By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Contributing oncologic centers: Southern region (Lund), Southeastern region (Linköping), Western region (Gothenburg), Stockholm-Gotland region, Uppsala-Örebro region, and Northern region (Umeå).
The objective of the current study was to compile prospective, population-based data on cutaneous invasive melanomas in Sweden during the period from 1990 to 1999, to describe and analyze survival data and prognostic factors, and to make comparisons with previously published Swedish and international data.
Twelve thousand five hundred thirty-three patients, which included 97% of all registered melanomas in Sweden, were included and described. Among these, 9515 patients with clinical Stage I and II melanoma were included in an analysis of survival and in a univariate analysis, and 6191 patients were included in a multivariate analysis of prognostic factors.
There was no significant change in melanoma incidence during 1990–1999. Favorable prognostic factors were found, especially in younger and female patients, resulting in a relative 5-year survival rate of 91.5%. In the multivariate analysis, significant factors that had a negative effect on survival were Clark level of invasion, Breslow thickness, ulceration, older patient age, trunk location, greatest tumor dimension, nodular histogenetic type, and male gender.
The Swedish Melanoma Study Group (SMSG) was established in 1976 with the objective of creating national guidelines for melanoma diagnosis, treatment, and follow-up and to conduct clinical studies. A national melanoma care program was initiated. Consensus was achieved regarding data to be collected on patients with newly diagnosed cutaneous melanoma. Studies focused on excision margins,1, 2 head and neck melanomas,3, 4 chemotherapy,5 and limb perfusion6 have been published previously.
Sweden, with 8.8 million inhabitants, is divided into 6 health care regions. In each of these regions, regional melanoma groups have been established with representatives in plastic and general surgery, dermatology, pathology, oncology, and the regional oncologic center. The latter is responsible for cancer registration and coordination of regional cancer care and is a part of the Swedish model of decentralized cancer care. Mutually agreed upon melanoma care programs were implemented successively, and these included recommendations about referral, diagnosis, staging, treatment, registration, and follow-up of all patients with melanoma in Sweden. Data are collected continuously and prospectively and are sent to and registered at each regional oncology center. Regular melanoma group meetings, as well as reports based on these data, have increased further the clinicians' awareness and knowledge of melanoma; thus enhancing the recruitment of patients to ongoing studies. A national data base was created by compiling these regional data on invasive melanoma. We have analyzed clinical and histopathologic data at diagnosis and have evaluated survival and prognostic factors in this prospective, population-based cohort of patients. The findings are compared with previously published Swedish and international data.
MATERIALS AND METHODS
The Swedish Prospective Melanoma Data Base
A working group of the SMSG and the oncologic centers compiled prospectively collected, population-based data during the period from 1990 to 1999 from the 6 Swedish regional melanoma registers. The patients with invasive cutaneous melanoma were accessed from the data bases of the Stockholm-Gotland and Western regions from 1990; from the Southeastern, Southern, and the Northern regions from 1992; and from the Uppsala-Örebro region from 1996 were included. The mean coverage rate of the registered melanomas (the number of patients in the regional registers/the number of patients in the Swedish Cancer Register) was 97% (range, 95–99%). During the study period, 12,533 primary invasive melanomas were registered and described: Among those, 9515 patients with clinical Stage I and II melanomas were used for the current survival and univariate analyses of prognostic factors, and 6191 patients with complete data available were used in the multivariate analyses of survival and prognostic factors. Only the first invasive melanoma diagnosed in each individual during the study period was included. Thus, patients who had a second melanoma during the study period were excluded. The following variables were registered for each patient; age at diagnosis, gender, tumor site, and maximum tumor dimension. The latter parameter was defined as reported by the clinician. The intent of operation(s), the surgical technique, and the extent of excision margins were registered. Histopathologic characteristics of the primary tumor, including histogenetic type, Clark level of invasion, Breslow thickness, and presence of ulceration, were registered. Factors that determine clinical stage, such as the occurrence of satellite, in-transit, regional lymph node, or distant metastases and their sites were registered. Follow-up data included the occurrence of a new primary melanoma, the type of recurrence, and the date and underlying cause of death.
