Carli P, De Giorgi V, Palli D, Maurichi A, Mulas P, Orlandi C, Imberti GL, Stanganelli I, Soma P, Dioguardi D, Catricala' C, Betti R, Cecchi R, Bottoni U, Bonci A, Scalvenzi M, Giannotti B. Dermatologist Detection and Skin Self-examination Are Associated With Thinner MelanomasResults From a Survey of the Italian Multidisciplinary Group on Melanoma. Arch Dermatol. 2003;139(5):607-612. doi:10.1001/archderm.139.5.607
Copyright 2003 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2003
To investigate patterns of detection and variables associated with early diagnosis of melanoma in a population at intermediate melanoma risk.
Hospital and university centers belonging to the Italian Multidisciplinary Group on Melanoma.
Eight hundred sixteen patients who were consecutively diagnosed as having melanoma and treated at 11 participating centers.
Main Outcome Measure
Relationship between patterns of detection and patient's and physician's delay with melanoma thickness, assessed by multivariate analysis.
A statistically significant association with early diagnosis was found for female sex (odds ratio [OR] for a lesion >1 mm in thickness, 0.70; 95% confidence interval [CI], 0.50-0.97), higher educational level (OR, 0.44; 95% CI, 0.24-0.79), residence in northern and central Italy (compared with southern Italy) (OR, 0.44; 95% CI, 0.30-0.65 and OR, 0.24; 95% CI, 0.15-0.37, respectively), and the habit of performing a skin self-examination (OR, 0.65; 95% CI, 0.45-0.93). When adjusted for all the previously mentioned variables, only melanoma detection made by a dermatologist, maybe incidentally, was associated with a statistically significant additional effect on early diagnosis (OR, 0.45; 95% CI, 0.28-0.73). No significant effect of anatomical site (trunk compared with other sites: OR, 0.83; 95% CI, 0.59-1.17), presence of atypical nevi (OR, 0.78; 95% CI, 0.52-1.17), and patient's delay (>3 months compared with ≤3 months: OR, 1.12; 95% CI, 0.78-1.60) was found.
Future melanoma early diagnosis strategies should adequately stress the role of skin self-examination among the adult population, and should recommend that dermatologists perform a total skin examination to identify suspect lesions (such an examination should also be performed during consultations for other reasons).
INCIDENCE RATES of cutaneous melanoma are continuing to increase in white populations worldwide.1 To the contrary, recent figures of melanoma-related mortality show a tendency to level off in many countries.2 This apparent discrepancy between incidence and mortality can be explained by the fact that only the incidence rates of "thin" melanomas, associated with a favorable prognosis, are increasing.3,4 This finding derives, at least in part, from the improvement of early diagnosis of melanoma achieved in past decades.5
Unfortunately, the incidence rates of "thick" forms—associated with a high risk of recurrence—are not declining.4,5 Therefore, an important objective of the fight against melanoma in upcoming years will be the improvement and refinement of criteria aimed at secondary melanoma prevention, including educational messages for the professional and adult population to better reduce the incidence of advanced forms. Therefore, knowledge about the demographic and personal factors associated with early diagnosis of melanoma may be involved in the attempt to improve educational and screening strategies for the future.
Most data about pattern of melanoma detection have been obtained from high-risk populations, predominantly those who are fair skinned. Little is known about such conditions in intermediate-risk populations, such as those living in Mediterranean areas, who are not spared by the so-called melanoma epidemic (5%-7% average annual increase of incidence rates).6
In this study, we report the pattern of detection of and variables independently associated with early diagnosis in a series of melanomas consecutively referred to centers belonging to the framework of the Italian Multidisciplinary Group on Melanoma.
This study is based on 816 cases of cutaneous melanoma consecutively identified from January 1, 2001, to December 31, 2001, in 11 Italian clinical centers belonging to the Italian Multidisciplinary Group on Melanoma, 1 of the 2 working groups on cutaneous melanoma of nationwide relevance in Italy. The Italian Multidisciplinary Group on Melanoma includes local multidisciplinary groups, based in geographically defined areas, composed of dermatologists, plastic surgeons, oncologists, pathologists, and other specialists involved in the fight against melanoma. The number of melanomas collected by each center ranged from 31 to 163. Centers located in northern Italy enrolled 306 cases (37.5%); those in central Italy, 268 cases (32.8%); and those in southern Italy, 242 cases (29.7%).
