2.1 Study population

The study population consisted of discovery and an independent validation cohort, both issued from recipients of a first kidney (or kidney–pancreas) graft followed at the Lyon University Hospital Transplantation Center (France).

The discovery cohort included 34 consecutive adult patients who participated in the flu vaccination campaign. The exclusion criteria included an age over 70 and the presence of circulating DSA or ongoing rejection, to eliminate potential bias on the flu vaccine response due to immunosenescence or the majoration of immunosuppressive therapy. Other exclusion criteria included any counter-indication to influvac® vaccine, ongoing infectious disease, pregnancy, intravenous immunoglobulins injection in the last 3 months. Patients with undetectable TTV DNAemia were excluded due to the lack of a reliable assay for screening anti-TTV IgG, making it unclear whether these patients effectively controlled TTV replication or were uninfected. The nature and trough levels of the immunosuppressive drugs were recorded at the time of vaccination, as well as proteinuria and estimated glomerular filtration rate (eGFR), estimated by the CKD-EPI formula. Patients were injected with influvac® vaccine according to the flu vaccination campaign guidelines. Assessments of antibody-response and T-cell activatability (see below) were conducted at the peak of the response (between 21 and 28 days after vaccination). The medical records of these patients were retrospectively screened for pre-defined events related to over-immunosuppression. These events included: (i) infections (bacterial, viral, or fungal) requiring or prolonging hospitalization and (ii) the development of de novo solid or hematological malignancies (including in situ basal and squamous cell skin carcinomas) within the 4 years following vaccination. Infections treated in the outpatient setting were not included as to limit recall bias. TTV DNAemia was measured retrospectively in serum samples collected prior vaccination. A control group for this discovery cohort comprised 124 consecutive healthy volunteers with detectable TTV DNAemia who were accepted for a living kidney donation in Groningen Transplantation center (The ethical approval number of this study is METc 2008/186).

The validation cohort consisted of a total of 93 consecutive kidney and kidney pancreas recipients transplanted in Lyon University Hospital (France) between January 1, 2014 and December 31, 2015 who were prospectively enrolled 12 months after transplantation. Exclusion criteria were the same as in the discovery cohort. Relevant clinical events related to over-immunosuppression (see above) and under-immunosuppression (appearance of de novo DSA, episode of biopsy-proven rejection, and graft loss) were recorded for 36 months after the inclusion (up to 48 months post-transplantation). The trough level of immunosuppressive drugs, proteinuria, eGFR, and presence of DSA were prospectively monitored during the follow-up period, at least annually, with additional checks conducted in immunizing situations (e.g., transfusions) or when rejection was suspected. “For-cause” biopsies were performed in case of allograft dysfunction (onset of proteinuria and/or decrease eGFR) or appearance of de novo DSA. TTV DNAemia was retrospectively measured in plasma samples prospectively collected at predefined post-transplantation timepoints (12, 24, and 36 months). Undetectable TTV DNAemia at each of the three follow-up timepoints was considered an exclusion criterion.

The clinical and research activities being reported are consistent with the Principles of the Declaration of Istanbul. All participants from both cohorts gave written consent. The study was approved by the Local Institutional Review Board (approval number: 2020-A02918-31).

2.2 Influenza vaccination

All patients from the discovery cohorts received a subcutaneous injection of the Influvac® vaccine. The vaccine was adjuvant-free and contained antigens from three distrinct influenza strains (A/California, (H1N1pdm09), A/Switzerland (H3N2), and B/Phuket). Since our interest was for primary humoral response, the analysis was focused on the response to Switzerland virus, the only of the three strains that was selected for the northern hemisphere influenza vaccine formulation for the first time.

