POD24 is a novel determinant of prognosis in patients with Waldenström macroglobulinemia

Key Points

• •CXCR4 mutations portend inferior outcome for patients with WM treated with frontline BR, whereas MYD88^L265P does not impact survival.
• •POD24 serves as an early surrogate end point by reliably identifying patients with unfavorable subsequent survival.

Visual Abstract

Abstract

Waldenström macroglobulinemia (WM) is characterized by recurrent MYD88 and CXCR4 mutations, whose prognostic value in patients treated with chemoimmunotherapy remains unclear. Moreover, the typically prolonged progression-free survival (PFS) correlates inconsistently with overall survival (OS), underscoring the importance of examining other surrogates. Progression of disease within 24 months (POD24), an established early end point, delineates functionally high-risk patients in other indolent lymphomas. This international study evaluated 253 patients receiving frontline fixed-duration bendamustine-rituximab (BR), a common chemoimmunotherapy for WM. At a median follow-up of 5.9 years, 5-year PFS and OS were 65% and 87%, respectively; 5-year PFS was similar between MYD88^L265P (90%) and MYD88^{wild-type (WT)} subcohorts (64% each; P = .4). Among 89 patients with known CXCR4 status, the subcohort with CXCR4^mutation (28%) had shorter PFS (median 3.3 vs 8.8 years; hazard ratio [HR], 2.8; P = .0036) and OS (HR, 2.6; P = .036) compared with CXCR4^WT. POD24 occurred in 11.5% of patients who demonstrated inferior subsequent OS (5-year OS, 71% vs 86%; HR, 3.1; P = .005) and higher mortality (standardized mortality ratio [SMR], 3.7), unlike the non-POD24 group, whose mortality was comparable to the matched general population (SMR, 1.1). In conclusion, BR is effective irrespective of the MYD88 status, but CXCR4 mutations and POD24 portend worse outcomes. Patients without POD24 represent a cohort with distinctly favorable outcome.

Introduction

Waldenström macroglobulinemia (WM) is a lymphoplasmacytic lymphoma, with distinct somatic alterations such as clonal MYD88^L265P mutation, detectable in most patients, and subclonal CXCR4 mutation(s), present in 25% to 40% of patients.¹ Our approach to WM has steadily evolved, although the anti-CD20 monoclonal antibody, rituximab, remains the linchpin and is frequently used in combination with limited-duration chemotherapy.¹ Bruton tyrosine kinase inhibitors (BTKi) are also highly effective, but their activity may be compromised in the absence of mutated MYD88.² Moreover, BTKi, in general, have reduced efficacy in the presence of mutated CXCR4 (CXCR4^MUT) and require continuous administration, potentially increasing not only the costs but the likelihood of cumulative toxicities and drug-drug inteactions: factors that make finite-duration chemoimmunotherapy still considerably appealing.^3,4 Bendamustinerituximab (BR) regimen gained popularity before ibrutinib’s approval, based on a subset analysis of the StiL NHL1-2003 trial, which showed durable remission (median progression-free survival [PFS] of 69.5 months) in treatment-naïve patients with WM.⁵ Subsequent studies confirmed ibrutinib’s efficacy, catapulting it to be among the preferred chemoimmunotherapeutic regimens, although the impact of recurrent molecular alterations, such as MYD88^L265P and CXCR4^MUT, on its efficacy remains less clear.^6,7 Furthermore, the typically prolonged PFS that attests to the efficacy of many treatments correlates inconsistently with overall survival (OS) in patients with WM,^4,5,8,9 underscoring the importance of examining other surrogate end points such as the progression of disease within 24 months of treatment initiation (POD24), an established early end point to delineate functionally high-risk patients in other indolent lymphomas.10, 11, 12, 13 In this study, we examined the implications of POD24, MYD88^L265P, and CXCR4^MUT in patients treated with BR in the frontline setting.

