Quantitative MYD88^L265P and CXCR4^S338X analysis to assess clinical trial performance in Waldenstrom macroglobulinemia

TO THE EDITOR:

Serum immunoglobulin M (IgM) measurements represent the gold standard for assessing disease response in Waldenström macroglobulinemia (WM).^1,2 Many agents used to treat WM can raise or lower serum IgM levels without impacting underlying disease, thereby affecting categorical response assessment.3, 4, 5, 6, 7, 8 Somatic mutations in MYD88 (MYD88^Mut) and CXCR4 (CXCR4^Mut) are found in 95% to 97% and 30% to 40% of patients with WM, respectively.9, 10, 11, 12, 13, 14, 15 Nearly all MYD88^Mut are the MYD88^L265P variant, whereas CXCR4^S338X is the most common CXCR4^Mut.^13,14 Like MYD88^L265P, CXCR4^S338X can be assessed by quantitative allele-specific polymerase chain reaction (PCR) (qAS-PCR).^12,15 Both MYD88^Mut and CXCR4^Mut status can impact treatment response in WM.^8,14,16 In particular, CXCR4^S338X clonality is an important determinant of ibrutinib (IBR) outcomes in patients with WM.¹⁷ B-cell selection can greatly augment the detection of MYD88^Mut and CXCR4^Mut in bone marrow (BM) and peripheral blood (PB) samples.^9,10,15 MYD88^Mut and CXCR4^Mut testing is recommended in the diagnostic workup of WM.¹⁸

Serial qAS-PCR measurements of BM or PB MYD88^L265P can be used to assess changes in underlying BM WM disease burden.^9,10 However, data on the use of qAS-PCR analysis for MYD88^L265P and CXCR4^S338X in prospective clinical trials remain limited. As such, we utilized BM and/or PB CD19–selected samples to perform qAS-PCR analysis for MYD88^L265P and CXCR4^S338X and compared outcomes across 5 prospective clinical trials: ixazomib, dexamethasone, rituximab (IDR; ClinicalTrials.gov: NCT02400437); IBR(ClinicalTrials.gov: NCT02604511); venetoclax (VEN; ClinicalTrials.gov: NCT02677324); IBR plus ulocuplomab (IBR/ULO; ClinicalTrials.gov: NCT03225716); and IBR plus VEN (IBR/VEN; ClinicalTrials.gov: NCT04273139).19, 20, 21, 22, 23 DNA from CD19-selected cells from BM (n = 96) and matched PB (n = 60) obtained at baseline and at best response were used. The median time to best response (in months) in these trials were: IDR, 17.2; IBR, 33.7; IBR/ULO, 10.5) IBR/VEN, 11.3; and VEN, 11.5. A standard curve was used to convert ΔCT values to percentage of MYD88^L265P in each sample.^9,10 The limit of detection was 0.1%.⁹ Changes in MYD88^L265P ΔCt were compared to changes in underlying BM disease burden and categorical response assessment using 11th International Workshop on Waldenstrom's Macroglobulinemia (IWWM-11) response criteria.² Because CXCR4^S338X can result from either c.1013C>A or c.1013C>G transversions, we used 2 qAS-PCR assays with limits of detection of 0.4% and 0.16%, respectively.¹⁵ The median follow-up and categorical response outcome for the 5 trials are shown in supplemental Table 1. Data analysis was performed in R and included Wilcoxon signed-rank test for continuous data and Pearson correlation for comparing qAS-PCR findings with underlying disease.

Across all 5 trials, serial changes in underlying BM disease burden showed a better association with BM CD19⁺ MYD88^L265P allele burden (r = 0.53; P < .001) vs serum IgM levels (r = 0.43; P < .001) at best response, although these correlations were modest (supplemental Figure 1A-B). The moderate correlation between changes in BM disease burden and either BM CD19⁺ MYD88^L265P allele burden or serum IgM levels is not a surprise. BM disease burden assessments in WM include “normal” B cells and plasma cells. Indeed, in our initial report of the discovery of MYD88^L265P mutation in WM, the median percentage of reads supporting the MYD88^L265P variant was 46.3% (range, 23.4-95.7)¹⁶; this percentage likely reflected the presence of polyclonal B cells in the CD19⁺ isolated cells used for whole genome analysis. Assessing quantitative changes in MYD88^L265P may therefore offer an advantage over pathological enumeration of lymphoplasmacytic cells (LPC) in BM samples.

