Incidence and Risk of Therapy-Related Myeloid Neoplasms in Solid Cancer: A Danish National Population-Based Matched Cohort Study

Introduction

Therapy-related myeloid neoplasms (tMN) is a detrimental late-effect to cancer treatment. With advances in treatments of solid cancers (SC) resulting in more SC survivors, the concern is increasingly relevant when planning treatment and opting for best possible long-term outcomes.

The present study examines risk of MN among SC patients and matched cancer-free individuals by varying cancer treatment.

Materials and Methods

We conducted a Danish nationwide, population-based matched cohort study, combining health and quality registries. All Danish inhabitants are assigned a unique identification number enabling linkage of data with complete follow-up. We included patients ≥18 years with a first diagnosis of SC (excluding non-melanoma skin cancer), between 2004 and 2016. Each patient was year-age-sex-matched (1:10) to cancer-free comparators from the general population. We used date of SC diagnosis as index date for patients and their matched comparators. The primary outcome was a subsequent diagnosis of a myeloid neoplasm (MN; AML, MDS or MPN). Patients and comparators were followed until MN, emigration, death, or end-of-follow-up (July 2018).

We quantified the risk of tMN/MN by cancer treatment using the following categories 1) no cytotoxic therapy (e.g. surgery), 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

Cox regression were used to estimate risk of tMN/MN for 19 cancer groups (e.g. lung and breast cancer) controlling for matching factors per design and adjusting for Charlson Comorbidity Index. We estimated 10-year cumulative incidence proportions of tMN/MN for each cancer.

Results
We identified 1,126 incident MN/tMN cases among 311,215 SC patients and 8,343 incident MN cases among 3,100,946 matched individuals. Median age at diagnosis ranged from 40 years in testicular cancer to 72 years in urological cancers. Half of SC patients were treated without cytotoxic therapy, e.g. surgery or antihormone therapy.

The overall 10-year cumulative incidence of MN/tMN among patients ranged from 0.1 to 3.1% with highest risks in patients treated with radio-chemotherapy. Among comparators, the risk ranged from 0.1-0.5%.

Overall, the risk of MN/tMN was higher across all treatment groups in patients with SC compared with matched individuals: No cytotoxic treatment hazard ratio (HR) 1.5 (95%CI 1.4-1.7), radiotherapy 1.6 (95%CI 1.4-1.9), chemotherapy 5.3 (95%CI 4.7-6.0), and radio-chemotherapy 5.8 (5.0-6.7).

Across SC subgroups, the highest risk of tMN was seen in patients treated with radio-chemotherapy but varied from HR 1.5 (95%CI 0.8-2.7) in colorectal to 4.2 (95%CI 3.0-5.8) in breast, to 7.5 (95%CI 3.9-14.4) in prostate, and 9.6 (95%CI 4.3-21.4) in female genital cancers.

The high risk of MN in surgically treated SC patients compared with matched cancer-free comparators was most pronounced in head/neck HR 2.8 (95%CI 1.4-5.6), lung 2.0 (95%CI 1.3-3.2), and renal 1.8 (95%CI 1.2-2.8) cancers.

The risk of tMN in radiotherapy-treated patients was comparable to the risk of MN in SC patients unexposed to cytotoxic treatment (e.g. breast cancer HR 1.3 (95%CI 1.0-1.9) vs. 1.4 (95%CI 1.0-1.9)).

We observed higher risks of tMN in younger SC patients exposed to cytotoxic therapy compared with older individuals: Overall radio-chemotherapy ≤ 60 years HR 10.5 (8.1-13.4) vs. >60 years 4.1 (3.3-4.9).

Results remained stable in sensitivity analyses.

Discussion

The risk of MN/tMN is much higher in patients with a history of SC compared with matched individuals especially among those treated with combination-cytotoxic therapy. Even SC patients treated with surgery alone have higher risks than cancer-free matched individuals comparable to that of radio-therapy-treated patients. This is especially evident in lifestyle-related cancers indicating shared risk factors, e.g., tobacco or alcohol. This may point to shared genetic predispositions and higher prevalence of clonal haematopoiesis among SC patients.

Our findings emphasize that though incidence of tMN is low, it is a considerable late effect of SC treatment and that less toxic options are warranted. Importantly, our study reveals that even SC patients not undergoing cytotoxic treatment, face a higher a risk of MN compared with cancer-free individuals mandating a broad scope in follow-up and advice of cancer patients.