Methods

Participants

Following the institutional review board’s approval, we included consecutively evaluated patients with active/symptomatic WM across medical centers in the United States and Europe who were diagnosed between 1 January 2012, and 31 July 2021, and were treated with up to 6 cycles of frontline fixed-duration BR. Consent and when applicable, a waiver of consent, was obtained as per the institutional review board guidelines. The study was conducted in accordance with the Declaration of Helsinki. Patients on rituximab maintenance therapy following BR were excluded to avoid biases, as this strategy is not universally embraced. Responses were categorized as per the 11th International Workshop on Waldenstrom's Macroglobulinemia (IWWM-11) consensus criteria.¹⁴

Statistics

Nonparametric tests were used to compare continuous variables, whereas χ² or Fisher’s exact test were used to assess nominal variables. Time-to-event outcomes were determined using the Kaplan-Meier method, with subgroup comparisons performed using the log-rank test. Reverse censoring was used for calculating the follow-up time. PFS was calculated from the time of initiation of BR until progression or death, whichever occurred earlier. OS was calculated from the initiation of treatment until last follow-up or death, with patients censored if alive at last follow-up. To estimate the prognostic impact of POD24, we performed a landmark analysis among patients alive at 24 months, excluding those with a follow-up <24 months (n = 201). Follow-up was then calculated from the 2-year mark for the 2 subcohorts, which were those who progressed (POD24 cohort) and those who did not progress within 24 months from initiation of BR therapy (the reference cohort). In addition, we used POD24 as a time-dependent covariate to evaluate its association with subsequent OS. For assessment of POD24 as a time-dependent covariate (n = 202), the end points were calculated from the time of progression for the POD24 group and from the 2-year mark for the reference cohort.^10,15 Patients who did not progress but died because of non-WM causes within 2 years of treatment initiation were excluded from the time-dependent analysis.

To further examine the superior outcome of patients without POD24, we compared their OS with a cohort of healthy age, sex, calendar year, and country of origin-matched population. Expected survival was calculated based on the death rate data obtained from the Human Mortality Database for France, the United Kingdom, Greece, and the United States (available at www.mortality.org, Human Mortality Database, Max Planck Institute for Demographic Research [Germany], University of California, Berkeley [USA], and French Institute for Demographic Studies [France]; data downloaded on 14 March 2025). The observed and expected survival rates were analyzed by calculating the standardized mortality ratio (SMR) and corresponding 95% confidence interval (CI).

Results

The baseline characteristics of the cohort (n = 253) are outlined in Table 1. The median follow-up was 5.9 years (95% CI, 5.3-6.5) and the median number of administered cycles of BR was 6 (range, 1-6). Major response rate was high (94%), despite 25 patients (10%) discontinuing BR prematurely (supplemental Tables 1 and 2). Median PFS was 6.7 years (95% CI, 5.5-8.8; 5-year PFS, 65% [95% CI, 61-75]). Median OS was not reached (NR; 5-year OS, 87% [95% CI, 83-92]; supplemental Figure 1A-B).

Table 1.

Baseline characteristics at initiation of fixed-duration BR

Parameter	Value
Age at treatment initiation, median (range), y	66.5 (40-87)
Sex, females, n (%)	91 (36)
Serum IgM, median (interquartile range), mg/dL	3000 (1315-5067)
Hemoglobin, median (interquartile range), g/dL	9.9 (8.3-11.5)
Platelet count, median (interquartile range), ×109/L	210 (119-311)
Bone marrow lymphoplasmacytic infiltrate, median, % (range)	54 (26-80)
β2 microglobulin, median (interquartile range), mcg/mL	3.6 (2.5-5.2)
MYD88L265P mutated, n (%)	154/172 (90)
CXCR4 mutated, n (%)	25/89 (28)
Elevated serum lactate dehydrogenase, n (%)	21 (18)
International prognostic scoring system for WM
Low risk, n (%)	25/136 (19)
Intermediate risk, n (%)	38/136 (28)
High risk, n (%)	73/136 (54)

Among 172 patients with known MYD88^L265P status, 154 patients (90%) harbored MYD88^L265P. Both subcohorts, with MYD88^L265P and MYD88^WT genotypes demonstrated comparable baseline characteristics (supplemental Table 3), response (supplemental Figure 2), PFS (5-year rate 64% in each group; P = .4), and OS rates (5-year 89% vs 74%, P = .44), respectively (Figure 1A-B). Among 89 patients with known CXCR4 status, 25 patients (28%) exhibited CXCR4^MUT genotype (supplemental Table 4). The response rates were lower for patients with CXCR4^MUT than for patients with CXCR4^WT (supplemental Figure 2). The 5-year PFS rate was 43% (median 3.3 years; 95% CI; 1.9 to NR) and 78% (median 8.8 years; 95% CI, 6.7 to NR), respectively (P = .0036; Figure 1C), which translated to significantly inferior OS for the CXCR4^MUT subcohort (Figure 1D). The PFS and OS outcomes were comparable between the cohorts with and without the MYD88 and CXCR4 mutation status related data available for analysis (supplemental Figure 3).

