Impact of Baseline and Week 2 and Week 4 Posttransplant CMV Cell-Mediated Immunity on Risk of CMV Infections and Mortality in Recipients of Allogeneic Hematopoietic Cell Transplant

Abstract

Background

Cytomegalovirus (CMV) infection is a common opportunistic infection after allogeneic hematopoietic cell transplant (alloHCT). We explored whether a change in CMV cell-mediated immunity during the first month after transplant predicts the risk of development of CMV infection and all-cause mortality.

Methods

This follow-up analysis is based on data from the REACT study, a multicenter prospective observational study of recipients of alloHCT who were CMV-seropositive. Production of interferon γ following ex vivo stimulation with CMV antigens IE1 (immediate early 1) and pp65 (phosphoprotein 65) was assessed by CMV ELISPOT assay at baseline and 2 and 4 weeks after transplant. Clinically significant CMV infection (CS-CMVi) was defined as CMV viremia and/or disease necessitating antiviral therapy. We evaluated the impact of CMV CMI changes on the risk of CS-CMVi and post transplant mortality.

Results

The analysis included 226 recipients of alloHCT with CMV cell-mediated immunity data at baseline and 2 and/or 4 weeks after transplant. CS-CMVi occurred in 64 patients (28%). On Cox regression analyses, independent predictors of CS-CMVi included a negative Δ change from baseline to week 2 of pp65 spot counts (hazard ratio, 3.65 [95% CI, 1.65–8.04]; P = .001) to week 4 of IE1 spot counts (hazard ratio, 2.79 [95% CI, 1.46–5.35]; P = .002), anti-thymocyte globulin conditioning regimen, type of transplant, female sex, and corticosteroid use. Kaplan-Meir analysis showed a significant association of a negative IE1 change from baseline to week 4 and increased all-cause mortality after transplant (log rank test = 0.041).

Conclusions

A decrease in CMV-specific T-cell responses during the first month after transplant may predict CS-CMVi and is associated with all-cause mortality in recipients of alloHCT.

Cytomegalovirus (CMV) infection is one of the most common infectious complications affecting recipients of allogeneic hematopoietic cell transplant (alloHCT) [1]. Control of CMV replication relies on cell-mediated immunity (CMI), especially CMV-specific CD4+ T-cell response, which precedes the development of CMV-specific antibodies and cytotoxic CD8+ T cells for control of viral replication [2, 3].

CD4+ and CD8+ T cells are implicated in controlling latent CMV infection through a complex interaction involving a CD8+ T-cell response producing interferon γ, interferon α, and a host of other cytokines in response to the presence of CMV. Measuring CMV CMI may be a useful tool to predict the risk of CMV infection or disease [4]. The routine use of CMV CMI assays is hindered by (1) limited access to commercially available assays, (2) lack of standardization across platforms (eg, enzyme-linked immunospot [ELISPOT] assays and enzyme-linked immunoassays), and (3) remaining questions related to the utility of information from CMV CMI assays in relation to posttransplant outcomes [5]. CMV ELISPOT assays assess CMV CMI by measuring the release of interferon γ by CD4+ and CD8+ T cells in vitro from isolated peripheral blood mononuclear cells [6]. Evaluation of the clinical utility of CMV ELISPOT assays showed promising results for predicting CMV infections in recipients of alloHCT and solid organ transplant [7–10].

We previously reported that among recipients of alloHCT who were CMV-seropositive and enrolled in a prospective observational study, low CMV CMI within the first 6 months after transplant (as measured by a CMV ELISPOT assay) was associated with an increased risk of clinically significant CMV infection (CS-CMVi) [9]. Furthermore, we noted that among patients with low CMV CMI, the mortality rate within 6 months posttransplant was significantly higher among those who did experience CS-CMVi than among those who did not (37% vs 12%) [9]. In the present analysis, we explored the relationships between the Δ changes in CMV CMI from baseline (before transplant) to specific time points during the first month after transplant and the occurrence of CS-CMVi as well as all-cause mortality.

