Pre-transplant immune factors may be associated with BK polyomavirus reactivation in kidney transplant recipients

BK polyomavirus (BKPyV) reactivation in kidney transplant recipients can lead to allograft damage and loss. The elements of the adaptive immune system that are permissive of reactivation and responsible for viral control remain incompletely described. We performed a prospective study evaluating BKPyV-specific T-cell response, humoral response and overall T-cell phenotype beginning pre-transplant through one year post-transplant in 28 patients at two centers. We performed an exploratory analysis of risk factors for the development of viremia and viruria as well as compared the immune response to BKPyV in these groups and those who remained BK negative. 6 patients developed viruria and 3 developed viremia. BKPyV-specific CD8+ T-cells increased post-transplant in viremic and viruric but not BK negative patients. BKPyV-specific CD4+ T-cells increased in viremic, but not viruric or BK negative patients. Anti-BKPyV IgG antibodies increased in viruric and viremic patients but remained unchanged in BK negative patients. Viremic patients had a greater proportion of CD8+ effector cells pre-transplant and at 12 months post-transplant. Viremic patients had fewer CD4+ effector memory cells at 3 months post-transplant. Exploratory analysis demonstrated lower CD4 and higher total CD8 proportions, higher anti-BKPyV antibody titers and the cause of renal failure were associated BKPyV reactivation. In conclusion, low CD4, high CD8 and increased effector CD8 cells were found pre-transplant in patients who became viremic, a phenotype associated with immune senescence. This pre-transplant T-cell senescence phenotype could potentially be used to identify patients at increased risk of BKPyV reactivation.

Currently, it is recommended that all KTRs be screened for BKPyV by PCR of urine or blood post-transplant [8,13]. The only treatment known to be efficacious is reduction in immune suppression (IS) [14], which carries with it the risk of acute rejection [15]. Previous studies have demonstrated low or negative anti-BKPyV antibodies [16,17] and low or absent BKPyV-specific T-cells prior to transplant [8,18,19] are risk factors for BKPyV reactivation. The development of BKPyV-specific T-cells without IS reduction has been associated with self-limited viremia, and failure to develop BKPyV-specific cellular response is associated with prolonged viremia and BKN [20,21]. Rising anti-BKPyV IgG and IgM antibody titers are associated with viral reactivation and correlate with severity of disease [22][23][24][25].
Although previous studies have evaluated the BKPyV-specific T-cell response, detailed longitudinal understanding of such response in context of clinical characteristics and outcomes is lacking. Furthermore, no studies have attempted to evaluate pre-transplant T-cell phenotypes in order to establish whether specific profiles may alter reactivation risk. We hypothesized that risk of developing BKV-associated diseases post-transplant may in part be determined by specific immune factors pre-transplant. In this exploratory study, we prospectively followed 28 patients who underwent renal transplantation at two local institutions. We assessed the presence of BKPyV-specific humoral and cellular immune response before transplant and for one year post-transplant to identify early BKPyV-specific immune alterations to identify those who were protected against BKPyV viremia or reactivation limited to the urine (viruria). Additionally, we performed an immuno-phenotype analysis of T-cells to identify pre-transplant phenotypic alterations which may be permissive of or protective against viral reactivation.

Subjects and sample collection
This prospective observational cohort study was approved by the internal review boards of Beth Israel Deaconess Medical Center and the Brigham and Women's Hospital. Patients were enrolled at the transplant clinics of both institutions from September 2012 to October 2014. Urine and peripheral blood samples were collected before kidney transplantation and 1, 3, 6 and 12 months post-transplant. Plasma and peripheral blood mononuclear cells (PBMC) were isolated and aliquots of PBMC, plasma and urine were stored at -80˚C. Demographic and clinical information, including BKPyV urine and serum PCR screening values, were collected from the medical record.
Cells were stained with fluorescently-conjugated antibodies specific for human CD4 (clone L200) and CD8 (clone SK1). They were then permeabilized, stained for IFN-γ (clone B27) and CD3 (clone SP34.2), and finally fixed. All antibodies were obtained from BD Biosciences. Data were collected using a LSRII flow cytometer (BD Biosciences) and analyzed with FlowJo software (Treestar Inc.). ICS results were considered positive if the percentage of IFNγ producing CD4 + or CD8 + T cells was equal to or greater than two times the baseline value of IFN-γ. ICS results are reported as the difference between the baseline and experimental IFN-γ values.

Humoral response to BKPyV
BKPyV pseudovirion-coated ELISA was used to quantify anti-BKPyV IgG against BKPyV serotype I as has been previously described [26]. Results were reported as mean values of duplicate reactions. The serum dilution in assays for anti-BKPyV IgG was 1:100. The cut-off for seropositivity was an OD value >0.20.

