The presence of multiple melanocytic naevi is a strong risk factor for melanoma. Use of the whole body naevus count to identify at-risk patients is impractical.
The presence of multiple melanocytic naevi is a strong risk factor for melanoma. Use of the whole body naevus count to identify at-risk patients is impractical.
To (i) identify a valid anatomical predictor of total naevus count; (ii) determine the number of naevi that most accurately predict total naevus count above 25, 50 and 100; and (iii) evaluate determinants of multiple melanocytic naevi and atypical naevi.
Clinical data from 292 consecutive Spanish patients consulting for skin lesions requiring debriding were collected throughout 2009 and 2010. Correlations between site-specific and whole body naevus counts were analysed. Cut-offs to predict total naevus counts were determined using the area under the receiver operating characteristic curve.
The studied population was young (median age 31 years, interquartile range 28–43). The naevus count on the right arm correlated best with the total nevus count (R2 0·80 for men, 0·86 for women). Presence of at least five naevi on the right arm was the strongest determinant of a total naevus count above 50 [odds ratio (OR) 34·4, 95% confidence interval (CI) 13·9–85·0] and of having at least one atypical naevus (OR 5·7, 95% CI 2·4–13·5). Cut-off values of 6, 8 and 11 naevi on the right arm best predicted total naevus count above 25, 50 and 100, respectively.
Our results support the arm as a practical and reliable site to estimate the total naevus count when screening or phenotyping large populations. Threshold values for the number of naevi on the arm are proposed to help identify patients for melanoma screening.
The number of common melanocytic naevi is a strong marker for the risk of cutaneous malignant melanoma (CMM). In a meta-analysis of 46 studies, Gandini et al. estimated a pooled relative risk of developing CMM of 6·9 in patients with more than 100 common naevi compared with those having fewer than 15. Similarly, an elevated risk of CMM (6·4) has been reported for patients with five atypical naevi compared with those having none.
In order to identify at-risk patients in a rapid and practical manner during medical check-ups, many studies have used naevus counts on selected body sites as a proxy for total body naevus count. However, information on the relationship between site-specific and total naevus counts in adults is scarce.[2-5] The correlation between site-specific and total body naevus counts appears to be highest for the arms, although results have not been consistent across studies or sexes. Other studies have suggested using the anterior surface of the thigh, particularly for women, or the back or the lateral arm in both sexes to estimate total naevus count.[3, 5] At a time when routine screening by whole body examination of high-risk individuals for CMM is recommended in a primary care setting, a site-specific naevus count that reliably predicts the total naevus count could be an important adjunct in defining high-risk subjects, and may also be useful when examining large number of patients in a research setting or in large-scale screening campaigns.
The main objectives of this prospective study were: (i) to identify a valid anatomical predictor of total naevus count, (ii) to estimate cut-off values in naevus counts for this anatomical site, once identified, that best predict the presence of more than 25, 50 and 100 common melanocytic naevi on the whole body and (iii) to determine epidemiological, phenotypic, clinical and environmental risk factors associated with multiple or atypical naevi in a Southern European population.
This prospective study included 292 successive patients who visited a public department of dermatology in Valencia (Spain) for the examination of skin lesions that required debriding. This inclusion criterion approximately reproduces the patient selection process in melanoma screening campaigns. Data were obtained between January 2009 and December 2010 by face-to-face interview and physical assessment performed by two trained dermatologists (E.N. and B.E). The study was approved by the institutional review board of the Instituto Valenciano de Oncología, and informed consent was obtained from each patient before entering the study. No patient refused to participate in the study.
The following characteristics were evaluated:
To avoid any confusion between naevi and ephelides, only naevi with a diameter > 2 mm were considered. Macular lesions with a tendency to confluence and fading during the winter, and which were located in photoexposed areas were considered ephelides. Clinically, atypical naevi were counted separately and defined as melanocytic lesions presenting a macular component and at least three of the following features: diameter ≤ 5 mm, an ill-defined edge, irregular borders, erythema or multiple colours.
The location of naevi was recorded on a schematic body drawing divided into nine areas (face, neck, right arm, left arm, anterior trunk, back, buttocks, right leg and left leg). The relationship between site-specific naevus counts and the whole body naevus count was determined using the Spearman correlation coefficient. The number of naevi counted on each area was correlated with the number of naevi over the remainder of the body (total body naevus count minus site-specific naevus count), as suggested by others. Sex-specific correlations were calculated to determine whether the correlation between site-specific and total naevus counts differed by sex.
