Development of prognostic factors and survival in cutaneous melanoma over 25 years

An analysis of the Central Malignant Melanoma Registry of the German Dermatological Society^†

Abstract

BACKGROUND

Recent studies revealed that incidence rates of cutaneous melanoma (CM) were leveling off predominantly among younger people and patterns suggested birth-cohort effects. The current study analyzed the development of prognostic factors and survival in incident CM over 25 years.

METHODS

All 45,483 patients with incident CM diagnosed between 1976 and 2000 recorded by the German Central Malignant Melanoma Registry were considered. Linear and logistic regression analyses were used to judge time trends. Trends of survival rates were tested with the multivariate Cox model.

RESULTS

Median tumor thickness decreased from 1.81 mm in 1976 to 0.53 mm in 2000 (P < 0.0001). The percentages of in situ and level II CM increased, respectively (P < 0.0001). The percentage of ulcerated CM decreased (P < 0.0001). The percentage of superficial spreading melanoma increased, whereas the percentage of nodular melanoma decreased (P < 0.0001). These time trends were all significant in the strata of gender, however, male patients presented in general with more advanced disease. Between 1976 and 2000, the average patient got older (P < 0.0001). The percentage of patients diagnosed with the primary tumor alone increased (P < 0.0001). Across the 25 years of observation, adjusted survival rates did not increase for females (P = 0.1561) but they increased for males (P < 0.0001).

CONCLUSIONS

The data demonstrated a strong trend towards prognostically more favorable CM most likely due to earlier diagnosis. Men and older people should be the focus of health promotion activities as they presented with more advanced disease. Cancer 2005. © 2005 American Cancer Society.

Between 1940 and the mid to late 1980s, the incidence of invasive cutaneous melanoma (CM) increased rapidly in populations of predominantly Caucasian origin.1 American and European studies on trends in cancer incidence reported that in those decades, CM had the most rapid incidence increases.2, 3 Since the mid-1980s, however, studies from Europe,4 Canada,5, 6 the U.S.,7, 8 New Zealand,9 and New South Wales, Australia10 revealed that incidence rates of CM were slowing, stabilizing, or even decreasing. Many of the more recent studies also showed that the observed leveling off in rates predominantly occurred in younger people and the patterns consistently suggest positive effects for cohorts born after 1945 or 1950.5–7, 9, 10

Consistent with the described trends for incidence rates, the most recent analyses of worldwide mortality data of CM concluded that rates started to decline in American, North European, and Australian cohorts born just before World War II and in more recent English and Canadian cohorts.11–13 However, analyses also showed that mortality rates were still increasing in several countries for certain subgroups, such as middle-aged men, and in southern European populations.11, 12 Despite these mostly encouraging trends, monitoring of incidence and mortality rates as well as changes in the characteristics of CM remains crucial.

Until 1995, cancer registration was not compulsory in Germany (population 82.2 million), by size of population the largest country in Europe. In 1995, a federal law made cancer registration mandatory for all German states. However, by 1998, still only a small fraction of the German population was monitored by population-based cancer registries and registration did not officially start before 2000 within several states.14 In 1983, the German Dermatological Society founded the Central Malignant Melanoma Registry (CMMR), a clinical-based melanoma registry that was developed to overcome the shortage of basic information on CM. Over the last 20 years, the CMMR has developed into a large multicenter project that records data retrospectively and prospectively for patients diagnosed with CM in 63 dermatologic centers located in the former Federal Republic of Germany, 16 centers located in the former German Democratic Republic, 2 centers in Switzerland, and 3 centers in Austria. In Germany, the majority of patients are referred to hospitals for treatment, and cooperative management of the patients between the dermatologist in private practice and the hospital is conducted. Thus, most patients are treated by specialists both in the hospital and in private practice. The registration of patients with melanoma in the CMMR in a given region with a participating center seems to be rather complete. However, only about one-half of all larger clinical centers in Germany actively participate in the registry and it is estimated that the CMMR records approximately 35–50% of all patients with melanoma in Germany.15

The CMMR database is currently one of the largest CM databases worldwide. It provides comprehensive information and exact documentation of prognostically relevant features of CM, which can be included in this type of specialized tumor registry. The current study analyzed trends of clinicopathologic features of CM recorded by the CMMR and survival rates of patients with CM over the 25 years from 1976 to 2000.