METHODS

Study Design

This post hoc analysis was based on data from the REACT study [9], a multicenter prospective observational trial conducted at 13 transplant centers in the United States, Canada, and Europe to determine the performance of a CMV ELISPOT assay in predicting immune competence against CMV reactivation after alloHCT. Full details of the trial design and the study population were previously published [9]. Briefly, all enrolled patients underwent an alloHCT from matched or haploidentical related donors or from matched or mismatched unrelated donors. Pediatric patients (<18 years) were excluded. Enrolled patients provided a blood sample within 14 days prior to alloHCT (baseline), at days 14 and 28, and then every 2 weeks for up to 6 months after transplant. At each time point, CMV ELISPOT assay and a central laboratory measurement of CMV viral load by polymerase chain reaction were performed through the Roche Cobas platform (lower limit of detection, 137 IU/mL). CMV reactivation and CMV disease were defined per Ljungman et al [11]. The primary end point was detection of the first CS-CMVi, defined as the first CMV reactivation or CMV disease after hematopoietic cell transplant (HCT) necessitating the start of anti-CMV therapy by each center's institutional protocols [12]. CS-CMVi was assessed subsequent to each time point. All-cause mortality was an exploratory end point. For all the analysis evaluating the impact of CMV CMI after week 2 or 4, CS-CMV infection and mortality that occurred after those time points were included in the analysis. CMV ELISPOT assay was performed in accordance with validated test procedures, which are described in the report of the REACT study [9]. None of the patients received primary prophylaxis for CMV.

In the REACT study, CMV CMI (CMV-specific T-cell activity) was assessed by determining the production of interferon γ following ex vivo stimulation with 2 CMV antigens: immediate early 1 (IE1) and phosphoprotein 65 (pp65). Results were reported in spot counts (SPCs) per 250 000 cells for IE1 and pp65. When the nil control SPCs were <10 per 250 000 cells, they were subtracted from the SPCs obtained for IE1 and pp65. For the current analysis, SPCs were categorized as low if they were ≤100 per 250 000 cells and high if >100 per 250 000 cells, according to our previous results [9]. Changes in SPCs for IE1 and pp65 between baseline and 2 and 4 weeks after transplant were categorized as a positive Δ change if the week 2 or 4 count was greater than or equal to the baseline count, and negative Δ change if the week 2 or 4 count was less than the baseline count. In addition to CMV CMI results, data obtained from medical records included patient demographics, medical history, relevant transplant-related data, corticosteroid use for other than premedication purposes, and laboratory results (including CMV polymerase chain reaction results).

Patient Consent Statement

The protocol was approved by the institutional review board of each site, and all patients signed an informed consent form.

Statistical Analysis

Categorical variables were compared with chi-square or Fisher exact test. Continuous variables were compared through the Wilcoxon rank sum test. Cox proportional hazards models were used to identify factors independently associated with CS-CMVi development and to evaluate the impact of CMV CMI change from baseline to posttransplant on the risk of CS-CMVi development and mortality. For analyses of the impact of CMV CMI on the risk of development of CS-CMVi, separate models were used for week 2 and 4 analyses. Correlations between IE1 and pp65 changes were evaluated via Spearman rank-order correlation analysis. Survival curves of the same patients were estimated by the Kaplan-Meier method and compared with the log-rank test. All tests were 2-sided with a significance level of 0.05. Statistical analyses were performed with SAS version 9.4 (SAS Institute).