Factors associated with the presence of either more than 50 naevi or at least one atypical naevus were identified by logistic regression analyses. A stepwise approach was used to build the multivariate models, which were systematically adjusted for the effects of sex and age. The body site with the strongest correlation with the total naevus count (the right arm) was selected as the covariable for multivariate analyses. To assess the influence of the selected body site on the results, the analysis was repeated for another body site highly correlated with the total body count (the left arm).
Sensitivity, specificity and area under the receiver operating characteristics (ROC) curve were computed for statistically significant determinants of the presence of more than 20, 50 and 100 common naevi. These values were also calculated for each cut-off value to dichotomize the number of site-specific naevi with each unit increase. Statistical analyses were performed using SPSS software (version 15.0 for Windows; IBM, Armonk, NY, U.S.A.).
A total of 292 patients, 94 males (32·2%) and 198 females (67·8%), were evaluated. The study population characteristics are shown in Table 1. The median patient age was 31 years (interquartile range: 28–43). Most subjects had brown or black hair and eyes, no ephelides (77·7%), had experienced fewer than five severe sunburns (89·5%), were not highly photosensitive (Fitzpatrick phototype I or II: 28·8%) and worked indoors or outdoors for < 10 years (88·7%). The total naevus count was significantly greater in men than women (84·4 vs. 62·4, P < 0·001, Table 2). A greater naevus count for men was also observed for the trunk (P < 0·001), whereas no statistically significant difference in number of naevi occurred between sexes for other anatomical areas.
|Variable||N = 292 (%)|
|≤ 30 years||143 (49·0)|
|31–60 years||137 (46·9)|
|> 60 years||12 (4·1)|
|Presence of ephelides||65 (22·3)|
|Presence of solar lentigines||138 (47·3)|
|History of at least five severe sunburns||22 (7·5)|
|Working outdoors for at least 10 years||30 (10·3)|
|At least one atypical naevus||57 (19·6)|
|More than one naevus on the dorsum of the feet or the buttocks||113 (38·8)|
|Naevus on the anterior region of the scalp||70 (24·1)|
|Presence of atypical naevus syndrome||62 (21·2)|
|Mean of total naevi by age group|
|≤ 30 years old (SD)||71·7 (52·6)|
|31–60 years old (SD)||70·2 (57·8)|
|> 60 years old (SD)||35·2 (51·3)|
|Anatomical area||Men (n = 94)||Women (n = 198)||P-value|
|Mean (SD)||Mean (SD)|
|Face and neck||6·2 (5·6)||6·8 (5·5)||0·405|
|Trunk||49·4 (36·0)||27·0 (23·3)||< 0·001|
|Upper extremities||20·0 (18·7)||18·2 (17·5)||0·408|
|Lower extremities||8·4 (10·9)||10·4 (15·1)||0·264|
|Total naevus count||84·4 (60·5)||62·4 (51·4)||< 0·001|
The correlation between site-specific and total naevus count was highest for the arms (right arm: 0·85 and left arm: 0·83), and the back (0·83), followed by the anterior trunk in both sexes (Table 3). The weakest correlations were observed for the neck, the face and the buttocks. There was overall lesser variability in correlations across body sites for women (0·57–0·86) than men (0·32–0·84).
|Body area||Both sexes||Male||Female|
As the naevus count on the right arm presented the strongest correlation with the total number of naevi, and this body site is easily accessible for clinical examination, it was selected for investigating the factors associated with the presence of more than 50 common naevi and at least one atypical naevus. After adjustment for sex and age, the following factors were strongly and significantly associated with a total naevus count above 50 (Table 4): low phototype (phototype I or II), having at least five naevi on the right arm, more than one naevus on the dorsum of the feet or the buttocks and the presence of naevi on the anterior area of the scalp. Having at least five naevi on the right arm [odds ratio (OR) 5·7, 95% confidence interval (CI) 2·4–13·5], at least one naevus on the scalp (OR 2·6, 95% CI 1·4–5·1) or history of five or more severe sunburns (OR 2·4, 95% CI 1·2–4·5) significantly increased the chance of having some atypical naevi (Table S1). Of note, after controlling for age, sex and other covariates, the phototype did not appear to be associated with the presence of atypical naevi.
|OR||95% CI||OR||95% CI|
|Blue/green eyes vs. brown/black eyes||1·1||0·7–1·7||–||–|
|Red/blonde vs. brown/black hair||1·6||0·8–2·9||–||–|
|Phototype (I–II vs. III–V)||6·0||3·3–11·1||3·7||1·5–8·5|
|Sunburns (> 5 vs. ≤ 5)||2·6||1·6–4·3||–||–|
|Presence of ephelides (yes vs. no)||3·2||1·7–5·9||–||–|
|Presence of lentigines (yes vs. no)||1·8||1·7–5·9||–||–|
|Outdoor vs. indoor work||1·0||0·5–2·2||–||–|
|More than one naevus on the dorsum of the feet/buttocks||6·7||3·9–11·5||4·5||2·1–9·7|
|Naevus on the anterior scalp||7·4||3·7–14·8||4·5||1·6–12·0|
|Presence of ≥ 5 naevi on the right arm||35·3||16·8–73·9||34·4||13·9–85·0|
A count of five or more melanocytic naevi on the right arm best predicted which patients had total body count above 50 and some atypical naevi (Table 4 and Table S2).
