Strengthening health data on a rare and heterogeneous disease: sarcoma incidence and histological subtypes in Germany

In Germany, 11 population-based cancer registries covering the entire country exist. At the time of the data call, diagnoses up to the year 2013 were considered as sufficiently complete to be eligible for analysis. The cancer registry of Baden-Wuerttemberg had to be excluded as it was still being established in 2013. For trend analyses considering the years of diagnosis from 2004 to 2013, Hessen and Nordrhein-Westfalen (except for the district of Muenster) had to be excluded as well as they were considered complete only after the year 2004 by the ZfKD.

All malignant sarcomas according to the RARECARE Project [6, 10] and the WHO classification [1, 3, 11] were considered for analysis (Table 1) and, above all, GIST of uncertain biological behaviour. They were classified into 16 histological groups (Table 1) according to the classifications of the ICD-O-3 [12] and the WHO 2002 [1]. The complex neoplasia group consists of tumours of uncertain differentiation according to the WHO classification 2002 [1] (complex mixed and stromal neoplasia (ICD-O-3 Morphology (ICD-O-3 M) 8930–8991), synovial-like neoplasms (9040–9044)) and the malignant glomus tumour (8711). Above this, sarcomas were grouped into three entities: soft tissue sarcomas (all sites except bone (ICD-O-3 Topography (ICD-O-3 T) C40.0-C41.9); all included histologies except GIST (ICD-O-3 M 8936)), bone sarcomas (ICD-O-3 T C40.0-C41.9; all included histologies except GIST (ICD-O-3 M 8936)) and GIST (ICD-O-3 M 8936/3, 8936/1). This is modified from the RARECARE project classification of sarcomas into four main groups (tiers) [6, 10]. In contrast, we grouped the malignant GIST (8936/3) together with the GIST of uncertain malignant potential (ICD-O-3 8936/1), as the behaviour. is often hard to define in these entities. Kaposi sarcomas were not grouped as a separate entity due to the small number of cases. However, they were considered as a separate group according to histology for the sub-analysis. Sensitivity analyses were performed excluding certain histologies according to the opinion of experts or because they are missing in the WHO classification from 2002 (see Additional file 1: Table S1).

To compare the distribution of histological groups between different regions of the body, soft tissue sarcomas were classified into six different sites: head and neck, limbs, trunk, thorax, abdomen, and pelvis (see Additional file 2: Table S2). We chose to classify tumours in accordance with other studies in order to achieve optimal comparability. Consequently, some codes may have been summarized into the same categories although treatment strategies show relevant differences (e.g. classification of C48 into trunk tumours).

Information on the general population numbers was taken from the German Federal Statistical Office [13]. In total, a population of 69.95 million people representing 87% of the German population (key date December 31st, 2012) were included.

To determine the incidence of all sarcomas and their subtypes for Germany, incidence rates were calculated as the number of cases per 100,000 inhabitants, age-standardized by the European standard population 1976. Rates were analysed by sex and by year (2004 to 2013). In addition, age-specific rates in 5-year age bands were estimated, and subgroup analyses according to the sarcoma entity, histological groups, most frequent histologies, and federal states were performed. For all analyses stratified by federal state, cumulated frequencies and rates for the years 2009 to 2013 including 95% confidence intervals (95% CI) were calculated, dividing the cumulated numbers of cases by the corresponding cumulated population under risk.

For trend analyses, the most recent ten-year-period (2004 to 2013) was considered in order to enable valid analyses. In order to detect statistically significant changes in incidence in the last 10 years, annual percentage changes (APC) in the trend of the incidence rates were calculated with the statistical software Joinpoint (Joinpoint Regression Program, Version 4.4.0.0. January 2017; Statistical Research and Applications Branch, National Cancer Institute). Joinpoint uses log-linear regression to fit models with a minimal number of joinpoints (where rates change significantly) and estimates APCs and 95% confidence intervals for each section (between joinpoints): With no joinpoint, one APC was derived, with up to one joinpoint, an AAPC (average APC) was calculated. The slope of the regression line was tested to see if it was significantly different from zero.

Other analyses were performed with the software SAS 9.4 (SAS Institute, Cary NC).

Cases with unknown or unspecific histologies had to be excluded as it is unknown whether they were sarcomas. Hence, the number of sarcomas excluded is not possible to determine.

In the year 2013, 6846 sarcoma cases were diagnosed in Germany (excluding Baden-Wuerttemberg) and reported to a German cancer registry (3404 men, 3442 women, Table 2).

Soft tissue sarcomas were the most frequent entity (70% in men and 74% in women), followed by GIST (22% in men and 18% in women) and bone sarcomas (9% in men and 8% in women). There was about the same number of male cases as female ones (ratio men/women 1:1). However, the age-standardized incidence rate per 100,000 was higher in men than in women (7.4 (95% confidence interval (CI): 7.2–7.7) versus 6.6 (95% CI: 6.3–6.8). For soft tissue sarcomas, there were slightly more female than male cases (ratio men/women 1:1.1). For bone sarcomas and GIST, it was vice versa (ratio men/women 1.1:1 and 1.2:1 respectively).The age-standardized incidence rates were higher for men than for women for all entities.

Excluding certain histologies according to expert opinion or because they were not mentioned in the WHO classification 2002 did not considerably alter these results in men (Table 2). In women, the exclusion of Mullerian mixed tumours (ICD-O-3 M 8950/3) resulted in a slightly lower incidence in women.