The patients, all white persons with newly diagnosed lesions, were included in the study at the first visit after surgery, when the diagnosis of melanoma was histologically confirmed. Each patient received a standardized questionnaire concerning questions about who first perceived the lesion, afterward revealed to be a melanoma, as being suggestive of a melanoma or potentially dangerous (the patient by skin self-examination [SSE], the spouse, a friend, the family physician, a dermatologist, another specialist, or others); the interval before a diagnostic confirmation by a dermatologist or another specialist who proposed removal (patient's delay); and the interval before the lesion was removed (physician's delay). Patients were asked about their knowledge of the criteria for early diagnosis of melanoma (asymmetry, border irregularity, color variation, and diameter >6 mm), their custom for SSE, and periodic medical consultation aimed to screen for melanoma.
The number of common and clinically atypical nevi, classified in accordance with preestablished numerical categories (≤10, 11-30, 31-50, and >50 for total count of common nevi; and absent, 1-5, and >5 for atypical nevi), was assessed during the same visit by a dermatologist expert in the clinical examination of pigmented skin lesions. Concerning atypical nevi, observers were asked to define them as clinically benign nevi showing some or all of the criteria (asymmetry, border irregularity, color variation, and diameter >6 mm) for cutaneous melanoma.7
Anatomical site was described according to the following broad areas: head and neck, upper limbs, anterior trunk, posterior trunk, and lower limbs. Because all patients were diagnosed as having melanoma and treated within formally constituted multidisciplinary groups on melanoma, including pathologists with expertise in the field of melanocytic lesions, no attempt in reviewing pathologic slides was made. The inference about pathologic data was limited to melanoma thickness according to the well-established categories of the revised American Joint Committee on Cancer classification.8 A thickness assessment can be expected sufficiently reproducible among expert pathologists.9
The mean (±SD) age of patients was 53.8 (±14.8) years for men and 49.6 (±15.7) years for women. The median thickness of melanomas was 0.62 mm (25th-75th percentile, 0.30-1.27 mm). For men, the median thickness was 0.75 mm (25th-75th percentile, 0.32-1.60 mm); and for women, the median thickness was 0.59 mm (25th-75th percentile, 0.30-1.00 mm). One hundred twenty-nine melanomas (15.8%) were classified as in situ. The thickness measurement was not available for 13 melanomas (1.6%).
The original questionnaires were checked, coded, and computerized after a linkage with information obtained from pathologic reports. To investigate the relationship between the thickness of each case of melanoma and individual characteristics reported in the interview, we performed comparisons by analysis of covariance to account for several confounders (including into each model terms for age, sex, study area, educational level, and SSE). A multivariate logistic analysis was also performed to quantify the association between several individual characteristics and a dichotomous category for thickness (≤1 mm or >1 mm) as the outcome variable, after adjusting for the same relevant confounders.
Patterns of detection of melanoma and data about time to diagnosis (patient's delay) and time to surgical excision once clinical diagnosis was made (physician's delay) are shown in Table 1 and Table 2.
Most patients self-detected melanoma. Their spouse detected 12.5% of the lesions, while physicians first detected 38.7% of the lesions. The percentage of melanomas detected by a spouse largely differed according to sex (18.5% [70/379] in male patients vs 6.4% [27/422] in female patients; χ2 test, P = .000).
More than half of the subjects (68.9%) waited no more than 3 months before obtaining a well-established diagnosis of possible melanoma to be excised. The main reasons for longer waiting were the feeling that it was not important, fear about a possible diagnosis of cancer, lack of time, and the wrong opinion—common among laypersons in the past—that to remove a nevus is dangerous. Fifty-two patients reported waiting more than 3 months because another physician, seldom the family physician, did not think it was really a lesion suggestive of being a melanoma.
Surgical excision was performed within 1 month in about 80% of the patients. Only 3 patients waited more than 1 year before having the lesion excised (Table 1).
In Table 3, a multivariate analysis based on difference in mean values of melanoma thickness (log transformed) according to major demographic and personal terms is shown. All the variables included in the model (geographic area, sex, age, educational level, and habit of performing an SSE) were independently associated with differences in mean melanoma thickness. A lower mean thickness was significantly associated with female sex, high school (or higher) educational level, and the habit of performing an SSE. Residence in southern Italy and age older than 60 years were associated with a higher mean thickness, compared with residence in northern Italy and age younger than 40 years, respectively.
Once separately introduced into the basic multivariate model, an additional factor associated with a statistically significant lower mean thickness was represented by a lesion first detected by a dermatologist, while no significant effect on mean thickness was found for anatomical site of the melanoma, presence of clinically atypical nevi, or presence of many (>30) nevi. Patients who waited less than 3 months for seeking diagnostic confirmation about the lesions showed a slightly lower mean thickness, although the difference was not statistically significant.