We first assessed treatment response by examining changes in serum IgM and BM LPC involvement at best response for all 5 trials (Figure 1A-B). While all trials demonstrated similar serum IgM reductions at best response (Figure 1A) and commensurate reductions in BM LPC burden (Figure 1B), BM CD19⁺ MYD88^L265P allele burden showed no change following treatment with IBR and was less pronounced for IBR/ULO vs IDR, IBR/VEN, or VEN (Figure 1C). Excluding those patients on IBR, who were both MYD88- and CXCR4-mutated, did not restore correlation of either BM or MYD88^L265P allele burden with serum IgM changes (data not shown), whereas all patients on IBR/ULO were both MYD88- and CXCR4-mutated. Conversely, marked reductions in BM CD19⁺ MYD88^L265P allele burden occurred following IDR, IBR/VEN, or VEN, with a >2 log-fold reduction at best response (Figure 1D). We next assessed changes in BM CD19⁺ MYD88^L265P allele burden at best response by major vs nonmajor response attainment, a key measure based on serum IgM reductions for determining clinical trial outcomes in WM.^1,2 Per Figure 1E, the BM CD19⁺ MYD88^L265P allele burden showed a >2 log-fold difference for major vs nonmajor responders who received IDR, IBR/VEN, or VEN. Conversely, a minimal difference for BM CD19⁺ MYD88^L265P allele burden was observed between major and nonmajor responders for patients treated with IBR/ULO, whereas no difference was evident for those treated with IBR alone despite a longer interval for best response assessment vs the other cohorts (Figure 1D).

Figure 1.

Response assessment utilizing serum IgM, BM disease burden, and quantitative MYD88^L265P and/or CXCR4^S338X burden in CD19-selected B cells at best response across 5 prospective WM trials. Shown are baseline and best response serum IgM levels (A); BM WM disease burden (B); Percentage of BM CD19–selected cells expressing mutated (L265P) as assessed by quantitative allele-specific PCR (C); absolute change in BM MYD88^L265P at best response (D); change in BM CD19⁺MYD88^L265P allele burden by IGM response status with nonmajor responders (<50% reduction in serum IgM) denoted by red dots and overlying red arrows and major responders (≥50% reduction in serum IgM) denoted by blue dots and overlying blue arrows (E); and change in BM CD19⁺MYD88^L265P vs CXCR4^S338X at best response (F). Response box plots with median values and interquartile ranges are depicted with an overlay of the individual data points. For panels A-E: IDR (N = 11), IBR (N = 26), VEN (N = 14), IBR/ULO (N = 10), and IBR/VEN (N = 30). For panel F: IDR (N = 1), IBR (N = 9), VEN (N = 3), IBR/ULO (N = 5), and IBR/VEN (N = 7). Pairwise Wilcoxon signed-rank test was performed for data in panels A-C,F; and unpaired Wilcoxon signed-rank test for data in panel E. ∗P < .05; ∗∗P < .005. Data are obtained from patients for whom samples at baseline and at time of best response were available. LFC, log-fold change; NA, not performed due to limited sample size; NS, not significant.

Given the importance of CXCR4 to treatment outcomes, we next assessed changes in BM CD19⁺ CXCR4^S338X allele burden at best response. Twenty-five of the 96 patients included in this study carried the CXCR4^S338X variant; an additional 24 patients carried other CXCR4 variants not amenable to CXCR4^S338X variant PCR testing. Patients with either CXCR4^S338X c.1013C>A or c.1013C>G variants showed nearly identical decreases for both BM CD19⁺ CXCR4^S338X and MYD88^L265P allele burden across all 5 trials (Figure 1F). However, decreases in BM CD19⁺ CXCR4^S338X allele burden were greatest at best response in patients receiving IDR, IBR/VEN, and VEN (Figure 1F). Patients who received IBR alone had a modest decrease in BM CD19⁺ CXCR4^S338X allele burden, whereas the addition of the CXCR4 antagonist Ibrutinib plus Ulocuplomab (IBR/ULO) resulted in a greater reduction in BM CD19⁺ CXCR4^S338X allele burden over IBR alone (P < .05).

