Fax: +358-3-215 607.
Incidence of bone and soft tissue sarcoma after radiotherapy: A cohort study of 295,712 Finnish cancer patients
Version of Record online: 8 SEP 2005
Copyright © 2005 Wiley-Liss, Inc.
International Journal of Cancer
Volume 118, Issue 4, pages 1017–1021, 15 February 2006
How to Cite
Virtanen, A., Pukkala, E. and Auvinen, A. (2006), Incidence of bone and soft tissue sarcoma after radiotherapy: A cohort study of 295,712 Finnish cancer patients. Int. J. Cancer, 118: 1017–1021. doi: 10.1002/ijc.21456
- Issue online: 13 DEC 2005
- Version of Record online: 8 SEP 2005
- Manuscript Accepted: 1 JUL 2005
- Manuscript Received: 17 MAR 2005
- bone neoplasms;
- cohort studies;
- radiation effects;
- soft tissue neoplasms
Radiotherapy is commonly used for treatment of malignant disease. As a consequence of radiotherapy, an increased risk of developing a second malignant neoplasm has been shown. However, little is known about the effects of radiation on developing sarcoma. The aim of this study was to examine the risk of developing a bone or soft tissue sarcoma after radiotherapy for a first primary cancer. The study population included all the patients with primary cancers of breast, cervix uteri, corpus uteri, lung, ovary, prostate, rectum and lymphoma diagnosed during 1953–2000 and identified from the Finnish Cancer Registry. Patients were followed up for subsequent sarcomas. The follow-up yielded 1.5 million person-years at risk and 147 sarcomas. Compared to the national incidence rates, after 10 years of follow-up sarcoma risk was increased among patients who had received neither radiotherapy nor chemotherapy (standardised incidence ratio (SIR) 2.0, 95% CI 1.3–3.0), radiotherapy without chemotherapy (SIR 3.2, 95% CI 2.3–4.3), chemotherapy without radiotherapy (SIR 4.9, 95% CI 1.0–14.4), as well as combined radiotherapy and chemotherapy (SIR 3.4, 95% CI 0.4–12.5). For radiotherapy in ages below 55 the SIR was 4.2 (95% CI 2.9–5.8). In the adjusted regression analysis the rate ratio was 1.5 (95% CI 0.9–2.6) for the radiotherapy group. In conclusion, radiotherapy appears to be associated with an increased risk of developing sarcoma especially among younger patients. Further investigation is needed to clarify the dose–response of the preceding ionizing radiation. © 2005 Wiley-Liss, Inc.
The risk of developing cancer is increased among persons who already have had cancer previously, which may be caused by both the joint causal factors for the first primary cancer and the treatment of the first cancer, as both chemotherapy and radiotherapy are known to increase the risk.1 Not much research has been done on the effect of ionising radiation on development of sarcoma. An epidemiological evaluation of sarcoma after exposure to radiation gave an average excess relative risk estimate of 1.42 at 1 Sv, being higher among men and younger age.2 This estimate is comparable to several other cancer types, with excess relative risk estimates ranging 0.1–0.2 per Sv.3
Bone and soft tissue cancers comprise less than 1.5% of all the cancers diagnosed in Finland during 1953–2000.4, 5 Histologically sarcomas, defined as malignant tumours originating from mesenchymal tissue, constitute nearly all the most common types of bone and soft tissue cancer.6 Ionising radiation, is in addition to specific chemicals and genetic pre-disposability, one of the few known factors to induce bone or soft tissue cancer.7, 8
The aim of this study was to assess the risk of developing a bone or soft tissue sarcoma after radiotherapy for cancer using a large, population-based cohort of cancer patients.
Patients and methods
The study cohort included all the primary cancers of the breast, cervix uteri, corpus uteri, lung, ovary, prostate and rectum, and all the Hodgkin's and non-Hodgkin lymphomas diagnosed and registered during 1953–2000 in the Finnish Cancer Registry database. These cancer sites constitute 43% of all the registered cancers in men and 55% in women during 1953–2000 in Finland.4, 5 These types of cancer were selected because they are relatively common and treated both with radiotherapy and other treatment modalities, mainly chemotherapy, surgery and hormonal therapy. Small-cell carcinoma of the lung was excluded because it is not commonly treated with radiation therapy. Finnish Cancer Registry is a population-based, nationwide cancer registry with an almost complete coverage of all cancer cases diagnosed in Finland since 1953.9 A permission to use the Finnish Cancer Registry database was granted by the National Research and Development Centre for Welfare and Health.
