Risk of Second Malignant Neoplasms After Childhood Leukemia and Lymphoma: An International Study

Milena Maule; Ghislaine Scélo; Guido Pastore; Paul Brennan; Kari Hemminki; Elizabeth Tracey; Risto Sankila; Elisabete Weiderpass; Jorgen H. Olsen; Mary L. McBride; David H. Brewster; Vera Pompe-Kirn; Erich V. Kliewer; Kee Seng Chia; Jon M. Tonita; Carmen Martos; Jon G. Jonasson; Franco Merletti; Paolo Boffetta


J Natl Cancer Inst. 2007;99(10):790-800. 

In This Article


In this study of second cancers after childhood leukemia and lymphoma, we found that survivors had an increased risk of developing a second malignant neoplasm compared with the general population. The highest risk was found in Hodgkin lymphoma survivors, confirming previous findings from population-based and clinical studies.[12,13,21,22,23,24,25,26,27,28,29,30,31,32,33] The most frequent second malignant neoplasms were, in descending order, brain cancer, non-Hodgkin lymphoma, and thyroid cancer after leukemia; thyroid cancer, female breast cancer, and non-melanoma skin cancer after Hodgkin lymphoma; and thyroid cancer and brain cancer after non-Hodgkin lymphoma.

Large cohorts of childhood cancer survivors have been studied to estimate the risk of developing a second malignant neoplasm and to disentangle the role of various risk factors (i.e., factors related to the host, initial cancer type, and cancer therapies).[21] The main risk factors for second malignant neoplasms in survivors of childhood cancer are radiotherapy and chemotherapy.[34] At least 80% of second malignant neoplasms in childhood cancer survivors develop in the radiotherapy field, with a median latency of at least 10 years and no risk plateau.[35] Different tissues are not equally sensitive to radiation, and host factors such as repair enzymes, rate of cell division, endocrine, and immune function may be important cofactors. For a given radiation dose, orthovoltage (voltage in the range of 140-400 kV) in use until the middle of the 1970s and the 1980s was more oncogenic than megavoltage (voltage > 1 MV).[35,36] Other possible risk factors include rare genetic/host predisposition to developing multiple cancers and underlying factors common to different diseases or treatment, such as immunosuppression due to infection with human immunodeficiency virus or bone marrow transplantation.[36] Environmental and behavioral shared risk factors are less likely to have played a role in the development of a second malignant neoplasm after childhood hematolymphopoietic neoplasms, given the short latency of childhood cancer.

Individual treatment information was not available from the registries contributing data to our study, and hence no causal inference on the effects of therapies can be drawn. Nevertheless, knowledge of temporal changes in treatment practice in pediatric hematology may help the interpretation of our results, if it is assumed that intercountry variations in treatment procedures were small among the high-income countries of the 13 cancer registries included in our analysis. Similar trends in survival after childhood cancer in European countries and the United States confirm a general uniformity of procedures and treatments.[37,38]

Among second malignant neoplasms that may be related to radiotherapy are brain, thyroid, breast, and nonmelanoma skin cancers.[36] Brain cancers are the most common second malignant neoplasms after lymphoid leukemia, with a median latency of 7-11 years.[25,39,40] A causal link between brain cancer and cranial irradiation has been suggested.[41] In the 1970s and 1980s, cranial and spinal radiotherapy was used with doses in the range of 18-24 Gy as prophylaxis to central nervous system involvement.[42,43] In addition, brain cancers have been reported in children who had lymphoid leukemia and were not treated with radiotherapy.[44] Thyroid cancer induced by radiation is frequently observed in long-term survivors of childhood cancer, although it is not a leading cause of death among them. Those treated for childhood cancer before age 10 years have the highest risk.[45] The frequency of thyroid cancer increases linearly with radiation doses up to 30 Gy and decreases at higher doses.[46] In the 1970s and 1980s, children treated for lymphoid leukemia received small quantities of radiation at the thyroid because of the unclear borders of the cranial irradiation field,[42,43] whereas children with lymphomas were often treated with supradiaphragmatic radiotherapy (including mediastinal, sovraclavear, and laterocervical nodes).[47] Patients treated with radiotherapy also have increased risk of developing nonmelanoma skin neoplasms in the treatment field.[48] Second malignant neoplasms are a leading cause of death among long-term survivors of Hodgkin lymphoma, and breast cancer is the most frequent second malignant neoplasm among female survivors.[32] A previous study found that the risk starts to increase 5-9 years after completion of radiotherapy, is highest among girls treated at an age of 10 years or older, and is lower for women diagnosed when they were more than 30 years of age.[49] The effect of radiation is likely to be stronger during puberty, when breast cells proliferate rapidly due to the hormonal stimulation, resulting in an increased risk of developing breast cancer. The main concern for Hodgkin lymphoma survivors is thus the risk of subsequent breast cancer among adolescents and young women treated with supradiaphragmatic radiotherapy.[47] Recently, it was estimated that cumulative absolute risks of breast cancer after Hodgkin lymphoma range from 0% to 39.6% (based on 3817 women diagnosed up to 30 years of age), depending on age at Hodgkin lymphoma diagnosis, duration of follow-up, and type of treatment. For patients diagnosed by 15 years of age who received a mediastinal dose of at least 40 Gy and no alkylating agents, the cumulative projected risk of breast cancer at 30 years of follow-up was 10.3% (95% CI = 6.8% to 15.2%).[50] Women treated with radiotherapy or alkylating agents and who had early menopause (i.e., before age 40 years) had a lower risk of breast cancer than those who maintained normal ovarian function.[51]

