Analysis of the Whole CDR3 T Cell Receptor Repertoire after Hematopoietic Stem Cell Transplantation in 2 Clinical Cohorts

Abstract

A major cause of morbidity and mortality for patients who undergo hematologic stem cell transplantation (HSCT) is acute graft-versus-host disease (aGVHD), a mostly T cell-mediated disease. Examination of the T cell receptor (TCR) repertoire of HSCT recipients and the use of next-generation nucleotide sequencing have raised the question of whether features of TCR repertoire reconstitution might reproducibly associate with aGVHD. We hypothesized that the peripheral blood TCR repertoire of patients with steroid-nonresponsive aGVHD would be less diverse. We also hypothesized that patients with GVHD who shared HLA might also share common clones at the time of GVHD diagnosis, thereby potentially providing potential clinical indicators for treatment stratification. We further hypothesized that HSCT recipients with the same HLA mismatch might share a more similar TCR repertoire based on a potentially shared focus of alloreactive responses. We studied 2 separate patient cohorts and 2 separate platforms for measuring TCR repertoire. The first cohort of patients was from a multicenter Phase III randomized double-blinded clinical trial of patients who developed aGVHD (NCT01002742). The second cohort comprised samples from biobanks from 2 transplantation centers and the Center for International Blood and Marrow Transplant Research of patients who underwent mismatched HSCT. There were no statistically significant differences in the TCR diversity of steroid responders and nonresponders among patients with aGVHD on the day of diagnosis. Most clones in the repertoire were unique to each patient, but a small number of clones were found to be both exclusive to and shared among aGVHD nonresponders. We were also able to show a strong correlation between the presence of Vβ20 and Vβ29 and steroid responsiveness. Using the Bhattacharya coefficient, those patients who shared the same HLA mismatch were shown to be no more similar to one another than to those who had a completely different mismatch. Using 2 separate clinical cohorts and 2 separate platforms for analyzing the TCR repertoire, we have shown that the sampled human TCR repertoire is largely unique to each patient but contains glimmers of common clones of subsets of clones based on responsiveness to steroids in aGVHD on the day of diagnosis. These studies are informative for future strategies to assess for reproducible TCR responses in human alloreactivity and possible markers of GVHD responsiveness to therapy.

Keywords: CDR3; Graft-versus-host disease; Stem cell transplantation; T cell; T-cell receptor.

Conflict of interest statement

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that be construed as a potential conflict of interest.

Copyright © 2020 American Society for Transplantation and Cellular Therapy. Published by Elsevier Inc. All rights reserved.

Figures

Figure 1.

Comparison of TCR repertoire diversity for patients who were either Responders or Non-Responders to steroid treatment of aGVHD. Patients are stratified by donor source HLA matching (Matched, Non-Matched, Partial-Matched) and relatedness (Related, Unrelated). In the dataset, there are no patients in the Related Non- Matched group who were non-responders. There were no statistically significant differences in repertoire. (A) Shannon Entropy; (B) Simpson Index; (C) Inverse Simpson.

Figure 1.

Comparison of TCR repertoire diversity for patients who were either Responders or Non-Responders to steroid treatment of aGVHD. Patients are stratified by donor source HLA matching (Matched, Non-Matched, Partial-Matched) and relatedness (Related, Unrelated). In the dataset, there are no patients in the Related Non- Matched group who were non-responders. There were no statistically significant differences in repertoire. (A) Shannon Entropy; (B) Simpson Index; (C) Inverse Simpson.

Figure 2.

Comparison of TCR repertoire diversity for patients enrolled in the BMT CTN 0802 clinical trial (NCT01002742) who were randomized to either MMF or placebo. There were no statistically significant differences in repertoire. (A) Shannon Entropy; (B) Simpson Index; (C) Inverse Simpson

Figure 2.

Comparison of TCR repertoire diversity for patients enrolled in the BMT CTN 0802 clinical trial (NCT01002742) who were randomized to either MMF or placebo. There were no statistically significant differences in repertoire. (A) Shannon Entropy; (B) Simpson Index; (C) Inverse Simpson

Figure 3.

