CMV reactivation drives posttransplant T-cell reconstitution and results in defects in the underlying TCRβ repertoire

One of the major obstacles associated with hematopoietic stem cell transplant (HSCT) is the fact that this treatment, although curative for many diseases, is associated with significant toxicity and resultant transplant-related mortality (TRM).^1-4  TRM takes on many forms, but is often related to dysfunctional immune reconstitution after HSCT. The extent to which cytomegalovirus (CMV) influences posttransplant immune dysfunction has been the subject of intense interest,^1,5-9  with several studies documenting the increased risk of TRM based on CMV serostatus, infection, and expansion of CMV-tetramer–positive cells.^10-22  However, although the phenomenology of CMV’s impact on TRM is well documented, the causative molecular immunologic mechanisms remain unknown. To address these questions, we have undertaken a detailed assessment of immunologic reconstitution after HSCT using new deep-sequencing technologies that have allowed us to investigate this issue at a level of molecular detail not previously possible. We present evidence that, in a cohort of patients undergoing unrelated-donor transplantation, CMV reactivation is capable of resetting CD8 T-cell homeostasis, resulting in the clonal expansion of CMV-specific CD8 effector memory T cells (Tem). Importantly, T-cell receptor β (TCRβ) deep-sequencing analysis allowed us, for the first time, to look beyond the clonal expansions, and to evaluate the remainder of the TCR repertoire and, thereby, to assess the impact of CMV reactivation on the integrity of the repertoire as a whole. This analysis has revealed that CMV reactivation was associated with the development of defects in the underlying Tem TCR repertoire, providing the strongest molecular evidence to date linking CMV reactivation and quantitative immune dysregulation after HSCT.

Seventeen patients underwent prospective, calendar-based clinical and immune monitoring after enrollment on 2 contemporaneous clinical trials: (1) The Bone Marrow Immune Monitoring Protocol and (2) The Abatacept Feasibility Study. Both studies were institutional review board–approved and conducted between 2010 and 2013, as previously described.²³  Patient and transplant characteristics are shown in Table 1. Patients were analyzed prospectively for CMV reactivation by polymerase chain reaction (PCR), and those that reactivated CMV (>300 copies per mL whole blood) were treated with antiviral therapy according to institutional standards. All patients reactivating CMV developed viremia that was responsive to either valganciclovir or ganciclovir. No patient developed CMV disease. In addition to patient analyses, 10 healthy adult controls also underwent single time-point immune analysis and 7 healthy adult controls underwent TCRβ repertoire analysis.

The patients and controls described in this manuscript were enrolled in clinical trials that were conducted according to the principles set forth in the Declaration of Helsinki, and which were approved by the institutional review boards. Written informed consent was received from all participants.

Patients underwent longitudinal multiparameter flow cytometric analysis of immune reconstitution, CMV functional response, tetramer analysis, and flow cytometric sorting of CD8⁺ naive T cells (Tnaive), Tem, and Tetramer⁺ cells as described in supplemental Methods (available on the Blood Web site) and supplemental Tables 1-2.

Total genomic DNA was extracted and quantified (Qiagen) from unsorted peripheral blood mononuclear cells (PBMCs) and from sorted CD8⁺ Tnaive, Tem, and CMV-tetramer⁺ cells; TCRβ chain sequencing was performed at Adaptive Biotechnologies using the ImmunoSEQ platform^24-26  with primers specific for all 54 known expressed Vβ regions and all 13 Jβ regions. TCRβ deep-sequencing depth was sufficient to achieve at least fivefold coverage of every original template, sufficient to prevent sampling effects.²⁵  An average of 3.3 × 10⁶ ± 3.4 × 10⁵ reads were generated for the PBMC, Tnaive, and Tem samples (supplemental Table 3).

Bioinformatic analysis was performed on the sequencing data using the ImmunoSEQ platform, which included a determination of: the number and sequence of each of the productive unique Vβ and Jβ genes identified within each sample, the degree of clone sharing between samples, and the identity of the shared clones. In addition, clonality and TCRβ repertoire hole analysis was performed as described in detail in the supplemental Methods.