When describing the material and analyzing survival and prognostic factors, we used the 2002 American Joint Committee on Cancer (AJCC) melanoma classification system.7 Elective or sentinel lymph node surgery was not practiced; thus, the staging was clinical. To make comparisons with previously published histopathologic data, we used the thickness intervals recommended by the International Union Against Cancer TNM Committee in 1977.8
The recommended treatment for primary melanoma during the study period was wide local excision with primary closure or skin transplantation. Excision margins differed due to ongoing studies. Melanomas that measure ≤ 2 mm in thickness usually are excised with a margin of 1–2 cm, and melanomas that measure > 2 mm in thickness usually are excised with a margin of 2–5 cm.
The Swedish National Cancer and Cause-of-Death Registries
Since 1958, all clinicians and pathologists in Sweden have been obligated legally to report every new malignancy independently to the Swedish Cancer Registry, which has a very high coverage rate of cancers.9 This makes the current registration population-based and nationwide. All cancers are registered with a unique patient identification number, which facilitates record linking.
Physicians in Sweden also are obliged to report every death and its cause to the Swedish Causes-of-Death Registry, which also has a very high coverage rate of cancers.10 In our study, data on mortality until 2000 were obtained from Statistics Sweden11 and were used to calculate the study endpoints of overall, melanoma-specific, and relative survival.
All patients with a melanoma diagnosed between 1990 and 1999 were followed until December 31, 2000. The annual age-standardized incidence rates per 100,000 person-years were calculated using the direct method of standardization with the world standard population and the Swedish standard population based on the census of 1970 as reference populations.12, 13 Clinical characteristics for continuous variables were compared using a standard t test. Associations between categorical variables were examined with chi-square tests for independence. For comparisons of median age, tumor thickness and dimensions, the median test of median values, the median test was used.14
Analyses of time-related trend changes in variables during the study period were performed using a regression model that related the logarithm of the yearly rates to a linear trend term. The linear regression model, thus, implies a constant annual change.15
Survival was calculated from the date of diagnosis (the date of diagnostic biopsy or the date of pathology report) to the date of death or the end of follow-up. The potential median follow-up was 60 months (range, 12–132 months). Overall survival and melanoma-specific survival rates were estimated using the method of Kaplan and Meier.16 In the analysis of the melanoma-specific survival rate, deaths due to causes other than melanoma were treated as censored observations. Relative survival was calculated according to the method of Hakulinen.17 The death rate in the reference population was based on information from national mortality data for Sweden for the period 1958–2000. The relative survival rate was calculated as the ratio of the observed survival rate in the patient group to the survival rate expected in a group similar to the group of patients at diagnosis with respect to all other factors (for this study, these factors included gender, age, and calendar time) that affected survival, except for the cancer in question. If this ratio is equal to 1, then the patients do not have any excess mortality due to their disease; whereas values < 1 generally indicate excess mortality attributable to the cancer. Distribution comparisons were made using the two-tailed log-rank test. A Cox proportional hazards regression model was used to assess the independent prognostic contribution of clinical variables either alone or after adjustment for other clinical variables. Cox models were adjusted for region and for year of diagnosis using the stratified version of the Cox method. In the multivariate analysis, both tumor thickness and Clark level of invasion were included. The prognostic impact of clinical variables was expressed as hazard rate ratios (RR) with 95% confidence intervals (95%CI).
There were no significant changes in the age-adjusted incidence rate of malignant melanoma among patients of either gender during the period 1990–1999 (if referring to the Swedish standard population of 1970, 17.1 in 100,000 males and 15.5 in 100,000 females) (Fig. 1).11, 18
All of the invasive melanomas and parameters analyzed are presented in Table 1.
Table 1. Distribution of Patient and Tumor Characteristics of Invasive Melanomas in Sweden, 1990–1999
No. of patients (%)
AJCC: American Joint Committee on Cancer; SSM: superficial spreading melanoma; LMM: lentiginous melanoma; NM: nodular melanoma; ALM: acral lentiginous melanoma.
This category was not included in the percentage distribution.