MATERIALS AND METHODS

By April 2003, 56,520 patients from over 80 clinical centers in Germany, Austria, and Switzerland were registered with the CMMR. In these countries, the follow-up of patients with CM is usually organized by clinical centers. Many patients with CM are transferred regularly to the dermatologic centers participating in the CMMR, even patients whose tumor was excised in private practice.

The current analysis included 45,483 patients with incident CM diagnosed between January 1976 and December 2000. CM of all clinical stages was included in the analysis. Multiple lesions were excluded. Written informed consent was obtained from all patients in the current study.

Statistical Analysis

Statistical analysis was conducted using SPSS for Windows, release 6.3 (SPSS, San Diego, CA). Depending on the type of the dependent variable, linear or logistic regression analyses were used to identify and judge time trends. The year of diagnosis was used as the independent time characteristic. Estimates resulting from linear and logistic regression analyses were reported in the current article.

Time trends with respect to survival rates were analyzed by the Kaplan-–Meier method and the multivariate Cox model in the periods 1976–1980, 1981–1985, 1986–1990, 1991–1995, and 1996–2000. Follow-up was defined as the date of last follow-up or the date of death minus the date of surgical resection. Follow-up was truncated after 10 years and was described as a median value with an interquartile range (IQR). Patients with a follow-up time of < 3 months were excluded from the survival analysis. Only deaths caused by CM as recorded in the CMMR were considered for survival analysis. Five-year survival probabilities with respective 95%-confidence intervals (95% CI) were calculated using the Kaplan–Meier method and compared with the log-rank test statistic. Survival probability values were based on results of life-table analyses. Multivariate Cox proportional hazards model was calculated to judge time trends of survival rates adjusted for other prognostic factors and results were presented as relative risks (RR) with respective 95% CIs.

RESULTS

The 45,483 patients in the current analysis were recorded by 82 collaborating dermatologic centers in Germany, Austria, and Switzerland. The mean age of the patients was 53.5 years (standard deviation ± 16.2); 55.7% were female; and 95.6% were of German, Austrian, or Swiss nationality. The number of patients in the current analysis increased from 133 in 1976 to 3538 in 1995.

Clinical Characteristics

The mean age of patients at diagnosis increased slightly from 48.8 years for male and 50.0 years for female patients in 1976 to 56.4 years and 54.7 years, respectively, in 2000 (P < 0.0001) (Table 1). The percentage of patients age ≤ 40 years decreased (1976–1980, 28.7%; 1995–2000, 21.7%) whereas the percentage of patients age > 65 years continuously increased (1976–1980, 19.7%; 1995–2000, 29.3%). The percentage of female patients decreased from 62.7% in 1976 to 51.6% in 2000 (P < 0.0001). The male-to-female ratio increased from 0.54:1 in 1976 to 0.92:1 in 2000 (P < 0.0001). Among males, most CMs were diagnosed on the posterior trunk (40.3%; 39.2% on the back). Among females, most CMs were diagnosed on the lower extremities (42.7%; 24.6% on the leg and 15.4% on the back). Among males, only the percentage of CMs diagnosed on the lower extremities changed during the 25 years of observation, from 22.0% in 1976 to 16.3% in 2000 (P < 0.0001). Among females, the percentage of CMs diagnosed on the lower extremities decreased from 49.8% in 1976 to 40.3% in 2000 (P < 0.0001), whereas the percentages of CMs diagnosed on the posterior trunk (1976, 14.7%; 2000, 19.3%; P < 0.0001), the anterior trunk (1976, 6.2%; 2000, 9.7%; P < 0.0001), and the head and neck (1976, 13.1%; 2000, 15.3%; P = 0.0381) increased.