To identify predictors of high CD8⁺ Tem status, 2 statistical tests were used: for binary variables, we calculated the nonparametric Kendall τ-b statistic using the cor.test function in R (http://www.R-project.org/). For the 2 continuous variables (recipient age and donor age), we carried out Spearman rank correlation, also using R. Additional statistical analyses were calculated with the Mann-Whitney t test with 95% confidence intervals, using GraphPad Inc software, Prism version 5.0.

This study was based upon prospective, calendar-based clinical and immune monitoring of HSCT recipients enrolled on 2 contemporaneous clinical trials (Table 1). All patients received myeloablative pretransplant conditioning and unmanipulated unrelated-donor allografts.²³  Immune reconstitution was monitored in detail in all patients on-study, and revealed an extremely wide variability in the pace and character of their immune reconstitution, for which the heterogeneity was most prominent in the CD8⁺ Tem compartment, with over 200-fold variability in the first year posttransplant (Figure 1A). To determine the factors that were contributing to this variability, we performed univariate analysis of the following patient characteristics and posttransplant complications (Table 2): patient sex, age, donor age, degree of donor/recipient HLA matching, relapse, posttransplant GVHD prophylaxis regimen (cyclosporine/methotrexate vs cyclosporine/methotrexate/abatacept),²³  grade II-IV aGVHD, moderate-severe cGVHD, infections (including respiratory viruses and bacterial infections), and CMV reactivation. Univariate analysis with the Kendall τ-b test (for binary variables) or with the Spearman rank test (for continuous variables) was used to determine which variables were positively correlated with elevated CD8⁺ Tem counts. This analysis identified only CMV reactivation as predictive of day +365 CD8⁺ Tem count (P = .009 for CMV reactivation, P = not significant for all other parameters, Table 2). Of note, none of the clinical variables listed in Table 2 were predictive of total CD3⁺, CD4⁺, or CD8⁺ T-cell counts (not shown). Although this study was not sufficiently large to build a full multivariate model, we did perform partial least squares (PLS) discriminant analysis to determine the primary predictors of high CD8 Tem counts,²⁷  and this analysis also identified CMV reactivation as the primary driver of CD8 Tem expansion (not shown).

Figure 1B shows the kinetics of CMV reactivation in the +CMV cohort. The remaining patients (−CMV) did not demonstrate any CMV reactivation during the 1-year period of follow-up. As shown in the figure (purple bar), the mean day of first CMV reactivation in the +CMV cohort was 28 ± 4 days.

Although CMV reactivation did not impact the reconstitution kinetics of total white blood cells (WBCs), granulocytes, monocytes, total natural killer (NK) cells, lymphocytes, total T cells, B cells (Figure 1C), or total CD4⁺ T cells (Figure 1D), CMV reactivation was associated with significant expansion of CD8⁺ T cells, with a total CD8⁺ T-cell count at 1 year posttransplant that was more than fivefold higher in the +CMV compared with –CMV cohort (Figure 1D). Indeed, as shown in Figure 1E, the canonical inversion of the CD4:CD8 ratio occurred exclusively in CMV reactivating patients, with an increasingly inverted ratio observed in the +CMV cohort throughout the first year posttransplant, whereas –CMV patients progressively normalized this ratio. The inversion of the CD4:CD8 ratio in CMV-reactivating patients was validated in a larger cohort of 46 patients who underwent unrelated donor HSCT at Emory University and who received CD4 and CD8 T-cell counts at day +100 as a part of their clinical care (Figure 1F).