The male-to-female ratio was 0.97. The median age at diagnosis for all patients was 61 years. Male melanomas were diagnosed 5 years later than female melanomas (Fig. 2). During the study period, there was no significant change in the median age at diagnosis in the total population; however, there was a significant decrease in the proportion of patients age < 50 years, and there was a corresponding increase for older patients (chi-square test for trend; P < 0.0001) (Fig. 3).
Men had more melanomas localized on the trunk compared with women, whereas the opposite was seen for melanomas located on the extremities. Tumor sites differed among age groups (Fig. 4). For men age ≥ 50 years, the proportion of head and neck melanomas was larger, and the proportion of lower extremity melanomas was smaller compared with younger men. No such age-related difference was found regarding trunk melanomas. Women age ≥ 50 years had a higher proportion of head and neck and upper extremity melanomas but a lower proportion of trunk melanomas compared with younger women. There were no significant time-related changes in melanoma location for men or women age < 50 years during the study period; however, there was a slight increase in trunk melanomas for women age ≥ 50 years (chi-square test for trend; P = 0.001).
Among the patients who had their disease staged at diagnosis, the melanomas were Stage I and II in 95.3% of patients, Stage III in 2.9% of patients, and Stage IV in 1.8% of patients.
During the study period, excision margins were reduced in 1992, as a result of a randomized SMSG trial1, 2 that focused on excision margins, to 2 cm for melanomas that measured 0.8–2.0 mm in thickness (1 cm in the Western region as a result of a World Health Organization trial19), and the need for skin transplantations in this group decreased from 25% in 1990 to 13% in 1999. For melanomas that measured > 2.0 mm in thickness, the excision margins also were reduced from 5 cm to 2 cm or 4 cm due to an ongoing, randomized trial (with a change to 3 cm in the Western region), and the need for skin transplantation was 39.6%.
The clinical estimation of the greatest tumor dimension demonstrated that 22% of patients had tumors with tumors that measured < 5 mm in greatest dimension, and the proportion was higher for females than for males, whereas the opposite was seen for tumors with a greatest dimension > 14 mm. The superficial spreading melanoma (SSM) histogenetic type was most frequent, followed by nodular melanoma, lentiginous melanoma, and other types. The distribution of histogenetic types of melanomas was unaltered during the study. The median melanoma thickness was 0.9 mm and was constant during the study period. The median melanoma thickness for males was 1.0 mm, which was significantly higher than the median 0.85 mm thickness for females (P < 0.0001). The differences in thickness distribution related to age and gender are shown in Figure 5A,B. A time-trend analysis for 1990–1999 demonstrated a significant linear decrease of thin (< 1.0 mm) melanomas in younger patients from 69% to 66% (chi-square test for trend; P = 0.003). Clark Level II melanomas were more frequent in females than in males, whereas the opposite relation was found for Clark Level IV tumors. From 1990 to 1999, the proportion of Clark Level II melanomas decreased from 39% to 29% (chi-square test for trend; P < 0.001).
Ulceration of melanomas was less frequent in females (20.7%) than in males (27.5%). There were no significant time-related changes in the frequency of ulceration. The relation between melanoma thickness and the presence of ulceration is illustrated in Figure 6.
The observed overall 5-year survival rate for the patients with clinical Stage I and II melanomas that were diagnosed during 1990–1999 with last follow-up in December, 2000, was 80.5% (95%CI ± 1%), their relative survival rate was 91.5% (95%CI ± 1%), and their melanoma-specific survival rate was 90.4% (95%CI ± 1%) (Table 2). The 5-year melanoma-specific survival rate for patients with Stage III melanoma was 35.6% (95%CI ± 7.5%); and, for patients with Stage IV melanoma, it was 24.6% (95%CI ± 7.5%). For patients who had clinical Stage I–II melanoma, the influence of melanoma thickness and gender are illustrated in Figure 7A–C. The 5-year survival was significantly better within each thickness group among the patients with T1a–T4a tumors (without ulceration) compared with each corresponding group among the patients with T1b–T4b tumors (ulceration present) (Fig. 8).
