Time trend analysis of the skin melanoma incidence of Finland from 1953 through 2003 including 16,414 cases
Article first published online: 13 FEB 2006
Copyright © 2006 Wiley-Liss, Inc.
International Journal of Cancer
Volume 119, Issue 2, pages 380–384, 15 July 2006
How to Cite
Stang, A., Pukkala, E., Sankila, R., Söderman, B. and Hakulinen, T. (2006), Time trend analysis of the skin melanoma incidence of Finland from 1953 through 2003 including 16,414 cases. Int. J. Cancer, 119: 380–384. doi: 10.1002/ijc.21836
- Issue published online: 27 APR 2006
- Article first published online: 13 FEB 2006
- Manuscript Accepted: 5 DEC 2005
- Manuscript Received: 8 SEP 2005
- German Research Foundation (Deutsche Forschungsgemeinschaft, DFG). Grant Number: Förderkennzeichen 418 FIN 111/1/04
- skin neoplasms;
- time trend;
- cancer registries;
- epidemiologic methods;
Site-specific analyses of the skin melanoma incidence show marked differences between men and women by site and over time. The aim of our study was to analyze long-term population-based incidence time trends of skin melanoma in Finland over a period of more than 50 years, with special emphasis on sex- and subsite-specific changes over time. We analyzed incidence data of the Finnish Cancer Registry from 1953 through 2003 including overall 16,414 cases. We calculated age-standardized incidence rates per 100,000 person years using the European Standard Population. From 1953 through 2003, the incidence of skin melanoma increased from 1.5 to 12.8 per 100,000 among men and from 1.8 to 10.4 per 100,000 among women. Incidence rates showed a constant increase from 1953 through the mid of the 1980s. Thereafter, the rate of increase leveled off. The highest relative incidence increases occurred on the trunk among men and on the legs and hips among women. Within the skin area of the head, melanoma of the ear showed the highest relative increase among both men and women. Subsite-specific sex differences in the early registration period tended to become more pronounced in the most recent period. The highest body surface adjusted incidence rates occurred on the head. Only skin melanoma of the head showed an exponential age-specific incidence pattern and the aetiology of these skin melanomas may differ from skin melanoma on other subsites. © 2006 Wiley-Liss, Inc.
Over the past few decades, the incidence of skin melanoma has been rising in both sexes in almost all western-civilized developed countries, especially in predominantly white populations.1 One of the most established risk factors for skin melanoma is ultraviolet radiation.2
De Vries et al. recently studied incidence and mortality trends of skin melanoma in Western Europe including Finland of the period 1953–1997, focusing only on the overall skin melanoma incidence and mortality.3 They found that the incidence increase among women was best explained by an age-period-drift model. Among men, an age-cohort-period model best explained the incidence increase. Joinpoints were observed for melanoma in the age groups 25–49 and 50–69 years and occurred between the late 1970s and the 1980s in Finland, Norway and Sweden. After the joinpoints, the estimated annual percentage change decreased and indicated a stabilization or decrease of incidence rates.
Site-specific analyses of the skin melanoma incidence show marked differences between men and women. It has been frequently observed that incidence rates are highest on the trunk among men and legs among women.4, 5, 6, 7, 8, 9, 10, 11 The different gender-specific site-distributions of skin melanoma are often explained by different clothings and sun-bathing behaviors between men and women.7, 8, 11, 12
The aim of our study was to provide insights in the etiology of skin melanoma by analyzing long-term population-based incidence time trends of skin melanoma in Finland over a period of more than 50 years, with special emphasis on sex- and subsite-specific changes over time.
Material and methods
Finland is situated in Northern Europe between 60 and 70 degrees north and 20 and 30 degrees east. About 76% of Finns live in urban and 23.6% in rural communities. Sixty-five percent work in services, 28% in secondary production and 7% in primary production. The Finnish Cancer Registry was founded in 1952 and it covers the whole territory of Finland including about 5.1 million inhabitants.