Sarcoma as an outcome event was defined as a cancer originating from bone or soft tissue, with histological confirmation as sarcoma. Bone cancers included chondrosarcomas, Ewing's sarcomas, fibrosarcomas, osteosarcomas, reticulum cell sarcomas and undefined sarcomas. Soft tissue cancers included fibrosarcomas, leiomyosarcomas, liposarcomas, myksosarcomas, rhabdomyosarcomas, reticulum cell sarcomas, synovial sarcomas and undefined sarcomas. These histological types constituted 72% of all the bone cancers and 85% of all the soft tissue cancers registered during 1986–1995 in the Finnish Cancer registry database.
Analyses were stratified by treatment for the primary cancer during 4 months since diagnosis: age during diagnosis of the primary cancer, sex, length of follow-up time (0, 1–4, 5–9, 10–14 or ≥ 15 years since diagnosis of primary cancer) and site of the sarcoma (bone or soft tissue). Treatment was classified as radiotherapy without chemotherapy, chemotherapy without radiotherapy, combined chemotherapy and radiotherapy, and treatment without chemotherapy or radiotherapy. Chemotherapy was considered separately because it may be a separate risk factor for developing sarcoma. As the risk of developing a second primary cancer decreases with increasing age,10 age at the time of diagnosis of the first primary cancer was classified as under 55 or 55 and more years.
Follow-up for sarcoma started at diagnosis of the first primary cancer and ended at death, emigration, or the common closing date (December 31, 2000) whichever occurred first. Information on vital status and emigration was obtained from the Population Register Centre using computerised record-linkage.
The observed sarcoma cases were identified from the Finnish Cancer Registry. The expected numbers of sarcoma cases were calculated by multiplying the person-years at risk with the national incidence rates of sarcoma stratified by sex, 5-year calendar period and 5-year age group. The standardised incidence ratio (SIR) was calculated as the ratio of observed to expected numbers of cases. The 95% confidence intervals (CIs) were estimated assuming that the observed number followed a Poisson distribution and the expected number being a constant.11 Also, cumulative incidence or incidence proportion estimating sarcoma risk at 10 years of follow-up was counted.12 Number needed to harm13 was estimated to illustrate the absolute effect of radiotherapy on subsequent risk of sarcoma. Number needed to harm is the inverse of the absolute risk difference and expresses the number of patients who should receive radiotherapy in order to induce one excess sarcoma during 10 years follow-up.
In the analysis, sarcoma cases were examined for the entire follow-up and separately for the follow-up of 10 or more years after diagnosis of the first primary cancer, because the minimal latency time for radiation-induced solid tumours is 10 years.14 Poisson regression modelling was used to adjust for the effects of different age, follow-up and sex distributions of different treatment categories on the incidence of sarcoma.15 Modification was assessed by adding interaction terms to the model containing the main effects. Stata 8.0 (Stat Corporation, College Station, TX, USA) software was used in the analysis.
A total of 295,712 patients with the selected first primary cancers were included in the cohort. Among the cohort members, 54% were women and 75% were aged 55 or more during time of diagnosis of primary cancer (Table I). The follow-up yielded 1,524,590 person-years at risk, and the mean length of follow-up was 5 years. The patients who received chemotherapy without radiotherapy provided 5%, those receiving radiotherapy but no chemotherapy 43%, and the patients receiving both chemotherapy and radiotherapy were 3% of the total person-time at risk. Treatment was received during 4 months since diagnosis.