Figure 2A-C demonstrates that the reconstitution of total CD8⁺ T cells in the +CMV cohort was driven by a massive expansion of CD8⁺ Tem and, unexpectedly, with a contraction of CD8⁺ Tnaive. As shown in the figure, the reciprocal expansion/contraction of these 2 CD8 subpopulations occurred both in their relative proportions (Figure 2B) and in their absolute numbers (Figure 2C). This resulted in an average of >10-fold more CD8⁺ Tem in the +CMV cohort at day +365 compared with the −CMV cohort (P = .03) and fivefold fewer CD8⁺ Tnaive in the +CMV cohort at day +365 compared with the −CMV cohort (P = .02 at day +365). Importantly, as shown in Figure 2B, the inflection point of the CD8⁺ Tem expansion and the concomitant CD8⁺ Tnaive contraction closely correlated with the mean time to initial CMV reactivation (28 ± 4 days; purple bar). Indeed, as shown in this figure, the reconstitution of CD8 Tem was indistinguishable between the 2 cohorts prior to day +28, with the resetting of CD8 T-cell subpopulation homeostasis occurring in concert with the mean time to first CMV reactivation. The expanding CD8⁺ Tem in the +CMV cohort expressed many of the canonical markers of viral-reactive effector memory cells, including loss of CD28, gain of Granzyme B (Figure 2D), and expression of CD57 (Figure 2E).

To determine whether the massive expansion of CD8 Tem in +CMV patients was due to viral reactivation, we first determined whether these T cells were derived from T-cell clonal expansions (TCEs).^28,29  To accomplish this, we used TCRβ deep sequencing, and compared the clonality of T cells purified from +CMV and −CMV patients. Given the dichotomous effect that CMV reactivation had on the size of CD8⁺ T naive vs CD8⁺ Tem subpopulations (Figure 2B-C), it was critical to examine the TCR repertoire and clonality of purified naive and Tem subsets individually (Figure 3A), rather than determining the repertoire in unfractionated T cells, as has been previously performed.^30-32  As shown in Figure 3, TCR deep-sequence analysis of the purified Tnaive and Tem samples enabled us to develop novel insights into the impact of CMV reactivation on posttransplant TCR repertoire dynamics. Thus, as shown in the representative TCR landscape (Figure 3B) and in Figure 3C-E, the clonality of the TCR repertoire of each sample was calculated using both normalized Shannon entropy number³³  and Gini coefficients.³⁴  This analysis confirmed that CMV reactivation was accompanied by a striking contraction of the TCR repertoire diversity in the CD8⁺ Tem population, which was directly linked to the global expansion of the Tem compartment: thus, as shown in Figure 3D, a linear relationship was observed between the CD8 Tem clonality and the percentage of Tem for the +CMV cohort. Taken together, these data provide strong evidence for TCEs in the +CMV cohort.

Having established that +CMV patients developed TCEs posttransplant, we then used 3 assays of increasing sensitivity to probe the degree to which CMV-specific T cells accounted for these TCEs: first, we enumerated CMV-tetramer⁺ cells in both the +CMV and −CMV cohorts (Figure 4A-B). Second, we measured the proportion of polyfunctional (interferon γ [IFNγ] + tumor necrosis factor [TNF]) CMV-peptide–responsive CD4⁺ and CD8⁺ T cells after transplant³⁵  in each cohort (Figure 4C-D). These 2 assays demonstrated that CMV-reactivating patients possessed significantly more CMV-specific and CMV-responsive T cells compared with nonreactivating patients.

