Comparison of intracellular cytokine staining versus an ELISA‐based assay to assess CMV‐specific cell‐mediated immunity in high‐risk kidney transplant recipients

The best method for monitoring cytomegalovirus (CMV)‐specific cell‐mediated immunity (CMV‐CMI) among high‐risk kidney transplant (KT) recipients remains uncertain. We assessed CMV‐CMI by intracellular cytokine staining (ICS) by flow cytometry and a commercial interferon (IFN)‐γ release assay (QuantiFERON®‐CMV [QTF‐CMV]) at posttransplant months 3, 4, and 5 in 53 CMV‐seropositive KT recipients that had received induction therapy with antithymocyte globulin (ATG) and a 3‐month course of valganciclovir prophylaxis. The discriminative capacity (areas under receiver operating characteristics curve [auROCs]) and diagnostic accuracy to predict immune protection against CMV infection from the discontinuation of prophylaxis to month 12 were compared between both methods. There was significant although moderate correlations between CMV‐specific IFN‐γ‐producing CD8+ T‐cell counts enumerated by ICS and IFN‐γ levels by QTF‐CMV at months 3 (rho: 0.493; p = 0.005) and 4 (rho: 0.440; p = 0.077). The auROCs for CMV‐specific CD4+ and CD8+ T‐cells by ICS were nonsignificantly higher than that of QTF‐CMV (0.696 and 0.733 vs. 0.678; p = 0.900 and 0.692, respectively). The optimal cut‐off of ≥0.395 CMV‐specific CD8+ T‐cells yielded a sensitivity of 86.4%, specificity of 54.6%, positive predictive value of 79.2% and negative predictive value of 66.7% to predict protection. The corresponding estimates for QTF‐CMV (IFN‐γ levels ≥0.2 IU/mL) were 78.9%, 37.5%, 75.0%, and 42.9%, respectively. The enumeration of CMV‐specific IFN‐γ‐producing CD8+ T‐cells at the time of cessation of prophylaxis performed slightly better than the QTF‐CMV assay to predict immune protection in seropositive KT recipients previously treated with ATG.


| INTRODUCTION
The monitoring of cytomegalovirus (CMV)-specific cell-mediated immunity (CMV-CMI) is being advocated as a promising method to tailor preventive strategies after solid organ transplantation (SOT) and to move towards personalized transplant medicine. 1 The implementation of these strategies, however, faces a number of uncertainties, including the choice of target risk group, the optimal timing for monitoring, and the best technical approach. 2 Various methods are currently available to quantify the magnitude and functionality of CMV-CMI, each of them offering a distinct set of advantages and limitations. 3 Intracellular cytokine staining (ICS) by flow cytometry is usually considered as the "gold standard" thanks to greater specificity and capacity for thorough functional assessment, although appears limited by its time-and labor-consuming nature and lack of standardization. 4 The QuantiFERON ® -CMV (QTF-CMV) test (Qiagen GmbH) is a commercial enzyme-linked immunosorbent assay (ELISA)-based interferon (IFN)-γ release assay that benefits from technical simplicity and validated cut-off values. 5 Nevertheless, our group has recently reported that the QTF-CMV assay exhibits low sensitivity and specificity to predict the occurrence of CMV infection among kidney transplant (KT) recipients that had been treated with rabbit antithymocyte globulin (rATG). 6 Only few studies have been conducted to directly compare these two approaches, 7,8 and none of them were specifically focused on any of the high-risk categories of KT recipients (donor-recipient mismatch [D+/ R-] or use of T-cell-depleting agents). Therefore, we analyzed the accuracy of the ICS method and the QTF-CMV assay to predict the presence of effective protection from CMV infection in a homogenous group of CMV-seropositive KT recipients that received rATG induction. Our ultimate aim was to identify the best strategy for CMV-CMI monitoring that would allow to individualize the duration of antiviral prophylaxis in this patient population.

| Study population and setting
We performed a subanalysis of a prospective observational cohort study focused on the QTF-CMV assay to investigate CMV-CMI, as published elsewhere. 6 Table S1) was representative of the original entire cohort, 6 with no significant differences in demographics, comorbidities, graft and patient outcomes, or occurrence of CMV events (data not shown).

| Study design and outcomes
Participants were enrolled at the time of transplantation and followed-up for at least 12 months, unless graft loss or death occurred earlier. Antiviral prophylaxis with valganciclovir (VGCV) (900 mg daily with renal dose adjustment if needed) was scheduled for 3 months, as per international and national guidelines. 2,9 Monitoring of CMV-CMI was carried out by ICS and QTF-CMV at months 3, 4, and 5 (±1 week), once the 3-month course of antiviral prophylaxis with VGCV was planned to be discontinued. CMV viral load was quantified fortnightly during the first 2 months (i.e., at days 15, 30, 45, and 60 approximately), on a monthly basis through month 6 (months 3, 4, 5, and 6), and bimonthly thereafter (months 8, 10, and 12), as well as at any time when clinically indicated, with a real-time polymerase chain reaction (RT-PCR) assay. The study outcome was the presence of immune protection from the occurrence of CMV infection (either asymptomatic infection or clinical disease) between the date of discontinuation of VGCV prophylaxis to posttransplant month 12. An overview of the study design is provided in Supporting Information: Figure S1.

