Identification and Monitoring of Pathogenic T Cell Clones Using Deep TCR Sequencing in T-Large Granular Lymphocyte Leukemia and Aplastic Anemia

Abstract 3477

The ability to identify and precisely quantitate self reactive T cell clones in immune-mediated bone marrow failure (BMF) may have wide ranging implications for the diagnosis and immune monitoring of these diseases. T cell large granular lymphocytic (T-LGL) leukemia is characterized by chronic lymphoproliferation of cytotoxic T cells (CTL) and is often associated with lineage-restricted cytopenias. As with idiopathic Aplastic Anemia (AA), it is generally believed that inhibition of hematopoiesis is CTL-mediated. Despite putative dissimilarity in target antigens and thereby cellular targets, T-LGL leukemia can serve as a monoclonal, and therefore simplified, model of otherwise polyclonal CTL-mediated AA. The recent identification of STAT3 mutations in both of these disorders further underscores the close pathogenic association (Jerez et al, ASH 2012 abstract). In the past, molecular analysis of the TCR repertoire using cloning, sequencing, and flow cytometric approaches allowed only a quantitatively limited insight into the clonotypic spectrum. Now, with the development of NGS technologies, the TCR repertoire can be analyzed at a previously unattainable level of sophistication and depth with potential to be adopted as a diagnostic tool.

Our laboratory has studied TCR clonal repertoires in 105 patients with LGL, 35 AA and controls (n=52) using traditional TCR VB sequencing. Beyond this, we have a subset of these patients (n=14, 4 samples analyzed serially) for which we have performed NGS of the TCR CDR3; 11 patients with CD8 T-LGL leukemia, 3 patients with AA, and 10 healthy controls. Among T-LGL patients, we selected 7 who shared HLA-B7 and erythroid suppression as presentation; 3 had Vβ17 (TRBV19) as the immunodominant clone by flow, and 5 tested positive for the STAT3 mutation. Several levels of bioinformatic analysis have been performed on TCR VB NGS data with an average depth of 1176823 sequences. First, we assessed the size of the dominant clones, which in T-LGL accounted for 57±34% of the TCR repertoire and 15±7% in AA. Next, using an index of diversity derived from the literature (Simpson, E. H. Nature 163:688, 1949), we were able to calculate an overall diversity score of the TCR repertoire for each patient and controls. As the formula to calculate diversity was essentially one minus the probability of selecting two identical sequences at random, a low diversity score indicated the presence of an expanded T cell clone. The diversity score was markedly different between healthy controls and T-LGL patients, mean 0.99±0.016 vs. 0.52±0.38, respectively (p=0.0004). Furthermore, in one T-LGL patient for which a baseline and a post therapy complete remission sample were available, the diversity score increased from 0.060 (pre) to 0.966 (post), illustrating the potential clinical utility of this approach to T cell repertoire diagnostics. Based on this finding, we next applied Vβ TCR NGS to three AA patients with the STAT3 mutation. The initial diversity score was well within the normal range (mean 0.99±0), however, post ATG therapy, measurements revealed that patients who relapsed demonstrated a decrease in the diversity index while the lone patient with a durable long term response to therapy maintained a consistent diversity score.

Assessment of antigenic specificity remains an elusive goal of TCR diagnositics. To address this issue, we first assessed the homology of the CDR3 regions between patients and controls and found that immunodominant clones present in T-LGL patients (>95% homology) were often present both in other patients as well as healthy controls, albeit at the basal level. Analysis by grouping amino acids based on side chain properties suggested that there are some similarities between CDR3 regions of the dominant clones in patients sharing HLAs. While these data are preliminary and require further investigation, it remains possible that the antigenic specificity of the dominant clone is not private, and, in some cases, there is a common antigen driving the initial expansion.

In sum, we conclude that deep TCR repertoire analysis of T-LGL and AA patients reveals the potential utility of a new technology in the diagnosis of disease and assessment of response to therapy. Further research in a much larger cohort of patients is warranted to comprehensively examine both the predictive and monitoring applications of Vβ TCR NGS in bone marrow failure syndromes.