Analysis of Chip Mutations Pre-and Post-Transplant in Multiple Myeloma (MM): Expanded Results from a Prospective Longitudinal Study

Background:
Despite the clinical significance of Clonal Hematopoiesis of Indeterminate Potential (CHIP) mutations in MM, there remains limited longitudinal research detailing how these mutations evolve under different therapeutic pressures. To address this gap, we are conducting a prospective study of CHIP in IMID-naïve patients (pts) from pre-transplant and at several time points post-transplant, encompassing the introduction of lenalidomide (LEN) maintenance.
Here, we expand our initial report of 66 pts (Khan, ASH 2023) with mutation testing performed pre-transplant in146 patients and 3 months (mos) post-transplant in 81 pts. Additionally, in a separate analysis we align our mutation calls with a high-fidelity list of candidate CHIP variants recently described by Vlasschaert et al.(Blood 2023), derived from data from 550,000 individuals in the UK Biobank and the All of Us Research Program.

Methods:
ARCH 001 trial is a prospective, longitudinal study evaluating evolution of CHIP in a transplant-eligible MM population testing at time points pertinent to therapy (pre-transplant using non-IMID containing induction, post-transplant at 3 mos before LEN maintenance, 1 year and 2 years on LEN maintenance). Using a minimum sequencing depth of 4000x, mutation calls produced by SmMIP-tools were subjected to various filters to reduce the likelihood of false positives (Medeiros et al. Bioinformatics, 2022). These filters include a minimum number of single-stranded consensus reads and the requirement that mutations be present in sequencing replicates. Moreover, variant allele frequency (VAF) thresholds of 1-30% were used for somatic variant calls (referred to hereon as M1 filters), with <1% allele frequency permitted for mutations detected above threshold in subsequent samples. In addition to the M1 filters, mutations were restricted to candidate drivers described by Vlasschaert et al. in a separate analysis (referred to hereon as M2 filters).

Results:
The total cohort is typical for a transplant-eligible population with median age at diagnosis 64 years (range 33-73), male predominance 60%, IgG subtype 63%, high-risk FISH cytogenetics 36% (35/98). All patients received the non-IMiD containing induction regimen CyBorD. Using M1 filters, a total of 65 CHIP mutations were identified pre-transplant in 52/146 pts (36%) with most common mutated genes: DNMT3a (48%), TET2 (17%), and PPM1D (11%). Nine patients (6%) had ≥1 mutation. Mean allele frequency was 6.8% (range 0.3-34.2%). At the 3-mo post-transplant time point, 41 mutations were detected in 29/81 pts (35%). 10% pts carried ≥1 mutation. The mean allele frequency was 4%. DNMT3a remained the most frequent mutation (61%), followed by TET2 (19%) and PPM1D (10%).
Of the 81 pts with pre-and 3 mos post-transplant samples, a total of 54 unique mutations were identified in 39 pts, of which 26 (48%) were shared across the 2 time points. Testing from additional 1 and 2 year post-transplant time points is ongoing.
Using the additional M2 filter, the mutation incidence pre-transplant and post-transplant were 39/146 (27%) and 25/81 (31%), respectively, with missense mutations in DNMT3A, ASXL1 and PPM1D being the most frequently filtered mutations.

Conclusion:
Our results using deep SmMIP sequencing and M1 filters reveal a mutation prevalence of 36% in IMiD-naive MM pts pre-transplant, and 35% in the early post-transplant period. Despite changes in individual mutations across time points, both mutation prevalence and affected genes, as well as the VAFs of shared mutations, remain consistent from pre- to post-transplant.
Although our mutation prevalence rates may appear higher than those typically reported, mutation rates vary in the literature, with some MM studies reporting similar rates. It is likely differences in assay sensitivity, variations in filtering approaches, selection of targeted genes, and deliberate use of low VAF threshold may impact results.
While mutation rates are reduced by a more stringent filtering, pattern of frequent mutations, as well as proportion of mutations that are shared across pre and post-transplant time points remains similar. Given the paucity of data exploring the evolution of CHIP mutations in MM over extended time and treatments, we feel that using a less stringent filtering strategy may identify small but clinically significant mutations important for longitudinal tracking.