Table 5 provides cut-off values for naevus counts on the right arm that most accurately predicted the presence of over 25, 50 and 100 common melanocytic naevi; the best cut-offs (with the highest ROC values) corresponded to at least 6, 8 and 11 naevi on the arm, respectively. For instance, a dermatologist applying the threshold of eight naevi on the right arm (area under the ROC curve 0·89, sensitivity 0·88, specificity 0·93) will correctly select 88% of patients whose total naevus count is over 50 and appropriately identify 93% of subjects whose total naevus count lies below 50.
|No. of naevi on right arm||> 25 naevi||> 50 naevi||> 100 naevi|
|Sensitivity||Specificity||AUC (95% CI)||Sensitivity||Specificity||AUC (95% CI)||Sensitivity||Specificity||AUC (95% CI)|
|≥ 1||0·97||0·32||0·64 (0·56–0·72)||0·99||0·20||0·60 (0·53–0·66)||1||0·13||0·57 (0·49–0·64)|
|≥ 2||0·95||0·44||0·69 (0·62–0·77)||0·99||0·30||0·64 (0·58–0·71)||1||0·19||0·60 (0·53–0·67)|
|≥ 3||0·90||0·68||0·79 (0·73–0·86)||0·96||0·47||0·72 (0·65–0·78)||1||0·32||0·66 (0·60–0·72)|
|≥ 4||0·85||0·81||0·83 (0·77–0·89)||0·94||0·60||0·77 (0·71–0·83)||1||0·42||0·71 (0·65–0·77)|
|≥ 5||0·80||0·89||0·85 (0·79–0·90)||0·94||0·71||0·82 (0·77–0·87)||1||0·49||0·74 (0·68–0·80)|
|≥ 6||0·73||0·97||0·85 (0·81–0·90)||0·90||0·83||0·87 (0·82–0·91)||0·99||0·58||0·78 (0·73–0·84)|
|≥ 7||0·68||1||0·84 (0·80–0·89)||0·87||0·88||0·88 (0·83–0·92)||0·99||0·64||0·81 (0·76–0·86)|
|≥ 8||0·64||1||0·82 (0·78–0·87)||0·88||0·93||0·89 (0·85–0·93)||0·99||0·68||0·83 (0·79–0·88)|
|≥ 9||0·58||1||0·79 (0·74–0·84)||0·77||0·95||0·86 (0·82–0·91)||0·92||0·72||0·82 (0·77–0·87)|
|≥ 10||0·54||1||0·77 (0·72–0·82)||0·73||0·95||0·84 (0·79–0·87)||0·90||0·75||0·83 (0·77–0·88)|
|≥ 11||0·42||1||0·71 (0·65–0·77)||0·58||0·99||0·78 (0·73–0·84)||0·83||0·86||0·84 (0·79–0·90)|
|≥ 12||0·34||1||0·67 (0·61–0·73)||0·49||1||0·74 (0·69–0·80)||0·77||0·91||0·84 (0·78–0·90)|
|≥ 13||0·31||1||0·65 (0·59–0·72)||0·44||1||0·72 (0·66–0·78)||0·73||0·93||0·83 (0·76–0·89)|
|≥ 14||0·30||1||0·65 (0·58–0·71)||0·42||1||0·71 (0·65–0·77)||0·73||0·94||0·84 (0·77–0·90)|
|≥ 15||0·26||1||0·63 (0·56–0·70)||0·37||1||0·68 (0·62–0·75)||0·69||0·96||0·83 (0·76–0·89)|
This study of 292 Spanish patients has confirmed the suitability of naevus count on the arm as a proxy for total naevus count at least in young people with a high Fitzpatrick phototype. Contrary to some previous series, our results were consistent for both sexes. Cut-off values for the number of naevi on the upper limb that enable accurate identification of patients with more than 25, 50 and 100 melanocytic naevi for CMM screening are proposed. These threshold values are commonly used in large melanoma prevention campaigns as well as in epidemiological studies, and therefore should be helpful for phenotyping patients or for selecting at-risk patients for full body skin examination.
