Patient risk stratification and tailored clinical management of post‐transplant CMV‐, EBV‐, and BKV‐infections by monitoring virus‐specific T‐cell immunity

Abstract Background Despite routine post‐transplant viral monitoring and pre‐emptive therapy, viral infections remain a major cause of allogeneic hematopoietic cell transplantation‐related morbidity and mortality. Objective We here aimed to prospectively assess the kinetics and the magnitude of cytomegalovirus‐(CMV), Epstein Barr virus‐(EBV), and BK virus‐(BKV)‐specific T cell responses post‐transplant and evaluate their role in guiding therapeutic decisions by patient risk‐stratification. Study design The tri‐virus‐specific immune recovery was assessed by Elispot, in 50 consecutively transplanted patients, on days +20, +30, +60, +100, +150, +200 post‐transplant and in case of reactivation, weekly for 1 month. Results The great majority of the patients experienced at least one reactivation, while over 40% of them developed multiple reactivations from more than one of the tested viruses, especially those transplanted from matched or mismatched unrelated donors. The early reconstitution of virus‐specific immunity (day +20), favorably correlated with transplant outcomes. Εxpanding levels of CMV‐, EBV‐, and BKV‐specific T cells (VSTs) post‐reactivation coincided with decreasing viral load and control of infection. Certain cut‐offs of absolute VST numbers or net VST cell expansion post‐reactivation were determined, above which, patients with CMV or BKV reactivation had >90% probability of complete response (CR). Conclusion Immune monitoring of virus‐specific T‐cell reconstitution post‐transplant may allow risk‐stratification of virus reactivating patients and enable patient‐tailored treatment. The identification of individuals with high probability of CR will minimize unnecessary overtreatment and drug‐associated toxicity while allowing candidates for pre‐emptive intervention with adoptive transfer of VSTs to be appropriately selected.


INTRODUCTION
Allogeneic hematopoietic cell transplantation (allo-HCT) is a potentially curative treatment for hematological disorders. Among its main hurdles however, is the profound T-cell deficiency, resulting in development of viral infections -most commonly by cytomegalovirus (CMV), Epstein Barr virus (EBV) and polyomavirus type I (BKV)-, and the substantial transplant-related morbidity and mortality [1,2]. The reconstitution of antiviral immunity post-allo-HCT is often delayed or/and severely impaired by the immunosuppression administered to prevent or treat the immunological complications of allogeneic transplantation, affecting both the quantity and the quality of virus-specific T-cells (VSTs) [3]. The importance of a specific and robust anti-viral immune reconstitution is emphasized by the fact that strategies boosting antiviral immunity, such as immunotherapy with donor-or thirdparty-derived VSTs, have provided protection with 70-90% response rates [1,[4][5][6][7].
Heretofore, clinicians relied exclusively on the viral load monitoring, to guide interventions in treating viral infections post-transplant.
Given that immunity is a dynamic process, close monitoring of VST immune reconstitution (VST-IR), may identify patients able to potentially self-control viral reactivations and, eventually, provide the basis for tailored management of antiviral therapy to those in real need avoiding unnecessary overtreatment or/and the outgrowth of drug resistant viral variants.
As opposed to single-virus-specific immune reconstitution, the recovery of viral immunity against multiple viruses has been limitedly investigated in HCT [8,9]. We here aimed, by concomitant monitoring of IR against the most common viruses post-allo-HCT, namely CMV, EBV, and BKV, to prospectively recognize correlations between viral reactivation and the timing and kinetics of VST-IR, compare the recovery of functional VSTs with virological outcomes and provide surrogate markers for identifying patients able to successfully clear the infection and fine-tuning the clinical-decision making.

Subjects
Fifty consecutive allo-HCT patients were included in this prospective study, approved by the Institutional Review Board of the George Papanikolaou Hospital and performed in accordance with the Declaration of Helsinki.

Virological monitoring
CMV and EBV viral loads were routinely monitored by quantitative PCR (Qiagen) on blood samples, once a week. BKV load was measured by quantitative PCR (Geneproof) in urine every other week at the presence of clinical signs/symptoms or/and at request of the treating physician.

Immunological monitoring by enzyme-linked immunospot
Immunological monitoring was performed in peripheral blood samples collected at days +20, +30, +60, +100, +150, +200 post-allo-HSCT and in case of viral reactivation, weekly for 1 month, as shown in Figure S1. Peripheral blood mononuclear cells (PBMCs) were stimulated with peptides spanning CMV and EBV antigens whereas BKV-STs were firstly expanded in culture (Supplementary Information) [4,5,10] To assess whether CR of viral reactivation was correlated with the VST kinetics and could be predicted based on certain cut-offs VST numbers or of the net increase of VSTs from the onset of reactivation up to 2 weeks later (Delta-SFC [ΔSFC:max-VSTs minus onset-VSTs]), a receiver operating characteristic (ROC)-curve analysis was performed.
The reactivating patients' cohort was then split at the optimal VST cutoff, and cumulative incidences of CR between the above or below cutoff sub-cohorts were compared with the competing risk model.
Data are presented as mean ± SEM or median (range) when values follow normal or abnormal distribution, respectively. ANOVA followed by Tukey's test or a nonparametric Mann-Whitney U test were used for analysis of differences between data sets or multiple comparisons, respectively.