Table 2. Five-Year Survival of Patients with Stage I and Stage II Melanoma in Sweden, 1990–1999: Overall, Relative, and Disease-Specific Survival
A univariate survival analysis of clinical Stage I and II melanomas demonstrated (in descending order) an increased risk ratio (RR) > 10 for patients with Clark Level V, Breslow thickness > 4 mm, Clark Level IV, and Breslow thickness 2.01–4.0 mm; an RR > 5–10 for patients who had ulceration and Clark Level III; and an RR of 1–5 for patients with the nodular histogenetic type, Breslow thickness 1.01–2.0 mm, age > 70 years, the acral-lentiginous type, greatest melanoma dimension > 14 mm, hand/foot/subungual location, and age 50–69 years. Low RRs were demonstrated for females, upper extremity locations, and lower extremity locations (Table 3). A univariate survival analysis that included all clinical stages demonstrated an RR of 11.3 (95%CI, 9.5–13.4) for patients with Stage III melanoma and an RR of 20.2 (95%CI, 16.7–24.6) for patients with Stage IV melanoma.
Table 3. Univariate Analysis of Prognostic Factors for Patients with Stage I and Stage II Melanoma: Disease-Specific Survival Adjusted for Region and Year of Diagnosis
The RR was adjusted for region and year of diagnosis.
< 50 yrs
≥ 70 yrs
Head and neck
Greatest tumor dimension
> 5–9 mm
> 9–14 mm
> 14 mm
≤ 1.0 mm
> 4.0 mm
A multivariate analysis of clinical Stage I and II melanomas demonstrated an almost similar order of RRs for the same factors from the univariate analysis. Patients who had ulceration, however, had a lower RR compared with patients who had Clark Level III and Breslow thickness 1.01–2.0 mm, and patients who had melanoma located on the trunk had an increased RR of 1.48 (95%CI, 1.10–1.98) (Table 4).
Table 4. Multivariate Analysis of Patients with Clinical Stage I and II Melanoma: Disease-Specific Survival Adjusted for Region and Year of Diagnosis
The RR was adjusted for region and year of diagnosis.
< 50 yrs
≥ 70 yrs
Head and neck
Greatest tumor dimension
< 15 mm
≥ 15 mm
< 1.0 mm
> 4.01 mm
During 1960–1979, the age-adjusted incidence rate of invasive cutaneous melanoma in Sweden increased annually by 5.5% in males and by 5.7% in females. The corresponding rates for 1970–1989 were 2.6% in males and 2.2% in females. During 1990–1999, no significant changes were found in the time-trend analysis.11 New rates from Sweden for 2001 and 2002, however, indicate a further increase in the incidence of melanoma that will be followed in our register. A decrease in the incidence of melanoma has been observed both in New South Wales20 and in the United States.21 Altered sun habits, resulting in a change and reduction of sun exposure, may be an explanation for this decrease, in addition to the effects of activities that lead to the early diagnosis and removal of premalignant lesions. There also may be a minor dilution effect from immigrants coming from areas with a low incidence of melanoma, which has been discussed in Australia22 and has been noticed in Sweden.23 Regardless, melanoma incidence in Sweden remains high from a European perspective.24
Age and Gender
In a previously published, population-based study from the southern region of Sweden,25 the male-to-female ratio increased from 0.43 in 1965 to 0.71 in 1985; in the Stockholm-Gotland region, the male-to female ratio increased from 0.79 in 1976–1979 to 0.89 in 1990–1994.26 The ratio of 0.97 that was found in our study supports this tendency. Our male-to-female ratio was higher than the central European ratio of 0.87 from the central malignant melanoma register of the German Dermatological Society (CMMR) but lower than the ratio of 1.29 reported by the Queensland Melanoma Register (QMR).27 This rise of the male-to-female ratio may reflect differences in outdoor activities in addition the fact that women are more informed about and interested in preventive actions. Women are more concerned about their skin lesions and have them removed at a younger age and at earlier stages of melanoma development. The mean age at diagnosis of the patients in the current study (59.5 years) was older than the mean ages reported in the CMMR and QMR studies (55 years in both). Differences in knowledge and awareness of melanoma and an elderly population in Sweden compared with Queensland28 may provide an explanation for the differences, along with the selection bias of the CMMR material, which comprises only 25% of the population.
