Ultra-Sensitive Next-Generation Sequencing Establishes the Prognostic Value of Very Low MRD in Adults with Acute Lymphoblastic Leukemia Undergoing Hematopoietic Cell Transplantation

BACKGROUND

The prognostic utility of measurable residual disease (MRD) in adult acute lymphoblastic leukemia (ALL) has been defined at a sensitivity of 10^-4. Measurement of MRD by ultra-sensitive next-generation sequencing (NGS) of Ig and/or TCR rearrangements has a sensitivity of 10^-6 and is increasingly used across the United States to evaluate treatment response. However, the clinical utility of very low level NGS-based MRD (<10^-4) is unknown. In this large multicenter cohort of adult ALL patients who underwent allogeneic hematopoietic cell transplant (HCT), we evaluated the prognostic value of ultra-sensitive NGS-based MRD.

METHODS

Adults (18+) with ALL who underwent HCT at Stanford University or Oregon Health and Science University (OHSU) between January 2014 and April 2021 with planned MRD monitoring using the NGS-based clonoSEQ assay (Adaptive Biotechnologies, Seattle, WA) were included. Pre-HCT MRD (MRD_pre) and post-HCT MRD (MRD_post) were followed for up to one-year post-HCT. MRD was evaluated as residual clonal cells per million nucleated cells and was defined as undetectable (0), low (<10^-4), high (≥10^-4 to ≤10^-3) and very high (>10^-3). Patients were followed for leukemia relapse for up to two years. Competing-risks analyses were used to model the hazard of relapse according to MRD_pre and MRD_post levels. A univariate analysis using a Fine-Gray subdistribution hazard competing risk regression model was used to estimate the effect of MRD_pre level on the cumulative incidence of relapse. A Cox regression model with time-varying covariates was fit to evaluate the association between MRD_pre and MRD_post levels with relapse.

RESULTS

158 ALL patients who underwent HCT (Stanford n = 116, OHSU n = 42) had a trackable clonoSEQ clone. The median age was 43 years (range 18-77); 136 patients (86%) had B-ALL, 22 (14%) had T-ALL, and 131 (83%) underwent myeloablative HCT. Of the 122 patients with MRD_pre assessment, MRD_pre was undetectable, low, high, and very high in 82 (67%), 24 (20%), 11 (9%), and 5 (4%), respectively. Of the 111 patients with both MRD_pre and at least one MRD_post assessments, 38 had at least one positive MRD_post result.

The cumulative incidence of relapse differed significantly according to level of MRD_pre (Figure 1), including among patients with low MRD_pre of <10^-4 where post-HCT relapse risk was strikingly higher than in patients with undetectable MRD_pre (HR 3.56, 95% CI 1.39-9.15, p = 0.01). A very strong association with post-HCT relapse was also seen in patients with detectable MRD_post at any level (including <10^-4), even after controlling for MRD_pre (HR 4.41, 95% CI 2.90-6.70, p < 0.0001).

Finally, we found that patients entering HCT with undetectable MRD_pre who remained undetectable through the first year of MRD_post monitoring had <10% likelihood of relapse in the second post-HCT year (Figure 2). In contrast, patients entering HCT with detectable MRD_pre had an ongoing risk of continued detectable of MRD_post without a clear plateau of relapse risk post-HCT.

CONCLUSION

In this analysis across two large transplant centers, we found that detection of MRD by ultra-sensitive NGS at a level of 10^-6 offers significant prognostic value in adults with ALL undergoing HCT. Detectable MRD_pre even at a low level of <10^-4 increases the risk of post-HCT relapse. The presence of MRD_post at any level following HCT also substantially increases the risk of relapse. Finally, we found that patients with undetectable MRD_pre who continue to have undetectable MRD_post through the first year of monitoring have excellent post-HCT outcomes.

*ECL, SD, and JS contributed to this work equally