Prognostic impact of CXCR4 mutation type on clinical features and outcomes of patients with Waldenström macroglobulinemia

Abstract

Abstract Introduction:Waldenström macroglobulinemia (WM), the most common subtype of lymphoplasmacytic lymphoma (LPL), is characterized by the secretion of IgM monoclonal protein and the presence of somatic mutations in MYD88 (MYD88mut) and CXCR4 (CXCR4mut) in 90% and 40% of patients (pts), respectively. Testing for these gene mutations is part of routine clinical workup and helps guide management since pts with MYD88mut/CXCR4wild-type (wt) WM have better response rates to BTK inhibitors. A previous analysis (Castillo et al., BJH 2019) reported differing responses to ibrutinib based on CXCR4 mutation subtype (nonsense (ns) vs. frameshift (fs)). However, the relative impact of different CXCR4 mutation subtypes, such as nonsense mutations (CXCR4ns) and insertions/deletions (indels), including frameshift mutations, on disease features and clinical outcomes during different stages of WM is not well characterized. We conducted a single-center retrospective analysis to investigate these knowledge gaps. Methods: We included all pts at our center with a diagnosis of WM or IgM monoclonal gammopathy of undetermined significance (MGUS) who had next-generation sequencing (NGS)-based CXCR4 mutation testing of bone marrow specimens. The multi-gene NGS assay used for this study had greater than 250x coverage of the coding region of CXCR4 and provided mutation status of other genes including MYD88 and TP53. For pts having more than one NGS panel, the earliest result was used. Overall survival (OS) was calculated from first NGS test date until the date of death or last follow-up; survival times were compared using log-rank tests. Results: Between 8/2015 and 5/2025, we identified 268 pts, of whom 158 (59%) were male, and 246 (92%) self-reported as White or Caucasian; median age at NGS testing was 69 years (range 35-91 years). Within this cohort, 185 pts (69%) had symptomatic WM, 79 (30%) had smoldering WM, and four pts (1.5%) had IgM MGUS. CXCR4 mutations were found in 38%, 27% and 0% of pts with symptomatic WM, smoldering WM, and IgM MGUS, respectively. Among those with mutations, CXCR4mut variant allele frequencies (VAFs) were in the range of <5%, 5-10%, and >10% in 43%, 19% and 38% of symptomatic WM pts vs. 52%, 29% and 19% of smoldering WM pts, respectively. With a median follow-up of 46 months, 35 of 79 pts with smoldering WM progressed to symptomatic WM. Among the 220 pts who had (n = 185) or developed (n = 35) symptomatic WM, 138 had CXCR4wt WM and 81 had CXCR4mutWM. The latter group was comprised of 48 CXCR4ns and 33 CXCR4 indels. When mutations were alternatively categorized by predicted CXCR4 functional impact, 58 pts had protein-truncating variants (PTVs), which were either nonsense or frameshift mutations leading to a premature stop codon and protein truncation, and 24 pts had non-PTVs (indels without premature stop codons). Comparisons of the three groups of pts within each of the two categorizations (CXCR4wt vs. CXCR4ns vs. CXCR4 indels) or (CXCR4wt vs. CXCR4 PTVs vs. CXCR4 non-PTVs) did not reveal significant differences in the frequencies of hyperviscosity, von Willebrand syndrome, cold agglutinin disease, neuropathy, bone lesions, or Bing Neel syndrome. The incidence of secondary malignancies or transformation from WM to large B-cell lymphoma was also not significantly different. There was a significant correlation between CXCR4wt status and concurrent MYD88wt and TP53wtstatus. Pts with MYD88wt WM were also more likely to be CXCR4wt than those with MYD88mut WM (95% vs 51%, p<0.001). Similarly, those with TP53wt WM were more likely to be CXCR4wt compared to those with TP53mut (66% vs 41%, p<0.025). Among pts with smoldering WM (N=79) at first NGS panel, OS was not significantly different based on CXCR4 mutation status, whether categorized by mutation subtype or predicted mutation functional impact (both p > 0.05). Similarly, for those with symptomatic WM (N=185) at first NGS panel (median follow-up = 38 months (95% CI = 31 – 47)), OS did not differ by mutation status. Conclusions: In total, 34% of pts in our cohort had CXCR4 mutations. Importantly, the presence of these mutations did not impact their overall survival. Whether the lower percentage of CXCR4 mutations noted in smoldering WM and MGUS pts is a true reflection of disease state vs. testing sensitivity remains to be determined. The impact of VAF and CXCR4 mutation subtype on response to different therapies also warrants further investigation.