In our study, all six breast cancers occurred 10-29 years after a Hodgkin lymphoma diagnosis among women who were 10-14 years of age at the time of the first diagnosis and who had originally been diagnosed before 1980 (when radiotherapy was used intensively). This pattern is consistent with the radiosensitivity of the breast at ages near puberty. The estimated cumulative incidence of any second malignant neoplasm among Hodgkin lymphoma survivors was 12.7% (95% CI = 8.29% to 17.2%) and 31.9% (95% CI = 20.1% to 43.7%) at 30 and 45 years of follow-up, respectively. These findings highlight the importance of careful follow-up of long-term Hodgkin lymphoma survivors. However, therapies that may be related to the high estimated risk of second malignant neoplasms in these women were in use two to three decades ago and do not reflect the optimal treatment protocols in use currently, which, by using lower doses and smaller radiotherapy fields, are expected to decrease the risk of late effects.[52,53] Radiation therapy may also be related to risk of head and neck cancers. In a recent report,[54] elevated risk of head and neck neoplasms (SIR = 20.9, 95% CI = 10.5 to 41.8, based on eight cases) was found in 4834 patients who survived childhood leukemia. In the group of patients for whom radiation therapy data were available, this treatment was associated with head and neck carcinomas.

With regard to chemotherapy, secondary acute myeloid leukemia is the most frequent hematologic second malignant neoplasm among Hodgkin lymphoma survivors.[55] In our study, we found an increased risk for myeloid leukemia among Hodgkin lymphoma survivors (SIR = 33.9, 95% CI = 9.24 to 86.8, based on four cases). MOPP (mechloretamine, vincristine, prednisone, and procarbazine), the standard chemotherapy for Hodgkin lymphoma in the 1970s and the early 1980s, has been linked to increased risk of secondary leukemias, with a plateau 10-14 years after the end of the therapy.[23,24] In the early 1980s, treatment procedures shifted from intense to less aggressive radiotherapy and the introduction of less leukemogenic chemotherapy (such as ABVD [adriamycin, bleomycin, vinblastine, and dacarbazine]).[36] Recently, an elevated risk of bladder cancer (SIR = 10.6, 95% CI = 2.7 to 42.3, based on two cases) was found in a cohort of 4834 childhood leukemia survivors.[54] An increased risk of bladder cancer in adults has been associated with cyclophosphamide, which is often used to treat non-Hodgkin lymphomas[56,57] and, more rarely, childhood leukemia.[58]

In our study, we found that childhood cancer survivors had increased risk for second malignant neoplasms that have been associated with radiotherapy (cancers of the brain, thyroid, skin [nonmelanoma], breast, and head and neck [the tongue and salivary glands in particular]) and chemotherapy (acute nonlymphocytic leukemia and bladder cancer). Although we cannot draw conclusions on the causative role of treatments on second malignancies, the risk pattern that emerges from our study is generally consistent with the late effects of therapies for childhood cancer in use during the study period.

This study allowed a precise and up-to-date estimation of cumulative incidence of second malignant neoplasms after childhood hematolymphopoietic neoplasms. The cumulative incidence of any malignancies for survivors ranged from 2.43% (95% CI = 1.09 to 3.78) 30 years after leukemia to 31.9% (95% CI = 20.1 to 43.7) 45 years after Hodgkin lymphoma. Compared with the general population, survivors of childhood Hodgkin lymphoma had an 11-fold increased risk of having any malignancy in 30 years since the first neoplasm diagnosis. Cumulative incidence of second malignant neoplasms was higher for children who had a leukemia or lymphoma diagnosed after 1980. This phenomenon may be related to the more intensive and aggressive chemotherapies that have been used in more recent years as opposed to less toxic although less effective therapies before 1980.

The data analyzed in this study came from 13 well-established cancer registries and form a large dataset for a population-based study of second malignant neoplasms in childhood cancer survivors. Nevertheless, several possible limitations have to be considered in interpreting our results, including the small number of second malignant neoplasms, the possibility of chance associations due to multiple comparisons, and misclassification due to the unspecialized coding used to classify childhood cancer. However, for the first primary malignancies considered in this work, conversion between ICD-9 and childhood cancer classification is straightforward, and, because second malignant neoplasms may occur several years after the first neoplasm, adult cancer classification may be the appropriate one to use for many of the second malignant neoplasms identified.

In addition, the analysis of lymphoid leukemia data was not possible in our study. Reliable diagnoses of lymphoid leukemia are available starting from the 1970s,[59] whereas some of the registries included cases diagnosed as early as 1943. The number of lymphoid leukemias, which generally account for approximately 80% of all leukemias in children, was therefore likely to be underestimated. The analysis of the lymphoid leukemia subgroup, however, yielded risk estimates of second malignant neoplasms that were essentially identical to those for all leukemias but less precise because of the smaller number of events (data not shown).

Sometimes more than one second malignant neoplasm occurs in childhood cancer survivors. In this study, underestimation of the total number of malignancies following a primary neoplasm should therefore be expected because the study design implied that subjects exited from the cohort after the occurrence of one second malignant neoplasm.

The major strength of this population-based study is the size of the cohort, which has provided a unique opportunity to evaluate both the relative and the absolute risks of second malignant neoplasms for long-term survivors. The geographical heterogeneity of the registries contributing to this study and the high quality of their data make the described pattern of second malignant neoplasm risk for childhood leukemia and lymphoma survivors valid for many areas in the world. Extensive efforts are still needed to collect information on the long-term risk of second cancers in the increasingly large and aging population of childhood cancer survivors. A thorough understanding of the epidemiology of second cancers is essential for developing preventive strategies[60] targeted at individuals who survived childhood cancer.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: