Preventive strategies to identify patients with B-ALL who are likely to experience relapse are important so that timely therapeutic decisions on the use of immunotherapy and/or transplantation can be made and overall survival can be improved. These strategies primarily focus on either fine-tuning risk stratification or using ultradeep and sensitive minimal residual disease (MRD) detection methods. Moreover, these interventions need to be performed considering ethnic population groups and regions and the different chemotherapy protocols, as both influence the frequency of the type of mutations and the underlying pathways involved. Hence, we extended our initial findings of clonal patterns of relapse and the landscape of somatic mutations encountered in paired-sample cohorts to the current digital PCR-based “DETECTOR” validation study8. We shortlisted recurrent mutations (encountered in 2 or more patients) in leukemogenesis-initiating or leukemia driver- or relapse-enriched genes (KRAS, NRAS, NT5C2, KMT2D, UHRF1, TP53 and PMS2) and analysed their mutation fractions and evolution patterns in sequential therapy samples as well as in a cohort of unpaired relapse samples. For the validation study, we standardized a triplex assay format on a microfluidics-based digital PCR platform.
With the current validation study, we confirmed that RAS clones have a dynamic evolutionary nature and can undergo both positive and negative selection under chemotherapy. In particular, the NRAS (c.38G > A and c.35G > A) and KRAS (c.38G > A) clones are present as both minor and major clonal populations (50% of patients; 7/14) in diagnostic samples and are either eliminated post chemotherapy (absence in sequential remission samples) or rise in later sequential samples (mid-maintenance time point) after being undetectable in earlier remission samples. In fact, two of the cases in which a rise in the major diagnostic clone was detected in the mid-maintenance samples subsequently relapsed 5 and 7 months later. Figure 5 highlights the clonal evolution plot for one of these relapsed patients who initially had an NRAS c.38G > A mutation with an MAF of 15% in the D0 sample, followed by the disappearance of the clone in the D35 and EOC samples, revealing a rise in the clonal MAF to 27.7% in the MM sample with subsequent early relapse at 11 months with a biallelic NRAS clone with a 97% VAF, as identified by deep sequencing. After 2 months of reinduction with high-dose chemotherapy, the same index patient had persistent clones with an MAF of 21% in the 14-month post relapse-treated sample. In addition to sequential samples, NRAS minor and major clonal populations were also present in 8/14 (57%) of the unpaired relapse samples, driving the relapse in 5/14 (36%) cases as a major clone. In fact, many studies have shown that RAS clones behave in a similar dynamic fashion, but no studies have utilized digital PCR to confirm their nature in sequential and relapse cohorts10–13. Similar to previous studies, RAS clones at diagnosis were more common in hyperdiploid patients, while in relapsing patients, no association was observed with the primary genetic event, and their distribution was more heterogeneous. In addition to RAS clonal predominance and dynamic nature, our validation study was also able to demonstrate the emergence of minor clones in relapse-specific or enriched genes of NT5C2 and PMS2 (4/14 cases; 28.5%) in mid-maintenance (MM) samples. This finding confirms their origin as therapy-acquired de novo events under chemotherapy pressure. Although all of these clones were minor in nature, it would be interesting to see how many of these actually relapse on longitudinal follow up. Since our study analysis was restricted to sequential samples until the MM phase and further time-point samples were not collected and tested, a conclusive comment on the frequency of patients who would relapse cannot be made. However, the unpaired relapse sample data also revealed similar mutant clones of NT5C2 and PMS2 in 6/14 (43%) patients, with 2/14 (14%) being major NT5C2 clones driving early relapse. This indicates that at least 50% of those with detectable minor subclones of NT5C2 in mid-maintenance samples might experience relapse. Waanders et al. demonstrated the reverse tracing of the rise of minor clones of NRAS (c.35G > A), KRAS and NT5C2 in 3/5 of patients who experienced relapse by performing droplet digital PCR on samples from different time points from diagnosis to remission, confirming the minor-major clonal pattern of relapse7. All of these samples in their study were in remission but revealed minor subclones in both bone marrow and peripheral blood samples on droplet digital PCR with MAFs between 0.006% and 0.4%, highlighting the importance of detecting these clones early to predict a pending relapse and hence providing timely intervention.
None of the cases revealed UHRF1 or KMT2D clones, and only one case had a minor TP53 clone in either the sequential or relapse samples tested, although these genes were recurrent genes in our published discovery cohort. This could be because despite being recurrent genes, the nature of the mutations in these genes was very heterogeneous, and it was difficult to include all the mutants in a cost-effective multiplex validation assay.
Overall, the digital PCR triplex assay on the Quant Studio Absolute Q system could reliably detect mutant copies in the range of 0.117–1.34 copies/µl for the 13 different mutant probes (refer to Supplementary material Table 3), which translates to a sensitivity range between 10− 5 and 10− 6. The internal ROX dye quality check ensured that all the microwells were loaded with sample mix precisely and that the dead volume of the system was < 5%. The reactions can be cost effective, with a per-well cost of approximately 10 USD. This immensely helped with utilizing the test as a screening tool for somatic clonal MRD detection in a resource constraint setting. This study provides a proof of concept on the clonal evolution of somatic mutants in pediatric B-ALL patients and adds to the paucity of data available on the use of digital PCR as a rapid, reliable and cost-effective MRD tool for clone monitoring and screening under chemical stress in acute lymphoblastic leukemia patients. However, the triplex assay used for validation in our study needs further curation with regard to mutants to be screened in our ethnic region, as in its current form, it can reliably detect only approximately 30% of all relapses.