Since 1961, the reporting of newly diagnosed cancers to the cancer registry is compulsory. Hospitals, pathology laboratories and general practitioners report cancer cases on a standardized form. The Central Statistical Office submits information whenever cancer is mentioned on a death certificate. The registry covers more than 99% of the solid tumors diagnosed in Finland.13
We extracted all primary invasive skin melanoma cases diagnosed between 1953 and 2003 from the Finnish Cancer Registry. Skin melanoma diagnoses were coded according to the International Classification of Diseases, 7th edition.14 If more than 1 skin melanoma (synchronous or at different points in time) was diagnosed during the registration period, we only counted the first diagnosis of skin melanoma. Between 1953 and 2003, overall 16,414 skin melanoma cases (7,869 men and 8,545 women) were reported. Overall, 99.5% of the skin melanoma cases were histologically verified. The proportion of death certificate only cases was 0.24% (40 cases). Overall, 35 skin melanomas (0.19%) were first diagnosed at autopsy. About 58.3% of skin melanomas were localized, 5.5% had regional lymph node metastases and 6.3% had distant metastases. In addition, about 2.5% were nonlocalized with missing information on the details of metastases. Staging was unknown in overall 27.5% and increased over time (1953–1962: 8.2%, 1993–2002: 37.8%). Until now, the T-staging of localized skin melanoma according to Breslow tumor thickness and Clark invasion level according to the TNM classification15 is not routinely available in the cancer registry data set and therefore we could not analyze incidence time trends of localized skin melanomas by T-stage.
We calculated age-specific (20–29, 30–39,…, 70–79, 80+ years) and age-standardized incidence rates per 100,000 person years, using the European Standard Population for the overall- and for subsite-specific (International Classification of Diseases, 7th edition (ICD7): 190.0–190.4 head, 190.5: trunk, 190.6: arms and shoulders, 190.7: legs and hips, 190.8–190.9: overlapping or unknown sites within the skin) skin melanoma incidence. Overall, 88 skin melanoma (0.5%) were coded as overlapping and 1,397 (8.5%) were coded as skin melanoma with unknown site.
For the study of the anatomical distribution of skin melanoma, we calculated relative site-specific age-standardized incidence rates per unit area of the skin for the early period 1953–1962 and the most recent period 1993–2002. We therefore divided the site-specific age-standardized incidence rates by the proportion of surface area of the corresponding body sites based on the estimates of the proportional surface area from Elwood and Gallagher12 who used estimates of the body surface from data of Lund and Browder16 and Pearl and Scott17: face 2.3%, ear 0.5%, scalp and neck 6.1%, trunk 32%, arms and hands 16.5%, legs 40%. We estimated the proportional surface area of the lips (0.1%) and the eye lids (0.5%). We calculated subsite-specific sex ratios (men: women) of the age-standardized incidence rates. We performed the incidence analyses with SAS 9.1.18
We ran joinpoint regression models19 to obtain formal estimates of points of change and to test the hypothesis that a change in trend occurred using the joinpoint regression program, version 3.0, from the National Cancer Institute.20 We fitted log-linear joinpoint regression models (0 up to 3 joinpoints) to the age-standardized rates with calendar year as the predictor and the standard error as the error term of the rate. For the estimation of percentage changes over time, we fitted regression lines to the natural logarithm of the subsite-specific age-standardized incidence rates using calendar year as a regressor variable, that is, y = a + bx, where y = ln(rate) and x = calendar year. The estimated annual percentage change (EAPC) is then estimated as 100*(eb−1). These models assumed that the logarithm of the rates changed at a constant rate over the periods.
From 1953 through 2003, the age-standardized incidence rates of skin melanoma increased from 1.5 to 12.8 per 100,000 among men and from 1.8 to 10.4 per 100,000 among women. Incidence rates showed a constant increase from 1953 through the mid of the 1980s. Thereafter, the rate of increase leveled off. Among men, we identified a joinpoint in the year 1987 (95%CI 1985–1989) with estimated annual percentage changes of +5.2% (95%CI 4.7–5.7%) from 1953 through 1987 and +1.2% (95%CI 0.6–1.9%) from 1987 through 2003. Among women, the joinpoint was 1982 (95%CI 1980–1986) with estimated percentage changes of +5.1% (95%CI 4.5–5.7%) from 1953 through 1982 and +1.4% (95%CI 0.9–1.9%) from 1982 through 2003. Joinpoint analyses by subsite revealed roughly similar joinpoints although considerably less precisely estimated (data not shown). Among men, the incidence of trunk melanoma is the highest throughout the registration period. Among women, the incidence of skin melanoma of the legs and hips is the highest incidence throughout the registration period. Several subsite-specific incidence increases appear to level off in the 1980s. The incidence increase of skin melanoma is accompanied by an increase of skin melanoma without subsite specification, especially during the most recent period (Fig. 1).