Paradoxically, a lower mean thickness was also found in those who waited more than 1 month before surgery once a definite diagnosis of a lesion suggestive of a melanoma was established.
Table 4 shows variables associated with the diagnosis of thin (≤1-mm) melanomas by a multivariate logistic model, including terms for residence, sex, age, educational level, and custom of performing an SSE. All of these variables were independently associated with the diagnosis of thin tumors (age was of borderline statistical significance). A low thickness was also associated with residence in northern and central Italy (compared with residence in southern Italy). The risk of having diagnosed a thick melanoma was inversely associated with the number of years of education (test for trend, P = .008), although specific education categories were not significantly associated with this binary outcome. As previously shown for the effect on mean thickness, the first detection of the lesion by a dermatologist was strongly associated with a diagnosis of thin lesions, while no effect was found for other nondermatologist detectors.
As previously reported in predominantly fair skinned populations, most (40.6%) of the Italian patients self-detected their own melanoma. Their spouse detected 12.5% of the lesions. As expected, the percentage of melanomas detected by a spouse was higher in male than in female patients. These data are similar to those reported by Epstein et al10 in a series of 102 white US patients seen between 1995 and 1997 (self-detection rate, 55%; and spouse detection rate, 12%) and to those reported by Brady et al11 in a series of 471 patients seen at the Memorial Sloan-Kettering Cancer Center in New York City (self-detection rate, 57%; and spouse detection rate, 11%).
We cannot state if melanoma detection by physicians, which occurred in 38.7% of Italian patients (24% in the study by Epstein et al10 and 16% in the study by Brady et al11), has occurred incidentally during routine consultation for an unrelated event (by family physicians or other specialists) or for another skin lesion (by dermatologists) or as a consequence of a consultation aimed to screen melanoma. Fewer than one third of the patients reported periodic or occasional consultation aimed to screen melanoma, by either family physicians or dermatologists. This custom was not, however, associated with earlier diagnosis when placed into a multivariate model (data not shown).
Previous studies12 showed that the depth of lesions identified by a physician was significantly lower compared with that of lesions noticed by the patients themselves, but little is known about the effect of specialization (being a dermatologist or not) in such incidental detection. In the study by Epstein et al,10 who best focused on the impact of physician detection on melanoma thickness, tumors detected by a physician were associated with a lower median thickness (0.23 mm) compared with those that were self-detected (0.90 mm) (P<.001). Physician detection was associated with an increase in the probability of detecting lesions less than 0.75 mm in thickness (odds ratio, 4.2; 95% confidence interval, 1.4-11.1).10 This was confirmed by Brady et al,11 who showed that physicians were 3.6 times more likely to detect thin lesions compared with nonphysician detectors. However, no attempt was made by these researchers to differentiate detection by family physicians from that by other specialists, including dermatologists. In our study, to our knowledge, we first showed that a differential effect on melanoma thickness should be expected in accordance with medical specialization because only lesions detected by dermatologists were associated—after adjustment for possible confounders—with lower thickness (adjusted mean thickness, 0.68 vs 0.90 mm [for other pattern of detection]). This effect was clinically relevant because it associated with a protective effect against detection of thick (>1-mm) lesions (odds ratio, 0.45; 95% confidence interval, 0.28-0.73). No effect on thickness was found for nondermatologist detectors, including the category "other specialists." The higher skill of dermatologists, compared with other professionals, in diagnosing pigmented skin lesions, previously reported in formal studies13 of clinical photographs, is, therefore, confirmed in practice by our data. This represents a formidable tool to play in melanoma prevention strategies. A recent survey of the American Academy of Dermatology reported that only a few (30%) respondent dermatologists customarily perform a total skin examination on all of their adult patients, and 49% reported screening only patients perceived to be at increased risk; 42% of the respondents reported lack of time as an impediment to screening, whereas 9% did not perform screening because of lack of financial reimbursement.14 Younger dermatologists were significantly more likely to screen all patients for skin cancer.14 The performance of a total skin examination should be strongly recommended to dermatologists, because this is clearly associated with a specific additional effect on melanoma prevention besides other variables associated with early diagnosis.