| Intracellular cytokine staining
The protocol has been previously described. 7,10 Whole blood specimens were collected in sodium heparin tubes and processed within 24 h after collection. Enumeration of CMV-specific IFN-γproducing CD8 + and CD4 + T-cells was carried out by flow cytometry for ICS (BD Fastimmune, BD Becton Dickinson and Company, and BD Biosciences). A volume of 0.5 mL of heparinized whole blood was simultaneously stimulated for 6 h with 2 sets of 15-mer overlapping peptides encompassing the entire sequence of pp65 (UL83) and immediate early (IE)-1 peptides (2 μg/mL per peptide) obtained from JPT Peptide Technologies GmbH in the presence of costimulatory monoclonal antibodies targeting CD28 and CD49d. Samples mock stimulated with phosphate-buffered saline (PBS)/dimethyl sulfoxide solution (without peptides) and costimulatory antibodies or stimulated with phytohemagglutinin (PHA) (1 mg/mL) (Sigma-Aldrich) were run in parallel. Brefeldin A (5 μg/mL) was added for the last 4 h of  Figure S2). The total number of each T-cell subpopulation was calculated by multiplying the corresponding percentage of CMV-specific cells (after background subtraction) by the absolute number of CD8 + and CD4 + T-cells. Therefore, the IFN-γproducing CD4 + and CD8 + counts enumerated comprised the sum of both pp65-and IE-1-stimulated T-cells.

| QuantiFERON ® -CMV assay
The assay was carried out according to the manufacturer's instructions. 11 One-milliliter aliquots of whole blood were obtained by venipuncture in a single lithium-heparin blood collection tube and immediately transferred into three QTF-CMV tubes: one of them contained a pool of 22 CMV peptides mapped within pp28, pp65, pp50, IE-1, IE-2, and glycoprotein B antigens and restricted by several widespread human leukocyte antigen (HLA) class I molecules ("CMV tube"); a second tube contained PHA as positive control ("mitogen tube"); and the third tube contained heparin only and served as negative control ("nil tube"). These tubes were shaken vigorously for 5 s and incubated for 18 to 24 h at 37°C. Supernatants were harvested and frozen at -80°C until analysis. IFN-γ levels were assessed by ELISA after generating a standard curve for each run by a single technician blinded to clinical details. As per the manufacturer's instructions, the IFN-γ levels were interpreted as follows: (a) reactive QTF-CMV assay at ≥0.2 IU/mL (CMV minus nil tubes) and ≥25% of the nil tube; (b) nonreactive assay at <0.2 IU/mL (CMV minus nil) and/or <25% of the nil tube, and ≥0.5 IU/mL (mitogen minus nil); and (c) indeterminate assay at <0.2 IU/mL (CMV minus nil) and/or <25% of nil, and <0.5 IU/mL (mitogen minus nil).

| CMV serostatus and RT-PCR for CMV monitoring
Pretransplant CMV IgG serostatus was investigated in donors and recipients by enzyme immunoassay. Whole blood CMV viral load was quantified using a commercial RT-PCR assay (RealStar ® CMV PCR kit 1.0; Altona Diagnostics GmbH), with a lower limit of quantification of 120 IU/mL. DNA was extracted from 200 μL of sample with the NucliSENS ® easyMag ® instrument (bioMérieux Diagnostics), according to the manufacturer's instructions.

| Statistical analysis
Quantitative data are shown as the mean ± standard deviation (SD) or the median with interquartile range (IQR). Qualitative data are expressed as absolute and relative frequencies. Categorical variables were compared with the χ 2 test. The Mann-Whitney U test was applied for continuous variables, whereas comparison between paired samples was performed with the Wilcoxon signed rank test. Correlations were analyzed using the Spearman's correlation coefficient. Scaterplotts with 95% confidence intervals (CIs) of the regression lines were generated.
The discriminative capacity of the ICS method and the QTF-CMV assay to predict the presence of protective CMV-CMI was estimated through the area under receiver operating characteristics curve (auROC).
The diagnostic accuracy was assessed by the sensitivity, specificity, positive (PPV) and negative predictive values (NPV), and likelihood ratios with 95% CIs. To these aims the monitoring points that coincided with the time of discontinuation of VGCV prophylaxis were selected. As stated above, prophylaxis was scheduled to be administered for 3 months. Therefore, we selected the CMV-CMI assessment performed at post-transplant month 3. Indeterminate QTF-CMV results were operationally considered as a "negative" assay (i.e., below the selected protective threshold for IFN-γ FERNÁNDEZ-RUIZ ET AL. | 3 of 11 production) for analysis purposes. The optimal threshold for CMVspecific IFN-γ-producing T-cells was established by means of the Youden's index (J = sensitivity + specificity -1). 13 The agreement between positive (i.e., protective) results in both methods was estimated by the Cohen's κ-coefficient. Time-to-event curves were plotted by the Kaplan-Meier method and intergroup differences were compared with the log-rank test. Statistical analysis was performed with SPSS version 20.0 (IBM Corp) and graphs were generated with Prism version 6.0 (GraphPad Software Inc).   CMV infection and CMV disease were analyzed as separate outcomes (Supporting Information: Table S2).