If one ignores skin melanoma with unspecified subsite, the estimated annual percentage changes (EAPCs) by subsite from 1953 to the gender-specific joinpoints show that skin melanoma on the trunk among men and on the arms and shoulders among women had the relatively largest increases. Among subsites of the skin of the head, skin melanoma on the ear showed the largest increase over time among both men and women. EAPCs differed by gender. Whereas the EAPC of skin melanoma on the trunk was higher among men than women, the EAPC of skin melanoma on the legs and hips were higher among women than men. After the gender-specific joinpoints (men: 1987, women: 1982), the EAPCs were considerably smaller in magnitude on all subsites. The EAPC of skin melanoma on the legs and hips among men for the period 1987–2003 showed a decreasing trend. Among women, skin melanoma on the scalp and neck and skin melanoma on the arms and shoulders showed continuous percentage increases during the recent period of 1982–2003 (Table I).
|Subsite (ICD7 code)||Men||Women|
|Head (190.0–190.4)||1168||3.5||2.4; 4.7||0.2||−1.9; 2.2||1457||4.3||2.8; 5.8||0.6||−0.6; 1.8|
|Lip, eye lid and face (190.0, 190.1, 190.3)||622||2.4||1.0; 3.8||−0.4||−3.1; 2.3||1048||3.8||2.1; 5.5||0.4||−1.0; 1.9|
|Ear (190.2)||156||16.3||5.3; 28.5||0.7||−3.0; 4.5||135||32.7||31.0; 34.5||−2.1||−6.5; 2.6|
|Scalp and neck (190.4)||390||9.5||3.8; 15.5||0.6||−2.5; 3.8||274||14.0||1.7; 27.8||3.0||0.8; 5.2|
|Trunk (190.5)||4044||7.3||6.6; 8.1||0.1||−0.7; 0.8||2291||5.0||4.0; 6.0||0.1||−0.7; 1.0|
|Arms and shoulders (190.6)||942||6.9||5.7; 8.2||1.7||0.2; 3.2||1350||7.9||5.9; 9.8||1.8||0.6; 3.1|
|Legs and hips (190.7)||941||2.6||1.5; 3.8||−1.3||−3.0; 0.5||2736||5.0||4.1; 6.0||−0.4||−1.2; 0.5|
|Multiple or unknown sites (190.8–190.9)||774||10.0||2.0; 18.7||9.7||8.0; 11.3||711||4.7||−3.0; 13.0||9.9||7.4; 12.5|
|All subsites combined (190.0–190.9)||7869||5.2||4.7; 5.7||1.2||0.6; 1.9||8545||5.1||4.5; 5.7||1.4||0.9; 1.9|
The incidence sex ratios by subsite show that differences between the sexes in the early registration period became more pronounced in the most recent registration period. The sex ratio of melanoma on the trunk continuously increased from 1.31 in 1953–1962 to 2.38 in 1993–2002. The sex ratio of melanoma on the legs and hips decreased from 0.56 in 1953–1962 to 0.43 in 1963–1972 and thereafter remained roughly constant. The sex ratio of skin melanoma on the ears and face including lips and eyelids only slightly changed over time. Incidence sex ratios of skin melanoma on the lips and eye lids are uninformative due to the instability of estimation (Table II).
|Lip, eye lid, and face (190.0, 190.1, 190.3)||0.97||0.64–1.48||0.83||0.59–1.15||0.86||0.66–1.11||0.93||0.76–1.13||1.05||0.88–1.24|
|Scalp and neck (190.4)||1.11||0.53–2.34||1.86||1.12–3.08||1.95||1.28–2.97||2.40||1.68–3.43||1.59||1.25–2.03|
|Arms and shoulders (190.6)||0.91||0.57–1.45||0.80||0.59–1.09||0.86||0.70–1.06||0.82||0.70–0.94||0.90||0.78–1.04|
|Legs and hips (190.7)||0.56||0.42–0.73||0.43||0.35–0.53||0.40||0.34–0.48||0.40||0.35–0.45||0.42||0.37–0.47|
|Multiple or unknown sites (190.8–190.9)||1.38||0.73–2.62||1.24||0.81–1.88||1.08||0.73–1.60||1.73||1.34–2.24||1.35||1.17–1.57|
|All subsites combined (190.0–190.9)||0.97||0.84–1.12||1.02||0.92–1.14||1.06||0.98–1.14||1.20||1.13–1.28||1.24||1.18–1.31|
The age-specific incidence rates of skin melanoma on the head follow approximately an exponential curve (i.e., a straight line on the logarithmic scale). The age-specific incidence increases among the other subsites roughly show a leveling-off at about ages 60–69, throughout the registration period (Fig. 2).