Table 1. Changes in Clinical Characteristics and 5-Year Survival Probabilities (Percent with 95% Confidence Intervals) of Primary Cutaneous Melanoma Recorded by the German CMMR between 1976 and 2000

Characteristics	1976–1980	1981–1985	1986–1990	1991–1995	1996–2000	Total
CMMR: Central Malignant Melanoma Registry.
No. of patients	1199	3870	11,461	15,202	13,751	45,483
Age (yrs)
≤ 40	28.7%	25.4%	23.0%	23.4%	21.7%	23.1%
41–55	30.9%	33.8%	33.5%	29.5%	24.9%	29.5%
56–65	20.7%	19.0%	20.4%	21.9%	24.0%	21.9%
> 65	19.7%	21.9%	23.1%	25.2%	29.3%	25.5%
Mean age	50.8	51.5	52.6	53.4	55.2	53.5
Percent female	62.5%	60.8%	56.9%	55.2%	53.3%	55.7%
Body site, male
Head and neck	19.4%	14.5%	13.9%	14.5%	15.6%	14.8%
Anterior trunk	14.7%	15.5%	16.0%	17.0%	15.9	16.2%
Posterior trunk	33.2%	39.4%	40.7%	40.5%	40.4%	40.3%
Upper extremities	11.7%	11.2%	12.3%	11.8%	12.3%	12.0%
Lower extremities	21.0%	19.4%	17.1%	16.3%	15.7%	16.6%
Body site, female
Head and neck	13.9%	14.1%	12.6%	13.2%	14.6%	13.6%
Anterior trunk	5.9%	6.2%	7.1%	8.1%	8.3%	7.7%
Posterior trunk	15.6%	15.4%	16.7%	18.5%	18.0%	17.5%
Upper extremities	18.7%	18.2%	19.1%	18.0%	18.6%	18.5%
Lower extremities	45.9%	46.1%	44.4%	42.1%	40.5%	42.7%
5-yr survival rates	83.2% (80.9–85.6)	86.6% (85.4–87.9)	91.7% (91.0–92.4)	90.7% (89.9–91.4)	91.2% (89.5–92.8)	90.2% (89.7–90.7)
Male	78.4% (74.1–82.7)	81.2% (78.9–83.5)	89.4% (88.2–90.6)	88.0% (86.8–89.3)	88.8% (86.1–91.5)	87.4% (86.7–88.2)
Female	86.1% (83.3–88.9)	90.0% (88.6–91.4)	93.4% (92.6–94.3)	92.8% (91.9–93.7)	93.5% (91.7–95.4)	92.3% (91.8–92.8)

Pathologic Characteristics

The percentage of patients diagnosed with the primary tumor alone increased from 87.8% in 1976 to 95.2% in 2000 (males in 1976, 84.0%; males in 2000, 94.1%; females in 1976, 90.1%; females in 2000, 96.3%; P < 0.0001, respectively), whereas the percentages of patients with CM diagnosed with satellite or intransit metastases (1976, 7.2%; 2000, 1.2%; P < 0.0001), with regional lymph node metastases (1976, 4.3%; 2000, 2.7%; P = 0.0001), and with distant metastases (1976, 1.6%; 2000, 0.95%; P = 0.0091) decreased (Table 2). These time trends were consistent and statistically significant within the strata of gender, but not for distant metastases in female patients (P = 0.3813).

Table 2. Changes in Pathologic Characteristics of Primary Cutaneous Melanoma Recorded by the German CMMR between 1976 and 2000

Characteristics	1976–1980	1981–1985	1986–1990	1991–1995	1996–2000	Total
CMMR: Central Malignant Melanoma Registry.
Clinical staging
Primary tumor alone	91.3%	89.6%	92.4%	94.5%	94.5%	93.5%
Satellite or intransit metastases	1.3%	4.0%	3.7%	1.6%	1.1%	2.2%
Regional lymph node metastases	6.4%	4.7%	2.8%	2.9%	3.3%	3.3%
Distant metastases	1.0%	1.8%	1.0%	0.9%	1.0%	1.1%
Tumor thickness
≤ 1.00 mm	35.0%	42.2%	53.2%	58.5%	59.8%	55.8%
1.01–2.00 mm	24.6%	24.5%	22.8%	19.7%	18.3%	20.5%
2.01–4.00 mm	26.0%	22.3%	16.0%	13.8%	13.4%	15.2%
> 4.00 mm	14.4%	11.0%	7.9%	8.0%	8.5%	8.5%
Median	1.60 mm	1.30 mm	0.97 mm	0.81 mm	0.80 mm	0.90 mm
Level of invasion
I	2.1%	3.1%	5.8%	6.9%	7.9%	6.4%
II	14.2%	19.1%	20.8%	21.2%	22.5%	21.1%
III	35.7%	35.3%	38.0%	34.8%	36.9%	36.3%
IV	41.0%	37.2%	31.3%	33.0%	28.7%	31.9%
V	6.9%	5.4%	4.1%	4.2%	4.0%	4.3%
Ulceration, yes vs. no	17.3%	18.5%	17.1%	14.1%	12.9%	14.8%
Regression, yes vs. no	6.3%	7.0%	10.8%	13.1%	12.2%	11.8%
Growth pattern
Superficial spreading	44.1%	53.4%	61.3%	62.5%	59.3%	60.0%
Nodular	33.7%	31.1%	23.0%	19.6%	17.9%	21.3%
Lentigo maligna	17.2%	9.3%	7.8%	9.5%	11.1%	9.7%
Acral lentiginous	3.0%	4.2%	4.2%	4.1%	4.1%	4.1%
Other	2.1%	2.1%	3.6%	4.3%	7.6%	4.9%

