Surveillance of the Immune Repertoire of Aplastic Anemia Patients Using Deep Sequencing

Cytotoxic T cells (CTL) responses have been implicated in the pathogenesis of idiopathic aplastic anemia (AA). In the past, CDR3 region of variable portion of B-chain (VB) of TCR has been analyzed by capillary electrophoresis and later by sequencing to identify clonal expansions or oligoclonal skewing of otherwise polyclonal TCR VB repertoire. CTL clonal expansions, most likely, a result of aberrant antigenic drive, lead to increased numbers of identical CDR3 sequences which can serve as clonal barcodes for individually expanded T cell clones. In AA oligoclonal skewing of VB TCR has been detected at presentation using manual clonotyping and interpreted as an overheated cellular response to putative autoimmune targets located within hematopoietic stem cell compartment. With the advent of NGS, the process of analysis of TCR repertoire has been improved: while manual sequencing allowed for characterization of only hundreds of VB CDR3 clonotypes, deep TCR sequencing affords incredible sequencing depth and therefore precision and resolution. We hypothesized that deep TCR sequencing (NGS clonotyping) can be used to identify changes in the T cell repertoire, including clonal expansions which would either qualitatively or quantitatively enable to us to predict subsequent responsiveness to immunosuppression (IST), propensity of relapse or potentially monitor the IST intensity.

We have identified 26 AA patients (56% severe AA) from whom samples were obtained at presentation and following IST with ATG/CsA including at recovery and relapse. In 10 patients serial samples were analyzed to allow for comparisons between clonal repertoire at presentation and in intervals following IST including when hematologic response was achieved. A hematologic response was seen in 88% of patients. PNH clone was present in 15/23 AA (AA/PNH).

Genomic DNA was sequenced according to the ImmunoSeq NGS protocols. A group of 587 controls with an average sequencing depth of 30± 11 x 10⁴ reads was used as reference. The average sequencing depth in AA patients was 74± 53 x10³ productive reads. The healthy individuals showed a Simpson's Diversity value (SDV) of .998 (i.e ., number of unique clonotypic reads/number of total clonotypic reads). A total of 341/587 (58%) healthy controls had >1 rearrangement >1% of total repertoire with a mean productive frequency of 2.4%± 2.1%.

The average SDV for AA was 0.993. Analyzing the frequency of productive rearrangements in patients at diagnosis, 22/26 patients had >1 rearrangement >1% of total repertoire with a mean productive frequency of 2.9%± 2.1%.

We then analyzed the homology of the amino acid sequence of each expanded TCRB clone >1%. If the corresponding CDR3 amino acid sequence was absent in any other individuals (AA patients or controls), we referred to it as a "private" clone. In contrast, "public" clones had CDR3 sequences, which were also present in other individuals; to that end, a clone was defined as "public expanded" if the corresponding productive frequency was >1% in any other control or patient. We estimated that of all the immunodominant clones found in AA patients, 41% were private, likely recognizing patient specific antigens, while 33% were shared only between AA patients (42% in 2, 32% between 3 and 5, 26%>5 patients, N=22) and thus were AA-related. Public specificities constituted 59% of expanded clonotypes. The public clonotypes likely recognize ubiquitous antigens, which also exist in a tolerant form in healthy individuals. Any expanded clonotypes present in controls even if also found in patients (at any frequency) are the least likely to be pathogenic.

We subsequently analyzed serial samples to assess the changes in size of the individual clones and establish a correlation between response and contraction of pathogenic clones. Average clonal contraction by 10.9 fold was observed in 50% of responders, and thus the remaining stable or expanding clones are unlikely pathogenic. In contrast, in non-responders only 33% of clones contracted by an average of 2.4 fold. Reappearance of originally expanded clones as well as new expanded clonotypes was observed upon relapse.

In sum, our studies indicate that deep TCR sequencing will provide intricate insights into the dynamics of autoimmune process in AA and in the future may be used for monitoring of the efficiency of IST in the clinic and for the study of potentially pathogenic clones in the laboratory.