Clonal Hematopoiesis in Patients Receiving Immune Checkpoint Inhibitor Therapy

Background. Clonal hematopoiesis (CH) is an aging associated phenomenon with potential to transform to overt myeloid disease at a rate of 1% per year. CH is hypothesized to represent a precursor state for leukemogenesis. CH is detected in 10% of healthy, elderly (>70 years) individuals at variant allele frequencies (VAFs) above 2%, but prevalence is higher in patients (pts) with malignancies (~25%). We recently demonstrated in a large population database study that recent changes in therapy for melanoma and non-small lung cancer (NSCLC) are associated with a decreased incidence of therapy-related myeloid neoplasms (tMNs).¹ This trend may reflect declining utilization of chemo-radiotherapies or even regression of CH clones in the context of immune checkpoint inhibitor therapy (ICI). To improve our understanding of the evolving clonal architecture in patients with CH receiving ICI we analyzed blood samples from pts with melanoma and NSCLC at baseline and after exposure to ICIs. We aimed to characterize (i) baseline prevalence of CH in pts with melanoma and NSCLC (ii) alterations in clonal architecture associated with ICI treatment in serial blood samples from these pts.

Methods. In this retrospective analysis, stored blood samples from pts with melanoma (N=32) and NSCLC (N= 109) treated with ICIs were analyzed. To detect CH, genomic DNA collected before treatment initiation and serially on ICI were analyzed using a custom panel targeting 93 genes. A 1% VAF cutoff was used to define CH. Relationships among clinical, laboratory and mutational variables were examined using chi-square test, at a significance level of 0.05.

Results. We present preliminary results on 25 pts with melanoma (Table 1). Median age was 61 years. 52% (13/25) were men. One/25 pts had previously received chemotherapy and 4/25 had prior radiation therapy exposure. 64% (16/25) had stage IV disease and 36% (9/25) had stage III disease. 20% (5/25) had a history of autoimmune disease. CH was present in 32% of cohort, consistent with prior reports in solid tumor pts. Mutations in DNMT3A (36%) were most common, followed by TET2 (18%). Mutations in PPM1D, ARID1A, SF3B1, LPA and TGFBR2 accounted for 46% of the mutations (Figure 1). The mutational pattern was similar to prior reports in other cancer cohorts. DNMT3A mutations were most common in this cohort, with only one patient harboring a mutation in the R882 AML/MDS hotspot, previously described in CH. Most common mutation type was single nucleotide variants (55%), followed by frameshift (27%), and splice site mutations and truncations. The mean VAF for somatic mutations was 9%, much higher than the usual VAF cutoff for CH. Mutations were more common in smokers (40% vs 26.7%), and in individuals older than 60 years (37.5% vs 22%). Pts with autoimmune disease did not demonstrate CH mutations (0/5), whereas 40% (8/20) of pts without autoimmune diseases harbored mutations, (p=0.08).

Conclusions. Our findings demonstrate a high baseline prevalence of putative CH mutations and high mean VAFs in a cohort of melanoma patients, even in the setting of minimal prior chemotherapy and radiation exposure. Despite this high CH prevalence, risk of tMN development after melanoma is declining with the use of newer therapies, suggesting a possible immune clearance or halt in progression of clonal hematopoiesis with newer therapies (ICI). Lack of CH mutations in individuals with autoimmune disease is an interesting finding and warrants further investigation. Ongoing serial sample analysis of the melanoma cohort after exposure to ICI will provide insight into the impact of ICI on the underlying clonal architecture. These findings are also being tested in the NSCLC validation cohort.

References

1. Singh A, Mrad C, Faber MG, et al. Evolving risk of therapy-related myelodysplastic syndromes and acute myeloid leukemia (tMDS/AML) following modern cancer therapies. Journal of Clinical Oncology. 2020;38(15_suppl):7516-7516.

Acknowledgement: Data and samples for this study were provided by the Data Bank and BioRepository (DBBR), which is funded by the National Cancer Institute (P30 CA016056) and is a Roswell Park Cancer Institute Cancer Center Support Grant shared resource.