RESULTS

This multicenter study followed 241 recipients of alloHCT who were CMV-seropositive. After 2 recipients of blood cord transplant were excluded, 226 patients had results for CMV CMI at baseline and at 2 and/or 4 weeks after transplant and were included in these analyses. The clinical and demographic characteristics of the 226 patients are summarized in Table 1. Sixty-four patients (28%) had a CS-CMVi during the 6-month follow-up period. Of these 64 patients, 10 (15%) had CMV end organ disease. Most cases of CS-CMVi (57/64) occurred within the first 75 days after HCT (median, 36; IQR, 30–57). Patients with and without CS-CMVi differed in terms of demographic and clinical characteristics, including female sex, receipt of anti-thymocyte globulin (ATG), acute graft-vs-host disease, and corticosteroid use, which were significantly more common in patients with CS-CMVi.

CMV CMI at Specific Time Points and CS-CMVi Risk

For the Cox regression analyses of the impact of CMV CMI at specific time points on CS-CMVi risk, 2 models were constructed: 1 for baseline and week 2 and 1 for baseline and week 4. In both models, low pp65 SPCs at week 4 were associated with an increased risk for CS-CMVi (adjusted hazard ratio, 2.57 [95% CI, 1.11–5.94]; P = .028). IE1 SPCs at any time point were not predictive of CS-CMVi (Supplementary Table 1).

Changes in CMV CMI and CS-CMVi Risk

To determine the effect of Δ changes in CMV CMI on the risk for CS-CMVi, changes were analyzed separately for each antigen (pp65 and IE1) from baseline to week 2 and baseline to week 4 and then correlated with CS-CMVi. Data were available for 152 patients for the week 2 analysis and 154 patients for the week 4 analysis. The median time from sample collection and HCT (baseline) was −3 days (IQR, −5 to −1).

Between baseline and week 2, IE1 had a median SPC change of 0 (IQR, −28 to 0), and pp65 had a median SPC change of −14 (IQR, −175 to 0). Seventy percent (30/43) and 79% (34/43) of patients with a CS-CMVi and 39% (42/109) and 45% (49/109) without a CS-CMVi had negative Δ changes in IE1 SPCs (P = .0005) and pp65 SPCs (P = .0001), respectively (Table 2). Between baseline and week 4, IE1 had the same median SPC change of 0 (IQR, −17 to 8), and pp65 also had a median SPC change of 0 (IQR, −146 to 41). In total, 63% (25/40) and 65% (26/40) of patients with a CS-CMVi and 33% (38/114) and 39% (44/114) without a CS-CMVi had negative Δ changes in IE1 SPCs (P = .001) and pp65 SPCs (P = .004), respectively (Table 2).

On Cox regression analyses, independent predictors of CS-CMVi were a negative Δ change for pp65 from baseline to week 2 and a negative Δ change for IE1 from baseline to week 4 (Tables 3 and 4). When CMV CMI was treated as a continuous variable, Cox regression analysis showed that for every decrease of pp65 SPCs by 50 per 250 000 cells from baseline to week 2, the hazard ratio of developing CS-CMVi increased by 12% (1.12; 95% CI, 1.05–1.19). Changes in IE1 from baseline to week 4 were not associated with CS-CMVi when CMV CMI was considered as a continuous variable on univariate or Cox regression analyses (P = .38 and P = .89, respectively). In addition, ATG use during the conditioning regimen was an independent predictor for CS-CMVi after week 4. Spearman correlation coefficient and box plots of the Δ changes in pp65 and IE1 SPCs between baseline and weeks 2 and 4 in patients with and without CS-CMVi are depicted in Figure 1A and 1B and Figure 2A–2D. The performance of the negative changes in IE1 and pp65 from baseline to week 2 or 4 and its prediction of CS-CMVi (sensitivity, specificity, positive predictive value, and negative predictive value) are summarized in Supplementary Table 2.

Changes in CMV CMI and Mortality

Overall, all-cause mortality at 6-month follow-up was higher in patients with CS-CMVi (P = .03019). Figure 1C and 1D shows the correlation between IE1 and pp65 changes from baseline to weeks 2 and 4 and mortality. Furthermore, we evaluated the impact of Δ changes in both antigens on patients’ mortality at 6 months using Kaplan-Meir curves for survival analysis (Figure 3A–3D). A negative IE1 change from baseline to week 4 was significantly associated with increased mortality after transplant (log rank test, P = .041; Figure 3D); however, type of transplant and prior corticosteroid use were the only independent factors associated with mortality by multivariate survival analysis.

