Clinical experience with a novel assay measuring cytomegalovirus (CMV)-specific CD4+ and CD8+ T-cell immunity by flow cytometry and intracellular cytokine staining to predict clinically significant CMV events

Background Cytomegalovirus (CMV) infection is one of the most common opportunistic infections following organ transplantation, despite administration of CMV prophylaxis. CMV-specific T-cell immunity (TCI) has been associated with reduced rates of CMV infection. We describe for the first time clinical experience using the CMV T-Cell Immunity Panel (CMV-TCIP), a commercially available assay which measures CMV-specific CD4+ and CD8+ T-cell responses, to predict clinically significant CMV events. Methods Adult (> 18-year-old) patients with CMV-TCIP results and ≥ 1 subsequent assessment for CMV DNAemia were included at Brown University and the University of Maryland Medical Center-affiliated hospitals between 4/2017 and 5/2019. A clinically significant CMV event was defined as CMV DNAemia prompting initiation of treatment. We excluded indeterminate results, mostly due to background positivity, allogeneic hematopoetic cell transplant (HCT) recipients, or patients who were continued on antiviral therapy against CMV irrespective of the CMV-TCIP result, because ongoing antiviral therapy could prevent a CMV event. Results We analyzed 44 samples from 37 patients: 31 were solid organ transplant recipients, 4 had hematologic malignancies, 2 had autoimmune disorders. The CMV-protection receiver operating characteristic (ROC) area under the curve (AUC) was significant for %CMV-specific CD4+ (AUC: 0.78, P < 0.001) and borderline for CD8+ (AUC: 0.66, P = 0.064) T-cells. At a cut-off value of 0.22% CMV-specific CD4+ T-cells, positive predictive value (PPV) for protection against CMV was 85% (95%CI 65–96%), and negative predictive value (NPV) was 67% (95%CI 41–87%). Conclusions The CMV-TCIP, in particular %CMV-specific CD4+ T-cells, showed good diagnostic performance to predict CMV events. The CMV-TCIP may be a useful test in clinical practice, and merits further validation in larger prospective studies.


Background
Cytomegalovirus (CMV) infection remains one of the most prevalent opportunistic infections (OI) following solid organ transplantation (SOT), and in individuals with hematologic malignancies or other immunocompromising conditions [1][2][3]. It is associated with significant morbidity due to its direct (CMV disease) and indirect (other OI, rejection, chronic allograft dysfunction) effects [1,2,4]. Prevention strategies after SOT consist of universal prophylaxis, preemptive therapy, or a combination of the two [1,2]. However, these strategies have their respective pitfalls. For example, the optimal duration of antiviral prophylaxis is uncertain, varying from as short as 3 months to > 1 year [1,2,5]. Despite antiviral prophylaxis, patients may still develop CMV infection following discontinuation of prophylaxis [5][6][7]. In addition, there are risks of medication side-effects from antiviral prophylaxis, risk of drug resistance with prolonged antiviral prophylaxis, and the cost of antivirals can be prohibitive. Preemptive monitoring strategies can be inconvenient due to the need for serial viral load (VL) monitoring (every 1-2 weeks) [1,2,[6][7][8].
The development of CMV infection and severity of CMV disease are largely influenced by the ability of the immune system to control viral replication. This generally requires intact humoral and cell-mediated immune responses, of which the latter is a frequent target of immunosuppressant therapy in SOT recipients. Despite treatment with T-cell inhibiting medications, most SOT recipients do not develop CMV infection, which suggests many individuals are able to maintain T-cell responsiveness against CMV. Much effort has been expended searching for a good measure of immune competency against CMV, including both global (non-pathogen specific) and CMV-specific assays [2]. An ideal diagnostic test that could provide a measure of immune competency to control CMV infection might allow for personalized anti-CMV care.
The Viracor® CMV T-cell Immunity Panel (CMV-TCIP) is the first commercially available assay measuring CMV-specific CMI in the US. CMV-TCIP is a flow cytometric assay that measures %CMV-specific CD4+ and CD8+ T-cells separately, following stimulation of whole blood with CMV peptides and lysates [27]. In this study, we describe for the first time patient-level experience with the CMV-TCIP, and assess the potential utility of CMV-specific CD4+ and CD8+ T-cell responses in predicting clinically significant CMV events.