We next assessed corresponding changes in PB CD19⁺ MYD88^L265P allele burden. Across all 5 clinical trials, PB CD19⁺ MYD88^L265P correlated with both BM disease burden (r = 0.38; P = .004) and BM CD19⁺ MYD88 allele burden (r = 0.64; P < .001) (Figure 2A-B). However, marked differences were observed by treatment type. Akin to changes in BM CD19⁺ MYD88^L265P allele burden, minimal changes in PB CD19⁺ MYD88^L265P occurred at best response in patients receiving IBR or IBR/ULO. In contrast, a >2 log-fold reduction occurred at best response in patients who received IBR/VEN or VEN (Figure 2C). Fewer events prevented a comprehensive assessment of absolute changes in PB CD19⁺ MYD88^L265P allele burden by major vs nonmajor response status. However, minimal differences in PB CD19⁺ MYD88^L265P allele burden were observed at best response between major and nonmajor responders for patients receiving IBR, whereas a >2 log-fold reduction occurred in those who received IBR/VEN (Figure 2D).

Figure 2.

Evaluation of PB CD19⁺MYD88^L265P allele burden at best response across all 5 clinical trials evaluated. Comparison of PB vs BM CD19⁺MYD88^L265P allele burden at best response (A); comparison of PB CD19⁺MYD88^L265P allele burden vs BM disease burden at best response (B); Absolute change in PB CD19⁺MYD88^L265P at best response (C); and change in PB CD19⁺MYD88^L265P allele burden by IGM response status with nonmajor responders (<50% reduction in serum IgM) denoted by red dots and overlying red arrows and major responders (≥50% reduction in serum IgM) denoted by blue dots and overlying blue arrows (D). Pearson R, adjusted r² values, and the root mean square error are shown for each comparison. Patients are colored by clinical trial participation. IDR: N = 4; IBR: N = 22; VEN: N = 5; IBR/ULO: N = 7; and IBR/VEN: N = 22. Data are obtained from patients for whom samples at baseline and at time of best response were available. Three samples were excluded from the analyses in panel B due to missing BM involvement data. LFC, log-fold change; NA, not performed due to limited sample size; NS, not significant.

In this exploratory study involving multiple prospective clinical trials, we provide evidence that BM and PB CD19⁺ MYD88^L265P analysis may contribute alongside BM disease burden assessments to better assessment of clinical trial outcomes vs changes in serum IgM levels. Indeed, per Figure 1A, serial changes in serum IgM levels would have predicted that all 5 trials evaluated in this series behaved analogously. The addition of BM and potentially PB CD19⁺ MYD88^L265P analysis to drug response evaluation could provide a more robust assessment of clinical trial performance by identifying treatments that produce more meaningful tumor reductions in WM patients. The findings also lend support for the advancement of more sensitive digital droplet PCR or cell free DNA analysis for quantitating BM and PB MYD88^L265P and CXCR4^S338X allele burden without the need for CD19 sorting, thereby enabling adoption by clinical laboratories.^11,12,24,25

The relative impact of BCL2 vs Bruton tyrosine kinase (BTK) inhibition in reducing WM tumor burden reduction was notable. The discordance between IgM and BM response for BTKi may be related to inhibition of BTK-driven STAT5 signaling, a driver of IgM production in WM.²⁶ Although multiple studies have demonstrated clinical benefit for BTKi despite their moderate BM response, our findings lend support for targeting BCL2 and advancing next-generation BCL2 inhibitors to achieve greater tumor reduction in WM. Importantly, a deeper reduction of BM CD19⁺ CXCR4^S338X allele burden was also achieved with VEN-based or IDR therapy vs IBR or IBR/ULO. The findings from this prospective, multitrial study therefore provide support for further investigating serial quantitative MYD88^L265P and CXCR4^S338X BM and PB allele burdens for disease response evaluation and clinical trial performance in WM.

Studies were approved by the Harvard/Dana-Farber Cancer Center institutional review board, and samples were obtained after written consent.

Conflict-of-interest disclosure: S.P.T. received research funding, and/or consulting fees from AbbVie/Pharmacyclics Inc, Janssen Oncology Inc, BeiGene Inc, Eli Lilly Pharmaceuticals, Bristol Myers Squibb, and Ono Pharmaceuticals. J.J.C. received research funds or consulting fees from AbbVie, AstraZeneca, Avilar, BeOne, Cellectar, Johnson & Johnson, LOXO, and Pharmacyclics. S.R.S. received research funding and/or consulting fees from ADC Therapeutics, BeiGene, Cellectar, and SOBI. A.R.B. received research funds and/or honoraria from Adaptive, BeiGene, CSL Behring, Genzyme, Karyopharm, Pharmacyclics, and Sanofi. J.M.H. receives consulting fees from Arbella Therapeutics. The remaining authors declare no competing financial interests.