Demierre M, Chung C, Miller DR, Geller AC. Early Detection of Thick Melanomas in the United StatesBeware of the Nodular Subtype. Arch Dermatol. 2005;141(6):745-750. doi:10.1001/archderm.141.6.745
Incidence and mortality of melanoma in the United States have risen steeply. Part of this mortality increase may be related to late detection of biologically aggressive nodular melanomas. We determined trends in distribution of thin and thick melanoma, with emphasis on the histopathologic subtype nodular melanoma.
Surveillance, Epidemiology, and End Results melanoma incidence data for whites were obtained for 1988 through 1999 and stratified according to histologic subtype: lentigo maligna melanoma, superficial spreading melanoma, nodular melanoma, other, and not otherwise specified (NOS); thickness: 0-0.99 mm, 1.00 mm-1.99 mm, and ≥2.0 mm; patient age (0-49, 50+); gender; and year (1988-1991, 1992-1995, 1996-1999). Comparison of tumor thickness between strata was defined by year of diagnosis, sex, age, and histologic subtype.
The number of new melanoma cases in a 3-year period increased 60% from 1988-1991 (n = 9132) to 1996-1999 (n = 14 575). The proportion of thick melanomas (≥2 mm) remained relatively stable during the 12 study years. Nodular melanoma comprised 9% of all recorded cases but 34% of melanomas 2 mm or larger, including melanoma not otherwise specified (NOS), and nearly 50% of all melanomas 2 mm or larger when NOS cases were excluded. In contrast, superficial spreading melanoma was almost uniformly diagnosed as an early tumor, mostly (77%) presenting as thin melanoma (<1 mm) and with only 7% presenting as thick melanoma (≥2 mm).
A substantial number of thick melanomas in the United States are of the nodular subtype, and median thickness of nodular melanoma has not changed during the 12 years of study. New strategies are needed to decrease the incidence of thick melanoma in the United States.
Cutaneous melanoma is one of the most rapidly increasing cancers in the United States.1 Recent incidence and mortality data among the white population in the United States derived from the Surveillance, Epidemiology, and End Results (SEER) database show a steep increase in incidence and mortality from 1969 to 1999.2 Overall melanoma mortality increased approximately 50% from 2 to 3 deaths per 100 000 population, but the increase was disproportionately greater in men 65 years and older, with an increase of 157%, which is 3-fold greater than the rate for women of the same age. A similar trend was observed in the incidence data, with the highest increase (5-fold) in men 65 years and older.
Melanoma mortality is strongly associated with thickness of the primary lesion.3 Although an earlier analysis1 of the SEER database found overall improved detection of thin tumors less than 1 mm thick, thick tumors of 4 mm or greater significantly increased only in men 60 years or older from 1988 to 1997.1 The increased incidence of thin tumors is presumably the result of increased awareness and earlier diagnosis.
The increase in thick melanomas among older men has remained unexplained but may be related to histopathologic subtype. In Australia, nodular type and older age are most frequently associated with thick melanomas,4,5 whereas in one Swedish study,6 nodular melanoma (NM) was the most significant determinant of thick melanomas. Because these data have important implications for secondary prevention,7 we sought to determine the trends in distribution of thick melanoma in the United States with particular emphasis on the histopathologic subtype NM.
The SEER program collects cancer incidence and staging information from many population-based cancer registries in the United States. For this study, we used information on melanoma collected from 9 registries in the SEER database from 1988 through 1999 and obtained from the National Center for Health Statistics (seer.cancer.gov/). The SEER melanoma incidence data were stratified according to histologic subtype (lentigo maligna melanoma [LMM], superficial spreading melanoma [SSM], NM, other [eg, acral lentiginous melanoma, desmoplastic melanoma, spindle cell melanoma], and melanoma not otherwise specified [NOS]); thickness (0-0.99 mm [thin], 1.00-1.99 mm [intermediate], and ≥2.0 mm [thick]); patient age (0-49 years and ≥50 years); sex; year of diagnosis (1988-1991, 1992-1995, and 1996-1999); and SEER site (all 11 sites). Classification by thickness followed the new American Joint Committee on Cancer staging system for cutaneous melanoma.3 We compared tumor thickness among strata defined by year of diagnosis, sex, age, and histologic subtype.
Observed differences in proportions between groups and over time were tested using χ2 tests. Differences in median thickness between groups were tested using the Fisher exact test of the proportion of cases above and below the combined median.8P = .05 was used as the critical test for all comparisons.