2.3 T-cell residual activatability

Blood was collected immediately prior influenza vaccination in patients of the discovery cohort. Peripheral blood mononuclear cells (PBMC) and plasma were isolated using Ficoll gradient centrifugation. PBMCs were cultured at 37°C in 5% CO₂ in 1 mL of patient's own plasma, with and without anti-CD3/CD28 beads (Gibco DynabeadsR). After 24 h, anti-CD3/CD28 beads were removed using a magnet and the cells were stained 30 min at room temperature in the dark with fluorescent antibodies. Anti-CD3 (UHCT1), anti-CD4 (SK3), anti-CXCR5 (RF8B2), anti-CXCR3 (1C3), anti-CCR6 (11A9), anti-CD25 (2A3), anti-CD40L (TRAP1), anti-ICOS (ISA-3) (all from BD Biosciences), and viability dye LIVE/DEAD Aqua (Invitrogen) were used. A FACS ARIA II flow cytometer was used for flow cytometry. Data were analyzed with BD FACS Diva (BD Biosciences) and FlowJoR software (FlowJo, LLC). All samples were run on the same instrument, which was calibrated with Rainbow Calibration ParticlesR before each acquisition (SpheroTech).

2.4 Anti-hemagglutin antibody titer

Blood was collected immediately prior influenza vaccination and between 21 and 28 days later in the patients of the discovery cohort.

Normalized titer of anti-hemagglutinin antibody was measured using hemagglutination-Inhibition assay. Briefly, a fixed amount of Switzerland viral strain was added to serial dilution of the serum of patients and the plate was incubated for 1 h in the presence of red blood cells. Influenza virus has the property to bind to sialic acid receptors on erythrocytes. When antibodies against influenza virus are present, they prevent the attachment of the virus to red blood cell and therefore inhibit the hemagglutination. The highest dilution of serum that prevents hemagglutination defines the HI titer of the serum and corresponds to the antibody titer against the tested influenza virus strain. The analysis focused on Switzerland viral strain because it was selected for the northern hemisphere influenza vaccine formulation for the first time in 2015 and we wanted to identify the patient at risk of primary humoral response.

2.5 TTV DNAemia quantification

TTV DNAemia was quantified the day prior influenza vaccination using the TTV R-GENE® kit following the manufacturer's (bioMérieux) instructions. Briefly, Internal Control 2 was added to all samples and to the negative control. Then, nucleic acids were extracted from 200 µL of samples and eluted in 50 µL elution buffer using the NucliSENS® easyMAG® platform (bioMérieux). Ten microliters of extracted DNA were added to 15 µL of the ready-to-use amplification mixture. In each run, one sensitivity control and four quantification standards were included. Thermal cycling was performed for 15 min at 95°C, followed by 45 cycles at 95°C for 10 s and 60°C for 40 s using the ABI7500Fast (Applied Biosystems). Results were recorded as copies/mL (cp/mL) as determined by the quantification standards range.

Analytical performances of the TTV R‑GENE® kit was assessed in serum samples by the manufacturer (BioMerieux). The LoD95% in serum was confirmed at 250 cp/mL on eight TTV species (1, 6, 7, 8, 10, 24, 27, and 29—linearized plasmids). The tested concentrations for precision and linearity allowed to demonstrate a quantification range from 2.4 to 9.0 log10 cp/mL. For the multiple-amplification platforms study, the maximum absolute difference between all tested platforms and the QS5 used as reference was 0.5 log10 cp/mL, with a maximum average of 0.2 log10 cp/mL.²³

2.6 Statistical analyses

Categorical variables were expressed as percentages and continuous variables were expressed as mean ± standard deviation (SD). Differences between the groups were evaluated by Mann–Whitney and Wilcoxon tests for quantitative variables, and chi-square tests for categorical variables. Spearman correlation was used for correlation matrix analyses. Survival data were compared using the Log-Rank test.