Figure 1.

PFS and OS by genotype. (A) Patients with an MYD88 mutated and WT genotype had comparable PFS; the 5-year PFS was 64% for both MYD88^L265P and MYD88^WT genotypes (P = .4); median PFS was NR in either cohort. (B) The median OS was NR in either MYD88^L265P mutated and MYD88^WT subcohorts, and 5-year OS rates were comparable (89% [95% CI, 83-94] vs 74% [95% CI, 55-100], respectively; P = .44). (C) Patients with a CXCR4 mutated genotype demonstrated an inferior PFS (5-year PFS rates 43% vs 78% for CXCR4^WT genotype, P = .0036; HR, 2.8; 95% CI, 1.4-5.7). (D) The median OS was NR for subcohorts with either CXCR4 genotypes; 5-year OS was 75% (95% CI, 58-98) for CXCR4^MUT compared with 91% for CXCR4^WT (95% CI, 83-99; P = .036; HR, 2.6; 95% CI, 1.02-6.9) subcohort. WT, wild-type.

Among 201 patients included in the POD24 landmark analysis, 23 patients (11.5%) progressed within 24 months of initiating BR. This subgroup had a significantly shorter OS from the 2-year landmark (Figure 2A). Most patients in the POD24 cohort harbored CXCR4^MUT (70%) compared with the reference cohort (20%; P = .003; supplemental Table 5). However, the POD24 status remained independently prognostic for the landmark analysis–calculated OS after adjusting for the CXCR4^MUT status (HR, 4.9; 95% CI, 1.2-21; P = .03; supplemental Table 6). Patients with POD24 had a higher SMR of 3.7 (95% CI, 1.6-7.4; P = .004) compared with an age, sex, calendar year, and country of origin-matched cohort of healthy individuals (Figure 2B), whereas OS of the non-POD24 cohort (n = 178) was comparable (SMR, 1.1; 95% CI, 0.7-1.6; P = .75; Figure 2C). Of 202 patients included in the time-dependent covariate POD24 analysis, 24 patients (12%) progressed within 24 months of starting BR. Five-year OS following progression was 74% for early progressors (n = 24) compared with 86% in the reference cohort (POD ≥ 24 months, n = 178), despite the follow-up starting at the 2-year time point for the latter (HR, 2.4; 95% CI, 1.1-5.1; P = .024; supplemental Figure 4).

Figure 2.

POD24. (A) 2-year landmark analysis (n = 201). Patients with POD24 had a higher risk of death compared to those without POD24 (5-year OS of 71% and 86%, respectively; HR, 3.1; 95% CI, 1.4-6.9; P = .005). (B) Comparison of OS for POD24 cohort with general population. When compared with an age, sex, calendar year, and country of origin-matched cohort, the OS for patients who were alive and progressed within 24 months of initiation of BR was significantly inferior to that of the general population (observed deaths = 8; expected deaths = 2.1; SMR, 3.7; 95% CI, 1.6-7.4; P = .004). (C) Comparable OS of non-POD24 cohort and the matched general population from the 2-year landmark. When compared with an age, sex, calendar year, and country of origin-matched cohort, the OS for patients who did not progress within 24 months of initiation of BR was comparable to that of the general population (observed deaths = 23; expected deaths = 21.2; SMR, 1.1; 95% CI, 0.7-1.6; P = .75).

