma: the Vancouver Mole Study. Arch Dermatol 1990; 126: Mickey RM, Greenland S. The impact of confounder selection criteria on effect estimation. Am J Epidemiol 1989; 129: Dixon WJ. BMDP Statistical Software Manual. Berkeley, CA: University of California Press, English DR, Armstrong BK. Melanocytic nevi in children. I. Anatomic sites and demographic and host factors. Am J Epidemiol 1994; 139: Brogelly L, De Giorgi V, Bini F, Giannoti B. Melanocytic naevi: clinical features and correlation with the phenotype in healthy young males in Italy. Br J Dermatol 1991; 125: Coombs BD, Sharples KJ, Cooke KR et al. Variation and covariates of the number of benign nevi in adolescents. Am J Epidemiol 1992; 136: Cooke KR, Spears GF, Skegg DC. Frequency of moles in a defined population. J Epidemiol Community Health 1985; 39: Pavlotsky F, Azizi E, Gurvich R et al. Prevalence of melanocytic nevi and freckles in young Israeli males. Correlation with melanoma incidence in Jewish migrants: demographic and host factors. Am J Epidemiol 1997; 146: Kelly JW, Rivers JK, MacLennan R et al. Sunlight: a major factor associated with the development of melanocytic nevi in Australian schoolchildren. J Am Acad Dermatol 1994; 30: Armstrong BK, de Klerk NH, Holman CDJ. Etiology of common acquired melanocytic nevi: constitutional variables, sun exposure, and diet. J Natl Cancer Inst 1986; 77: Harrison SL, MacLennan R, Speare R, Wronski I. Sun exposure and melanocytic naevi in young Australian children. Lancet 1994; 344: