Incidence of New and Changed Nevi and Melanomas Detected Using Baseline Images and Dermoscopy in Patients at High Risk for Melanoma FREE

Cutaneous surveillance has been well described as an effective method for early detection of melanoma.^1- 5 An important aid to this screening process is baseline cutaneous photography. This has been reported to be an efficient method of detecting changes that are suggestive of melanoma and as a means to minimize unnecessary surgery.^1,3,5- 6

With photographic cutaneous surveillance, a high incidence of melanoma and early detection of melanoma in high-risk patients has been reported.^1,3,5,7 In following up such patients, many new and changed pigmented lesions are detected. Not all of these lesions are suggestive of melanoma, and along with dermatoscopy, many can be managed with reimaging and further follow-up. This study aims to identify the proportion of new and changed lesions that are melanomas, as well as the incidence of melanomas and new, changed, and regressed nevi.

We recruited patients at the private consulting rooms of 1 dermatologist (J.W.K.) in conjunction with the Victorian Melanoma Service, a state-based multidisciplinary consultative treatment service for cutaneous melanoma (J.W.K., J.P.B., and J.P.D.). Between January 1, 1992, and December 31, 1997, patients 16 years or older who presented to the consulting rooms for full-body cutaneous examination were assessed for eligibility. Baseline skin surface photography was offered if a patient had at least 1 of the following risk factors for melanoma: 4 or more clinically dysplastic nevi,^8- 9 100 or more melanocytic nevi,^8- 9 a personal history of melanoma,^10- 11 or a family history of melanoma.^12- 13

All 309 (168 female and 141 male) patients who had baseline photographs and had at least 1 follow-up visit by December 31, 1998, were included in the study. Patients were divided according to their age at the time of photography: younger than 30 years, 30 to 39 years, 40 to 49 years, and 50 years or older. At entry into the study, patients were advised about their high risk of melanoma, 3-month self-examination was recommended, clinical features of early melanoma were discussed, and appropriate sun protection measures were described.

Patients had a set of 14 baseline cutaneous photographs, each view defined by easily located anatomical reference points (Figure 1). The scalp, dorsum of hands, palms, soles, and genitalia were not routinely photographed. A 35-mm single-lens reflex camera (Nikon FM2; Nikon Corporation, Tokyo, Japan) equipped with a 105-mm macromodel lens capable of focusing at a magnification of 1:2 or greater was used. The patient was positioned in front of a gray background, and 2 electronic studio flashes (1200 W-seconds), each diffused using large soft boxes, were positioned at 45° to the patient. This softened any coarse shadows. Color slides were developed from Fujichrome 100 film (Fujifilm, Tokyo, Japan) and projected using a Kodak Ektagraphic Projector (Kodak, Rochester, NY).

The baseline was the date of the patient’s initial cutaneous photography. At a patient’s first examination, the number of dysplastic nevi was recorded in ranges of 0 to 10, 11 to 20, and 21 or more. In addition, the total number of melanocytic nevi was estimated and recorded in ranges of 100 or less and more than 100.

A nevus was considered dysplastic if it had both clinical¹⁴ and dermatoscopic¹⁵ features consistent with the diagnosis. Clinically, a dysplastic nevus had a macular component and showed at least 3 of the following 5 clinical features: ill-defined border, irregularly distributed pigmentation, background erythema, maximum diameter greater than 5 mm, and irregular border.¹⁴

Follow-up of the patients was scheduled at 3-, 6-, or 12-month intervals. At each visit, total body cutaneous examination was performed by the study dermatologist (J.W.K.), and all melanocytic nevi were compared with baseline photographs. Using these images, the number of new, changed, and completely regressed nevi was recorded at each visit. No distinction was made between common nevi and dysplastic nevi and between nevi detected at different anatomical sites when recording the incidence of new, changed, and regressed nevi.

The specific changes were classified according to gross macroscopic clinical changes in color, size (diameter), and shape that were evident by comparison with the baseline photographs of large body regions (Figure 2). Minor changes of clinically insignificant magnitude were not recorded. A new nevus was one that appeared after baseline (Figure 3). A nevus was recorded as regressed if it had completely disappeared.

Biopsy specimens were not obtained of all changed and new pigmented lesions. If melanoma could not be confidently excluded by clinical examination and dermatoscopy, an excisional biopsy was performed. All new and changed pigmented lesions not excised were recorded. Pigmented lesions removed for cosmetic reasons that were not clinically suggestive of melanoma were not counted.

All melanomas diagnosed on follow-up were recorded as new or changed lesions when compared with baseline photographs. All melanomas were examined by a single dermatopathologist (J.P.D.). Tumor type was classified according to McGovern et al¹⁶ as superficial spreading melanoma, nodular melanoma, or lentigo maligna melanoma, and the levels of invasion were recorded as I through V according to Clark et al.¹⁷ Histologic evidence of an associated nevus and the thickness of the melanomas were also recorded.

Incidence rates were calculated for melanomas, new nevi, nevi that changed, and regressed nevi. Duration of follow-up was calculated from the date of photography to the date of the last follow-up visit. All relevant lesions were included in the calculations, regardless of how many each patient had during follow-up. We calculated overall rates and rates separately by the baseline age (<30, 30-39, 40-49, and ≥50 years), sex, number of dysplastic nevi at baseline (<4, 4-10, 11-20, and ≥21), and total number of nevi (<100 and ≥100). Negative binomial regression was used to estimate rate ratios, confidence intervals, and P values for each of these variables.

The melanoma rate among patients was compared with the rate in all residents of Victoria by calculating the standardized incidence ratio with adjustment for calendar year, sex, and age in 5-year groups. All in situ and invasive melanomas in the population of Victoria from 1992 to 1998 and in the patients were included in this analysis (data provided by the Victorian Cancer Registry on July 26, 2002).¹⁸ A jackknife confidence interval was constructed to allow for the multiple melanomas per patient. All analyses were performed with Stata statistical software, version 7 (Stata Corporation, College Station, Tex).

Table 1 gives the number of patients who met each of the eligibility criteria. The median age at baseline was 38 years (age range, 16-74 years). The median number of follow-up visits following photography was 3 (range, 1-18). The median length of follow-up was 34 months (range, 2-79 months).

A total of 311 changed nevi were detected. Seventy-one nevi had no data for the type of change detected. Of those nevi with the change documented, the most frequent changes were in size (67%), color (15%), and both size and color (14%). Changes in shape (2%) and all 3 changes together (2%) were not common. Fifty-three (17%) of the changed nevi were excised (Table 2). Of these, almost half were dysplastic nevi and 14 (26%) were melanomas.

The incidence of changed nevi was 329 per 1000 person-years (Table 3). The rate decreased with increasing age and was more than 2-fold greater for patients younger than 30 years. The incidence was also higher in patients with higher numbers of dysplastic nevi at baseline but showed no association with the patient’s sex or the number of common nevi at baseline (Table 3).

Two hundred sixty-two new pigmented lesions were detected (Table 4). Of these, biopsies were performed on 18 (7%), yielding 4 melanomas (Table 2). The incidence of new pigmented lesions was 277 per 1000 person-years (Table 4). The rate of appearance of new nevi was highest in patients younger than 30 years at baseline. The rate was not associated with the patient’s sex or number of dysplastic nevi at baseline but was highest, albeit not significantly, in those with the largest number of common nevi at baseline.

Eighty-six nevi regressed completely (Table 5). The incidence of disappearing pigmented nevi was 91 per 1000 person-years (Table 5). The rate decreased with increasing age and was higher in men but showed no consistent relationship with the number of dysplastic nevi at baseline and no association with the number of common nevi at baseline (Table 5).

Eighteen melanomas were detected in 16 patients (Table 6). Seventeen melanomas were detected at follow-up visits; 4 were new lesions and 14 were changed lesions. One melanoma was found by the patient between visits (Table 6). Two melanomas were detected within 6 months of the baseline photography as changes in both size and color. Fifteen superficial spreading melanomas, 2 lentigo maligna melanomas, and 1 nodular melanoma were diagnosed during the study period. The median time to diagnosis of each patient’s melanoma after photography was 26 months (range, 5-72 months).

The presence of an associated nevus histopathologically showed little relationship with the clinical history of the pigmented lesion. Of the 4 melanomas that presented as new lesions, 2 showed histologic evidence of an associated nevus compared with 8 of the 14 melanomas in changed lesions (Table 6). Only 1 of the 6 in situ melanomas among the changed lesions had histologic evidence of an associated nevus (Table 6).

Assuming that none of the lesions that did not undergo biopsy was a melanoma, then 1 (0.4%) of 252 new lesions identified in patients younger than 50 years at photography was a melanoma compared with 3 (30%) of 10 new lesions identified in older patients. Similarly, 9 (3%) of 288 changed lesions in patients younger than 50 years were melanomas compared with 5 (22%) of 23 changed lesions in older patients. The overall ratio of benign to malignant biopsy results was 2.8. It was 4.5 in patients younger than 50 years and 0.75 in older patients.

Most (77%) of the melanomas detected were Clark level I or II. For the invasive melanomas, 80% were no more than 0.75 mm thick; the median thickness was 0.39 mm. One melanoma detected was a nodular melanoma, Clark level IV, and 2.45 mm thick, with an associated nevus. This melanoma was detected by the patient as a changing lesion on self-examination between visits. On comparison with the photographs, this melanoma had increased in size during the 6 months between follow-up visits.

The overall incidence of melanoma was 19 per 1000 person-years (Table 7), which was 34 (95% confidence interval, 21-54) times higher than that in the Victorian population of the same age and sex. The rate was highest in patients 50 years or older. Patients with greater numbers of dysplastic nevi had a higher melanoma incidence, but there was no positive association with the number of common nevi at baseline.

Our study confirms that nevi are dynamic lesions.^19- 23 In these high-risk patients, increased rates of melanoma were found in all age groups. Patients younger than 50 years, and especially those younger than 30 years, had more new, changed, and regressed nevi and fewer melanomas detected. In contrast, patients older than 50 years had more melanomas detected and had fewer new nevi appear or existing nevi change or regress. Therefore, a new or changed nevus in an older patient is much more likely to be a melanoma than in a younger patient. In patients younger than 50 years, less than 1% of all new lesions were melanomas, whereas in patients older than 50 years, 30% were melanomas. Similarly, for all changed lesions detected in patients younger than 50 years, 3% were melanomas, whereas in patients older than 50 years, 22% were melanomas.

Two previous studies support the observation that nevi become more stable in patients with increasing age. Halpern et al²⁰ reported that 33% of patients had new dysplastic nevi appear on their backs during a mean follow-up of 89 months, with the highest rate occurring in patients younger than 30 years (76%), which then decreased with increasing age. Histologically, Lund and Stobbe²² demonstrated that with increasing age, nevi become less active and more mature so that fewer changes and less regression occur.

In our patients, approximately one third had a changing nevus each year. This rate of change is far less than that found by Halpern et al,²⁰ with 51% of all evaluated nevi in the mean 89-month follow-up showing clinical signs of change. This difference is likely to be explained by the different thresholds for detecting change in each study. We recorded only the gross clinical changes in nevi as determined using baseline photographs of large body regions. Halpern et al,²⁰ on the other hand, examined nevi with macrophotography and recorded minor changes.

Many cross-sectional studies^21,24- 27 have reported that total nevus counts decrease with increasing age. This has generally been attributed to regression of nevi with age.^19- 21 However, in our study the rate at which nevi regressed was less than the rate of appearance of new nevi in all age groups, suggesting that declining numbers of nevi with age may be a cohort effect, perhaps reflecting different sun exposure patterns in childhood years. There has been no recent study of nevus numbers by age through adult life, and it is possible that current data would not show the (cross-sectional) decline with age. It is also possible that our short study time frame may have been insufficient to demonstrate the expected decline in numbers of nevi with age. Furthermore, the patients in our study had large numbers of nevi and dysplastic nevi, and in such patients, the natural history of nevi may be different from that in the general population.

The incidence of melanoma was 34-fold higher than in the general population. Surveillance bias may have contributed to this, and similar rates have been demonstrated in other studies of high-risk populations.^1,5

The number of dysplastic nevi was positively associated with the incidence of melanoma, confirming the widely described^1,8- 9,28 role of dysplastic nevi as a major risk factor for melanoma and the dose-response relationship previously observed between numbers of dysplastic nevi and melanoma incidence.¹ Somewhat surprisingly, we found no association between the total numbers of common nevi and melanoma rates. However, the broad ranges used to record nevus numbers may have masked this expected relationship.

A consensus exists that only new and changed nevi that are clinically suggestive of melanoma require excision. Using dermatoscopy, Kittler et al²⁹ detected 516 changes in nevi, of which only 75 were removed. Halpern et al²⁰ excised only 48 of 593 changed nevi. Only 17% of all changed and 7% of all new nevi detected in our patients were removed. Of these biopsy specimens, 26% of changed and 22% of new lesions were melanomas.

