Is next-generation sequencing the future of measurable residual disease assays for Philadelphia chromosome-positive acute lymphoblastic leukemia?

KEYWORDS:

• Measurable residual disease
• BCR::ABL1
• next-generation sequencing
• polymerase chain reaction
• prognosis
• risk stratification

1. Introduction

The outcomes of patients with Philadelphia chromosome-positive (Ph+) acute lymphoblastic leukemia (ALL) have significantly improved over the past several years, due largely to the development of better BCR::ABL1 tyrosine kinase inhibitors (TKIs) such as ponatinib [1]. Safe and effective chemotherapy-free regimens (e.g. blinatumomab plus dasatinib or ponatinib) have also resulted in promising outcomes, which may even be superior to standard chemotherapy-based approaches [2,3]. These newer regimens for Ph+ ALL achieve high rates of measurable residual disease (MRD) negativity, a response endpoint that is associated with improved long-term survival. Although it is well established that MRD negativity plays a vital prognostic role in Ph+ ALL, the optimal method to assess MRD and how to use this information to guide treatment decisions are still being established.

2. Methods to assess MRD in Ph+ ALL

Common methods to assess MRD in ALL include multiparameter flow cytometry (MFC), polymerase chain reaction (PCR) for immunoglobulin (IG) or T-cell receptor (TR) gene rearrangements, PCR for BCR::ABL1, and next-generation sequencing (NGS) for IG/TR rearrangements [4]. MFC is commonly used in Philadelphia chromosome-negative B-cell ALL or T-cell ALL, whereas PCR for BCR::ABL1 has historically been the MRD assay of choice for Ph+ ALL. While PCR for IG/TR is not typically available in the United States, it is commonly used in many European countries where it is well standardized. MFC is generally capable of achieving a sensitivity of at least 10⁻⁴, while PCR assays can achieve a sensitivity of 10⁻⁵. NGS for IG/TR has a sensitivity of 10⁻⁶ (1- to 2-logs better than MFC or PCR) and is available through a commercial assay (clonoSEQ®, Adaptive Biotechnologies, Inc.), which is the only Food and Drug Administration-cleared MRD assay for ALL in the United States.

3. Prognostic and therapeutic importance of MRD in Ph+ ALL

MRD is highly prognostic in ALL, as shown in a large meta-analysis [5]. In Ph+ ALL, several studies have shown that achievement of a complete molecular response (CMR) – typically defined as an absence of detectable and/or quantifiable BCR::ABL1 transcripts by PCR – is associated with superior outcomes. In a retrospective analysis of adults with Ph+ ALL who received frontline hyper-CVAD plus a TKI and did not undergo allogeneic stem cell transplantation (allo-SCT) in first remission, achievement of 3-month CMR resulted in a 4-year overall survival (OS) rate of 66%, as compared with only 36% for those with lesser responses (p = 0.0009) [6]. In a follow-up study of patients with Ph+ ALL who achieved 3-month CMR, the best outcomes were observed in those who received ponatinib as part of their frontline treatment regimen (5-year OS 84% versus 60–65% with first- or second-generation TKIs, respectively) [7]. Achievement of CMR can also be used to guide therapeutic decisions regarding allo-SCT in first remission. In a large, retrospective, propensity score analysis, subsequent receipt of allo-SCT did not significantly impact relapse-free survival (RFS) or OS in patients with Ph+ ALL who achieved 3-month CMR after frontline therapy [8]. While allo-SCT reduced the risk of relapse, this was offset by introduction of non-relapse mortality. Some small studies have suggested better quantification of low-level BCR::ABL1 transcripts with digital droplet PCR (ddPCR) assays and possibly slightly better sensitivity, as compared with real-time quantitative PCR [9,10]. However, whether ddPCR offers a clinically significant improvement in risk stratification in Ph+ ALL is largely unknown.

4. Is PCR for BCR::ABL1 the optimal MRD assay in Ph+ ALL?

While several studies have shown the benefit of achieving MRD negativity using PCR for BCR::ABL1, this may not be the optimal assay for MRD in Ph+ ALL. Assessment of MRD using IG/TR gene rearrangements (either by PCR or NGS) may provide a more accurate measure of MRD in Ph+ ALL. Discrepancies between MRD assessed by PCR for BCR:ABL1 and by PCR or NGS for IG/TR have been described [11–14]. One reason for this discordance may be related to different sensitivities of the MRD assays. This is illustrated in cases where very low-level disease – generally at a level of 0.001–0.0001% – is detected by NGS for IG/TR but not by PCR for BCR::ABL1 [14]. However, in most cases of discordance, PCR for BCR::ABL1 is positive for MRD, while either PCR or NGS for IG/TR is negative. So what phenomenon accounts for this latter type of discrepancy? Cell sorting assays have identified two different subtypes of Ph+ ALL, those in which BCR::ABL1 transcripts are restricted to the ALL blast compartment (‘typical Ph+ ALL’) and those in which BCR::ABL1 transcripts can be detected in non-blast cells, including myeloid or nonmalignant lymphoid cells (‘CML-like Ph+ ALL’) [12,13]. In cases of typical Ph+ ALL, BCR::ABL1 or IG/TR rearrangements track together in parallel, and thus, either may be an appropriate MRD target. However, in patients with Ph+ ALL with CML-like biology, persistence of BCR::ABL1 may not be reflective of true residual leukemic cells, thus limiting the applicability of PCR for BCR::ABL1 in this subgroup. For the approximately 20–25% of patients with Ph+ ALL who have CML-like biology [12,14], tracking IG/TR-based MRD provides a more accurate assessment of residual leukemia and is more appropriate for risk stratification and therapeutic decision-making.