|Number of Sarcomas||No RT, No CT||No RT, CT||RT, No CT||RT, CT||All|
|Men||38||288,100 (95,821)||19,200 (8,390)||64,700 (27,908)||8,700 (4,483)||380,700 (136,602)|
|Women||109||458,300 (73,907)||50,900 (12,940)||592,800 (64,127)||41,900 (8,136)||1,143,900 (159,110)|
|Age < 551||71||249,100 (28,419)||35,900 (7,286)||347,900 (31,315)||32,400 (5,944)||665,300 (72,964)|
|Age ≥ 551||76||497,300 (141,309)||34,200 (14,044)||309,600 (60,720)||18,300 (6,675)||859,300 (222,748)|
|Breast2||44||240,700 (33,672)||9,900 (2,447)||304,800 (35,057)||26,200 (4,673)||581,600 (75,849)|
|Cervix uteri3||18||24,500 (2,668)||400 (118)||105,700 (8,883)||2,700 (479)||133,300 (12,148)|
|Corpus uteri||24||53,800 (6,325)||1,800 (398)||122,100 (11,255)||2,500 (573)||180,200 (18,551)|
|Lung4||8||74,100 (43,911)||4,400 (4,891)||31,200 (22,551)||3,800 (3,482)||113,500 (74,835)|
|Lymphoma5||14||53,200 (12,997)||23,500 (6,104)||49,700 (6,391)||8,700 (1,930)||135,100 (27,422)|
|Ovary||10||47,800 (8,272)||26,000 (6,045)||25,200 (2,605)||5,400 (838)||104,400 (17,760)|
|Prostate||18||161,900 (42,420)||2,500 (555)||11,500 (3,276)||100 (60)||176,000 (46,311)|
|Rectum||11||90,300 (19,463)||1,700 (772)||7,300 (2,017)||1,300 (584)||100,600 (22,836)|
|Total||147||746,400 (169,728)||70,100 (21,330)||657,500 (92,035)||50,600 (12,619)||1,524,600 (295,712)|
For the total study group, 147 sarcomas were found vs. 88.5 expected. Soft tissue sarcomas comprised 86% of the sarcomas. The SIR of sarcoma tended to increase with time reaching 2.5 after 10–14 years (Figure 1). At 10 years of follow-up, the cumulative incidence of sarcoma was 0.03% (95% CI 0.017–0.037%) among patients who received radiotherapy and 0.02% (95% CI 0.017–0.032%) among patients who did not receive radiotherapy. The number needed to harm by radiotherapy was 45,000 patients. At that time the probability of soft tissue sarcoma was 0.02% (95% CI 0.012–0.031%) for radiotherapy without chemotherapy and 0.02% (95% CI 0.017–0.031%) for no radiotherapy or chemotherapy. At 10 years the cumulative incidence was highest among patients who received both treatments: 0.03% (95% CI 0.001–0.063%) for soft tissue sarcoma.
Among the patients who were not treated with chemotherapy or radiotherapy, 69 sarcomas were observed vs. 49.9 expected (SIR 1.4, 95% CI 1.1–1.8) (Table II). Furthermore, after 10 years of follow-up, a statistically significant excess risk of sarcoma was found in the subgroup of women (SIR 2.1, 95% CI 1.2–3.3) and the younger patients (SIR 2.9, 95% CI 1.6–4.9).