In our study, the benign-malignant ratio of lesion biopsy specimens was almost 3:1. This ratio compares with a previous study¹ conducted in the same private practice, in which the benign-malignant ratio was 9:1. Our lower ratio in this more recent study can be explained by the fact that we used dermatoscopy in addition to photography. The combination of photography to demonstrate the stability of many nevi that might otherwise be excised and dermatoscopy to help rule out melanoma in many new and changed nevi greatly reduced the number of excisions required in our patients. Overall, benign-malignant ratios of 12:1 for dermatologists and 30:1 for general physicians have also been reported.³⁰

When consultations and biopsies are priced at the current Australian government’s Medicare schedule rates (November 2003) and AUD $125 is allowed per set of photographs, the cost of diagnosing each melanoma in this study was AUD $4840. However, if we had a benign-malignant ratio of 12:1, as has been previously reported for dermatologists,³⁰ then the total number of biopsies would have increased to 234 and the cost of diagnosing each melanoma would have risen to AUD $6323. Thus, significant savings are likely to be made in reduced biopsy rates alone from the use of the combination of baseline photography and dermatoscopy.

In our study, 44% of melanomas were in situ and 80% of the invasive melanomas were less than 0.75 mm thick. Thirty-five percent of melanomas in Victoria for 1995 were in situ. The median thickness of invasive melanomas detected in this study was 0.39 mm, compared with 0.60 mm for all invasive melanomas in Victoria diagnosed between 1992 and 1998 in persons younger than 72 years. Surveillance played a major role in detection in this study, with 17 of the 18 melanomas detected by comparison with baseline photographs.

The literature disagrees about the proportion of melanomas that arise from preexisting benign nevi. At least two thirds of patients state that their melanomas arose from abnormal growth of an existing mole.³¹ Friedman et al³² reported that 70% of patients with histologic evidence of a nevus-associated melanoma had a history of a preexisting lesion. Early histologic studies reported an associated benign nevus in 53% of melanomas,³³ whereas later studies have shown this association to be lower, on the order of 20% to 30%.^32,34- 36 In our study, 44% of the melanomas had no histologic evidence of an associated nevus, whereas in our previous study¹ this figure was 65%. Thus, most melanomas appear to arise de novo according to histologic studies, yet the clinical histories suggest that melanomas arise from preexisting nevi.

In our study, the clinical impression of de novo melanoma vs melanoma that originated from a preexisting nevus based on comparison with baseline photographs was not always supported by the histologic findings. We expected that melanomas that presented with change in a preexisting pigmented lesion would show histologic evidence of an associated nevus, whereas melanomas that appeared as new lesions on normal skin would not. Although for invasive melanoma it is possible that evidence of a preexisting nevus has been destroyed by the melanoma, it is unlikely that a level I melanoma would replace a preexisting nevus.

Five of the 6 in situ melanomas that presented as change in a preexisting lesion were not associated with a nevus. These melanocytic lesions were present at the first consultation and initially appeared benign on clinical and dermatoscopic examination. During follow-up, these lesions underwent suspicious change and excisional biopsies were performed. It is likely that these pigmented lesions were in situ melanomas from the outset and were clinically stable or slow growing. In addition, 2 of the 4 melanomas that were absent on baseline photographs and evolved as new lesions during follow-up showed histologic evidence of an associated nevus. For these lesions, the melanoma and nevus may have evolved simultaneously during the interval between follow-up visits.

This study shows that nevi continue to change and appear throughout life, at least in high-risk patients. Using baseline total body photography and dermatoscopy to monitor pigmented lesions greatly reduces the normally high biopsy rates required to evaluate high-risk patients and provides large savings in the cost of biopsies and the morbidity associated with them. In addition, melanomas are detected earlier than would otherwise be expected, thus reducing the potential for mortality from melanoma.

It is clear that only a small proportion of new and changed lesions are likely to be melanomas. This is relevant to physicians in setting their thresholds for performing biopsies. New and changed lesions are much more likely to be melanoma in patients older than 50 years.

Correspondence: John W. Kelly, MDBS, Victorian Melanoma Service, The Alfred, Commercial Road, Prahran, Victoria 3181, Australia (kellyderm@bigpond.com).

Accepted for Publication: December 7, 2004.

Disclaimer: The authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Previous Presentations: This study was presented in part at the Clinical Oncological Society of Australia 29th Annual Scientific Meeting; November 26, 2002; Sydney, Australia; and the British Association of Dermatologists’ 83rd Annual Meeting; July 3, 2003; Brighton, England.

Additional Information: This study received the Schering-Plough Research Award (Dr Banky) for trainees of the Australasian College of Dermatologists.

Acknowledgment: We thank Amanda P. Henham, BApp-Sci(Photog), for designing the skin surface views and photographing all patients in this study.

Financial Disclosure: None.