The lower and upper thresholds of TTV DNAemia defining under- and over-immunosuppression were determined using different approaches. The lower threshold, under which patients are considered as non-immunocompromised, and therefore expected to present normal antibody responses (risk of rejection), was established using the data from the healthy volunteers of the Groningen cohort, defined as the average TTV DNAemia plus 2 standard deviations. The upper threshold, above which patients are expected to experience more complications of over-immunosuppression, was defined using a receiver operating characteristic (ROC) curve approach on data from the longitudinal follow-up of the discovery cohort.

Multivariate analyses were performed using logistic regression for categorical outcomes and Cox Proportional Hazard regression for survival data. Variables that were associated with the outcome in univariate analyses (p < 0.2), or those suspected to be associated with the outcome based on literature, were included in the multivariate analyses. Of note, transplantation vintage was, by design (see above), the same for all patients from the validation cohort and was therefore not included in the corresponding multivariate analysis.

Longitudinal TTV DNAemia levels in the validation cohort were evaluated by quantifying the percentage of values within the target range established with the discovery cohort. After descriptive analyses, a 50% cutoff was determined as an appropriate threshold, as it conferred sufficient statistical power by dividing the validation cohort into two equal subgroups, while remaining clinically relevant.

All the tests used were two-sided. The test used for comparison is indicated in the figure legends. The differences between the groups were considered statistically significant for p < 0.05.

Statistical analysis and graphs were performed using Prism software 8 (GraphPad), CIRCOS software, and R software.

3.1 TTV DNAemia correlates with residual activatability of circulating T cells of renal transplant recipients

Calcineurin inhibitors (CNI, i.e., ciclosporine and tacrolimus), which currently represents the corner stone of maintenance immunosuppression, block the early steps of TCR signaling cascade, thereby preventing the upregulation of many genes, including the α chain of IL-2 high-affinity receptor (CD25).²⁴ In a previous study, our group has reported that despite having adequate trough levels of CNI, up to 20% of renal transplant recipients displayed normal expression of CD25 on T follicular helper (Tfh) cells following in vitro stimulation (i.e., “residual activatability”), a status that was associated with an increased risk to develop de novo DSA.²⁵

To determine whether TTV DNAemia could be a potential biomarker for insufficient maintenance immunosuppression in renal transplant recipients, we first evaluated how it correlated with the residual activatability of the various T-cell subsets. Briefly, the PBMCs from 34 kidney transplant patients (Table 1) on CNI-based immunosuppression [28 (82.3%) on tacrolimus and 6 (17.7%) on ciclosporin] were collected at 23 ± 17 months after transplantation and cultured 24 h with anti-CD3/CD28 microbeads in the presence of patient's own plasma containing relevant concentration of IS drugs (Figure 1A). The PBMCs of seven healthy volunteers (51 ± 9.96 years old, 2/7 [28.6%] male) served as controls. Flow cytometry was used to (i) identify the seven distinct subsets of CD3+ T cells: four subsets of CD8+ cytotoxic T cells involved in TCMR (Figure 1B) and three subsets of CD4+ CXCR5+ Tfh²⁶ involved in donor-specific antibodies (DSA) generation (Figure 1C), and (ii) to measure the level of expression of the activation marker CD25 on each subset at baseline and post-activation (Figure 1B–D). As expected, a drastic upregulation of CD25 was observed on the cells of the seven T-cell subsets for all (7/7, 100%) the healthy volunteers (Figure 1D). Although the values were reduced in immunosuppressed kidney transplant patients (Figure 1D), the post-activation expression of CD25 was not null, indicating a residual T-cell activatability, which was highly variable between kidney transplant patients (Figure 1D).

The statistical relation between TTV DNAemia and the residual activatability of the seven T-cell subsets was then analyzed. An inverse correlation between TTV DNAemia and residual activatability of effector cytotoxic T cells was observed (Figure 1E). Furthermore, this was also observed for all three subsets of Tfh (Figure 1E), particularly Tfh2 and Tfh17, which are key for providing help to B cells.²⁶ These results suggest that low TTV DNAemia transplant recipients could be at higher risk to develop de novo DSA and therefore antibody-mediated rejection (AMR).^{17, 27} Interestingly, in contrast with TTV DNAemia, there was no statistical correlation between tacrolimus trough levels and the residual activatability of any of the seven T-cell subsets (Figure 1E).