Overall, 53.1% of male and 58.0% of female patients had a tumor thickness of ≤ 1.00 mm, whereas 9.9% of males and 7.4% of females were diagnosed with CM tumors > 4.00 mm (P < 0.0001). The percentage of thin CMs (≤ 1.00 mm) increased significantly from 39.0% in 1976 to 65.5% in 2000 (P < 0.0001), whereas the percentages of CMs 1.01–2.00 mm thick (1976, 28.9%; 2000, 17.3%; P < 0.0001), 2.01–4.00 mm thick (1976, 23.9%; 2000, 10.8%; P < 0.0001), and > 4.00 mm (1976, 11.4%; 2000, 8.0%; P < 0.0001) decreased. These trends were consistent and statistically significant within the strata of gender. The median tumor thickness decreased from 1.81 mm in 1976 to 0.53 mm in 2000 (P < 0.0001) (Fig. 1). This trend was significant within the 4 strata of gender by age (≤ 55 and > 55 years; P < 0.0001, respectively), with women age ≤ 55 years showing the lowest median tumor thickness across the observation time (P < 0.0001) (Fig. 2).

Figure 1. Time trends of mean and median tumor thickness of 45,483 primary cutaneous melanomas recorded by the German Central Malignant Melanoma Registry between 1976 and 2000.

Figure 2. Median tumor thickness for categories of time of observation (1976–1980, 1981–1985, 1986–1990, 1991–1995, and 1996–2000) stratified by gender (A: males and B: females) and age (age ≤ 55 years and age > 55 years).

Overall, 37.5% of male and 35.1% of female patients had CM with a Clark IV or V level of invasion (P < 0.0001). The percentages of in situ CM (1976, 3.0%; 2000, 10.0%; P < 0.0001) and of Level II CM (1976, 17.4%; 2000, 24.7%; P < 0.0001) increased, whereas the percentages of Level IV (1976, 39.1%; 2000, 27.2%; P < 0.0001) and Level V CM (1976, 6.1%; 2000, 3.8%; P < 0.0001) decreased. The percentage of Level III CM remained unchanged over the 25 years of observation (P = 0.7156). These time trends were significant within the strata of gender, but the results for Level V were significant for female (P < 0.0001) but not for male patients (P = 0.1533).

The percentage of ulcerated CMs decreased from 23.6% in 1976 to 13.0% in 2000 (males in 1976, 24.1%; males in 2000, 13.8%; females in 1976, 20.3%; females in 2000, 10.2%; P < 0.0001, respectively) and the percentage of patients with CM with signs of disease recurrence increased from 8.3% in 1976 to 14.8% in 2000 (P < 0.0001). Both of these time trends also were significant within the strata of gender. The percentage of patients with superficial spreading (1976, 54.8%; 2000, 64.4%; P < 0.0001) and lentigo CM (1976, 9.1%; 2000, 10.8%; P = 0.0146) increased, the percentage of patients with nodular melanoma (1976, 37.5%; 2000, 16.5%; P < 0.0001) decreased, and the percentage of of patients with acral lentigenous melanoma remained unchanged (P = 0.8666). The time trends relating to the growth pattern were significant in the strata of gender with minimal differences, but the result for lentigo maligna melanoma was significant for female (P = 0.0224) but not for male patients (P = 0.1051).