| Study population and outcomes
By selecting those values at which the Youden's index 13 was maximized, we established at 0.23 and 0.395 cells/μL the optimal thresholds for CMV-specific IFN-γ-producing CD4 + and CD8 + T-cell counts, respectively. CMV infection-free survival following discontinuation of prophylaxis was significantly higher for those KT recipients with a CMV-specific CD8 + T-cell count ≥0.395 cells/μL (logrank p = 0.008). A similar nonsignificant trend was also observed in patients exhibiting a CMV-specific CD4 + T-cell count ≥0.23 cells/μL (log-rank p = 0.079). In contrast, we found no differences in CMV infection-free survival according to the presence of a reactive or nonreactive QTF-CMV assay interpreted using the cut-off value established by the manufacturer (IFN-γ ≥ 0.2 IU/mL) (log-rank p = 0.315) (Figure 4). There was poor agreement between reactive results in the QTF-CMV assay and the presence of CMV-specific CD4 + and CD8 + IFN-γ-producing T-cell counts above these thresholds (κcoefficients = 0.156 and 0.171, respectively).

| Diagnostic accuracy of ICS and QTF-CMV
We also assessed the diagnostic accuracy of both approaches to predict the presence of protective CMV-CMI at the time of discontinuation of antiviral prophylaxis (Table 1)  On the basis of a homogeneous population of CMV-seropositive KT recipients that received induction therapy with rATG and a 3month course of VGCV, we found that the IFN-γ levels in the QTF-CMV assay were positively correlated with CMV-specific IFN-γproducing CD8 + (but not CD4 + ) T-cell counts enumerated by ICS. The discriminative capacity (auROC) was slightly higher for ICS than for IGRA, although the difference did not reach statistical significance.

| ICS and QTF-CMV at fixed time points
The accuracy to predict effective protection of the optimal threshold established for the CMV-specific CD8 + response (≥0.395 cells/μL) was better than that yielded by QTF-CMV interpreted as per the T A B L E 1 Diagnostic accuracy of the ICS method and QTF-CMV assay performed at the time of discontinuation of VGCV prophylaxis to predict immune protection from the subsequent development of CMV infection through posttransplant month 12. More importantly, since one of the intended applications of the assessment of CMV-CMI would be to prolong antiviral prophylaxis beyond the standard duration in selected patients lacking protection, assays with higher NPVs would be preferred. In our experience, the CMV-specific IFN-γ-producing CD8 + T-cell count measured by ICS showed a meaningfully better NPV to identify protected recipients (66.7%) than the QTF-CMV assay regardless of the cut-off level for IFN-γ (42.9% to 45.5%).
Despite the increasing number of observational [16][17][18][19] and intervention studies 20  and IE-1). Therefore, it is not surprising that we and others 7,8 observed a positive correlation between IFN-γ levels and CMV-specific CD8 + T-cell counts only, but not for CD4 + responses. In addition, the QTF-CMV assay targets viral epitopes whose presentation is restricted through certain widespread HLA class I molecules. It has been reported that the accuracy of the assay improves in the subgroup of patients harboring HLA-A1 and/or HLA-A2 alleles, at the expense of a decrease in specificity. 6,8 Some advantages of the commercial QTF-CMV would include its easiness in use-since it is performed on whole blood in an ELISA reader-and the existence of This finding is in contrast with previous studies that found a relative predominance of the CD4 + response in the long-term protection against CMV reactivation among KT recipients without T-cell depleting induction. 25 We and others have shown that the rATGinduced lymphodepletion disproportionately impacts the CD4 + subset, while the CD8 + subset is relatively preserved. 26 immunodominant viral epitopes (pp65 and IE-1). 29 Finally, QTF-CMV results were lacking for some patients at certain points. Innovation, also co-funded by the European Union. Funding sources had no involvement in the study design and conduction, data analysis, or manuscript preparation.