Berwick et al15 first reported that SSE is associated with reduced melanoma incidence (there was a lower percentage of individuals performing SSE in cases vs controls); moreover, the mean Breslow depth of lesions on the backs of patients who conducted rigorous SSE (by a mirror) was lower (1.09 mm) than that of patients not performing SSE (1.65 mm) (P = .01). According to these researchers, however, only a minority of the population studied, residing in Connecticut, was performing SSE (13.2% of patients and 17.5% of controls). From our data, 45.8% of Italian patients diagnosed as having melanoma in 2001 performed SSE, but only 20.4% performed them regularly (at least once every 3 months). Performing an SSE was strongly associated with early diagnosis, with an adjusted mean thickness of 0.77 mm for subjects who perform SSE, either regularly or occasionally, and of 0.95 mm for those who do not. An SSE was associated with a protective effect against detection of thick (>1-mm) lesions (odds ratio, 0.65; 95% confidence interval, 0.45-0.93). Therefore, an SSE should be recommended as an effective preventive measure to the general population, including those living in southern Europe.
Only 31.2% of the patients reported a time to diagnosis of longer than 3 months (patient's delay). Such a percentage is lower than that (82%) reported by Krige et al12 in a prospective study of 250 patients seen in Cape Town, South Africa, between 1987 and 1989. A more recent period of inclusion of our case series, with an improved awareness of the general population about melanoma prevention over time, may account for this difference.
Regarding physician's delay, only a few patients (20.8%) waited more than 1 month from definite clinical diagnosis to surgery. When placed into the multivariate model, a physician delay of more than 1 month was, paradoxically, associated with a better prognosis (adjusted mean thickness, 0.74 vs 0.89 mm). This latter finding, inexplicable in terms of tumor growth, could be interpreted cautiously because of the well-known conflicting results in the literature on the effect of delay on melanoma thickness. It is plausible that lesions perceived as early or equivocal by clinicians and, therefore, more likely thin, may have to wait longer before excision than those clinically suggestive of advanced melanomas (more probably thick).
No significant effect of a patient's delay of longer than 3 months was found on melanoma thickness. Our data parallel those of Krige et al,12 who found no correlation between delay in diagnosis and thickness for the study population overall, while an effect was found in patients with nodular melanoma. A detailed study on patient and physician delay in melanoma diagnosis was performed by Richard et al.16 Regarding the delay between the moment when the patients first noticed a lesion on the site where the melanoma was diagnosed and the time when they believe that the lesion was potentially dangerous, no significant correlation was found with Breslow thickness distribution. Regarding the interval between when the lesion was believed to be potentially dangerous and the moment at which the lesion was examined for the first time by a physician, a negative correlation with tumor thickness was found (ie, the longer the period, the thinner the tumor).16 Because of the previously mentioned conflicting results, the exact role of patient's and physician's delay in melanoma thickness remains to be established. In this view, melanoma looks quite different from breast cancer, for which delays of more than 20 weeks between an abnormal breast cancer screening result and diagnosis of the disease in asymptomatic women are associated with increased size of tumors and increased risk of metastases of the lymph nodes.17 This discrepancy can be explained by the fact that melanoma is not discovered when asymptomatic by a technique with known sensitivity, as for breast cancer, but by a heterogeneous manner, and at a different microstaging level, in accordance with the perception of the lesion by the patient or by others. The clinical features of the lesion probably affect the capability of the subject in the perception of the lesion as potentially dangerous and in separating melanoma from other benign lesions. The existence of rapidly growing lesions, generally of a nodular histotype, represents a further variable strongly affecting the thickness at diagnosis.18
Accordingly, a poor prognosis of melanoma can be accounted for by the earliness of perception of harboring a potentially dangerous lesion, which could be improved, and by the more or less aggressive behavior of the tumor rather than by delays between time of detection and excision.16
In conclusion, this study provides new insights into the fight against melanoma; it demonstrates that the pattern of detection in the Mediterranean population is similar to the patterns found in predominantly fair skinned people, and that detection by a dermatologist, maybe incidentally, is associated with a better prognosis. Moreover, an SSE is an independent predictor of early diagnosis after allowance for other confounders, such as age, sex, anatomical site, and educational level.
Future melanoma prevention strategies should adequately stress these 2 factors, one at the level of the general population (promotion of SSE among adults) and the other among health professionals (promotion of total skin examination, possibly on the occasion of consultation for other reasons, among dermatologists).
Corresponding author and reprints: Paolo Carli,MD, Department of Dermatology, University of Florence, Via degli Alfani 31, 501221 Florence, Italy (e-mail: CARLI@unifi.it).
Accepted for publication November 20, 2002.
We thank all the study participants for their cooperation, and Marco Ceroti, MSc, for statistical support.