3.2 Low TTV DNAemia correlates with antibody response against T-cell-dependent antigens in renal transplant recipients

A control cohort consisting of 124 healthy volunteers (56 ± 10 years old, 55/124 male [44,3%]) was used for comparison with the 34 kidney transplanted patients of the discovery cohort. As expected, TTV DNAemia was significantly lower in healthy controls (Figure 2A).

Interestingly, TTV DNAemia was highly variable among transplant recipients, some having TTV DNAemia within the range of what observed in healthy volunteers (red dots in Figure 2A). Setting the Z-score to 2, a classical approach to identify outliers,²⁸ we defined a threshold of 3.75 log₁₀ cp/mL under which renal transplant patients could be considered under-immunosuppressed (Figure 2A). Interestingly, this value is consistent with what recently observed with the same assay in a large cohort of healthy blood donors.²⁹

The incidence of de novo DSA is low (estimated 2%–10% per year²⁰) and only four patients from the discovery cohort developed de novo DSA during follow-up (and only one biopsy-proven antibody-mediated rejection). This low incidence does not allow to directly test the hypothesis that transplant recipients with a TTV DNAemia <3.75 log₁₀ cp/mL are at higher risk to develop de novo DSA. Instead, we reasoned that, since DSA are directed against the donor-specific (HLA or non-HLA) polymorphic proteins, their generation results from the same prototypical T-cell-dependent humoral response at stake for the response to any protein-based vaccine. Therefore, we took advantage of the campaign of influenza vaccine³⁰ that can be considered a standardized synchronized immune stimulation with hemagglutinin, a prototypical thymo-dependent antigen.³¹

The normalized titer of anti-hemagglutinin antibody was measured just prior vaccination and at the peak of the response and compared between patients whose TTV DNAemia was below versus above 3.75 log₁₀ cp/mL (Figure 2B). In line with our hypothesis patients whose TTV DNAemia was below 3.75 log₁₀ cp/mL developed significantly higher antibody titers (Figure 2C). In contrast, the magnitude of the antibody response could not be predicted from the routine monitoring of the trough levels of CNI (Figure 2D).

3.3 High TTV DNAemia correlates with complications associated with over-immunosuppression

To evaluate whether TTV DNAemia could serve as a biomarker for over-immunosuppression, we retrospectively reviewed the clinical records of the 34 renal transplant recipients of the discovery cohort focusing on serious infectious complications (defined as any infection requiring or prolonging hospitalization) and cancerous complications (diagnosis of solid or hematological malignancies). Over a follow-up period of 50 ± 16 months postvaccination, 21 events linked to over-immunosuppression were identified (Table 1). These included: 12 (35.3%) serious infectious events (n = 8 bacterial, n = 3 viral, and n = 1 fungal infections) and 5 (14.7%) cancers (n = 3 skin cancers and n = 2 other solid organ tumors).

The optimal cutoff value for TTV DNAemia (5.1 log10 cp/mL), which best distinguishes patients who have experienced complications of over-immunosuppression from those who have not, was established using ROC curve analysis (Figure 3A,B). The multivariate survival analysis identified TTV DNAemia >5.1 log10 cp/mL (HR: 9.41 [2.52-35.3]; p < 0.001; Table 2 and Figure 3C) as the only baseline variable that correlates with the subsequent occurrence of complications due to over-immunosuppression (Table 2).

3.4 Independent validation of the value of TTV DNAemia as a biomarker for long-term complications of inadequate immunosuppression

Having defined the lower (3.75 log₁₀ cp/mL) and upper (5.1 log₁₀ cp/mL) cut-off values of TTV DNAemia associated with the occurrence of complications due to, respectively, under- and over-immunosuppression, we went on validating these thresholds in an independent prospective cohort of 93 patients.

