uMRD as a surrogate endpoint for clinical trials, regulatory approvals, and therapy decisions in CLL

When analyzing patient outcomes in clinical trials, survival is often considered to be the gold standard of outcome measures; however, in many cancers, such as chronic lymphocytic leukemia (CLL), measuring outcomes by survival can be suboptimal. In practice, examining a survival endpoint can take many years, potentially delaying the acquisition of critical scientific data which is vital for clinical decision making and regulatory approvals. Surrogate endpoints, such as progression-free survival (PFS) and quality of life (QoL), are sometimes used in lieu of survival, although these can also have drawbacks and have been shown to be unreliable in practice.¹

Another promising surrogate endpoint is the achievement of a negative measurable residual disease (MRD) test, with multiple clinical trials and regulatory approvals using such data in recent years. However, its acceptance is not ubiquitous; the European Medicines Agency (EMA) accepted MRD-test data from clinical trials for the approval of blinatumomab in acute myeloid leukemia, while the US Food and Drug Administration (FDA) did not.

This raises the question of whether MRD-negativity, or the achievement of unmeasurable MRD (uMRD) status, is an appropriate endpoint for clinical trials in CLL and a subsequent indicator for regulatory approval. Below, we summarize commentary from Yang et al. published in Leukemia in 2022.¹

Overall, they suggest that use of MRD-test results to inform regulatory decision making may be premature at this time and that careful consideration is required before using MRD as a surrogate for PFS and survival.

PFS-survival correlation

An important consideration is whether PFS represents an acceptable surrogate for survival in CLL. Multiple studies of chemoimmunotherapy (CIT) in CLL indicate that it does not.

• Results from the CLL7 study (NCT00275054) of fludarabine, cyclophosphamide, and rituximab in high-risk early-stage CLL showed better PFS but not better survival compared with a watch and wait approach.
• The phase III UNITY-CLL study (NCT02612311) comparing ublituximab/umbralisib with chlorambucil/obinutuzumab in patients with untreated or advanced CLL demonstrated better PFS but worse survival for the former.

These discordances may reflect the hidden impact of subsequent therapies, insufficient follow-up, or fatal treatment-related adverse events.

Limitations of MRD-testing¹

International Workshop on Chronic Lymphocytic Leukemia (iwCLL) response criteria are based on laboratory evaluations of the blood and bone marrow alongside clinical evaluations of lymph nodes and the spleen. MRD-testing is typically performed using laboratory techniques, such as multi-parameter flow cytometry, quantitative real-time polymerase chain reaction, or next-generation sequencing technologies.

• Differences between iwCLL response criteria and MRD-testing can result in patients with unfavourable response by iwCLL criteria having a favourable uMRD status
• Current MRD-testing does not utilize circulating tumour DNA, which can be indicative of residual leukemia cells in lymph nodes, spleen and/or other sites
• There are often inconsistencies between MRD-testing of blood and bone marrow samples

Further research is required to fully understand and address these discrepancies.

PFS/survival correlation with MRD-testing¹

Treatment of CLL with CIT and/or venetoclax based regimens is more likely to achieve uMRD compared with Bruton’s tyrosine kinase inhibitors (BTKis) alone. However, BTKis prolong PFS and likely prolong survival.

• In the ECOG-E1912 trial (NCT02048813) of continuous ibrutinib plus rituximab vs fludarabine, cyclophosphamide, and rituximab in patients with CLL, the former was associated with improved PFS and survival but a lower frequency of uMRD.
• Combined data from the CLL8 and CLL10 (NCT00281918;NCT00769522) trials indicated that in patients with uMRD, those who had persistent lymphadenopathy had worse PFS compared with those who had not. The data also show no impact of persistent splenomegaly.
• The CLL14 (NCT02242942) and MURANO (NCT02005471) studies of venetoclax/obinutuzumab and venetoclax/rituximab showed no difference in PFS among patients with uMRD with complete vs partial responses by iwCLL criteria.

The above examples further illustrate the lack of correlation between PFS and/or survival and MRD-testing.

PFS correlation with residual leukemia cell concentration¹

There is much data to indicate that posttreatment concentration of residual leukemia cells correlates with PFS in CLL.

• Patients with MRD of <10⁻⁴ show better PFS rates than those with residual leukemia cells >10⁻⁴ and patients with residual leukemia cells <10⁻² have better outcomes than those with residual leukemia cells >10⁻².
• The MURANO study showed that patients with MRD of <10⁻⁴ had improved survival compared with those who had >10⁻⁴ residual leukemia cells. In addition, patients with MRD of <10⁻⁴ had no significant survival advantage over patients with >10⁻⁴ and <10⁻² residual leukemia cells at the 3-year follow-up post-venetoclax treatment.

The above data suggests that there are increasing improvements in PFS at lower residual leukemia cell concentrations up to a certain point. However, it must be noted that the reliability of such estimates can depend on multiple covariates such as the sensitivity, specificity, accuracy, and precision of MRD-testing.

Biological features of residual leukemia cells¹

Residual leukemia cells can proliferate after therapy cessation, resulting in disease recurrence and progression despite previously achieving uMRD.

• Data from the CLL14 and MURANO studies showed differing rates of uMRD loss between the venetoclax/anti-CD20 antibody and CIT cohorts.
• The CAPTIVATE and GLOW trials (NCT02910583; NCT03462719) demonstrated more patients remained MRD-negative when treated with ibrutinib/venetoclax.
• The GLOW trial also showed that patients with residual leukemia cells >10⁻⁴ but MRD <10⁻² were less likely to have disease progression after cessation of therapy compared with patients receiving chlorambucil/obinutuzumab. PFS of patients not achieving uMRD of <10⁻⁴ was similar to that of patients achieving uMRD in both arms, indicating that uMRD is an unreliable surrogate for PFS.

This suggests that while concentration of residual leukemia cells can correlate with PFS and survival, for patients with MRD <10^-4 the rate at which residual leukemia cells proliferate could offer a better correlate of these endpoints, indicating the potential advantage of sequential MRD-testing.

Two issues regarding the predictive accuracy of uMRD are confounded:

• Firstly, not all residual leukemia cells are alike, they can have differing biological potential and propensity to cause disease recurrence.
• Secondly, there are differences in the ability of different therapies to eliminate residual leukemia sub-clones, both within individuals and across patient populations.

Lymphocyte doubling rate correlates with outcomes in patients with CLL treated with chlorambucil and with modern CLL therapies. Additional confounding co-variates associated with the rate at which uMRD is lost include age and risk of tumour lysis syndrome. Taken together, these data indicate that the predictive accuracy of uMRD can depend on therapy context and individual patient characteristics.

Conclusion

Overall, the information presented above demonstrate the complexity of this topic, with many perplexing covariates and impacting factors. The authors conclude that while uMRD may be a suitable endpoint for informing clinical decisions regarding certain patients, it is oftentimes unsuitable for informing clinical decision making and regulatory approvals in the context of CLL. They also indicate that MRD-testing results of are a predictive rather than prognostic co-variate and suggest that sequential monitoring using an accurate and precise MRD-test in large scale phase III patient trials may be required to fully elucidate the relationship between uMRD and PFS/survival.