Patients
We retrospectively studied adult (> 18-year-old) patients with CMV-TCIP results and ≥ 1 subsequent CMV-VL at Brown University and the University of Maryland Medical Center (UMMC)-affiliated hospitals, between 4/2017 and 5/2019. Clinically significant CMV events were defined as CMV DNAemia (> 1000 copies/mL [28]) or any detectable VL with symptoms suspicious for CMV infection by clinician assessment, prompting initiation of treatment.
We excluded from analysis indeterminate (due to background positivity) results, allogeneic hematopoetic cell transplant (HCT) recipients, or patients who were continued on antiviral therapy against CMV, irrespective of the CMV-TCIP result, since ongoing antiviral therapy could prevent a CMV event.
As a rule, CMV IgG donor seropositive/recipient negative (D+/R-) kidney and heart transplant recipients receive 6 months, CMV R+ kidney and heart transplant and all liver transplant recipients 3 months of valganciclovir prophylaxis. The duration of prophylaxis is extended in lung transplant recipients (D+/R-: lifelong, R+: 6-12 months). Patients with CMV infection have CMV-VL tested every 1-2 weeks for the first 2 months after discontinuation of valganciclovir. All patients with CMV-TCIP results included in analyses had ≥1 subsequent CMV-VL and were followed clinically for 6 months after discontinuation of valganciclovir, or after the CMV-TCIP for patients who did not receive antiviral medications. CMV-TCIP were ordered at the time (within a week) of discontinuation of valganciclovir or low-level CMV viremia for patients not receiving valganciclovir. The study was approved by both Institutional Review Boards, in agreement with the Declaration of Helsinki (Brown IRB study approval # 1346550, UMMC IRB study approval # HP-00082131).

Statistical analyses
Data are presented as mean (standard deviation-S.D.) or median (25th-75th inter-quartile percentile range (IQR)) for variables of normal or non-normal distribution (by Kolmogorov-Smirnoff test), respectively. We compared continuous variables with Student's t, or Mann-Whitney tests for independent samples, and paired t or Wilcoxon signed-rank tests for related samples. We compared categorical variables between groups using χ 2 or Fisher's exact tests. We generated receiveroperating characteristic (ROC) curves to summarize diagnostic performance of tests in predicting protection from CMV. We calculated sensitivity, specificity, positive (PPV, protected against CMV) and negative (NPV, CMV event) predictive values, and their 95% confidence intervals (95%CI), for cut-off values that were closest to assay validation cut-off value (0.2%). Correlations were
We analyzed 44 samples (Brown 35, UMMC 9) from 37 patients (Brown: 28, UMMC: 9). Thirty-one (31) were SOT recipients, of which the majority (20/31) were kidney transplant recipients. Four patients had hematologic malignancies (2 multiple myeloma, 1 cutaneous T-cell lymphoma treated with alemtuzumab, 1 diffuse large Bcell lymphoma and HIV) with CMV DNAemia. One patient had autoimmune colitis treated with high-dose steroids and infliximab, after which he developed proven CMV colitis. Another patient had systemic lupus erythematosus and CMV pneumonia. Clinical features of study patients at the time of CMV-TCIP are summarized in Table 1.
As the % of CMV events in our study was rather high, we plotted positive (PPV) and negative (NPV) predictive values of %CMV-specific CD4+ T-cells for different frequencies of CMV events. For a pre-test probability between 15 and 25% (the CMV event rate in most studies [6,8,11,29]), we estimated a PPV between 90 and 95% and NPV between 39 and 54% (Fig. 5).

Repeat CMV-TCIP
Six patients had repeat CMV TCIP following their initial CMV event. All six were SOT recipients and 5 were high-risk for CMV (CMV D+/R-). We found a significant increase in %CMV-specific CD4+, %CMV-specific CD8+ (Fig. 6) and %SEB CD4+ responses (P = 0.028); the increase in %SEB CD8+ response was not statistically significant (P = 0.345). Five patients, all with > 0.22% CMV-specific CD4+ T-cells on repeat testing, did not experience subsequent CMV events after discontinuation of valganciclovir.