A total of 35 028 cases of white patients with invasive melanoma were recorded by the SEER registry in 1988-1999. Men made up 54% of the cases, and the median patient age was 54 years. Overall, the number of new cases increased 60% from 1988-1991 (n = 9132) to 1996-1999 (n = 14 575).
Of the 35 028 cases, nearly 48% were SSMs, followed most closely by melanoma NOS (32%) (Table 1). Strikingly, NM comprised 9% of all recorded cases but 34% of melanomas 2 mm or larger (including melanoma NOS) and nearly 50% of all melanomas 2 mm or larger when melanomas NOS were excluded.Among all NMs, tumor distribution by thickness was thin (18.7%), intermediate (25%), and thick (56.3%). Conversely, SSM was almost uniformly diagnosed at less than 2 mm, and only 7% of all SSM was diagnosed at 2 mm or greater. Median tumor thickness varied sharply among all other subtypes and NM, ranging from 0.39 mm for LMM to 2.19 mm for nodular lesions (P < .001). Median thickness of the LMM subtype was consistently thinner compared with the other subtypes for all periods of study. The distribution of tumor thickness in those classified as melanoma NOS appears to be approximately equal to a weighted average of the distributions among the cases with known histologic subtypes, suggesting minimal bias in classification of each type to melanoma NOS. Rarer histologic subtypes that included acral lentiginous (1.2%), desmoplastic (0.88%), and spindle cell melanomas (0.85%) had a median thickness of 1.28 mm for the 12-year period of study.
Median tumor thickness decreased significantly from 0.64 mm in 1988-1991 to 0.63 mm in 1992-1995 to 0.59 mm in 1996-1999 (P < .001), although the differences in tumor thickness by categories were slight (P = .04) (Table 2). Thin tumors (0-0.99 mm) comprised 66% to 68% of all cases in the 3 periods, intermediate tumors (1.00-1.99 mm) represented 17% to 18%, and thick tumors (≥2.00 mm) constituted 14% to 16% of all cases. When melanoma NOS was included, NMs comprised 36.5% of all lesions 2 mm or larger in 1988-1991, 35.7% in 1992-1995, and 31.1% in the most recent period (1996-1999) (P < .001 for trend). Excluding melanoma NOS, NMs comprised 52.6% of all lesions 2 mm or larger in 1988-1991, 49.9% in 1992-1995, and 45.2% in the most recent period (1996-1999) (P < .001). Among all NMs, those greater than 2 mm constituted 55% in 1988-1991, 59% in 1992-1995, and 58% in 1996-1999. Median tumor thickness decreased in the 3 periods for SSM (P = .01), LMM (P = .03), and melanoma NOS (P = .001) but remained stable or rose slightly for NM and those classified as other.
Important differences also exist in tumor thickness by sex and age (Table 3). Median tumor thickness was as follows: younger women, 0.54 mm; older women, 0.64 mm; younger men, 0.64 mm; and older men, 0.67 mm. Men 50 years and older were most likely to be diagnosed as having thick (≥2.0 mm) tumors for all histologic subtypes other than LMM (P < .001). In contrast, younger women had the least proportion of thick melanomas in all histologic subtypes. Among men 50 years and older, 19% of all melanomas were 2 mm or larger, more than doubling the 8% rate among younger women.
Several key findings emerge from the analysis of more than 35 000 cases recorded in the SEER registry. First, the precipitous increase of invasive melanoma reported in the 1970s and 1980s has continued unabated during the most recent period (1988-1999), with the number of new cases increasing by 60% in the last 12 years. Second, the proportion of thick melanomas (≥2 mm) did not substantially decline during the 12 study years. Third, the relatively rare NM represents nearly 50% of all histologically classified thick melanomas diagnosed in the United States, excluding melanoma NOS, a proportion that may not be widely appreciated. In contrast, SSM is almost uniformly diagnosed as an early tumor, mostly (77%) presenting as less than 1 mm and rarely presenting as greater than 2 mm. Older men were most commonly diagnosed as having melanoma, including NM.
Although clinical studies have long recognized that NMs are often thicker tumors, this is the first population-based study, to our knowledge, to document the high proportion of this tumor subtype among thick melanomas diagnosed in the United States. The median thickness of NM did not change during the 12 years. These data highlight the need for additional strategies to reduce the proportion of thick tumors. Discovering the biological and behavioral antecedents of NM will be central to improving the early detection campaign that has existed in the United States since 1985. Generally, NMs are known to be red or pink and raised.9 When pigmentation is present, it is usually evenly distributed throughout the lesion as opposed to SSMs, which show irregular pigmentation.
