Short-term Gain But Long-term Pain? Clonal Hematopoiesis and the Risk of Therapy-related Myeloid Neoplasms After Treatment for Solid Tumors

Therapy-related myeloid neoplasms (tMNs), including acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS), are second primary malignancies that may result after treatment for the original cancer. These disorders are of clinical importance for patients and of growing academic interest, especially as patients live longer owing to more effective therapies. TMNs were traditionally thought to develop from the mutagenic effects of cancer therapy. However, recent studies show that tMN-initiating mutations can predate cancer therapy.^1,2  Both patient-specific and treatment-specific factors likely impact the risk of t-MN development after diagnosis and treatment of another disorder. One of these patient-specific features could be clonal hematopoiesis (CH).³  Although CH can be an incidental finding in patients with prior solid malignancies, its presence is more common than in matched controls and confers a higher risk for the development of tMN.^4,5  Whether or not this should influence paradigms at the time of diagnosis of the primary malignancy is an open question.

Dr. Kelly Bolton and colleagues took a systematic approach to characterize the relationships between CH and environmental exposures, namely smoking and treatment for a primary malignancy. This group used targeted sequencing to study samples from nearly 25,000 patients of various ages with a primary solid tumor. The study population included patients with nonhematologic cancers who underwent matched tumor and blood sequencing. The authors extracted data on ancestry, smoking, date of birth, and cancer history including therapeutic exposure data through their cancer registry and clinical notes. The statistical models were robust with rational adjustments. Briefly, multivariable logistic regression was used to evaluate for an association between CH and therapy, age, sex, and smoking history. Additionally, for the AML/MDS risk models, there were univariate and multivariate analyses and cause-specific Cox proportional hazards regression. An R package aided in the model for absolute risk of AML/MDS in women with breast cancer.

Ultimately, the authors identified more than 11,000 unique mutations (median variant allele frequency [VAF] of 5%) consistent with CH in nearly 30 percent of the cohort. The most frequently identified mutations were DNMT3a, TET2, and ASXL1, in keeping with previous reports. A strength of this manuscript is the assessment of patient age, the role of cancer therapy, and personal environmental factors in driving selection of CH clones. The authors extracted and curated detailed clinical data (including smoking history and previous cancer treatments) for more than 10,000 of the patients who had received their cancer care in a single center. Fittingly, older age strongly correlated with the presence of CH in those patients with an odds ratio of nearly 2. Sixty percent of the cohort received treatment prior to blood draw for mutational analysis for this study, whereas the remaining were treatment naïve.

Patients who smoked or had received previous cancer treatment were more likely to have CH compared with treatment-naïve patients. The mutational spectrum of CH was similar across cancer types, except for DNA damage response gene mutations (TP53, PPM1D, CHEK2) being more frequent in patients with gynecologic malignancies. Mutations in ASXL1 were enriched in smokers. Treatment for malignancy with radiation, platinum, and topoisomerase II inhibitors preferentially selected for mutations in TP53, PPM1D, and CHEK2. Targeted therapies and immunotherapeutic agent exposure were not significantly associated with CH. Sequential sampling provided definitive evidence of both positive and negative changes in clone size across treatment modalities. Among mutations detected pre- and post-treatment, the majority (62%) of CH mutations remained stable, and 28 percent had evidence of growth while 10 percent decreased in clonal size. The group analyzed 35 cases for which paired samples were available at the time of molecular profiling for primary cancer and at the time of leukemic transformation for tMN. Among cases in which CH was previously detected, the CH mutation was present at tMN diagnosis; however, in 91 percent of cases, transformation was associated with acquisition of additional somatic mutations and TP53 was common. This suggests the baseline CH mutation alone was insufficient to cause tMN without the treatment for the primary malignancy.

Lastly, the authors created a cohort of 9,437 patients (from their center and the literature^4,6,7 ) with treated primary malignancies, of whom 75 developed tMN. Cause-specific Cox proportional hazards analysis showed that CH present at a VAF greater than 2 percent was associated with an increased tMN (hazard ratio, 6.9). They went on to model some potentially clinically applicable scenarios in breast cancer for discussion based on age, CH, and blood counts. Ultimately, most patients with breast cancer have a low 10-year absolute risk for tMN such that deferment of adjuvant chemotherapy would not affect their absolute tMN risk. However, for women at the highest risk in the hypothesized model, adjuvant chemotherapy increased the absolute risk of tMN by approximately 9 percent, exceeding the predicted absolute benefit in overall survival of chemotherapy in many women with early-stage breast cancer.