Logarithmic frequency distribution for TCR clonotypes categorized by the number of patients in which they occur for a given stratum. TCR clonotypes with InDegree=1 occur in only one patient and are consistently the most numerous category. TCR clonotypes with InDegree>1 occur in more than one patient and as a group are consistently a small fraction of the total. The set of patients consisting of both Responders and Non-Responders to steroid treatment of aGVHD is stratified by donor source HLA matching (Matched, Non-Matched, Partial-Matched) and relatedness (Related, Unrelated). The Related Non-Matched stratum is excluded because it consists solely of Responders. (A) Related Matched; (B) Unrelated Matched; (C) Related Partial-Matched; (D) Unrelated Partial-Matched; (E) Unrelated Non-Matched.

Figure 3.

Logarithmic frequency distribution for TCR clonotypes categorized by the number of patients in which they occur for a given stratum. TCR clonotypes with InDegree=1 occur in only one patient and are consistently the most numerous category. TCR clonotypes with InDegree>1 occur in more than one patient and as a group are consistently a small fraction of the total. The set of patients consisting of both Responders and Non-Responders to steroid treatment of aGVHD is stratified by donor source HLA matching (Matched, Non-Matched, Partial-Matched) and relatedness (Related, Unrelated). The Related Non-Matched stratum is excluded because it consists solely of Responders. (A) Related Matched; (B) Unrelated Matched; (C) Related Partial-Matched; (D) Unrelated Partial-Matched; (E) Unrelated Non-Matched.

Figure 3.

Logarithmic frequency distribution for TCR clonotypes categorized by the number of patients in which they occur for a given stratum. TCR clonotypes with InDegree=1 occur in only one patient and are consistently the most numerous category. TCR clonotypes with InDegree>1 occur in more than one patient and as a group are consistently a small fraction of the total. The set of patients consisting of both Responders and Non-Responders to steroid treatment of aGVHD is stratified by donor source HLA matching (Matched, Non-Matched, Partial-Matched) and relatedness (Related, Unrelated). The Related Non-Matched stratum is excluded because it consists solely of Responders. (A) Related Matched; (B) Unrelated Matched; (C) Related Partial-Matched; (D) Unrelated Partial-Matched; (E) Unrelated Non-Matched.

Figure 4.

Jittered point plot of the geometric mean abundance for TCR clonotypes that appear in two or more patients (InDegree>1) categorized by the proportion of Responder patients to the total of those with the TCR clonotype. These shared TCR clonotypes may appear in Non-Responders only (0), Responders only (1), or both (between 0 and 1). The set of patients consisting of both Responders and Non-Responders to steroid treatment of aGVHD is stratified by donor source HLA matching (Matched, Non-Matched, Partial-Matched) and relatedness (Related, Unrelated). The Related Non-Matched stratum is excluded because it consists solely of Responders. (A) Related Matched; (B) Unrelated Matched; (C) Related Partial-Matched; (D) Unrelated Partial-Matched; (E) Unrelated Non-Matched.

Figure 4.

Jittered point plot of the geometric mean abundance for TCR clonotypes that appear in two or more patients (InDegree>1) categorized by the proportion of Responder patients to the total of those with the TCR clonotype. These shared TCR clonotypes may appear in Non-Responders only (0), Responders only (1), or both (between 0 and 1). The set of patients consisting of both Responders and Non-Responders to steroid treatment of aGVHD is stratified by donor source HLA matching (Matched, Non-Matched, Partial-Matched) and relatedness (Related, Unrelated). The Related Non-Matched stratum is excluded because it consists solely of Responders. (A) Related Matched; (B) Unrelated Matched; (C) Related Partial-Matched; (D) Unrelated Partial-Matched; (E) Unrelated Non-Matched.

Figure 4.

Jittered point plot of the geometric mean abundance for TCR clonotypes that appear in two or more patients (InDegree>1) categorized by the proportion of Responder patients to the total of those with the TCR clonotype. These shared TCR clonotypes may appear in Non-Responders only (0), Responders only (1), or both (between 0 and 1). The set of patients consisting of both Responders and Non-Responders to steroid treatment of aGVHD is stratified by donor source HLA matching (Matched, Non-Matched, Partial-Matched) and relatedness (Related, Unrelated). The Related Non-Matched stratum is excluded because it consists solely of Responders. (A) Related Matched; (B) Unrelated Matched; (C) Related Partial-Matched; (D) Unrelated Partial-Matched; (E) Unrelated Non-Matched.

Figure 5.