5. Next-generation sequencing-based MRD in Ph+ ALL

NGS-based MRD assessment of IG/TR rearrangements can achieve a sensitivity of 10⁻⁶ and outperforms both MFC and PCR as an MRD tool in Ph-negative ALL [15–17]. However, data using NGS-based MRD in Ph+ ALL are relatively scant. In one retrospective analysis of patients with Ph+ ALL, MRD assessments using PCR for BCR::ABL1 and NGS for IG/TR were compared [14]. Among 25 patients in the primary cohort in whom MRD was assessed within 6 months from the start of frontline treatment, eight patients (32%) had at least one discordant sample. In the three patients who were NGS+/PCR-, the median level of NGS MRD was 0.0006% (range, 0.0003–0.004%), which is below the sensitivity of the PCR assay. In the five patients who were PCR+/NGS-, the median level of PCR MRD was 0.03% (range, 0.01–0.07%). There was a trend toward better outcomes in those patients who achieved NGS MRD negativity within 6 months of treatment (5-year OS rate: 81% versus 57% if MRD-positive; p = 0.07). Among 81 patients in the entire cohort, approximately 20% had at least one MRD assessment that was PCR+/NGS-, which is consistent with a previous report that found a discordance rate of 23% between PCR for BCR::ABL1 and PCR for IG/TR [12]. Among those who achieved MRD negativity by NGS for IG/TR, the presence of persistently detectable BCR::ABL1 transcripts did not impact outcomes. The 5-year RFS for patients who were PCR+/NGS- and PCR-/NGS- was 84% and 73%, respectively (p = 0.84), and 5-year OS was 94% and 80%, respectively (p = 0.87). With a median follow-up of over 5 years, only 1 of 20 patients who were PCR+/NGS- subsequently relapsed. Taken together, these data suggest the NGS-based MRD assessment is more accurate than PCR for BCR::ABL1. In addition to its superior sensitivity, NGS MRD assessment for IG/TR may be particularly useful for patients with Ph+ ALL with CML-like biology, where persistent PCR for BCR::ABL1 may not carry prognostic information.

6. Unanswered questions

Several questions remain about how best to assess MRD in Ph+ ALL and use this information for therapeutic decision-making. First, although the available data suggest that assays that track IG/TR rearrangements provide a more accurate assessment of ‘true’ MRD than does PCR for BCR::ABL1, the most clinically relevant timepoints for achievement of MRD negativity using these assays remain to be determined. For example, while 3-month CMR is well established to be associated with superior outcomes in Ph+ ALL and may allow for allo-SCT to be safely deferred, the optimal time to achieve NGS MRD negativity in order to avoid allo-SCT is unknown. This timepoint will also likely vary according to the regimen used (e.g. intensive chemotherapy + TKI versus blinatumomab + TKI) and therefore needs to be prospectively assessed in the context of a specific treatment. Future studies are also needed to determine whether very early achievement of NGS MRD-negativity in Ph+ ALL – for example, after 1 cycle of frontline therapy – may identify particularly favorable risk disease in which treatment de-escalation can be considered.

The growing acknowledgment of the existence of Ph+ ALL with ‘CML-like biology’ also raises several important and clinically relevant questions. While initial data suggest that the long-term survival of typical and CML-like Ph+ ALL are similar [13], we still lack an understanding of the nature of these presumably non-leukemic cells in which these BCR::ABL1 transcripts are present. For example, what is the relapse potential of these residual cells? Do they require continuous exposure to a TKI or is it safe to stop TKI therapy despite their persistence? How does Ph+ ALL with CML-like biology different from CML in lymphoid blast phase? Translational studies are needed to better understand the different biologies of these two subgroups of Ph+ ALL and their relationship with CML in lymphoid blast phase. Clinical studies with long follow-up are also needed to confirm the relapse risk of patients with persistently detectable BCR::ABL1 who achieve NGS MRD negativity.

Declaration of interest

NJ Short has received honoraria from Adaptive Biotechnologies, Inc. E Jabbour has received research funding and honoraria from Adaptive Biotechnologies, Inc. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

NJ Short wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version.

Additional information

Funding

This research is supported in part by the MD Anderson Cancer Center Leukemia SPORE [CA100632], and the NIH/NCI Cancer Center Support Grant [P30 CA016672].