The median age at diagnosis for all sarcomas was similar in men and women (67.4 and 67.7 respectively, Table 2). While GIST cases were about 2 years older at the time of diagnosis than all other sarcoma patients, bone sarcoma patients were much younger (49.8 (men) and 54.0 years (women) respectively).

For all sarcomas combined, there were more male cases than female ones in the two youngest age groups (Table 2). Above the age of 30, the ratio was reversed or balanced. The incidence steadily increased with age. Regarding 5-year age groups stratified by sarcoma entity (Fig. 1), it appeared that for all entities, the incidence decreased again at ages above 85 (except for soft tissue and bone sarcomas sarcomas in men). As well, the age distribution differed for bone sarcomas, showing a first peak at ages 10 to 19 and a second peak at ages above 85 for men and of 75 to 79 for women (Fig. 1).

The five most important histological groups were fibrosarcoma, liposarcoma, complex neoplasia, non-uterine leiomyosarcoma, and GIST, altogether representing more than half of all sarcomas (60% in men and 65% in women) (Table 3). Sarcomas which were not otherwise specified accounted for a considerable percentage of cases in both sexes (19% in men and 14% in women). The group of complex neoplasia was much more common in women than in men (sex ratio men/women 1:4.4) as carcinosarcomas including the Mullerian mixed tumour and endometrial stromal sarcomas are exclusively gynaecological tumours. The median age at diagnosis and the sex ratio differed considerably between the histological groups.

The distribution of histological groups depended on the site (see Additional file 3: Figure S1). While fibrosarcomas and angiosarcomas were the most predominant types in the head and neck, fibrosarcomas, liposarcomas and non-uterine leiomyosarcoma were most prominent in the limbs and trunk. In the thorax, 39% (men) and 36% (women) of sarcomas were not otherwise specified, and 22% (men) or 15% (women) had complex neoplasia. Angiosarcoma and fibrosarcoma were also common. In the abdomen, GIST represented 80% of sarcomas in both sexes. In the pelvis, distribution of histological groups differed considerably by sex. While in men, liposarcomas, non-uterine leiomyosarcomas, and rhabdomyosarcomas represent 64% of sarcoma cases, in women, complex neoplasia accounted for 71% of histologies. For bone sarcomas, about 34% were chondrosarcomas, about 26% osteosarcomas, and about 16% belonged to the Ewing family of tumours (data not shown).

Regarding the trend analysis for the years 2004 to 2013 (Fig. 2), there was a significant increase in incidence for men from 6.2 to 7.3 (APC 1.8% (95%-CI: 1.0 to 2.6)) and from 6.4 to 6.7 (AAPC 0.4% (95%-CI: 0.1 to 0.8)) for women (Table 4). It was due to a significant increase in incidence of GIST for both sexes: from 0.8 to 1.6 (APC 8.1% (95%-CI: 6.0 to 10.2)) in men and from 0.6 to 1.1 (APC 7.6% (95%-CI: 5.7 to 9.5)) in women. The incidence of soft tissue sarcomas (men) (APC 0.9% (95%-CI: -0.2 to 1.9)) as well as of bone sarcoma (both sexes, APC -1.1% (95%-CI: -3.1 to 0.9) for men and APC -0.5% (95%-CI: -2.0 to 1.1) for women) stayed constant over time (all incidence rates per 100.000, age-standardized according to European standard). As to soft tissue sarcoma in women, the incidence stayed constant up to the year 2009 and significantly decreased afterwards (one joinpoint detected). This lead to a decrease in the overall incidence of sarcoma after 2009 in women.

The sarcoma incidence differed significantly between federal states in Germany (Table 5). For all sarcomas, it ranged between 5.5 (95%-CI: 5.0–6.1) (Sachsen-Anhalt) and 7.8 (95%-CI: 7.5–8.0 (Nordrhein-Westfalen) or 7.0–8.6 (Hamburg)) in men and between 5.2 (95%-CI: 4.7–5.8) (Sachsen-Anhalt) and 8.7 (95%-CI: 7.9–9.5) (Hamburg) in women (all incidence rates per 100,000, age-standardized). Within the sexes, the confidence interval of the federal state with the lowest incidence did not overlap with the one of the federal state with the highest incidence. Analyses for sarcoma subtypes were not sensible due to low case numbers in small federal states.

To our knowledge, this is the first study on sarcoma incidence in Germany based on high-quality data from population-based cancer registries. Analyses were based on the latest data available at present. The study has high external validity as it covered 87% of the German population. In addition, the high number of cases allowed for detailed stratified analyses, e.g. by individual histologies. Finally, the study included all sarcomas based on histology and not only sarcomas emerging from soft tissue.

A lack of completeness cannot be ruled out. However, there are only estimates for ICD-10 C41- to C49. As well, different coding habits between federal states may exist. Moreover, there is a high proportion of sarcomas not otherwise specified, which limits the analysis of histological subgroups. Finally, cases with unknown histologies were excluded. Thus, some sarcomas may have been missed. It has to be mentioned that all cases had been coded according to the WHO classification of 2002 but in 2013 a novel classification has been published. Furthermore, only local pathology reports were available and to date it is unclear if and in which cases reference pathology was performed which would increase the reliability of subtyping sarcomas.