Table III presents detailed subsite-specific analysis of the age-standardized incidences of the late registration period (1993–2003). Regardless of the period, subsite-specific incidence rates without adjustment for body surface are highest on the trunk among men and on the legs and hips among women. Because of the different sizes of body surface, these incidences distort the interpretation of subsite-specific preponderances. After adjustment for body surface, there is a preponderance of skin melanoma on the head throughout the registration period among both men and women. We observed the highest body surface adjusted incidence on the head and its subsites (Table III).
|Site (ICD7-Code)||Surface (%)||Males||Females|
|N||Age-adj. rate||SE||RSA||N||Age-adj. rate||SE||RSA|
|Lip, eye lid, and face (190.0, 190.1, 190.3)||2.7||233||0.91||0.06||0.34||357||0.87||0.05||0.32|
|Scalp and Neck (190.4)||6.1||132||0.51||0.04||0.08||108||0.32||0.03||0.05|
|Arms and shoulders (190.6)||16.5||389||1.49||0.08||0.09||515||1.65||0.08||0.10|
|Legs and hips (190.7)||40.0||311||1.19||0.07||0.03||886||2.86||0.10||0.07|
|Multiple or unknown sites (190.8–190.9)||398||1.54||0.08||357||1.14||0.06|
|All subsites combined (190.0–190.9)||3034||11.69||0.21||3001||9.93||0.18|
The Finnish Cancer Registry enabled us to analyze skin melanoma incidence trends virtually for the whole population of Finns over a period of 51 years. Only few populations can provide incidence time trend analyses over such a long registration period, including Norway, Denmark and Connecticut, USA.21 We found a steady increase of the skin melanoma incidence in Finland since 1953. The estimated annual percentage increase was about 5% among both men and women, until the mid 1980s. Thereafter, the incidence increase leveled off. Although we analyzed a longer registration period and used the overall age-standardized incidence rate of skin melanoma for the joinpoint analysis, our joinpoints are practically the same as those derived from the period of 1953 through 1997 among the age groups 25–49 and 50–69 years, by De Vries et al.3
Our subsite-specific analyses show that annual percentage increases were highest on the trunk among men and on the legs and hips among women. Even in the very early registration period, the subsite distribution between men and women differed. Women had a higher risk of skin melanoma on the legs and hips and a lower risk of skin melanoma on the trunk than men, throughout the 51-year registration period. More detailed anatomic analyses from British Columbia, Canada, showed that trunk melanoma concentrate on the back, and melanoma of the legs and hips typically occur on the legs, with considerably lower incidences of melanoma of the hip/thigh.12 Recent detailed analyses of the exact location of 869 skin melanoma in Saarland, Germany (1978–1989), revealed that skin melanoma on the legs among women occurred in about 80% on the lower legs (unpublished data). The site-specific differences in Finland became larger with the increase of the skin melanoma risk over time. It is interesting to note that the highest annual percentage increases occurred on the skin of the ear among both men and women as has also been observed in East Germany from 1961 through 1989.11
The comparison of subsite-specific incidence rates adjusted for body surface shows that skin melanoma incidence rates among both men and women concentrate on the head throughout the registration period as has also been observed in other populations including New Zealand, Canada and Germany.11, 12, 22 The skin of the head is more likely to be chronically uncovered and exposed to UV radiation. Interestingly, also basal cell carcinoma of the skin has the highest body surface adjusted incidence rate on the head (including face, lips, eye lids, ears, scalp and neck) in Finland and Germany.23, 24 However, the magnitude of preponderance of skin melanoma to occur on the head is lower than that of non-melanotic skin cancer. For example, about 59% of all basal cell carcinoma and 67% of all other non-melanotic skin cancers diagnosed from 1991 through 1995 in Finland occurred on the head.23 During the same registration period, 14% out of 2,585 newly diagnosed skin melanoma occurred on the head.