One patient with undetectable TTV DNAemia at all time points of analysis was excluded from analysis. The characteristics of the 92 remaining patients of the validation cohort are presented in Table 3. Briefly, all 92 patients received a kidney or a kidney-pancreas grafts (n = 16; 17,4%) at Lyon University Hospital and were prospectively enrolled at 12 months post-transplantation. All received an induction therapy with either thymoglobulin (n = 58; 63%) or basiliximab (n = 34; 37%). Of note, the nature of the induction therapy did not have an impact on the TTV DNAemia of patients during the late follow-up period of the study (data not shown). All patients except 2 were on CNI-based maintenance immunosuppression (n = 11, 11.9% on ciclosporin A and n = 79, 85.9% on tacrolimus). The target trough levels for tacrolimus were 5–7 ng/mL (80–120 µg/mL for ciclosporin). Complications due to over-immunosuppression (infections and cancers) and under-immunosuppression (appearance of de novo DSA, episode of biopsy-proven rejection, and graft loss) were recorded over a follow-up period of 3 years (Table 3).

TTV DNAemia was measured retrospectively in prospectively collected samples at 12, 24, and 36 months post-transplantation (Figure 4A). As expected, the mean TTV DNAemia was higher at M12 (5.26 ± 1.8 log₁₀ cp/mL) than at M24 and M36 (4.11 ± 1.8 and 3.92 ± 1.9 log₁₀ cp/mL, respectively; p < 0.001, one-way ANOVA test), which reflects the tapering of maintenance immunosuppression beyond the first year of transplantation. Interestingly, while all these patients were managed according to the same trough level-based rules for adjusting the dose of CNI, massive interindividual variations in TTV DNAemia could be evidenced, suggesting that trough levels of CNI poorly reflect the individual depth of therapeutic immunosuppression. In fact, only a minority of patients had a TTV DNAemia within 3.75 and 5.1 log₁₀ cp/mL, the target range defined from the discovery cohort (n = 27 [30%], n = 32 [36.7%], and n = 34 [39.5%] at, respectively, 12, 24, and 36 months post-transplantation; Figure 4B). Interestingly, and consistently with our hypothesis, patients with TTV DNAemia within the target range at >50% of time points were those with the lowest incidence of complications due to “inadequate” (i.e., under- or over-) immunosuppression (Figure 4C). Furthermore, the total number of complications due to inadequate immunosuppression was also lower in patients with TTV DNAemia within target at >50% of time points (Figure 4D).

We conducted a multivariate analysis using multinomial logistic regression to determine the variables associated with the occurrence of a complication due to “inadequate” immunosuppression. TTV DNAemia and CNI trough levels within the target range >50% of controls were forced into the model (Table 4). TTV DNAemia within target at >50% of timepoints was the sole variable associated with a reduced risk (OR: 0.27 [0.09–0.77]; p = 0.019; Figure 4E; Table 4) to develop a complication due to “inadequate” immunosuppression. Of note, despite that CNI trough levels were controlled prospectively and much more frequently than TTV DNAemia (9.8 ± 3.6 vs. 3 times per patients, respectively) over the 3 years follow-up period, CNI trough levels within the target range at >50% of controls were not associated with a reduced risk to develop a complication due to “inadequate” immunosuppression (p = 0.93; Table 4).

Finally, additional analyses were conducted using a higher upper TTV DNAemia cutoff (6.1 log10 cp/mL) to assess changes in biomarker sensitivity. This new cutoff was proved less effective than the 5.1 log10 cp/mL cutoff in identifying patients at higher risk of complications (Figure S2A). Moreover, multivariate logistic regression analysis using the 6.1 log10 cp/mL cutoff no longer demonstrated an association between TTV DNAemia and the risk of complications due to inadequate immunosuppression (Figure S2B).