False positive results
Four (4) CMV R+ transplant recipients with positive %CMV-specific CD4+ responses (3 of whom also had positive CMV-specific CD8+ responses, Figs. 2 and 3) had subsequent CMV events, defined as DNAemia and initiation of treatment with valganciclovir by clinicians. Two patients (both renal transplant recipients) were restarted on valganciclovir for asymptomatic low-level CMV DNAaemia (CMV-VL were 600 and 534 copies/mL).
One heart transplant recipient with positive CMVspecific CD4+ and CD8+ responses was restarted empirically on valganciclovir for CMV DNAemia (1100 copies/mL) and oral ulcers after stopping primary prophylaxis for CMV. A second CMV-VL was also detectable at 700 copies/mL prior to starting treatment. No testing was performed to evaluate etiology of oral ulcer as it was presumed to be related to CMV. Valganciclovir was eventually stopped following resolution of oral ulcers and after achieving undetectable CMV-VL. There was no recurrence of CMV DNAemia. Interestingly, the oral ulcers recurred, and after appropriate testing (PCR), were found to be due to herpes simplex virus.
The fourth patient had a positive CMV-specific CD4+ but negative CD8+ response and had chronic diarrhea for 2 years after kidney transplant. Colonoscopy showed rare cells that were positive for CMV by immunochemistry, without cytopathic changes. He had low-level