Scatterplot of TCR repertoire diversity on the day a patient was diagnosed with aGVHD versus the number of days post-transplant. The TCR repertoire diversity estimate and its bootstrap standard error are computed using the method of Chao et al. (20). The radius of each dot equals one standard error in the estimated value for each sample. The patients are from the GVHD cohort. (A) Shannon Entropy; (B) Simpson Index; (C) Inverse Simpson.

Figure 5.

Scatterplot of TCR repertoire diversity on the day a patient was diagnosed with aGVHD versus the number of days post-transplant. The TCR repertoire diversity estimate and its bootstrap standard error are computed using the method of Chao et al. (20). The radius of each dot equals one standard error in the estimated value for each sample. The patients are from the GVHD cohort. (A) Shannon Entropy; (B) Simpson Index; (C) Inverse Simpson.

Figure 6:

The patients are from the GVHD cohort. Plots to visualize trends over time in various diversity measures. In each panel, the horizontal axis is the post-transplant inception of GVHD measured in days and discretized into four time periods: 0 to 30 days, 30 to 60 days, 60 to 90 days, and 90 days or more. We transformed the Simpson and Inverse Simpson indices to symmetrize the respective distributions. (A) Number of samples representing the entire cohort; (B) Number of unique TCR clonotypes per sample; (C) Total number of TCR clonotypes per sample; (D) Shannon Entropy; (E) Simpson Index transformed by arcsine-square-root; (F) Inverse Simpson transformed by base-10 logarithm.

Figure 6:

The patients are from the GVHD cohort. Plots to visualize trends over time in various diversity measures. In each panel, the horizontal axis is the post-transplant inception of GVHD measured in days and discretized into four time periods: 0 to 30 days, 30 to 60 days, 60 to 90 days, and 90 days or more. We transformed the Simpson and Inverse Simpson indices to symmetrize the respective distributions. (A) Number of samples representing the entire cohort; (B) Number of unique TCR clonotypes per sample; (C) Total number of TCR clonotypes per sample; (D) Shannon Entropy; (E) Simpson Index transformed by arcsine-square-root; (F) Inverse Simpson transformed by base-10 logarithm.

Figure 6:

The patients are from the GVHD cohort. Plots to visualize trends over time in various diversity measures. In each panel, the horizontal axis is the post-transplant inception of GVHD measured in days and discretized into four time periods: 0 to 30 days, 30 to 60 days, 60 to 90 days, and 90 days or more. We transformed the Simpson and Inverse Simpson indices to symmetrize the respective distributions. (A) Number of samples representing the entire cohort; (B) Number of unique TCR clonotypes per sample; (C) Total number of TCR clonotypes per sample; (D) Shannon Entropy; (E) Simpson Index transformed by arcsine-square-root; (F) Inverse Simpson transformed by base-10 logarithm.

Figure 7:

The GVHD cohort patients are those for whom HLA allele mismatch data are available (24 patients). They are stratified into two groups: those with an HLA-0201 mismatch (6 patients) and those with a mismatch other than HLA-0201 (18 patients). (A) Logarithmic frequency distribution for TCR clonotypes categorized by the number of patients in which they occur. TCR clonotypes with InDegree=1 occur in only one patient and are consistently the most numerous category. TCR clonotypes with InDegree>1 occur in more than one patient and as a group are consistently a small fraction of the total (i.e. 2.5% or 2998/120282). (B) Jittered point plot of the geometric mean abundance for TCR clonotypes that appear in two or more patients (InDegree>1) categorized by the proportion of HLA-0201 mismatch patients to the total of those with the TCR clonotype who have any HLA mismatch. These shared TCR clonotypes may appear only in patients with a mismatch other than HLA-0201 (0), only in patients with an HLA-0201 mismatch (1), or both types of patients (between 0 and 1).

Figure 8:

The patients are from the Replicated HLA Mismatch cohort. Boxplots of Bhattacharyya coefficient (BC) distributions are plotted on a logarithmic scale to compare relatedness for TCR VβCDR3. The distributions for the case and negative control are similar to one another but both are shifted to substantially smaller values relative to the positive control. (A) Within the same patient at different time points (positive control, 59 BC values), patients with the same HLA mismatch (case, 189 BC values), and patients with different HLA mismatch (negative control, 5530 BC values); (B) Within the same patient at different time points (positive control, 59 BC values), patients with the same donor allele (case, 321 BC values), and patients with different donor allele (negative control, 1591 BC values).