|Treatment||Entire follow-up||Follow-up ten or more years|
|O/E||SIR (95% CI)||O/E||SIR (95% CI)|
|Chemotherapy||0/1.1||0.0 (0.0–3.2)||0/0.1||0.0 (0.0–30.7)|
|Radiotherapy||7/4.1||1.7 (0.7–3.5)||4/0.8||5.1 (1.4–13.1)|
|Both||4/0.5||8.9 (2.4–22.8)||1/0.0||25.0 (0.6–139.3)|
|Neither||27/25.7||1.1 (0.7–1.5)||7/3.6||1.9 (0.8–4.0)|
|Total||38/31.4||1.2 (0.9–1.7)||12/4.6||2.6 (1.4–4.6)|
|Chemotherapy||4/2.3||1.8 (0.5–4.6)||3/0.5||6.1 (1.3–17.9)|
|Radiotherapy||61/29.1||2.1 (1.6–2.7)||40/13.0||3.1 (2.2–4.2)|
|Both||2/1.7||1.2 (0.2–4.4)||1/0.5||1.9 (0.1–10.3)|
|Neither||42/24.1||1.7 (1.3–2.4)||16/7.8||2.1 (1.2–3.3)|
|Total||109/57.1||1.9 (1.6–2.3)||60/21.9||2.7 (2.1–3.5)|
|Age < 55|
|Chemotherapy||1/1.1||1.0 (0.0–5.3)||0/0.4||0.0 (0.0–10.5)|
|Radiotherapy||45/13.2||3.4 (2.5–4.6)||34/8.2||4.2 (2.9–5.8)|
|Both||4/1.0||4.2 (1.1–10.7)||2/0.4||5.4 (0.7–19.5)|
|Neither||21/9.1||2.3 (1.4–3.5)||14 /4.8||2.9 (1.6–4.9)|
|Total||71/24.3||2.9 (2.3–3.7)||50/13.7||3.7 (2.7–4.8)|
|Age ≥ 55|
|Chemotherapy||3/2.3||1.3 (0.3–3.8)||3/0.3||11.5 (2.4–33.7)|
|Radiotherapy||23/19.9||1.2 (0.7–1.7)||10/5.7||1.8 (0.9–3.3)|
|Both||2/1.1||1.8 (0.2–6.3)||0/0.2||0.0 (0.0–17.6)|
|Neither||48/40.8||1.2 (0.9–1.6)||9/6.6||1.4 (0.6–2.6)|
|Total||76/64.2||1.2 (0.9–1.5)||22/12.8||1.7 (1.1–2.6)|
|Chemotherapy||4/3.4||1.2 (0.3–3.0)||3/0.6||4.9 (1.0–14.4)|
|Radiotherapy||68/33.2||2.1 (1.6–2.6)||44/13.8||3.2 (2.3–4.3)|
|Both||6/2.1||2.9 (1.1–6.2)||2/0.6||3.4 (0.4–12.5)|
|Neither||69/49.9||1.4 (1.1–1.8)||23/11.4||2.0 (1.3–3.0)|
|Total||147/88.5||1.7 (1.4–2.0)||72/26.4||2.7 (2.1–3.4)|
Among the patients who received chemotherapy without radiotherapy, 4 sarcomas were observed vs. 3.4 expected (Table II). A statistically significant excess risk of sarcoma after 10 years of follow-up was found among women (SIR 6.1, 95% CI 1.3–17.9) and among the older patients (SIR 11.5, 95% CI 2.4–33.7).
Among the patients who received radiotherapy without chemotherapy, more sarcomas were observed than expected, 68 observed vs. 33.2 expected (SIR 2.1, 95% CI 1.6–2.6) (Table II). After 10 years of follow-up, a statistically significant excess risk was observed among men (SIR 5.1, 95% CI 1.4–13.1), women (SIR 3.1, 95% CI 2.2–4.2), and the younger patients (SIR 4.2, 95% CI 2.9–5.8).
Among the patients who received both chemotherapy and radiotherapy, 6 sarcomas were observed vs. 2.1 expected (SIR 2.9, 95% CI 1.1–6.2) (Table II). However, the risks were not pronounced if the follow-up of 10 or more years was examined.
When examining separate sarcoma types of the entire follow-up time, more sarcomas of both bone and soft tissue were observed than expected among younger women who received radiotherapy alone or in combination with chemotherapy (SIR for bone sarcoma 3.0, 95% CI 1.2–6.1, and SIR for soft tissue sarcoma 3.1, 95% CI 2.2–4.4) (Table III). More soft tissue sarcomas were also observed among younger men, SIR for radiotherapy with or without chemotherapy 10.0, 95% CI 4.0–20.6. In addition, the subgroup of younger women who were not treated with radiotherapy developed more soft tissue sarcomas than expected, SIR 2.9, 95% CI 1.7–4.5.
|Age < 55|
|Age ≥ 55|
|Soft Tissue Sarcoma|
|Age < 55|
|Age ≥ 55|
Among the patients with cancer of the breast, cervix uteri, corpus uteri and lymphoma, statistically significant excess risks for the entire follow-up were found in the radiotherapy groups (Table IV). Also, highly significant excess risks of sarcoma were found among patients treated for cancer of corpus without chemotherapy or radiotherapy and among patients treated for lymphoma with combined chemotherapy and radiotherapy.