To most rigorously define the degree to which Tem clonal expansion was driven by CMV-specific T cells, we performed a tetramer-specific sorting experiment, followed by high-throughput TCRβ repertoire sequencing of these highly purified CMV-specific populations (supplemental Table 4; Figure 5A-B). Tetramer⁺ cells were sorted from day +365 samples from 4 CMV-reactivating patients (no cells were available from 1 patient): 001-001 (B*08 tetramer), 001-008 (A*24 tetramer), 0001-009 (A*02 tetramer), and 002-002 (A*02 + B*07 tetramers) (supplemental Table 4; supplemental Table 2). The analysis demonstrated that although each patient expanded a highly unique set of CMV-specific T-cell clones (with little clone sharing between patients) for each individual patient, there was an extremely high degree of overlap between the expanded Tem and tetramer⁺ clones (Figure 5A). Moreover, as shown in Figure 5B, enrichment for the Tetramer⁺ TCRβ clones in the most highly expanded Tem clones occurred. To further delineate this overlap, we mapped the 10 most highly expanded Tem clones onto the 10 most highly expanded tetramer⁺ clones (Table 3). This analysis also documented high clone sharing, which is perhaps best illustrated for patient 001-008, in which 7 of the 10 most highly expanded Tem TCRβ clones were also tetramer+. Taken together, these data provide strong molecular evidence that the expansion of CD8⁺ Tem in +CMV patients was driven by the clonal expansion of CMV-specific T cells. Moreover, the TCRβ sequences of the expanded and shared Tem and Tetramer⁺ T-cell clones from each patient (Table 3), identified 17 novel TCRβ sequences for which this study provides new evidence for linked functional expansion and CMV specificity. Although these clones likely represent only a portion of the entire CMV-specific repertoire,^36,37  to our knowledge, this comparative TCRβ deep-sequencing data constitutes the first quantitative molecular link between posttransplant CD8 Tem reconstitution and CMV-specific T cells, and strongly supports a model of oligoclonal T-cell expansion in the setting of CMV reactivation.

Although previously available technologies were able to document the impact of CMV reactivation on the clonality of the T-cell repertoire,^38,39  these methods (including flow cytometry–based V-J tracking and spectratyping) are not sensitive or specific enough to interrogate the TCR landscape that underlies the highly clonally expanded CMV-specific T-cell clones. In the current study, we were able to intensively probe, for the first time, the impact of CMV reactivation on the CD8 Tem TCR landscape in toto and, specifically, to determine whether CMV reactivation was associated with specific deficits or “holes” in the repertoire. To test this hypothesis, we first created 2 reference V-J family distributions: (1) a reference repertoire derived from unique CD8 Tnaive clones of 7 healthy adults between the ages of 26 and 57 years (supplemental Table 5) which was termed the VJ^r and (2) a reference derived from naive CD8 T cells of the transplanted patients at day +365 (which was termed the VJ^r-transplant). The rationale for the choice of Tnaive to create the VJ^rs was that a healthy adult Tnaive repertoire would be expected to be the most representative of the breadth of V-J fractions (rather than deriving the VJ^r from memory subsets, which, even in normal individuals, would be expected to display skewing for selected V-J fractions given previous antigen experience). To estimate the number of holes in the Tem compartment of each patient, we subtracted the V-J family distribution for each of the patient-specific query repertoires (VJ^q, see supplemental Methods) from either the VJ^r or the VJ^r-transplant. This exposed deficiencies in the former relative to each reference, and holes were deemed to exist in V-J families where such deficiencies exceeded the mean for VJ^r. Hole analysis was confined to the Tem compartment (instead of the Tnaive) because it was only in the CD8 Tem that TCEs occurred (Figures 3 and 5; Table 3) and, thus, it was in this compartment that the issue of holes was most relevant. As shown in Figure 6A-D, CMV reactivation was associated with a compromised CD8 Tem TCR repertoire, in which the TCR repertoire of +CMV patients demonstrated significantly more holes than were present in −CMV patients.

To investigate the potential confounding contribution of GVHD to holes in the repertoire, 2 additional analyses were performed. First, as shown in Figure 6E, −CMV patients were divided into those that did not develop significant aGVHD or cGVHD (patients 001-002, 002-001, 003-001, 004-001) and those that did develop significant GVHD (patients 001-005, 001-006, 003-003). As shown in the figure, this analysis documented no difference in the number of holes in −CMV patients based on GVHD status. Second, as shown in Figure 6F, we compared the number of holes in the subset of patients who all developed significant aGVHD or cGVHD, based on their CMV status, comparing +GVHD/+CMV patients (001-001, 001-008, 001-009, 002-002, 003-002) to +GVHD/−CMV patients (001-005, 001-006, 003-003). As shown in the figure, this subset analysis recapitulated the results in Figure 6C-D, documenting more holes in the +CMV/+GVHD patients compared with the −CMV/+GVHD patients. This constitutes the first exhaustive analysis of the underlying TCR repertoire after CMV reactivation, and provides the strongest evidence to date for significant molecular immune compromise in these patients.