Discussion

This multi-institutional study demonstrates robust outcomes with BR in WM that are unaffected by the MYD88^L265P status, in contrast to observations with BTKi therapy.² However, similar to findings with BTKi, the presence of CXCR4^MUT was associated with inferior outcomes. The C-terminal domain alterations of CXCR4 result in the loss of inhibitory serines, preventing CXCR4 receptor downregulation/internalization and in turn inducing drug-resistance through persistently activated downstream prosurvival signaling.¹⁶ Previous studies involving BR were underpowered to detect CXCR4^MUT associated outcome disparity.⁶ This largest report to date substantiates preclinical data suggesting CXCR4^MUT associated resistance to BR, akin to the ASPEN trial findings with BTKi,¹⁷ wherein lower very good partial response (VGPR) rates in the CXCR4^MUT subcohort receiving zanubrutinib (45% vs 21% with CXCR4^WT) or ibrutinib (31% vs 10%) translated to shorter PFS.⁴

Our study identified POD24 as an early surrogate end point for OS. Notably, CXCR4^MUT, but not high-risk International Prognostic Scoring System (IPSS)-WM enriched the POD24 cohort. By estimating survival probabilities from the 24-month landmark, we attempted to account for the biases arising from the time-dependent covariate POD24 based grouping. Leveling the field, the landmark analysis generated a more accurate report of the POD24-effect on mortality, eliminating the impact of early deaths. Patients with active WM are known to have a markedly inferior outcome compared with the corresponding survival of the matched general population (SMR, 4.6-6.3).¹⁸ Our study interestingly suggests that if patients are alive and progression-free for 2 years following BR therapy, their subsequent OS is equivalent to that of the matched general population, underscoring the usefulness of POD24 end point in patient counseling.

Similar outcomes of the subcohorts, with and without the molecular data, suggest that the biases resulting from the missing, retrospectively gathered information are minimal. A small sample-size precluded the examination for specific CXCR4 (nonsense vs frameshift) mutational impact. Additional limitations include the lack of centralized molecular marker assessment, absence of non-L265P MYD88^MUT data, and sparse information on deletion 17p/TP53 status, which is increasingly recognized as unfavorable.¹⁹ The absence of a standardized follow-up strategy is another limitation of the study. These gaps highlight the need for comprehensive molecular profiling to further refine the prognostic models. Nonetheless, our findings emphasize the importance of routine MYD88 and CXCR4 mutational testing. The utility of POD24 as a prognostic marker in patients treated with upfront BTKi or other infrequently used frontline chemoimmunotherapy regimens remains an open question. Specifically, its ability to accurately capture functionally high-risk patients may be compromised with regimens demonstrating short PFS.²⁰ Importantly, prospectively collected, trial-level data needs to be analyzed to identify determinants of POD24 with various anti-WM therapies at diagnosis itself that could potentially guide the frontline risk-adapted approaches to circumvent the POD24 issue altogether.

In conclusion, mutated CXCR4 is associated with inferior outcome with fixed-duration BR for WM. POD24, experienced by ∼12% of patients, may serve as an early surrogate end point by reliably identifying patients with inferior subsequent survival. In contrast, non-POD24 cohort demonstrates survival equivalent to that of the matched general population.

Conflict-of-interest disclosure: S.D. reports congress support and research funding from BeiGene, advisory board participation for BeiGene, Cellectar, Ouro Medicines, Recordati, and Sanius Health Ltd; and speaker bureau participation for BeiGene and Janssen. M.A.D. received honoraria from participation in advisory boards and satellite symposia for Amgen, Sanofi, Regeneron, Menarini, Takeda, GSK, Bristol Myers Squibb (BMS), Janssen, BeiGene, Swixx, and AstraZeneca. P.K. is the principal investigator of trials for which Mayo Clinic has received research funding from Amgen, Regeneron, BMS, Loxo Pharmaceuticals, Ichnos, Karyopharm, Sanofi, AbbVie, and GlaxoSmithKline; has received honorarium from Keosys; and has served on the advisory boards of BeiGene, Mustang Bio, Janssen, Pharmacyclics, X4 Pharmaceuticals, Kite, Oncopeptides, Ascentage, Angitia Bio, GlaxoSmithKline, Sanofi, and AbbVie. R.F. reports consulting for AbbVie, Adaptive, Amgen, Apple, BMS/Celgene, GSK, Janssen, Karyopharm, Pfizer, RA Capital, Regeneron, and Sanofi; scientific advisory board participation for Caris Life Sciences; participation in the board of directors for Antengene; and royalties of ∼$2000 per year from a patent for fluorescence in situ hybridization in multiple myeloma. The remaining authors declare no competing financial interests.