Viral prophylaxis in all patients included administration of acy-
clovir/valacyclovir (letermovir as primary anti-CMV prophylaxis became available later during the study). Pre-emptive treatment for CMV and EBV antigenemia included ganciclovir or foscarnet and rituximab, respectively. BKV hemorrhagic cystitis was treated with cidofovir. All patients received anti-viral treatment, except otherwise indicated.  (Table S3).

Viral episodes, IRs, and association with transplant outcomes
By looking at the kinetics of VSTs, at different time points and in association with the recipient serostatus, we observed significantly earlier and higher CMV-specific immune reconstitution between days +30 and +100 in seropositive over seronegative hosts, probably reflecting boost immune responses to the increased rate of reactivations in seropositive patients ( Figure S6). Similar, albeit not significant, kinetics of EBV-ST reconstitution was observed in EBV seropositive recipients (Table S3, Figure S6).  Figure 2C).

Viral reactivation and VST-IR
To better understand the temporal relationship between viral reac-

Risk factors for impaired VST-IR
To identify risk factors for delayed or/and impaired VST-IR, the VST levels at an early time point (day +20), were checked against different variables in univariate and multivariate analysis. ATG administration was strongly associated with significantly lower CMV-, EBV-, and BKV-ST levels both in the univariate and multivariate analysis whereas donor type (mismatched vs matched) did not retain significance in the multivariate analysis ( Figure 4). Notably, steroid administration raised as a contributing factor toward delayed reconstitution of EBV-specific T-cell immunity only in the multivariate analysis, probably due to a confounding effect of the other covariates in the univariate analysis.

Predicting CRs and guiding therapeutic decision based on certain levels of virus-specific immunity
To determine a threshold for protective VST immunity, thus avoiding overtreatment or stratifying patients to different therapy options,  Figure   S8C).

DISCUSSION
Despite improvements in anti-viral pharmacotherapy, CMV, EBV, and BKV reactivations remain a leading cause of morbidity and mortality In total, 74% of patients developed CMV or/and EBV or/and BKV reactivations, with a median of two viral episodes/patient. Apart from the substantial morbidity, and in agreement with other reports [11][12][13][14][15][16][17], viral reactivations from these three viruses, correlated also with adverse outcomes as regards OS and NRM, thus underscoring the magnitude of the indirect mortality induced by viral infections and their treatment. The huge human and financial cost associated with the management of viral infections post-transplant provides both a scientific and economic rationale for patient risk-stratification, so as to avoid unnecessary treatment in those patients who have acquired functional virus-specific immunity or to proceed with interventions such as VST immunotherapy for those at high-risk.
Although the correlation between VST presence and viral control has been demonstrated post-solid organ transplantation for all three viruses [18][19][20][21][22][23][24][25][26][27], reports in the HCT setting mostly have focused on CMV [16,. In our study, by simultaneously monitoring functional T-cell immunity against a broad spectrum of clinically problematic viruses after allo-HCT, in conjunction with viral load, we showed The VST-IR was serially measured by the levels of CD3+ CMV-, EBV-, and BKV-STs in PBMCs. Others have proposed monitoring of CD8+ CMV-STs as a prognostic tool to identify allo-HCT patients at high risk for CMV-infections [52,57], however, by evaluating only CD8+ cells, and given the importance of CD4+ STs in controlling infections [58], the clinical response may be misinterpreted or/and underestimated.
Hematopoietic stem cell transplant physicians will more accurately guide their therapeutic decisions if immune competence against viruses could be precisely estimated, enabling the discrimination of patients with solid antiviral immunity and high probability of CR from those having low probability of clearing the infection and being dependent on pre-emptive interventions. Several groups, focusing on CMV, have tried to identify VST thresholds predictive of protection from viral reactivation [59][60][61][62]. Instead, we here provide, a prediction model that could identify with at least 94% probability, among CMV and BKV reactivating patients, those who could successfully clear the infec-tion, based on certain VST levels at reactivation or their max ΔSFC 1-2 weeks later. Although ROC analysis couldn't provide CR predictive cut-offs in EBV reactivating patients, three patients for whom high EBV-VSTs at reactivation were detected, and a considerable expansion was measured in the following 2 weeks, cleared EBV infection without receiving rituximab.
Notwithstanding the need for validation of the predictive cut-offs for viral complete response in a larger, multicenter study, our data stress out the importance of monitoring VST-IR by functional assays that can be used to risk-stratify transplanted patients with viral infections and guide their clinical management. The identification of patients at low risk for morbidity and mortality from viral reactivations will minimize unnecessary overtreatment and drug-associated toxicity.
On the other hand, identification of patients at-risk might lead to pre-emptive intervention with intense antiviral pharmacotherapy or virus-specific immunotherapy. Guided pre-emptive therapeutic choices based on the actual individual risk for controlling viral infections will potentially result in lower morbidity and better survival chances in allo-HCT patients as well as substantially decrease the post-transplant care cost.

ACKNOWLEDGMENTS
We would like to thank Nikolaos Savvopoulos, Ioanna Vallianou, and Irene Deligianni (G. Papanikolaou Hospital) for technical assistance.