After controlling for age and sex, the estimated sarcoma risks were higher than 1 among the patients treated with radiotherapy alone or chemotherapy alone compared to that of the group of patients receiving neither treatment (Table V). The crude rate ratio (RR) for the radiotherapy group was 1.6 (95% CI 1.0–2.6) and for the chemotherapy group 2.4 (95% CI 0.7–8.1). After adjustment for age, sex and type of first cancer, the RR was 1.5 (95% CI 0.9–2.6) for the radiotherapy group and 2.2 (95% CI 0.7–7.5) for the chemotherapy group.
|Adjusted for age|
|Adjusted for age and sex|
|Adjusted for age, sex and type of first primary cancer|
No statistically significant difference was found in the effect of radiotherapy on bone and soft tissue sarcoma (p-value < 0.75). Furthermore, the effect of treatment did not vary significantly by age groups (p-value < 0.25).
The results of this study are consistent with the few previous studies published on sarcoma following radiotherapy. The excess risk of sarcoma after radiotherapy was comparable to or somewhat lower than the previously reported results. The excess risk was found among younger patients. This may suggest that there is an excess sarcoma risk in association with a particular cancer type that is more common among the young.
In a study of Hodgkin's disease survivors, the risk of solid tumours was increased after treatment by radiotherapy and chemotherapy.16 The results of this study indicate that chemotherapy is a risk factor for developing sarcoma particularly among women and older patients.
A high risk of developing bone cancer has been associated with exposure to high-dose radiotherapy.6 Patients treated with radiotherapy for retinoblastoma had a risk ratio of 1.9–10.7 compared with those not receiving radiotherapy.1 Excess risks of developing bone cancer have also been reported in relation to radiotherapy for breast cancer and Hodgkin's lymphoma.6 In a study of childhood radiotherapy, the risk of developing osteosarcoma was a linear function of local radiation dose.17
Excess risks of developing soft tissue sarcoma have been reported in relation to radiation treatment for cancers such as of breast, ovary, cervix, and Hodgkin's and non-Hodgkin lymphoma.7 For breast and ovary cancer, during a follow-up of over 10 years, the soft tissue cancer risk increased to 8–25-fold. After radiation, soft tissue cancers of several cell types have been found, and latency time has varied between 2 and 40 years.7
No previous study has addressed the latency time of developing a sarcoma as a second primary neoplasm. Our results suggest that the risk of sarcoma is increased after 10 years from irradiation. This is consistent with effects of radiation on the risk of other solid tumours.18 The minimal latency time for developing a solid tumour caused by radiation has been estimated as 10 years.11
In clinical decision making, absolute measures of risk provide most information. We conducted a cohort study that provides clinically most useful a readily applicable measures of risk, i.e. cumulative incidence that can be used for estimating the risk for a patient undergoing radiotherapy. This information is not available in case–control studies that provide only odds ratio as a measure of risk. Yet, this approach is limited in the depth of detail, for example, risk relative to doses to specific organs or tissue sites could not be estimated. The results are not derived from a randomised trial, and therefore, the patient groups may not be comparable. This may induce confounding by indication, i.e., patient or disease characteristics other than treatment may affect the results. Patient selection for radiotherapy may depend on the aim of the treatment, patient's age, tumour location, histology and stage as well as on other treatments.19 These variables were not controlled in this study, and more detailed case-control designs are needed.
Detailed characterisation of the treatment such as radiation dose to the site of origin of sarcoma is beyond the scope of this study. The suggestion that angiosarcoma may be a more prevalent sarcoma type after radiotherapy (for breast cancer) cannot be evaluated from these results.20 Also, malignant fibrous histiocytoma has been found to be a frequent type of postirradiation sarcoma.21, 22
We did not evaluate the concordance between the irradiated target volume and location of subsequent sarcoma. Including all sarcomas regardless of anatomic site is likely to cause differential misclassification biasing the results towards unity. This means that our results provide a credible lower limit for the true risk within the irradiated site. Nevertheless, our results do provide relevant information as they can be used to estimate the overall risk of sarcoma, which is of obvious clinical relevance.