Overall, 21.3% of CMs were nodular CM (NM). Among thin CM (≤ 1.00 mm), 4.9% were NM, and 28.2% of CM 1.01–2.00 mm thick, 50.3% of CM 2.01–4.00 mm thick, and 60.2% of CM > 4.00 mm were NM (P < 0.0001). The percentage of NM in thick CM (> 4.00 mm) decreased from 74.0% in 1976 to 49.4% in 2000 (P < 0.0001). This time trend also was consistent and significant for male and female patients separately.

Survival Rates

Overall, 25,737 patients with invasive CM had recorded follow-up data. The median follow-up time was 4.0 years (IQR = 2.0–7.2) and 8.5% of patients with follow-up died of CM. The 5-year cumulative survival probabilities increased overall from 83.2% for the patients diagnosed between 1976 and 1980 to 91.2% for patients diagnosed between the 1996 and 2000 (P <0.0001) (Table 1). This trend was significant for both genders separately (P <0.0001, respectively). Survival probabilities for men were lower compared with women in all 5 categories of years of diagnosis separately (P < 0.001, respectively). Comparisons of changes in cumulative survival probabilities between patients diagnosed in 1990 or before and patients diagnosed after 1990 stratified in 8 categories defined by tumor thickness and gender showed that only male patients with tumors ≤ 1 mm thick demonstrated significantly improvement (P = 0.0025) (Fig. 3). All other time comparisons within the 8 strata were statistically nonsignificant.

Figure 3. (A) Cumulative survival probabilities of male patients with primary cutaneous melanoma stratified by tumor thickness and year of diagnosis. The bold line represents patients diagnosed between 1976–1990 and the fine line represents patients diagnosed after 1990. (B) Cumulative survival probabilities of female patients with primary cutaneous melanoma stratified by tumor thickness and year of diagnosis. The bold line represents patients diagnosed between 1976–1990 and the fine line represents patients diagnosed after 1990.

The multivariate Cox model based on 8882 female patients with complete data did not confirm the influence of the year of diagnosis on survival (P = 0.1561), but the model based on 6996 male patients confirmed the influence of the year of diagnosis on survival (P < 0.0001). The RR of death caused by CM was decreased during the years 1986–2000 when compared with 1976–1985 (RR = 0.60, 95% CI, 0.50–0.72; P < 0.0001). Both Cox models were adjusted for age, gender, body site, clinical staging, tumor thickness, level of invasion, ulceration, and histologic subtype.

DISCUSSION

The CMMR is not population based. However, it can be argued that these data are largely representative of the situation in Germany, as regional incidence studies using registry data confirm (regions of Jena and Tuebingen, unpublished data). Registration of patients with melanoma in the CMMR in a given region with a participating center seems to be rather complete. However, only approximately one-half of all clinical centers actively participate in the registry. In Germany, the management of melanoma, including primary surgery, is nearly exclusively performed by dermatologists. In the 1970s and 1980s, melanoma was resectioned with wide safety margins of 5 cm, independent of the tumor thickness. At that time, nearly all patients with melanoma were referred to hospitals for treatment. During the 1990s, management changed and 1–3 cm safety margins are now applied depending on tumor thickness. As a consequence, a growing number of patients with melanoma might now receive surgery in private practice. Thus, it is possible that a lower frequency of patients with thin melanomas are now referred to hospitals. Despite this potential referral bias towards thicker tumors, we still see the clear trend towards thinner tumors within the registry. Thus, the time trends discussed below can be considered conservative and it could be argued that unbiased trends toward prognostically favorable CM might be even more pronounced.

The results of the current analysis, which was based on > 45,000 patients with incident CM diagnosed in Germany, Austria, and Switzerland between 1976 and 2000 confirm a continuous ongoing trend towards prognostically favorable CM. Prognostically, the most influential tumor characteristics16–20 (e.g., clinical staging, tumor thickness according to Breslow, ulceration, level of invasion according to Clark, and histologic subtype) all expressed strong trends towards more favorable CM. Nearly all of these developments have been identified before with data collected in other European countries,20–23 the U.S.,7, 8, 24–26 Canada,6 Australia,10, 24, 27 and New Zealand9, 28 and, as the current study suggests, these worldwide trends apparently continue.