We observed differences with respect to primary tumor location depending on age and gender. Head and neck melanomas were three times more frequent in older patients. The site distribution was almost the same as that registered in Norway during 1976–1985.29 For women age < 50 years, 41% of tumors were localized to the trunk, proportions that were equivalent to those for men in Queensland during 1986–1989.27 There were no time-trend changes in melanoma location. Differences in trunk melanoma between younger and older women indicate differing sun-tanning habits and clothing styles.
Data obtained on the maximum tumor dimension are not exact, because ill-defined borders can interfere with measurements by clinician; however, our findings are interesting in two aspects. First, many melanomas measure < 5 mm, indicating that melanoma size, as such, is of no use when diagnosing early melanomas and that anamnestic changes of pigmented lesions and episcopy should be used. Second, we found that patients with large melanomas (> 15 mm) had a significantly worse prognosis in the multivariate analysis.
The histopathologic data from the current study were compared with earlier Swedish data30, 31 and with data from central Europe and Australia.27 The proportion of SSM melanomas has increased in Sweden from 46% in 1960–1964 to 68% (unspecified and other types are excluded in the calculation to make a comparison possible), which is close to the European rate (CMMR) but lower than the 74% rate reported from the Australian (QMR) material. The median thickness of Swedish melanomas decreased from 3.0 mm in 1960–1964 to 0.9 mm, compared with 0.8 mm in the CMMR study and 0.6 mm in the QMR study. The proportion of thin melanomas (< 0.75 mm) has increased 4-fold to 40% since 1960–1964 but is lower than the proportion reported in the CMMR and QMR studies. The proportion of melanomas that measured > 1.5 mm was reduced to 32% but still was higher than the proportions reported in the CMMR and QMR studies. The proportion of Clark Level II melanomas has increased almost 10-fold to 31% since 1960–1964, but this remains low compared with the 61% proportion from the QMR study. The proportions of Clark Level IV and V melanomas were s halved to 33% but remained higher than the proportions reported in the CMMR and the QMR studies. An interesting observation is that 22% of the melanomas measured only < 5 mm in greatest dimension. We found a somewhat worrying decreasing trend in the proportion of thin melanomas in young patients and also found a 10% decrease in Clark Level II melanomas during 1990–1999. We can only speculate whether there have been fewer preventive campaigns, whether health care resources have declined, or whether patient attitudes have changed. The frequency of ulcerated melanomas in Sweden has decreased from 53% to 24% since 1960–1964. The proportion of ulcerated melanomas is high and may be lower; because, sometimes, pathologists previously did not report negative findings concerning ulceration of thin melanomas. However, our findings are similar to those that were reported in a study from the Stockholm-Gotland region during the period from 1976 to 1994. A retrospective histopathologic study investigated missed reports about ulceration. Twenty-four percent of the melanomas had ulceration, and 3% were unclassifiable. Previously, 12% of these melanomas had been categorized as unclassifiable.26 Our data from 1990 to 1999 indicate that many prognostic factors have changed in a positive direction in Sweden and are comparable to those reported by the CMMR, although the latter may have a positive selection bias. In addition, it is not unlikely that, in Germany, a greater proportion of melanomas have been diagnosed and treated at dermatologic clinics with better diagnostic resources, compared with Sweden, where patients usually are diagnosed in primary care, and surgery is performed by general surgeons at local hospitals. There also is a great difference compared with Australia, where there is a very low median melanoma thickness and a high frequency of SSM and Clark Level II melanomas, probably due to systematic, preventive melanoma campaigns. The older age of the Swedish study population may explain further the more common occurrence of thick melanomas.