BKPyV DNA Detection
DNA extraction was performed with the QIAamp MinElute Virus Spin Kit (Qiagen, CA) following kit protocol. BKPyV DNA was quantified by Quantitative PCR (qPCR) using 7300 Real Time PCR System (Applied Biosystems, CA). The primer pair 5 0 -AGTGGATGGGCAGCCTA TGTA-3 0 (nt 2511-2531) and 5 0 -TCATATCTGGGTCCCCTGGA-3 0 (nt2586-2605), and probe 6FAM-AGGTAGAAGAGGTTAGGGTGTTTGATGGCACA-TAMRA(nt 2546-2578) (Applied Biosystems, CA), located in the VP1 gene, were used for qPCR detection, as previously described [27], with a C to G modification of nucleotide 2569. For each serum or urine sample, the extraction volume was 200μl and the elution volume was 150μl. Each qPCR reaction was run in triplicate and all results were expressed in copies/ml.

Data analysis
Categorical independent variables were compared using Fisher's exact test and data are presented as number (N) and percent (%) of population. Continuous and ordinal independent variables were compared using the Kruskal-Wallis test and data are presented as means and standard deviations. Longitudinal variables were analyzed using linear or logistic generalized estimating equations (GEE) [28] in order to control for intra-individual correlations introduced by repeated measures. To identify variables associated with viremia and viruria, an exploratory univariable logistic regression screen was utilized with an inclusion P value cut off of 0.15 [29] for multivariable analysis. A stepwise selection process starting with the candidate variable with the lowest P value was undertaken to identify variables for multivariable analysis. Due to a low event number, a multivariable analysis for viremia as the outcome was not performed, rather a combined outcome of viruria and viremia together was analyzed. Variables were added in a step-wise fashion based upon their P value and retained in the model if their P value was 0.10. This process proceeded until all variables had a P value 0.10 or were identified as significant confounding variables. Logistic regression data are presented as an odds ratio with 95% CI and P value. All P values presented are two-tailed P values with P < 0.05 considered statistically significant. Statistical analysis performed on Stata version 14 statistical software, StataCorp LP, College Station, Tx. Graphics performed on GraphPad Prism Version 5.0 software, GraphPad Software, La Jolla CA.

Patient characteristics and outcomes
29 patients were enrolled and 28 underwent transplantation. One transplant was canceled due to donor complications. Patients were retrospectively stratified into groups with no viral reactivation (BK Negative), viruria only and viremia. No significant differences were found in recipient, donor or transplant characteristics between the 3 groups (Table 1). 7/28 (25%) patients had BKPyV reactivation in either urine or blood by standard of care clinical screening. Two additional patients were found to be viruric by research lab qPCR, making 9 (32%) total reactivations. 6 (21%) developed viruria only and 3 (11%) developed viremia ( Table 2). 1/29 (4%) patient was found to have viruria prior to transplant. 3 patients developed viruria within the first month post-transplant and all reactivations occurred within 6 months. Viremia developed between 3 and 6 months in all 3 cases with peak viral loads of 27,022 copies/ml, 2,111 copies/ml and 19,031 copies/ml. There were no cases of sustained viremia or BKN in the study period. All 3 viremic patients were treated with IS reduction and 2/3 (67%) were treated with Leflunomide. 1/6 (17%) viruric patient was treated with IS reduction. Viremia patients had a trend to higher serum creatinine at 12 months post-transplant at 1.60 (±0.26) mg/dL compared to viruria patients at 1.22 (±0.41) mg/dL and BK negative at 1.31 (±0.30) mg/dL (P = 0.20), but this did not attain statistical significance. There were 4 (14%) patients who developed acute rejection, none of whom developed BKPyV reactivation.
Each of the subtypes of both CD4 + and CD8 + T-cells were analyzed for the expression of activation markers CD38 and HLA-DR as well as the exhaustion marker PD-1. At one month post-transplant, viremic patients were found to have fewer CD4 + central memory cells  42) and BK negative (33.84±7.77) (P = 0.03). There were no other significant differences among activation markers found in any other subset. There were no significant differences among any subset expressing exhaustion marker PD-1.

Factors associated with BKPyV reactivation
In an exploratory analysis, the clinical and experimental variables were screened utilizing GEE logistic regression to evaluate their association with BKPyV reactivation, a combined endpoint of viremia and viruria (   (Table 5). No clinical characteristics or immune suppression medications or troughs were significant predictors. A multivariable model was not attempted due to the low number of viremia patients.

