Development and Evaluation of a Multiplex Digital PCR Assay for Detecting CXCR4 S338X Mutations in Waldenstrom's Macroglobulinemia

Background CXCR4 mutations are found in 30-40% of Waldenstrom's Macroglobulinemia (WM) patients and impact disease presentation, time to treatment, and outcomes to BTK-inhibitors. Moreover, antagonists targeting CXCR4 are being developed for use in CXCR4 mutated WM patients. Next generation sequencing (NGS) is most often used to detect mutated CXCR4 but shows low sensitivity in WM patients with <20% (Gustine et al, BJH 2021). The CXCR4 S338X (CXCR4S338X) mutation is the most common and impactful variant. We previously developed quantitative allele-specific PCR (qPCR) assays that offered sensitivities to 0.4% for the CXCR4S338X C>A variant and 0.16% for the CXCR4S338X C>G variant (Xu et al, BJH 2016) though required use of a standard curve, and separate reactions for each variant. Multiplex digital PCR may offer greater sensitivity, as well as simultaneous detection and quantification of both variants without need for a standard curve. We therefore sought to develop a dPCR assay for both CXCR4S338X variants for clinical utilization. Methods A TaqMan SNP genotyping assay was designed, consisting of a set of amplification primers, a VIC-labeled probe for detecting the wild-type allele, a FAM-labeled probe for mutant allele G, and an ABY-labeled probe for mutant allele A. An intentional mismatch was introduced in the ABY probe to reduce its binding to wild-type DNA, allowing for greater discrimination. DNA from CD19+ selected BM mononuclear cells (BM19+) and unselected BM mononuclear cells (BMMC) from previously identified CXCR4S338X mutated patients was used that consisted of 24 with CXCR4S338X C>A mutation and 45 with CXCR4S338X C>G mutation. This included samples that were near the limit of detection of the qPCR assays. All samples were evaluated by both qPCR and dPCR using the QuantStudio 6 Flex and QuantStudio Absolute Q Digital PCR machine, respectively. Results For both the C>A and C>G assays, detection was improved in samples with increased disease burden (>20% BM involvement) by either qPCR and dPCR. Estimates of mutant allele burden were highly correlated between qPCR and dPCR for both the C>A (r = 0.99) and C>G (0.96) assay. In BM19+ samples, dPCR detected CXCR4S338X C>A mutations in 23/24 (96%) versus 17/24 (71%) WM patients by qPCR (P<0.0001). The one wild-type sample by dPCR showed a CXCR4S338X C>G mutation. Conversely, CXCR4S338X C>G mutations were detected in all 45 BM19+ samples by both assays. Among BMMC samples, CXCR4S338X C>A mutation was detected in 17/24 (71%) and 13/24 (54%) by dPCR and qPCR, respectively (p=0.37). In BMMC, the CXCR4S338X C>G mutation was identified in 33/44 (75%) and 27/45 (60%) of samples evaluated by dPCR and qPCR, respectively (p=0.56). When restricted to BM samples with <20% WM disease burden, CXCR4S338X C>A was detected in 7/7 (100%) by dPCR and 4/7 (57%) by qPCR (p=0.19) in CD19+ samples; and 4/7 (57%) and 3/7 (43%) in BMMC samples (p=1.0). CXCR4S338X C>G was detected in all 17 BM19+ samples by either dPCR or qPCR, and 11/17 (65%) and 9/17 (53%) of BMMC samples when restricted to patients with <20% BM disease involvement. The cutoffs used for determining CXCR4 positivity by qPCR were 0.5% for CXCR4S338X C>A and 0.2% for CXCR4S338X C>G, in line with our previously published data. To determine the sensitivity of the dPCR assay, a CXCR4wild-type control was run on each plate. Across twelve plates, a total of 135,220 VIC positives were detected in the control sample versus just 1 FAM and 3 ABY positives. The highest mutant allele burden measured in the control sample of any plate for CXCR4S338X C>A was 0.011% (median = 0%, range = 0-0.011%, upper 95% CI estimate = 0.046%) and for CXCR4S338X C>G was 0.005% (median = 0%, range = 0-0.005%. upper 95% CI estimate = 0.04%). Based off these measurements, cutoffs of 0.05% for both the CXCR4S338X C>A and CXCR4S338X C>G dPCR assays were established, representing a 10x and 4x increase in sensitivity versus qPCR, respectively. Conclusion dPCR can simultaneously detect both the CXCR4S338X C>A and C>G mutations without need for running standard curves, while showing enhanced performance for identifying CXCR4S338X C>A mutations versus qPCR including in WM patients with <20% BM disease burden.