Given the observed defects in the TCR repertoire in the setting of CMV reactivation, we queried our data set to investigate whether a connection existed between reactivation and naive T-cell reconstitution as well as thymic output. We found that although patients reactivating CMV demonstrated specific expansion of CD8 Tem (Figure 2B-C), they demonstrated a global contraction of both CD4 and CD8 Tnaive (Figures 2B and 7A). In addition, as shown in Figure 7B, these patients demonstrated a progressive loss of CD31⁺CD4⁺ Tnaive, a subpopulation that has been previously shown to be enriched in new thymic emigrants.^40,41  These results are consistent with thymic compromise in CMV-reactivating patients, an observation that has previously been shown in humanized murine models.^42,43  Although Figures 2B and 7A-B document the impact of CMV reactivation on Tnaive reconstitution, when patients were dichotomized by the presence or absence of GVHD, no such correlation with Tnaive reconstitution was observed (Figure 7C-E).

We have performed an analysis of the impact of CMV reactivation on posttransplant immune reconstitution at an unprecedented level of cellular and molecular detail. Given that immune reconstitution must occur de novo in HSCT patients, and that these patients are at high risk for CMV reactivation, studying this patient population with the tools of multiparameter flow cytometry, cell sorting, tetramer-based cell purification, and TCRβ deep sequencing provides a rigorous time-collapsed experiment on the impact of CMV on global immune phenotype and function. Moreover, the ability to correlate changes in T-cell composition and reconstitution with the inflection point of viral reactivation in these patients further strengthens the mechanistic links that can be made between CMV reactivation and immune homeostasis. Although other reactivating viruses (eg, Epstein-Barr virus, herpes simplex virus) were not studied, and may also impact immune reconstitution, these experiments provided quantitative molecular evidence for CMV-specific T-cell clonal expansion and, importantly, for the development of holes in in the underlying T-cell repertoire in the setting of CMV reactivation.

In this study, 17 patients underwent detailed longitudinal immune analysis, which included both phenotypic and functional assays, as well as exhaustive TCRβ deep sequencing of sorted Tem, Tnaive, and Tetramer⁺ cells. In addition, 10 healthy controls were used as a comparator group for the longitudinal flow cytometric analysis and an additional 7 healthy controls were used to create the reference TCRβ repertoire. Although the number of transplanted patients that were exhaustively analyzed in this study is still relatively small, several factors strengthen the conclusions that we have been able to draw: first, each of the patients underwent detailed longitudinal analysis, with 10 time points analyzed per patient (Figures 1-2). This provided significant intrapatient controls for the longitudinal evaluation, and allowed a detailed correlation of changes in T-cell reconstitution with the inflection point of CMV reactivation. Second, a larger cohort of 46 patients, who underwent standard day 100 CD4 and CD8 counts at our institution, were also evaluated for CD4:CD8 ratio (Figure 1F), and confirmed the results that were obtained with the 17 exhaustively analyzed patients. Given the pivotal importance of the expansion of CD8 T cells to the results in this study, this external validation was of critical importance. Finally, the depth of the analysis that we performed on the patients enrolled provided a great deal of statistical power: high statistical significance was obtained for the flow cytometric as well as the deep-sequencing results where a mean of >3 million reads per sample was obtained. Together, these support the high fidelity of the data set used in the current analysis.

Our results support a model wherein CMV reactivation can drive posttransplant T-cell reconstitution. Indeed, several of the traditional hallmarks of T-cell reconstitution, often attributed to post-HSCT immune reconstitution per se,^44-51  occurred exclusively in those patients that reactivated CMV. These included the inversion of the CD4:CD8 T-cell ratio, as well as the enhanced expression of the terminal differentiation marker CD57^52-55  on CD8⁺ T cells in the +CMV cohort.

TCRβ deep sequencing enabled measurement of the impact of posttransplant CMV reactivation on immune reconstitution at a level of detail unavailable with previous technologies. This included the identification of T-cell clonal expansions in the setting of CMV reactivation, as has been previously documented.^32,56  Importantly and unique to the current study, the linkage of CD8 Tem deep sequencing with CMV Tetramer⁺ deep sequencing provided incontrovertible evidence that the (often extreme) expansion of CD8 Tem (Figure 2B-C) in the setting of CMV reactivation was due to patient-specific CMV-directed clones (Figure 5A-B), rather than to nonspecific global expansion of Tem.

Although the analysis of the TCEs provided several novel insights into the impact of CMV on immune reconstitution posttransplant, perhaps the most important insights concerned the TCR repertoire that lied “underneath” the clonal expansions. Thus, detailed molecular analysis of this repertoire in toto has been previously unattainable, given technical limitations of TCR spectratyping or targeted sequencing. However, the deep-sequencing techniques applied in this study have allowed us to evaluate in detail the structure of the entire TCR repertoire and the impact of CMV reactivation on this structure. Given the documented existence of CD8 Tem TCEs in the setting of CMV reactivation, 2 mutually exclusive scenarios with respect to the remainder of the TCR repertoire were possible: In the first scenario, TCEs exist, but do not alter the topology of the underlying repertoire, implying intact T-cell immunity despite clonal expansions. In the second, TCEs exist concomitant with defects in the underlying reservoir: this scenario predicts defects in T-cell immunity. As shown in Figure 6, in the current study, we were able to measure, for the first time, the impact of CMV-specific TCEs on the underlying CD8 Tem immune reservoir and find that there were significantly more holes in the Tem reservoir of CMV-reactivating patients than in those that did not reactivate virus. This provides the first molecular link between CMV reactivation and a detrimental impact on subsequent protective immunity in transplant patients.

The underlying cause of the defects in the TCR repertoire in +CMV HSCT patients remains unknown, but 2 possible hypotheses have been considered. In the first, the TCEs caused by CMV reactivation result in competition for T-cell “space” in a scenario in which there is a zero-sum game for the number of T-cell clones that can exist in any individual.⁵⁷  Although this scenario is possible, 3 lines of evidence suggest that it is unlikely. First, mouse models have shown that the number of CD8 Tem in the mammalian host adapts according to immunologic experience.⁵⁸  Second, holes in the repertoire were not observed in aging CMV-seropositive individuals (R.M., Paul Lindau, C.D., Jeanne DaGloria, Heidi Utsugi, Edus H. Warren, Stanley R. Riddell, Karen W. Makar, Cameron J. Turtle, H.S.R., manuscript submitted April 2015), providing evidence that TCEs alone do not necessitate defects in the repertoire. Finally, CMV-reactivating transplant patients demonstrated significant growth in their total CD8⁺ and CD8 Tem compartment (Figures 1D and 2C), suggesting a large capacity for CD8 expansion in these patients.

The second hypothesis concerns the health of the thymus in CMV-reactivating patients, and suggests that thymic damage may result in defects in T-cell immunity. Indeed, one of the unexpected observations made in the present study was that, although CMV reactivation drove a specific expansion of CD8⁺ Tem cells, it was associated with a global contraction of both CD4 and CD8 Tnaive, accompanied by significantly fewer CD31⁺CD4⁺ putative recent thymic emigrants.^40,41  Although the addition of T-cell receptor excision circles (TREC) analysis^59,60  (which was not feasible given sample constraints in the current study) would strengthen this data set, the flow cytometric observations suggest that a defect in thymopoiesis may have been linked to CMV reactivation and to the CMV-driven expansion of CD8⁺ Tem. Indeed, there is supportive precedent in the literature for this, as previous studies have documented that CMV can infect thymic epithelium and that activated and effector T cells can directly infiltrate and damage the thymus.^43,61-63  Moreover, a previous study of young adults, thymectomized as children, showed that the combination of the lack of thymic function and CMV infection lead to dramatic changes in T-cell homeostasis.⁵⁵  These data suggest the novel hypothesis that CMV-associated thymic damage may be one of the contributory mechanisms for the immune compromise that accompanies CMV reactivation after transplant.

In determining the impact of CMV reactivation on posttransplant immune reconstitution, it is critical to also consider the potential impact that GVHD can make on thymic and immune dysfunction, given the fact that GVHD and CMV reactivation are often linked. Indeed, previous studies often^59,60  but not always^64,65  have documented decreased TRECs in patients that have developed cGVHD. In the current study, we have documented that (1) +CMV patients had more CD8 Tem repertoire holes than −CMV patients (Figure 6A-D); (2) when a subanalysis of only patients that developed GVHD was performed, there were more holes in the CD8 Tem in the +CMV/+GVHD patients compared with the −CMV/+GVHD patients (Figure 6F); (3) in contrast, there was no increase in holes when −CMV/−GVHD patients were compared with −CMV/+GVHD patients (Figure 6E); and (4) when flow cytometric analysis of Tnaive reconstitution was dichotomized on the basis of CMV reactivation, defective reconstitution in CMV-reactivating patients was apparent (Figures 2B-C and 7A-B), but these differences were not observed when patients were dichotomized based on GVHD (Figure 7C-E). It is important to note, however, that our study was not sufficiently powered to perform the final subanalysis, to specifically determine whether patients that experienced both CMV reactivation and GVHD (often linked together) displayed more repertoire defects than patients who had reactivated CMV but had not developed GVHD. Given the evidence in the current study for CMV-induced TCEs and holes in the CD8 Tem reservoir, and previous data supporting the impact of GVHD on thymic dysfunction,^59,60  we believe that it is possible that there could be a combinatorial effect of both GVHD and CMV reactivation, such that patients that experience both complications would demonstrate the most severe defects in protective immunity. This represents a critical area for future analysis with a larger patient cohort.

This study has provided convincing molecular evidence that CMV is a major driver of CD8⁺ Tem expansion posttransplant, and for a linked appearance of defects in the CD8 Tem TCRβ repertoire amid contraction of the Tnaive compartment. The implications of this study are provocative, given newly available agents being tested for primary CMV prophylaxis during HSCT.¹⁵  Our results predict that preventing CMV reactivation will profoundly impact immune reconstitution after transplant, and that although quantitative CD8 reconstitution may be slower with CMV prophylaxis, qualitative T-cell reconstitution should be significantly improved.

The authors thank Dr Stan Riddell (Fred Hutchinson Cancer Research Center) and Dr Louis Picker (Oregon National Primate Research Center) for helpful discussions about the data.

This work was supported by the Katz Foundation, the CURE Childhood Cancer Foundation, a Food and Drug Administration grant FD004099 (L.S.K.), the National Institutes of Health, National Cancer Institute grants R01 CA72669 and P01 CA 065493 (B.R.B.), and National Institute of Allergy and Infectious Diseases grant R01 AI081918 (B.R.B.).

Contribution: Y.S. performed experiments, analyzed data, and wrote the paper; R.M., A.D.K., B.R.B., A.K.M., and H.S.R. analyzed data and wrote the paper; B.W., K.F., C.D., and J.R. analyzed data; D.T.K. performed the clinical trial and performed experiments; L.S., J. A. Cheeseman, J. A. Conger, and A. Garrett performed experiments; A. Grizzle performed the clinical trial; J.T.H., A.L., M.Q., H.J.K., and E.K.W. performed the clinical trial and wrote the paper; and L.S.K. conceived the study, performed the clinical trial, analyzed data, and wrote the paper.

Conflict-of-interest disclosure: H.S.R. owns stock and receives consulting fees from Adaptive Biotechnologies. K.F. and C.D. are employees of Adaptive Biotechnologies with salary and stock options. The remaining authors declare no competing financial interests.

The current affliation for Linda Stempora, Jennifer A. Cheeseman, and Allan D. Kirk is Department of Surgery, Duke University School of Medicine, Durham NC.

Correspondence: Leslie S. Kean, Ben Towne Center for Childhood Cancer Research, Seattle Children’s Research Institute, Seattle, WA 98101; e-mail: leslie.kean@seattlechildrens.org.