Discussion
BKN is an important and preventable cause of allograft dysfunction and loss. We have performed a longitudinal analysis of the adaptive immune response to BKPyV reactivation, including the timing and magnitude of the BKPyV-specific T-cell response in both CD4 + and CD8 + cells and the humoral immune response. Additionally, we quantified T-cell phenotypes to attempt to identify a profile which would be permissive of viral reactivation. Lastly, we evaluated these experimental factors alongside clinical data to fully assess risk of viral reactivation. Previous work has demonstrated that a robust CD4 + T-cell response to viral capsid proteins is associated with previously resolved BKPyV reactivation [24,[31][32][33], and an early post-transplant CD8 + T-cell response [34], specifically polyfunctional CD8 + T-cells [35] is associated with viral control. Our data demonstrate that the presence of CD4 + and CD8 + BKPyV-specific Tcells is common in the pre-transplant population, and that the prevalence of BKPyV-specific T-cells drops immediately after transplant. Our data echoes prior studies which suggest the absence of CD4 + , CD8 + , or both BKPyV-specific T-cells pre-and early post-transplant may be a risk factor for viral reactivation [19]. We have shown an increase in both CD4 + and CD8 + BKPyV-specific T-cells in viremic patients and that this is associated with clearance of viremia in all cases. Our data also show a significant, though less substantial, CD8 + BKPyV-specific Tcell response in those with viruria, also associated with viral control. Our findings suggest a role for monitoring the emergence of BKPyV-specific T-cells in those with BK reactivation identified by screening. This would serve as a guide to identifying those with an adequate response and likely, therefore, to have self-limited reactivation and in whom IS reduction is unnecessary. Additionally, tracking BKPyV-specific T-cells could be used to guide the magnitude and monitor the effectiveness of step-wise IS reduction to prevent BKPyV progression and prevent excess risk of allograft rejection.
Our data adds to prior clinical observations indicating that having experienced a humoral response does not give full protection from post-transplant viral replication [24,36]. It also confirms the course of BKPyV-specific antibody response follows the level and duration of BKPyV replication [22,24,25,37]. By the first month post-transplant, all patients with viral reactivation had positive anti-BKVPyV antibodies, and we demonstrated those titers increased in viremic and viruric patients with time, but remained unchanged in BK negative patients.
We further evaluated circulating T lymphocytes by phenotype, which revealed several important distinctions between the groups. The proportion of T-cells that are CD4 + was lower and CD8 + T-cells were higher in viremic patients than viruric or BK negative patients before transplant. Furthermore, a significantly greater proportion of CD8 + T-cells had a terminally differentiated effector phenotype prior to transplant in viremic patients, a phenotype they regained several months after transplant. Studies have identified an increased fraction of terminally differentiated CD8 + T-cells along with low proportions of CD4 + and high proportions of CD8 + T-cells as a marker of both age [38,39] and uremia associated [40,41] immune senescence. Our results suggest that prior to ever receiving an immune suppressive regimen, patients who later become viremic demonstrate a T-cell profile associated with immune senescence.
Our analysis of clinical and experimental factors associated with BKPyV reactivation and viremia identified markers of increased immune suppression, such as the presence of diabetes mellitus or GN, conditions associated with immune dysfunction or immune dysregulation [42][43][44][45], as the cause of renal failure, and of immune senescence, such as lower proportions of CD4 T-cells and higher proportions of CD8 T-cells. They also include potential markers of increased BKPyV burden, such as increased anti-BKPyV IgG titers.
Our exploratory analyses demonstrate lower CD4 and higher CD8 proportions were risk factors for viremia and viruria and was also seen pre-transplant. CD4 T-cells are typically twofold higher than CD8 T-cells, and a drop in CD4/CD8 ratio has been associated with immune senescence and mortality in elderly individuals [38]. This same profile has been described in CKD patients irrespective of whether they are yet on RRT or RRT modality, and does not improve post-transplant [40,41,[46][47][48]. Immunologic aging and senescence are also associated with reduced naïve cells and expansion of terminally differentiated effector CD8 + T-cells [48,49], a feature which was also present in viremic patients in our analysis. This leads to estimates that the adaptive immune system of ESRD patients is equivalent to healthy individuals 20-30 years older [46,50]. Furthermore, these phenotypic changes seem to be driven or exacerbated by persistent CMV infection [38,49]. The combination of low CD4/CD8 ratio and CMV seropositivity has been identified as an "immunological risk profile" and recently shown to predict severe bacterial infections and opportunistic infections after kidney transplant [51]. It has also been associated with a decreased risk of acute rejection [52]. Our study demonstrates for the first time in the literature that patients with features of an immune senescence phenotype can be identified pre-transplant and that these persons may be at increased risk of BKPyV reactivation. Additional clinical information, such as causes of renal failure associated with immune suppression may further inform the risk profile.
Limitations of our study include a small number of patients, reducing the power of the study to statistically identify all apparent differences. The small event number is particularly important for our exploratory logistic analysis of risk factors for BKPyV reactivation and viremia. The screening of large numbers of clinical and experimental variables for these analyses increases the risk of type 1 error, and these analyses should be viewed as hypothesis generating. Our patients were from two centers, utilizing mostly rATG and an early corticosteroid withdrawal protocol, almost all from living donors and a high percentage were preemptive transplants, potentially limiting generalizability. The frequency of BKPyV-specific T cells is low, requiring ex-vivo stimulation to study, limiting the clinical applicability of this method, though it is a common method used for study and is thought to represent the in vivo conditions [53]. Our longitudinal data were collected prospectively and reflect "real world" conditions of screening and response to intervention; they also incorporate both experimental and clinical data. Our study serves as a hypothesis generating pilot study, and results need to be replicated in larger studies. Further research on larger numbers of patients, in patients with non-living donors and in BKN patients is needed.
Supporting information S1 Table. CD4 T-cell phenotype. Means and standard deviation of CD4 + T-cell phenotypes as described in Fig 2A. (DOCX) S2 Table. CD8 T-cell phenotype. Means and standard deviation of CD8 + T-cell phenotypes as described in Fig 2B. (DOCX) S1 Dataset. All obtained data. All collected and de-identified data used for analysis.