Our study design was based on comparison between patients with similar first primary cancer receiving different forms of treatment. The Finnish cancer registration system is based on numerous data sources, and especially in the case of second primary cancers the coding principle is conservative: if there is even a slight chance that the new cancer might be a metastatic one, this is not coded as new primary. Therefore, since histological confirmation of sarcomas was required, the proportion of metastatic tumours was likely to be small. Also, adjustment for age, sex and type of first cancer did not affect the results of the regression analysis. Hence, we have eliminated the effect of factors causing the first primary cancer as a possible cause of the subsequent one. It is possible that shared etiologic factors, e.g. environmental or genetic, could affect the risk of subsequent sarcoma following a first primary cancer. However, it would not affect our analyses unless this association would occur for several types of cancer and furthermore these factors would need to be associated with type of treatment received.
A further strength of the study was the representative study population, consisting of practically all patients fulfilling the eligibility criteria in Finland in 1953–2000. Furthermore, we were able to achieve complete follow-up for death or emigration by using a record-linkage with the population registry. This minimises the possibility of selection bias and ensures the generalisability of the results.
Our results indicate an increased risk of sarcoma among patients treated for cancer previously, relative those not receiving radiotherapy or chemotherapy. Both radiotherapy and chemotherapy were associated with excess risk of sarcoma. The excess related to radiotherapy was attributable to patients below the age of 55 at the time of treatment. Further analyses focusing on the irradiated tissue and allowing assessment of the dose–response are needed.
- 1International Agency for Research on Cancer. Ionizing radiation, part 1: X- and gamma (γ) radiation and neutrons. IARC monographs on the evaluation of carcinogenic risks to humans, vol. 75. Lyon: IARC, 2000. 138–223.
- 2UNSCEAR. Report to the General Assembly with scientific annexes: Sources and effects of ionizing radiation. New York: UNSCEAR, 2000. 400.
- 3UNSCEAR. Epidemiological evaluation of radiation-induced cancer. New York: UNSCEAR, 1999. 40–1.
- 4Finnish Cancer Registry. Cancer incidence in Finland 1998 and 1999: Cancer statistics of the National Research and Development Centre for Welfare and Health, Cancer Society of Finland publication 63. Helsinki: Cancer Society of Finland, 2002. 8–9.
- 5Finnish Cancer Registry. Cancer incidence in Finland 2000 and 2001: Cancer statistics of the National Research and Development Centre for Welfare and Health, Cancer Society of Finland publication 65. Helsinki: Cancer Society of Finland, 2003. 8–9.
- 6WHO Classification of Tumours Pathology and Genetics. Tumours of soft tissue and bone. Lyon: IARC Press, 2002. 9–367.
- 7Bone cancer. In: Schottenfeld D, Fraumeni JF, eds. Cancer epidemiology and prevention, 2nd ed. New York: Oxford University Press, 1996. 971–83., , .
- 8Soft tissue sarcomas. In: Schottenfeld D, Fraumeni JF, eds. Cancer epidemiology and prevention, 2nd ed. New York: Oxford University Press, 1996. 984–99., ,
- 11Statistical methods in cancer research: The design and analysis of cohort studies, vol 2. Lyon: IARC, 1989. 48–118., .
- 12Epidemiology: An introduction. New York: Oxford University Press, 2002. 24–8..
- 14Ionizing radiation. In: Schottenfeld D, Fraumeni JF, eds. Cancer epidemiology and prevention, 2nd ed. New York: Oxford University Press, 1996. 319–54., , .
- 15Statistical models in epidemiology. New York: Oxford University Press, 1993. 227–48., .
- 19General principles of radiotherapy. In: Pointon RCS, ed. The radiotherapy of malignant disease. Berlin: Springer-Verlag, 1991. 111–29..
- 21Sarcoma after radiation therapy: Retrospective multiinstitutional study of 80 histologically confirmed cases, For the Radiation Therapist and Pathologist Groups of the Fédération Nationale des Centres de Lutte Contre le Cancer. Radiology 2000; 216: 197–205., , , , , , , , , , , , et al.