The clinical/epidemiological profile of the studied population appears largely to reflect that of the Spanish target population attending nationwide Euromelanoma screening campaigns over the same period (2009–2010): predominance of women (68% in this study vs. 65% in Euromelanoma), young subjects (median age: 31 vs. 38 years) with dark hair colour (83% vs. 81%) and a low proportion of long-term outdoor workers (10% vs. 11%) and people with a highly sun-sensitive Fitzpatrick phototype I or II: 29% vs. 45%. This profile could also explain the high mean values for naevi, atypical naevi and naevi on unusual sites, such as the scalp and dorsum of the feet. These similarities between patients attending a public outpatient clinic for the evaluation of cutaneous lesions that require debriding (generally, recurring benign skin lesions or melanocytic naevi) and patients screened within a public health campaign that highlighted risk factors, such as evolving naevi, suggest that our findings can be applied to populations currently targeted by skin cancer screening initiatives and may be particularly useful in phenotyping large numbers of subjects, for example, in a research setting. In this context, our population purposely differs from the general population. It is uncertain whether this method could apply in older patients as it is well known that naevus counts decrease with age.
The importance of the total number of naevi and of atypical naevi as strong risk markers for melanoma development is widely accepted. Our study describes the epidemiological and clinical features associated with the presence of more than 50 naevi and of atypical naevi. Patients of phototype I or II, with several naevi on the feet or the buttocks, having one or more naevi on the anterior area of the scalp and at least five on the right arm had the highest probability of presenting over 50 naevi upon clinical examination. Occupational sun exposure was not significantly associated with having either more than 50 naevi or some atypical naevi. Experiencing several severe sunburns, an indicator of intense sun exposure on inadequately protected skin, did not predict having more than 50 naevi but was significantly associated with the presence of an atypical naevus. Taken together, these results suggest that, in a darker-skinned population, the influence of environmental factors, such as sun exposure, is less important than in fairer-skinned populations. In our Mediterranean population, the total naevus count is probably mainly driven by the genetic background.
In accordance with previous reports, we found a lower total number of naevi in women than in men. This difference appears to be largely accounted for by the larger mean body area of males (2·1 m2) than females (1·7 m2) and naevus density did not differ materially between sexes in our study (ca. 40 per m2). On the other hand, and in contrast with other populations, we found no significant differences in the number of naevi on the legs between men and women. We have no definitive explanation for this observation, although, as Spain is a sunny country, it is quite common for both sexes to expose lower limbs during daily activities for many months of the year.
The presence of naevi on the dorsum of the feet or the buttocks, and of one naevus on the anterior scalp form part of the proposed criteria to define an atypical naevi syndrome. The presence of naevi on the anterior scalp as an independent predictor of having either more than 50 naevi or at least one atypical naevus is consistent with the results of De Giorgi et al. who reported the same findings in a prospective study of 795 Italian patients.
The arm has been previously documented as a valid site-specific predictor for the total naevus count. Byles et al. found a correlation of 0·71 between the left arm naevus count and the whole body naevus count in 131 Australian subjects. Gallus et al. found comparable results in an Italian paediatric population. Fariñas et al. analysed seven body sites in 146 patients and observed that the best site-specific predictors of total naevus count, after controlling for age and photoexposure, were the arms in men and the legs in women. In our study, although the naevus count on the legs showed a higher correlation with the total naevus count in women than in men, the naevus count on the arms (right or left), consistently yielded the highest correlations.
To our knowledge, this is the first study to propose statistically validated threshold values for the number of naevi on the arm for use in predicting total number of naevi (25, 50 and 100). The three chosen cut-offs provide dermatologists with various degrees of patient risk selection and can be adjusted on an individual basis, for instance, in the presence of other CMM risk factor(s). They could serve to better triage the target population for melanoma screening, such as in large pan-European screening campaigns. Furthermore, the ROC values for each cut-off permit the thresholds to be tailored according to the objectives set out (i.e. favouring specificity or sensitivity).
Strengths of the present study include the comparatively large sample size, the prospective design, the number of covariates examined and the potential application of our findings in subpopulation candidates for CMM screening. The examination of all patients by two experienced dermatologists allowed the collection of homogeneous and reliable information across all patients examined. However, recall bias in remembering sun exposure cannot be excluded and could possibly explain the inconsistency in the observed effect of sunburn history.
In conclusion, our study confirms the suitability of naevus count on the arm as a reliable and practical method for estimating total naevus count, and provides useful cut-off values that may be useful in a research setting to phenotype large numbers of subjects and facilitate selection of candidates for melanoma screening.