DISCUSSION

In this follow-up analysis, we aimed to determine if changes in CMV ELISPOT results during the first month after transplant predicts CS-CMVi and all-cause mortality in the first 6 months after transplant. We found that decline (negative Δ change) in CMV CMI, as measured by CMV ELISPOT for pp65 from baseline to week 2 and for IE1 from baseline to week 4, was a predictor of CS-CMVi, along with receiving ATG as part of the conditioning regimen, matched and mismatched unrelated donors, haploidentical transplant, corticosteroid use, and female sex. Furthermore, all-cause mortality at 6-month follow-up was higher in patients with CS-CMVi.

The relationship between negative Δ changes in CMV CMI and CS-CMVi can be explained in part by delayed reconstitution of donor-origin CMV-specific T cells after transplant and the loss of the previously activated recipient CMV-specific T cells as a result of conditioning chemotherapy. The utility of CMV ELISPOT assays in predicting progression of CMV infection has been demonstrated [8, 13–15]; however, few studies have examined the utility of early dynamic changes in CMV CMI in predicting progression of CMV infection. Jung et al [16] reported results similar to ours in a prospective study of 84 recipients of HCT who were CMV-seropositive. The authors found that dynamic changes (day 30 – day 0) in pp65 SPCs (<42 per 200 000 cells) and IE1 SPCs (<4 per 200 000 cells), as measured by ELISPOT assay, were predictors of CMV infection. The authors proposed an algorithm based on their findings denoting that the Δ change in pp65 between days 0 and 30 is sensitive and the Δ change in IE1 between days 0 and 30 is specific for predicting CMV infection, similar to our previous report on diagnostic accuracy [9]. Likewise, our data highlight the utility of pp65 and IE1 Δ changes within a month from transplant to predict the risk for CS-CMVi. Overall, negative changes in IE1 or pp65 values between the time points had good negative predictive value to identify patients at risk for CS-CMVi. Both biomarkers were strong predictors of CS-CMVi, as shown in the univariate analysis and with strong correlation. On multivariate analysis, likely due to small numbers, only one biomarker per time point prevailed as significant: negative changes on pp65 at values at week 2, and negative changes on IE1 antigen at week 4. Our results underscore the dynamic changes of CMV-specific T-cell immunity early after transplant as a predictor, of CS-CMVi rather than at specific time points. Similarly, Chiereghin et al [7] reported on CMV immunomonitoring using CMV CMI during the first month after small bowel/multivisceral transplant to identify patients at risk for CMV infection. In that study, response was considered positive when the number of spot-forming cells was ≥5 in the wells stimulated with peptides. Patients who were CMV-seropositive were classified as “early responders” if they had CMV T-cell reconstitution within the first month after transplant and “late responders” if they had it after the first month. The early responders had asymptomatic or mild CMV infections with lower mean and peak CMV viral loads, and the late responders had moderate or severe CMV infections.

Furthermore, this study also reveals a very interesting result: that a decrease in IE1 SPCs from baseline to week 4 was associated with an increase in all-cause mortality risk at 6 months after transplant. The correlation between lymphocyte recovery and transplant outcomes, including increased risk for infections, is well recognized [17]. The relationship between lymphocyte recovery after transplant and graft-vs-host disease, disease relapse, and mortality is a topic of high interest [18]. We described the relationship between CMV immune recovery and poor outcomes in the REACT study [9]. The data from our current analysis suggest that CMV CMI values from as early as the first month after transplant could serve as predictors of not only CS-CMVi but also mortality after transplant, thus warranting further evaluation [18, 19], especially in the era of CMV chemoprophylaxis with letermovir. Additional studies are needed to determine whether the Δ changes or the absolute CMI values are more relevant to predict CS-CMVi or mortality.