In an previously published Swedish report, the relative 5-year survival rate for males with melanoma increased from 54.5% in 1960–1964 to 73.1% in 1980–1982.32 The corresponding rates for females with melanoma were 65.8% and 84.9%, respectively. In the current study, we found a further increase in the relative male survival rate to 88% and an increase in the relative female survival rate to 94%: a considerable improvement. Furthermore, the gender differences with respect to survival have decreased. Our findings are similar to what has been reported from Europe.33, 34 The gender difference regarding 5-year relative survival is explained by lower age at diagnosis, fewer trunk melanomas, lower clinical stage, smaller and thinner melanomas, and less nodular and ulcerated melanomas in women. The relative survival of patients age < 50 years, 96%, also is high. Our findings correspond to the fact that the decrease in mortality from melanoma during the period 1985–1995 was more pronounced for women in the United States, Australia, New Zealand,35 and Sweden.36 For Stage I and II melanoma, we found a 5-year disease-specific survival rate similar to data reported on patients with T1a–T4b melanomas, as presented in the AJCC 2002 report,7 although no sentinel lymph node staging or elective lymph node dissections were done in our material. Our higher survival rates may be explained by our shorter follow-up. Conversely, our material included patients with up to 10% of undiagnosed lymph node micrometastasis if using the figures regarding histopathologically positive lymph nodes found in patients with clinically negative lymph nodes reported in the same AJCC 2002 material.7 We had expected a worse outcome in our material. We found a surprisingly high 5-year disease-specific survival for patients with Stage IV melanoma that merits further investigation.
In the multivariate analysis of patients with clinical Stage I and II melanomas, we found high risk rates for Clark level of invasion and Breslow thickness and found somewhat lower risk rates for ulceration, trunk sites, hand/foot/subungual sites, increasing age, nodular melanomas, and greatest tumor dimension. In contrast, the risk rate was reduced in females. Because both Clark level of invasion and Breslow thickness were included in the multivariate analysis, the calculated risk rates for both variables were reduced, but they still remained high. We were surprised to find such a strong impact of Clark level in our analysis, because the factor is used only for thin melanomas in the new AJCC 2002 classification system. We were aware that pathologists may evaluate Clark level differently37 and that Breslow thickness is more reproducible; however, the Clark level still may provide additional biologic information. Therefore, we recommend that the Clark level still should be reported to melanoma registries, although it is used only for T1 melanomas in the current AJCC 2002 classification system. We also made a multivariate analysis of our clinical Stage I and II melanomas using the same model as the AJCC group. The significant parameters found in the AJCC 2002 analysis were (in order) thickness, ulceration, age, gender, melanoma site, and Clark level.38 We found that all of those parameters were highly significant, although the Clark level was third on our list, and age was sixth; however, compared with the AJCC 2002 material, we found a higher risk rate for male patients, which also has been reported in most other materials.
The previous, steep increase in melanoma incidence in Sweden has leveled off, and no further significant change was found during 1990–1999. The age of patients at diagnosis is older in Sweden compared with patients in Central Europe and in Queensland, Australia. Differences in melanoma location due to gender were the same as described in previous Scandinavian reports. It is noteworthy that there was a high proportion (41%) of trunk melanomas among young females, a finding that may indicate changes in clothing and tanning habits. Compared with earlier periods, the proportion of SSM melanomas has increased along with the proportion of thin melanomas and of Clark Level II melanomas. In contrast, there was a decrease of nodular, thick, Clark Level IV–V melanomas and ulcerated melanomas. The median melanoma thickness has decreased in Sweden but remains higher compared with the median melanoma thickness in Queensland, Australia. Because of the changes mentioned above, a relative 5-year survival rate of 91.5% has been achieved. The 5-year survival rate for patients with T1a–T4b melanomas who had clinically negative lymph nodes was similar to the data from the final AJCC 2002 report, although the patients in the AJCC report had pathologically negative lymph nodes after undergoing elective lymph node surgery or sentinel lymph node staging. In a multivariate analysis of patients with clinical Stage I and II melanomas, significant factors that had a negative impact on survival were the Clark level of invasion, Breslow thickness, the presence of ulceration, melanoma with a hand/foot/subungual location or a trunk location, older patient age, nodular histogenetic type, and a greatest tumor dimension > 14 mm. Female patients had an improved prognosis.
The increased survival during the study period may have been because there was an improvement in secondary prevention of malignant melanoma in Sweden, probably due to greater knowledge and awareness among both the Swedish general population and Swedish health care professionals. Furthermore, the more favorable prognostic signs noted for the younger population may indicate the beginning of a change in tanning habits in Sweden. A worrisome trend appears to be the decrease in the incidence thin tumors and of Clark Level II melanomas in young patients. Our comparison with data from Queensland, Australia, however, demonstrates that there still is room for further improvements in melanoma prevention.
The authors thank all of the Swedish clinicians who sent melanoma data to their regional oncologic centers.