**1. Introduction**

Immunosuppressive therapy is vital for organ transplantation medicine; in the last 20 years, antirejection treatment has improved enormously thanks to the increased availability of new antirejection drugs. All these drugs, however, lead to some degree of immunosuppression, and subsequently increased infection risk. Measuring trough levels of antirejection drugs is currently standard of care in determining the optimal dosing of these drugs, but is well recognized that these trough levels do not accurately reflect the risk of under-immunosuppression, potentially resulting in rejection, or over-immunosuppression, potentially resulting in increased infections [1–3]. It is, therefore, important to identify the parameters that reflect the net immune status and have predictive capacities for long-term outcomes. Torquetenovirus (TTV) is a single stranded, negative sense, non-encapsulated DNA virus; it was first discovered in 1997 by Nishizawa et al. [4–6] and is present in 46%–100% of healthy people [7]. The international commission on taxonomy (ICTV) recognizes 29 di fferent genotypes, but their relative circulation has not been researched su fficiently [8]. Attempts to discover viable antibody assays have been hampered by the hypervariable nature of the viral capsid protein [9]. A few assays have been described by various groups; however, these assays have not proven to be scalable for large-scale clinical use [9–11]. In recent years, TTV has been studied by various research groups as a potential marker of immunosuppression following transplantation. TTV levels have been shown to increase at the start of antirejection treatment, reaching a relative plateau phase between 3 and 6 months after transplantation [12]. It is currently thought that an ideal viral loads exists for each type of organ transplantation, signifying optimal antirejection dosing. Viral loads above this ideal level have been shown to increase the chance of infections, whereas low viral loads have been shown to be associated with increased chance of rejection [13]. No formal cut-o ff TTV levels for optimal immunosuppression have been established, however, since these levels show grea<sup>t</sup> variation between di fferent research groups, even in seemingly similar patient populations. These variations may be due to di fferences in circulation of TTV genotypes or due to di fferences in the PCR test used in these studies.

Additionally, most studies have looked at longitudinal TTV measurements relatively shortly after transplantation. The follow-up periods have also been brief, usually up to one year after solid organ transplantation (SOT). The immunosuppressed condition of these patients and the increased risk of infection nevertheless persists for a lifetime. To date, very little is known about TTV levels past the first year after transplantation, and if these TTV levels are associated with increased risk of infection or rejection in the long term. In this study, we focused on TTV measurements after the first year following renal transplantation, and hypothesized that TTV levels are associated with outcome over several years. We studied this in a cohort of 666 renal transplant recipients.

#### **2. Materials and Methods**

#### *2.1. Study Population*

This cohort study was based on a previously described set of 706 renal transplant recipients [14,15]. Included were patients (aged ≥ 18 years) who visited the outpatient clinic of the University Medical Centre Groningen (UMCG), Groningen, the Netherlands, between November 2008 and June 2011, and who had a graft that had been functioning for at least one year with no history of alcohol and/or drug addiction. Of 706 renal transplant recipients that provided written informed consent, we excluded subjects with missing biomaterial (40 cases) from further analyses, which resulted in 666 cases eligible for study. The study protocol was approved by the UMCG institutional review board (METc 2008/186); clinical trials number NCT02811835 and adhered to the Declarations of Helsinki and Istanbul.

#### *2.2. TTV Viral Load Measurements*

Serum samples were stored at −80 ◦C, until analysis. One serum sample per patient was used; this was collected at enrollment. DNA was extracted from thawed serum samples using the eMAG Nucleic Acid Extraction System (bioMerieux, Marcy, France). The Argene R-Gene TTV quantification kit (bioMerieux, Marcy, France) was used to perform qPCR on an Applied Biosystems 7500 (Thermo fisher, Waltham, MA, USA) according to the manufacturer's instructions. Due to limited sample volumes, 100 μL, a 1 in 4 dilution using DMEM, was performed prior to sample extraction (ThermoFisher, Waltham, MA, USA). A control experiment (data not shown) showed no significant di fferences in the Ct values. The R gene assay is designed to detect TTV genotypes 1, 6, 7, 8, 10, 12, 15, 16, 19, 27 and 28 [16–18].

#### *2.3. Clinical End Points*

The primary endpoint of this study was all-cause mortality, death due to infectious causes and death-censored graft failure as secondary aims. Deaths due to infectious causes were defined using the previously specified list of International Classification of Diseases, Ninth Revision, codes 1–139 [19,20]. For example, a patient meeting the criteria, which is positive culture or PCR for *Pneumocystis jiroveci* infection, would be given the code 136.3 and would therefore be classified as dying due to an infection. Graft failure was defined as return to dialysis therapy or re-transplantation. The cause of graft failure was obtained from patient records and was reviewed by a blinded nephrologist. Endpoints were recorded until the end of September 2015 and there was no loss of subjects to follow-up.

#### *2.4. Data Analysis*

Data analyses were performed using SPSS 23.0 for Windows (SPSS Inc., Chicago, IL, USA) and R (R Foundation for Statistical Computing, Vienna, Austria). As no cut-o ff values for low, medium and high TTV load have been established, and to avoid bias, renal transplant recipients were stratified into three equally sized groups based on serum TTV. This created four groups, named undetectable-TTV, low-TTV, medium-TTV, and high-TTV, which were further analyzed. Di fferences in all-cause mortality, death due to infectious causes and graft failure between the four groups were compared using Kaplan–Meier plots and log rank tests. Data are presented as mean ± SD for normally distributed data, as median [interquartile range (IQR)] for non-normally distributed data, and as number (percentage) for nominal data. *t*-Tests, or one-way ANOVA tests with Tukey post-hoc tests, were performed on normally distributed data. Each group was compared with the other three groups. Kruskal–Wallis or Mann–Whitney U tests were performed on non-normally distributed data. Chi-square tests were used on categorical data. A two-sided *p* < 0.05 was considered to indicate statistical significance in all analyses.

Prospective associations of TTV on study endpoints were explored by means of Cox regression analysis. Risk of all-cause mortality, death due to infectious causes and graft failure are presented as HR [95% confidence interval]. In these analyses, associations were adjusted in a cumulative fashion for potential confounders, including age, sex, eGFR, proteinuria (model 1), time since transplantation (model 2) and the number of immunosuppressant medications taken (model 3). Cox regression models were built in a stepwise fashion to avoid over-fitting. The proportionality of hazards for covariates was investigated by inspecting the Schoenfeld residuals. eGFR and age were included as categorical variables with equal numbers of events in each group, as eGFR and age breached the proportionality of hazards assumption as continuous variables.

The optimal cuto ff values for death due to infectious causes was identified by using Youden's Index [21] in the area under receiver operating characteristics (auROC) curve. This approach was not used for all-cause mortality as the threshold was biologically unrealistic, therefore a sensitivity of 75% was set and the threshold was calculated. Sampling by replacement was used to create 1000 bootstrapped samples of equal size from within the study population, this was then used within our multivariate Cox regression models to validate the association of the TTV DNA thresholds with risk.

To assess in further detail how TTV levels change over time, and to validate our TTV test in patients with older transplants, the patients were subdivided into two groups. The first group had had a transplant 12 to 24 months prior to analysis. The 2nd group had had transplants over 24 months prior to analysis. Cox regression analysis was also performed as previously described.
