Comprehensive Evaluation and Validation of a Next-Generation Sequencing Assay for Minimal Residual Disease Detection in T-Lymphoblastic Leukemia/Lymphoma

Introduction: Detection of minimal residual disease (MRD) in T-lymphoblastic leukemia/lymphoma (T-ALL) has been used in post-therapy monitoring and outcome prediction. Currently, multi-color flow cytometry is widely used for MRD detection with limit of detection (LOD) of 0.1 to 0.01% depending on the clinical implementation. However, flow cytometry assays require extensive training and experience for accurate interpretation and standardization is challenging. Next generation sequencing (NGS) assays with panels targeting the inherent polymorphism of the immunoglobulin genes have been applied for MRD detection in B cell neoplasms, including an assay for myeloma and B-ALL MRD detection developed by Adaptive Biotechnologies (Seattle, WA) that was granted FDA approval for clinical testing. We have published a similar approach for T-ALL MRD detection using the polymorphic T-cell receptor (TCR) genes. These assays offer higher sensitivity, easier standardization, and more objective interpretation. Here we present our comprehensive evaluation and validation of a T-ALL MRD assay using the Adaptive Biotechnologies immunoSEQ kit for potential clinical application.

Methods: NGS was performed using the Adaptive Biotechnologies immunoSeq kit and illumina NextSeq 500 according to manufacturer's instruction. Quantification of T cell repertoire was performed using the Adaptive Biotechnologies analysis pipeline. A single TCRβ sequence with ≥19% frequency of the T-cell population in diagnostic samples was defined as a T-ALL clone. Positive MRD was defined as detection of one or more such T-ALL sequences having a 100% DNA sequence match in the corresponding MRD samples. T-ALL leukemic cells and normal T cells were sorted from 2 bone marrow aspirates and peripheral blood from 20 healthy donors, respectively. T-ALL cells were spiked into pooled normal T cells to create a series of diagnostic T-ALL standards and MRD T-ALL standards down to 0.001%. Prior to assaying the standards, a T-ALL specimen was assessed at 11 different DNA input levels ranging from 3-1000ng, and interference of non-T-cell DNA was tested. For all other assays, 800ng DNA was used in two independent hsTCRB replicates for T-ALL clone identification in diagnostic samples and 2,400ng DNA was used in six independent hsTCRB replicates for T-ALL clone detection in MRD samples. Within run precision was assessed by assaying all standards and 3 pairs of diagnostic and MRD samples in triplicate. Between run precision was assessed by assaying 1 standard and 4 samples on two different days by the same technologist. Intermediate precision was assessed by assaying 3 pairs of diagnostic and MRD samples by two different technologists. LOD was determined as the lowest frequency at which MRD could be detected in MRD standards and Limit of Quantitation (LOQ) determined as the level below which the coefficient of variation (CV) >25% in MRD standards.

Results: The %CV for T-ALL quantification across the range of DNA input was 4.5%, suggesting independence of quantification from DNA input amount. Non-T-cell DNA has almost no impact on T-ALL quantification. T-ALL clones were detected with 100% concordance in diagnostic and MRD standards. The %CV of within and between run precision and intermediate precision were < 25% in T-ALL identification assay and all MRD detection assays except at 0.001% MRD, in which %CV was 43.8%. Quantification of T-ALL in both diagnostic and MRD standards were highly correlated with expected values (R² = 0.96, slope = 0.89 and R² = 1.00, slope = 0.88, respectively). T-ALL identification and MRD detection in specimens stored at 4°C or at room temperature for up to 14 days were within 25% CV as compared to fresh specimens. The assay was then performed on 40 diagnostic clinical T-ALL samples and T-ALL clonal sequences were identified in 34 samples (85%).

Conclusions: The immunoSEQ assay is evaluated and technically validated for clinical application, but requires validation in a larger cohort of clinical samples with concurrent outcome data. The assay is suitable for roughly 70% T-ALL where TCRβ is rearranged. The assay LOD for T-ALL MRD detection is 0.001% and LOQ is 0.01%, offering the possibility of improved sensitivity and more objective interpretation.