There is general agreement that these encouraging trends most likely were caused by the earlier detection of CM.21, 24, 25, 29, 30 In 2000, > 65% of all CMs recorded by the CMMR were ≤ 1 mm, whereas 44% were ≤ 0.5 mm and 10% of CM were diagnosed as in situ CM. It can be assumed that patients have been seeking medical attention at an earlier stage, which, in turn, is most likely attributable to an increased public awareness about the dangers of sun exposure and suspicious skin lesions. In Germany, melanoma awareness campaigns were introduced in the 1980s, which contributed to this increase in public and most likely also medical attention.31–34

Similar to previous studies,21, 22, 24, 30, 35, 36 the current analysis also demonstrated a trend towards improved survival with marked differences between the genders clearly favoring females. This trend most likely can be attributed to earlier diagnosis.24, 30, 35, 36 Among females, this overall time trend towards improved survival was practically and completely explained through prognostic factors and their own trends towards prognostically favorable CM. In contrast, among males, survival since 1986 seemed to be improved beyond the effects of early diagnosis as expressed with prognostic factors. This result could be explained by changes in unmeasured biologic features of CM or by changes in treatment procedures.21

Although most of the identified trends towards prognostically more favorable CM were true for male and female patients, male patients presented in general with more advanced disease and worse prognosis. For example, survival rates were lower for males compared with females, and a significant trend toward fewer patients with CM diagnosed with distant metastases was realized for male patients. However, the proportion of males with CM with distant metastases was > 2% in the early 1980s (1982, 3.6%; 1985, 3.4%) and decreased to approximately 1% during the time of observation. The proportion of females with CM with distant metastases was ≤ 1% between 1976 to 2000. For example, a significant trend toward fewer Level V CMs was observed for females but not for males, although in 19 of the 25 years, the proportion of patients with Level V CM was found to be higher among males than among females. For example, in 23 of the 25 years of observation, the median tumor thickness was higher among males compared with females (0.9 mm versus 0.8 mm overall). Compared with females and younger patients, the median tumor thickness was highest in males age > 55 years across the years of observation. Thus, health promotional efforts aiming for early detection should be targeted toward males, particularly older males. Those efforts will be even more relevant in light of the observed trends toward more males and older patients.

A tendency toward older patients has been reported in other studies.24, 27 This trend lends support for the assumption that declining incidence rates in more recent birth cohorts, as reported from other countries, also occur in Germany.5–7, 9, 10 At least the previously reported stabilizing and declining trends in mortality rates of CM in recent birth-cohorts in Germany follow the described changes for northern European countries.12, 13, 36 It has been speculated that health promotional efforts concerning the damaging effects of sun exposure were effective for cohorts born after World War II and initiated those changes in incidence rates.4, 6, 9, 10 Several studies from other countries6, 8–10, 28, 37, 38 and from Germany36 have reported that the increase in CM incidence was more pronounced among males than among females. As a likely consequence, the current study found an increasing male-to-female ratio, as did earlier studies.24, 27, 29, 39, 40 For both genders, the proportions of CMs located on the lower extremities decreased, whereas in female patients, trunk CMs increased. Among males, back CMs were most frequent, whereas in females, they were the second most frequent. A reduction in the proportions or incidence rates of CM on the lower extremities (or legs)10, 24, 29 and an increase on the trunk6, 8, 9, 27, 39, 41 in males or females or both were reported for > 20 years. Etiologic studies have identified intermittent rather than chronic sun exposure as the important environmental risk factor for CM.42, 43 The observed results stress the importance of intermittent sun exposure for the development of CM. This hypothesis is supported by the observation that since the 1960s, the increasing affluence of western European countries, such as Germany, resulted in lifestyle changes that gave an increasing number of residents more leisure time and the possibility to enjoy holidays in sunnier climates.

The results of the current study confirmed an ongoing worldwide trend towards prognostically more favorable CM, demonstrated as significant for all relevant prognostic features such as Breslow tumor thickness, Clark level of invasion, ulceration, and histopathologic subtypes of CM. This development is most likely due to earlier detection of CM. This trend was true for both male and female patients. However, during the 25 years of observation, males on average presented with more advanced disease and had a worse prognosis. The study also found a tendency toward increasing age and proportionally more male patients. These trends most likely reflect respective changes in incidence rates. In particular, the increase in average age could be evidence that declining incidence rates in more recent birth cohorts have been occurring in Germany.