These observations may question the current preference of the intermittent sun exposure theory of skin melanoma causation. It has been found that skin melanomas on the head are more often lentigo maligna melanoma than skin melanomas on other subsites.10, 12, 25 Our analyses and analyses from Germany11 show that only skin melanoma on the head show an exponential pattern of the age-specific incidence. In his registry-based population-based cohort study of 109,000 working Finns, Pukkala found only very little social class variation in the incidence of skin melanoma on the head and neck. In contrast, the trunk and the limbs showed a clear social class gradient with higher incidences among the higher social class.26 These findings may indicate that skin melanoma on the head may be etiologically and biologically different from skin melanoma on other subsites.
Although our study has several strengths, there are also factors that limit our results. First, the proportion of skin melanoma with missing information on subsite (overall: 8.5%) increased over time (1953–1962: 5.4%, 1993–2002: 12.1%) and therefore may bias our interpretation of subsite-specific incidence time trends. The earlier registration period therefore may more validly represent the subsite-specific distribution of skin melanoma than the more recent period. The comparison of skin melanoma with and without subsite specification shows that melanoma without subsite specification are more often reported to the registry in an advanced stage (regional or distant metastases: without subsite information: 37%, with subsite information: 12%) and were more often based on the histological examination of metastases instead of the primary tumor (25% vs. 1% respectively) which might have made the subsite allocation more difficult. If missing value of subsite was not associated with either subsite, the incident cases among the specified subsites would be proportionally reduced resulting in an attenuation of the observed subsite-specific incidence increases. Second, some of our subsite analyses suffered from small overall numbers of melanoma resulting in imprecise estimates.
In conclusion, we observed a considerable overall incidence increase from 1953 through the mid 1980s, with the highest relative increase on the trunk among men and on the legs and hips among women. Within the skin area of the head, melanoma of the skin of the ear showed the highest relative increase among both men and women. The subsite distribution of skin melanoma differed by gender throughout the registration period. Sex differences in the early registration period became more pronounced in the most recent registration period. The sex ratio (male/females) of melanoma on the trunk increased from 1.31 in 1953–1962 to 2.38 in 1993–2002. The sex ratio of melanoma on the legs and hips decreased from 0.56 in 1953–1962 to 0.43 in 1963–1972 and thereafter remained roughly constant. The highest body surface adjusted incidence rates occurred on the head, especially the face. Only skin melanoma on the head showed an exponential age-specific incidence pattern and the etiology of these skin melanomas may differ from skin melanomas on other subsites.
Andreas Stang was a recipient of a grant from the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG), Förderkennzeichen 418 FIN 111/1/04.
- 1Trends in cancer incidence and mortality. IARC Scientific Publications No. 121, International Agency for Research on Cancer, Lyon:IARC, 1993., , , , .
- 2World Health Organization (WHO). Solar and ultraviolet radiation. IARC monographs on the evaluation of carcinogenic risks to humans, vol. 55. International Agency for Research on Cancer, Lyon: IARC, 1992.
- 3Changing epidemiology of malignant cutaneous melanoma in Europe 1953–1997: rising trends in incidence and mortality but recent stabilization in Western Europe and decreases in Scandinavia. Int J Cancer 2003; 107: 119–26., , , .
- 14World Health Organization (WHO). International classification of diseases, 7th revision. Geneva: WHO, 1957.
- 15SobinLH, WittekindC, eds. TNM classification of malignant tumours, 6th ed. New York: Wiley-Liss, 2002.
- 16Estimation of area of burns. Surg Gynecol Obstet 1944; 79: 352–61., .
- 18SAS Institute. SAS for windows, version 9.1. Cary, NC: SAS Institute, 2002.
- 20Statistical Research and Applications Branch. Joinpoint regression program, version 3.0, Statistical Research and Applications Branch, National Cancer Institute, 2005.
- 21ParkinDM, WhelanSL, FerlayJ, RaymondL, YoungJ, eds. Cancer incidence in five continents, vol. 7 IARC Scientific Publications No. 143. Lyon: IARC, 1997.
- 25Epidemiology of cutaneous malignant melanoma. Clin Oncol 1984; 12: 407–37..
- 26Cancer risk by social class and occupation: a survey of 109,000 cancer cases among Finns of working age. Basel: Karger, 1995. Contributions to epidemiology and biostatistics, vol. 7, 82–90.