Our findings suggest that the histologic nodular subtype may be driving the later detection of thick melanoma in white patients in the United States, and greater emphasis needs to be concentrated on early detection of this subtype, if possible. Earlier analyses in the United States have found that older men have thicker tumors1 and higher mortality rates than women.2,10 Data from Australia and Europe have suggested that thick lesions are more often nodular4- 6,11,12 and predominantly observed in older men.4- 6,11- 13 However, US population–based databases have not previously been examined for the interrelationships of tumor thickness, histologic subtype, sex, and age.
The high rates of thick NM found in this study are similar to reports from Sweden, Australia, and Italy. In Sweden, Bergenmar et al6 identified “nodular histogenetic type” as the most significant predictor for thick melanomas (≥2 mm). In Australia, both Hersey et al5 and Chamberlain et al4 demonstrated that thick melanomas (≥3 mm) were predominantly nodular. In a population-based study in Italy (1995-1997), nodular type and older age were the only variables significantly associated with thick melanomas (>3 mm).14 Our data from this population-based study in the United States confirm that a substantial percentage of thick melanomas are of the nodular histologic subtype.
Because melanoma thickness prevails over histologic subtype in regard to prognosis, pathologists and dermatopathologists have been known to disagree on the “distinctiveness” of NM.15- 20 It is also important to emphasize that vertical growth phase does not necessarily mean NM, although it is sometimes taken to mean the same thing (exact definitions are given in Table 4). The absence of consensus may explain in part the absence of histologic subtype for a third of the melanomas in the SEER database (the melanoma NOS category). Although some may argue that NMs are lesions that have overgrown an SSM, it does not appear to be the case, since NMs have been reported to have a smaller diameter than SSMs.22 The mitotic rate could provide additional evidence for differences between NMs and SSMs; however, this information is not captured by the SEER database.
Although our understanding of NM is incomplete, NM is likely the result of an additional “hit” in the multistep model of melanoma genesis.23,24 Recent data appear to support possible biologic and genetic differences between NM and SSM. In one study,25 NMs were more likely to have loss of p16 expression compared with SSMs, in support of the hypothesis that loss of p16 was associated with more aggressive tumors. To date, data on BRAF (B-raf proto-oncogene) mutations in melanoma have shown uneven distribution with high mutation frequency in melanomas that arise on intermittently sun-exposed skin,26 suggesting distinct genetic pathways leading to melanomas. Furthermore, in Japan, BRAF mutational analyses of a group of primary sporadic melanomas showed significant differences in mutational frequency among the 4 melanoma histologic subtypes: 50% of SSMs but no NMs had BRAF mutations.27 Confirmation of these results is needed.
Nodular melanomas have biological aggressiveness (ie, rapid growth). This may help explain why investigators have found little delay in seeking medical attention. In one study, Richard et al28 reported a shorter delay in seeking medical care for patients with NM (median, 46 days) than for patients with SSM (median, 78 days; P = .03). The rapid growth of NM helps to explain why patients might not delay seeking medical attention. These data suggest that the inherent biologic aggressiveness of a tumor could determine prognosis.29
In contrast to SSMs, NMs are most often symmetric and can have a diameter less than 5 mm.22 Bergenmar et al22 observed that NMs smaller than 2 mm thick were smaller in diameter compared with SSMs smaller than 2 mm thick (P < .05) and that NMs were more frequently new lesions (P < .05) compared with SSMs of similar thickness that more often had developed from preexisting nevi. The degree to which NMs exhibit color change is disputed. A recent Australian study30 observed that NMs (compared with SSMs) infrequently showed color change (29.1% of NMs vs 50% of SSMs; P = .007). The NMs were less likely to be pigmented (39.3% of NMs vs 82.3% of SSMs), and compared with SSMs, asymmetry was also relatively uncommon among NMs (20% were asymmetric vs 65% of SSMs; P = .001).30 However, Bergenmar et al22 found in their retrospective study that 63% of patients with NM observed a change in the lesion (eg, darkening color, diameter change, and increasing height). In the study by Chamberlain et al,30 which used different methods (both telephone surveys and in-person surveys), a change in elevation was most specific for NMs compared with SSMs.
