Incidence and Risk of Therapy-Related Myeloid Neoplasms in Non-Myeloid Hematological Malignancies: A National Population-Based Matched Cohort Study

Introduction

Therapy-related myeloid neoplasm (tMN) is a serious late-effect following cancer treatment affecting long-term survival. However, risk factors for tMN after non-myeloid hematological malignancies (nMHM) are not well understood. The present study examined the risk of myeloid neoplasms (MN) among patients diagnosed with nMHM and treated with different types of cancer treatment.

Materials and Methods

We conducted a nationwide, population-based matched cohort study combining Danish health and quality registries. The Danish health care system is tax-funded and all inhabitants are assigned a unique civil registration number enabling linkage of all administrative data enabling complete follow-up. We included patients ≥18 years with a first-time diagnosis of nMHM, between January 2004 and December 2016. Each patient was year-age-sex-matched in a 1:10 ratio to cancer-free individuals from the general population. We used date of nMHM diagnosis as index date for patients and their matched comparators. Primary outcomes were incident MN (AML, MDS, and CMML) between January 2004 to July 2018. All patients and matched comparators were followed until MN/tMN, emigration, death, or end-of-follow-up (July 2018).

We quantified the risk of MN by cancer treatment into the following categories 1) no cytotoxic therapy (e.g. watch and wait or surgery only), 2) radiotherapy, 3) chemotherapy, and 4) radio-chemotherapy. Treatment category could only change from a less to a more intensive treated category during follow-up. We used Cox regression analyses to estimate risk of MN/tMN, controlling for matching factors per design. We further estimated 10-year cumulative incidence proportions for MN/tMN. Results were given by disease (non-Hodgkin (NHL) and Hodgkin lymphomas (HL), multiple myeloma (MM) and lymphatic leukemia), crude and adjusted for Charlson Comorbidity Index, age, sex, and calendar year. We performed sensitivity analyses to assess robustness of results.

Results

We identified 490 MN/tMN cases among 44,478 nMHM patients and 510 MN cases among 443,178 comparators. Median age at first diagnosis of MN varied from 45 years in HL to 70 years in MM patients. Most cases of MN/tMN among nHMN patients were preceded by chemotherapy or combined radio-chemotherapy.

The overall 10-year cumulative incidence of MN/tMN amongst patients ranged from 0.76 to 5.19% with highest risks in patients treated with cytotoxic therapy. Amongst matched comparators, the risk ranged from 0.12-0.37%.

The risk of MN/tMN was elevated across treatment groups in patients with hematological malignancies compared to matched controls. For NHL, HRs were 7.3 (CI=4.5-11.6) for the no-cytotoxic therapy group vs radiotherapy group 5.1 (CI=2.1-12.7), chemotherapy-alone group 9.2 (CI=6.9-12.3), and radio-chemotherapy group HR 12.6 (CI=8.5-18.4).

In MM, HRs for the no-cytotoxic therapy group were 11.05 (CI=5.6-21.7) vs chemotherapy group 18.42 (CI=10.5-32.2), and radio-chemotherapy group 14.7 (CI=6.1-35.3).

For lymphatic leukemia, HRs for the no-cytotoxic therapy group were 3.4 (CI=2.4-4.9) vs 33.2 (CI=20.2-54.5) in chemotherapy-only patients.

We observed a trend towards higher risks in younger individuals (≤ 60 years) exposed to cytotoxic therapy compared to older individuals (>60 years): NHL+chemotherapy, HR 14.9 (8.3-26.9) vs 7.80 (5.6-10.9) and lymphatic leukemia+chemotherapy: HR 123.2 (CI=27.6-549.3) vs 23.6 (CI=13.6-40.9). We were unable to report effect estimates in age-stratified analyses for patients with HL and MM owing to patient confidentiality.

Results remained consistent and unchanged in sensitivity analyses.

Discussion

The risk of MN/tMN is higher in patients with a history of nMHM compared to the risk of MNs in matched individuals. The risk across subgroups was highest in patients treated with chemotherapy and radio-chemotherapy. Still, nMHM patients not exposed to DNA-altering treatments had higher risks than cancer-free comparators.

These results indicate that development of tMN caused by cancer treatment in patients with non-myeloid hematological disease is augmented by shared environmental, genetic, or other acquired risk factors, including features in bone marrow microenvironment. Our findings emphasize that therapy related myeloid malignancies are still a considerable late effect of treatment when treating other hematological malignancies.