Our study is not without limitations. First, as a multicenter trial the study has missing data, such as absolute lymphocyte count, short-term follow-up (up to 6 months from HCT), and cause of death. Also this study was observational; therefore, management of CS-CMVi was not uniform but up to the provider, which may have affected outcomes.

Future directions from this research include evaluating (1) the role of CMV CMI measurement on when to initiate therapy for low CMV viral loads during chemoprophylaxis (eg, letermovir), (2) whether to safely discontinue primary or secondary prophylaxis as well as CMV PCR testing, and (3) the impact of long-term CMV primary or secondary prophylaxis on CMV CMI and transplant outcomes [20, 21].

In summary, changes in CMV-specific T-cell responses from pretransplant to 1 month posttransplant may serve as predictors of CS-CMVi and all-cause mortality in recipients of alloHCT who are CMV-seropositive. Future clinical trials should examine whether such recipients who do not exhibit an increase in CMV CMI would benefit from closer monitoring of CMV viral loads and possibly from prolonged antiviral prophylaxis to improve CMV- and transplant-related outcomes.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank Stephanie Deming, Research Medical Library, MD Anderson Cancer Center, for editing the article.

Author contributions. E. J. A.-H. and R. F. C. participated in the research design and wrote the original draft. Y. J. did the statistical analysis. S. D. R. and S. D. help with the acquisition of data. F. K. and T. B. helped with editing and reviewing the manuscript. Acquisition of data: D. J. W., K. M., P. C., P. H., R. K. A., K. S. P., D. K., R. N., P. L., S. B. M., D. P. S., E. J. A.-H., and L. E. H. All authors contributed to reviewing and editing the article.

Financial support. This work was supported in part by research funding from Oxford Immunotec USA, Inc. Specifically, Oxford Immunotec provided ELISPOT CMV responses and assistance with study design. The sponsor did not participate in the analysis and/or the interpretation of the data. The work was also supported by the National Institutes of Health/National Cancer Institute under award P30CA016672.

Potential conflicts of interest. E. J. A.-H. has received research funding paid to her institution from Oxford Immunotec USA and Merck. D. J. W. has received research funding from Oxford Immunotech, Takeda/Shire, Chimerix, and Ansun Pharmaceuticals. S. D. R. serves as a consultant to SIRPant Immunotherapeutics and ReAlta Life Sciences. P. H. is an employee of Iovance Biotherapeutics (no conflicts with the published work) and reports honoraria from BMS, Janssen, Karyopharm, Sanofi, and Amgen. R. K. A. has received research funding paid to her institution from Aicuris, Astellas, Chimerix, Merck, Oxford Immunotec, Qiagen, Regeneron, and Takeda/Shire. D. K. has received research grants from Roche, Takeda, Atara Bio, and Qiagen and advisory fees from Roche, Merck, and Takeda. P. L. has served on endpoint committees for Takeda and Merck and as an expert for AiCuris. S. B. M. has received research funding paid to his institution from Chimerix, Oxford Immunotec, and Merck. F. K. received honorarium from Medscape. S. D. has received research funds paid to his institution from Allovir and Merck and served on an advisory board for Merck, as a consultant for Allovir, and on a speaker bureau for Merck and Takeda. T. B. is an employee of Oxford Immunotec (study sponsor). R. F. C. has served as a consultant to Oxford Immunotec USA, Merck, Takeda, Adagio Therapeutics, Karius, Astellas, Janssen, Ansun Pharmaceuticals, Genentech, Molecular Partners, QIAGEN, and ADMA Biologics and has received research funding paid to his institution from Oxford Immunotec USA, Merck, Ansun Pharmaceuticals, Takeda, Janssen, Genentech, Viracor, Karius, and AiCuris. All other authors report no potential conflicts.

References