Discussion
In this study, we found a strong correlation between the results of the CMV-TCIP, specifically low CMV-specific CD4+ T-cells measured by ICS and FC, and subsequent CMV events. The association between CMV events and CMV-specific CD8+ T-cells did not reach statistical significance, although P-value was 0.06 (Figs. 2 and 4). In patients with repeat CMV-TCIP, CMV-specific CMI became stronger over time, facilitating discontinuation of valganciclovir (Fig. 6). Our report provides the first realworld data on the predictive value of this commercially available assay, that is supportive of its potential clinical utility.
A diagnostic test of immune competency against CMV can be utilized in different scenarios: at the end of primary prophylaxis, to determine if extended prophylaxis or close VL monitoring might be of benefit [10-16, 18, 22, 25]; at the end of treatment of CMV infection, to support the need for secondary prophylaxis [6,8]; finally, in patients with asymptomatic CMV DNAemia, to determine if antiviral treatment is truly indicated [12,13,17,23,29]. Herein, the CMV-TCIP performed well in a relatively small (but comparable in size to other similar studies [8,12,29,30]) case series, including all three potential scenaria. Larger-scale prospective studies should evaluate clinical utility of the assay in each.
When comparing CMV-CMI assays, one study demonstrated that an ELISPOT-based assay performed similar to the Quantiferon®-CMV assay [31], while a recent meta-analysis suggests that ELISPOT-based CMV-CMI testing might perform better than Quantiferon®-CMV in predicting CMV events [32]. While the CMV-TCIP has been studied less than the above assays, the methods on which it is based (ICS/FC) have served as the goldstandard of immunoassays for years [30,33]. Also, the predictive performance of the test in our study was comparable to previous reports, especially after adjustment for pre-test probability (Fig. 5).
The relevance of individual CD4+/CD8 + T-cell subpopulations in the immune response to CMV infection in SOT recipients has been extensively studied, primarily via ICS/FC [30]. One early study of renal transplant recipients showed that presence of CMV-specific CD4+, not CD8+, T-cells was protective against development of CMV disease [34]. On the contrary, a study of heart and lung transplant recipients suggested that CMV-specific CD8+, not CD4+, T-cells were protective [35]. More recently, a study of CMV-specific T-cell subpopulations in renal transplant recipients demonstrated that low pretransplant CD8+ T-cells, low post-transplant day (PTD) 15 CD4+ or CD8+ T-cells, and low PTD60 and PTD180 CD4+ T-cells were predictive of subsequent CMV events [36]. Given the complex interactions between subpopulations of the cellular immune system, it is likely that both CMV-specific CD4+ and CD8+ T-cells play a role in the immune response to CMV infection [30].
A limitation of the Quantiferon®-CMV assay is that is seems to be more skewed towards CD8+ response. ELISPOT-based CMV-CMI testing, while measuring both CD4+ and CD8+ responses, does not provide detailed analysis of the individual components. The CMV-TCIP is the only clinical test of CMV-specific CMI to date that analyzes CD8+ and CD4+ T-cell responses separately. CMV-specific CD4+ T-cells are necessary to generate a pool of memory cytotoxic CD8+ T-cells, that can potentially prevent disease, by controlling recurrent CMV viremia, in the absence of antiviral medication [30,37]. This mechanism could account for the strong association of %CMV-specific CD4+, more than CD8+ Tcells with protection against CMV in our study, and the better performance of ELISPOT-based CMV-CMI testing, compared to the Quantiferon®-CMV in predicting CMV events [32]. Importantly, there is also evidence that CD4+ T cells have direct antiviral properties against CMV and play an essential role in abrogating reactivation and controlling primary CMV infection [38][39][40]. Given the importance of both CD4+ [4,30,34,36,41,42] and CD8+ [35,41,42] T-cells in the immune response to CMV, and the variety of clinical scenarios in which CMI assays may be used, detailed information regarding both CD4+ and CD8+ specific responses could be of clinical utility.
Recently, investigators have also examined measures of global immunity to predict subsequent CMV events, focusing on inexpensive tests, that most clinicians order routinely [6,43]. One such study in SOT recipients showed that the absolute lymphocyte count (ALC) at the completion of CMV treatment was independently associated with risk for subsequent recurrent CMV events [6], in agreement with another report studying SOT and HCT recipients [43]. We did not find a significant correlation found between ALC and risk for subsequent CMV event (Fig. 4). This discrepancy may be related to the higher rate of lymphocyte depletion (use of antilymphocyte induction therapy and HCT) in these reports, compared to our patient population.
Limitations of our study are its retrospective design, small sample size, host diversity and clinical scenarios in which this test was ordered. Our endpoint was initiation of antiviral for CMV guided by DNAemia or symptoms at clinician discretion, rather than CMV disease as defined by consensus criteria [44]. Nevertheless, clinicians frequently initiate treatment at high-level or rising CMV DNAemia before symptoms develop, therefore investigators have previously used this "real-world" outcome as a clinical endpoint when studying a CMV-CMI assay [26]. Last, providers were not blinded to test results, which might have influenced clinical thresholds to treat CMV DNAemia and choose observation over treatment with valganciclovir in patients with positive CMV-TCIP values. However, this does not seem to be the case, since patients with false positive CMV-TCIP results were classified as such because clinicians decided to start treatment. None of these patients had symptoms and signs clearly attributable to CMV infection, and at least two had evidence of controlling the infection (downtrending viremia), prior to initiation of valganciclovir. Also, peak CMV-VL was higher in cases with TCIP negative results, which argues against the clinicians having a lower threshold to treat CMV in such patients.
It should be noted that, besides CD69 and IFN-γ, other activation/memory markers like CCR7, CD45RO, CD27, CD62L and cytokines like TNF-α and IL2 of CD4+ and CD8+ T-cells may be helpful to further delineate CMV-specific TCI. In one small study of CMV R+ lung transplant recipients at risk for CMV infection, sensitivity was numerically lower but specificity higher for % TNF-α-producing CD8+ T-cells, while specificity was numerically higher for %TNF-α-producing CD4+ Tcells with the same sensitivity, compared to IFN-γ. IL2 had lower sensitivity and specificity compared to IFN-γ, whilst combining IFN-γ and IL2 did not improve predictive performance [45]. Future studies might help identify even more sensitive and specific CD4/8 "deep immunophenotyping" [46] and "polyfunctional signatures" [47], to predict protection against CMV.

Conclusions
The CMV-TCIP assay, in particular %CMV-specific CD4+ T-cells, demonstrated good performance in predicting subsequent CMV events in immunocompromised patients at risk for CMV infection. Given the potential clinical utility of this assay, further validation in larger prospective studies is warranted.