With regard to site differences between SSMs and NMs, in Australia, NMs are reportedly found more often on the lower limbs or head and neck.4 The head and neck predominance of thick melanomas has been noted by other investigators5,31 However, in an Italian population–based study, tumor site was not related to thickness.14
Since NMs are smaller compared with SSMs, frequently appear as new lesions, and are often symmetric and elevated, it has been recommended that physicians involved in the management of cutaneous melanoma “adjust their threshold for excisional biopsy sufficiently to allow for the features that characterize early nodular melanomas.”30(p700) However, reversing the late detection of NM will be a formidable challenge. To date, a decade of comprehensive early-detection educational programs in the United States has had little impact on the rates of “biologically” aggressive thick NM. It is possible that the ABCD rule of clinical changes suggestive of melanoma (A, asymmetry; B, border change; C, color change; D, diameter >5 mm)32 has been more relevant for the more common and indolent SSM.9 Readjusting our thresholds for biopsy specimens with less reliance on size or asymmetry may be a worthwhile strategy. Important tradeoffs to consider in the earlier detection of NM would include a higher proportion of false-positive diagnoses stemming from findings of benign lesions, such as inflamed seborrheic keratoses, pyogenic granulomas, and cherry angiomas.
Our study has several limitations. The first is the absence of information on site and survival. However, 10-year survival for patients with SSM and NM is the same when tumor thickness is considered.33 Second, pathologic test results were not uniformly reviewed before coding by local SEER registries, similar to the way data were recorded in the Victorian Cancer Registry.4 The findings of Chamberlain and colleagues from another database in which histopathologic review of all cases was performed by a single expert dermatopathologist were actually similar to those of the Victorian Cancer Registry. Notably, the proportion of thick and nodular tumors was consistent between the current study and the study conducted in Victoria, indicating the reproducibility and biological significance of our study. Ideally, future studies should incorporate the verification of histologic coding.
Third, in the early 1990s, reports from state and SEER registries34- 37 found increased underreporting because population-based registries were not recording all cases from outside the hospital setting. To our knowledge, no studies have reported differences in tumor thickness, as a result of new reporting, but we cannot determine if modest improvements in tumor thickness, found in this study, reflect actual changes or variations that result from changes in obtaining new non–hospital-based cases. Tumor histologic findings remained relatively consistent during the 12 years of study. In regard to the nodular histologic data, a possibility of a bias exists, in the more recent period, toward NM being read as other types. However, no changes occurred in median thickness of NMs between the early (1988-1991) and late periods (1996-1999). Finally, 32% of recorded melanomas were melanoma NOS. Interestingly, the distribution and median thickness of melanoma NOS are far closer to those of SSM than to NM. This remained so for the 12 years of the study. Although this suggests a high representation of SSM in the NOS group, histologic type is likely to be represented in the unknown group in a nonbiased way.
In conclusion, apparently the current ABCD of melanoma, the core of the early-detection educational programs, may not suffice to permit the early detection of NM. A recent review by Abbasi et al38 recommended that ABCD be expanded to ABCDE to emphasize the significance of evolving pigmented lesions. Although we do not know the exact differences between early and late NMs, we must focus our efforts on identifying pathways that will lead to their early detection and/or understanding the molecular alterations involved in aggressive biological behavior (ie, dysregulation of apoptosis, BRAF mutations, or loss of p16 expression). Meanwhile, since SSMs are being detected earlier, ensuring that dermatologists perform total skin examinations will be relevant because physician detection is associated with thinner melanomas.39- 41 Teaching about skin self-examination will be critical as melanomas appear to be self-detected in most cases.41,42 Targeting our screening efforts to those with risk factors for melanoma (eg, older age, presence of solar [actinic] keratoses, nonmelanoma skin cancer, history of nevi [large number of nevi or clinically atypical nevi], and family history of melanoma) will also continue to be important. For example, it is apparent that targeting of middle-aged and older men is warranted. This observation has been supported by the Institute of Medicine, which conceded that “clinicians and patients should continue to be alert to the common signs of skin cancer—with a particular emphasis on older white males and on melanoma.”43(p62) Finally, as we gain greater understanding of melanoma carcinogenesis and molecular alterations leading to the development of biologically aggressive thick melanomas, other strategies such as chemoprevention44 may play a role in reducing the risk of developing thick melanoma and decreasing mortality from melanoma.
Correspondence: Marie-France Demierre, MD, FRCPC, Department of Dermatology and Medicine, Boston University School of Medicine, Skin Oncology Program, Boston Medical Center, 720 Harrison Ave, DOB 801A, Boston, MA 02118 (email@example.com).
Accepted for Publication: February 17, 2005.
Previous Presentation: This study was presented as a poster at the Second International Melanoma Research Congress meeting “From Bench to Bedside”; November 15, 2004; Phoenix, Ariz.
Acknowledgment: We are indebted to H. Randolph Byers, MD, PhD, for his critical review of the manuscript.
Financial Disclosure: None.