Torque Teno Virus as a Novel Biomarker Targeting the Efficacy of Immunosuppression After Lung Transplantation

Abstract

Torque teno viruses (TTV) are small DNA-viruses, of the genus Alphatorquevirus, whose replication is linked to immune status. TTV load may be an indicator for efficacy of IS in lung transplant recipients (LTRs). In a prospective single-center-study 143 LTRs were followed up and tested by quantitative TTV-DNA PCR. Using multivariate Cox-regression contribution of TTV-load to the occurrence of severe infections, chronic lung allograft dysfunction (CLAD), acute cellular rejection (ACR), and death was assessed. During follow-up 28 (20%) patients developed infections with a rate of 7.7 per 100 patient-years (PY). The hazard-ratio (HR) associated with a one-log10 increase of TTV-load before the event was 5.05. CLAD occurred with a rate of 6.0%-PY. HR for a 1 log10 increase of the lowest TTV level before the event was 0.71 (CI: 0.54–0.93). TTV-load predicts clinical events and may be useful to optimize IS during the first years of follow-up of LTRs.

lung transplantation, biomarker, torque teno virus, individualized immunosuppression

Monitoring of immunosuppressive therapy after lung transplantation is still one of the major challenges in posttransplant care and is so far based on measuring immunosuppressive drug blood levels to achieve concentrations within the established therapeutic range. While this strategy largely prevents drug toxicity, it is insufficient to determine the individual status of immunosuppression and, consequently, to tailor the dose to each patient’s requirements. The inverse relationship between the level of immunosuppression and the risks of infection and rejection leaves only a narrow therapeutic window that is optimal for treatment. Therefore, new approaches, suitable to reflect each patient’s individual status of immunosuppression, are warranted to monitor immunosuppressive therapy and to improve the efficacy and safety of treatment.

Noninvasive biomarkers would be helpful tools to individualize the immunosuppressive medication and may predict acute rejection, chronic rejection, or infection episodes [1].

Torque teno viruses (TTVs) are persistent, small, nonenveloped, single-stranded DNA viruses that belong to the family Anelloviridae [2–5]. TTVs are ubiquitous and prevalent in about 90% of the human population. A large number of genetically distinct TTV strains exist and coinfections with different TTV strains are common [4, 6, 7]. TTVs often cause chronic low-level viremia in the immunocompetent host [8] but, to date, no clear association of TTV to any human illness has been found [9–11]. The different TTV strains were summarized in the genus Alphatorquevirus [12].

A close relationship between alphatorquevirus replication and immune status modification exists. Adaptive immune responses play an important role in controlling alphatorquevirus replication and reducing the level of alphatorquevirus viremia [13]. Maggi et al have shown that immunocompromised individuals, such as human immunodeficiency virus–infected persons with low CD4⁺ T-cell levels and patients under drug-induced immunosuppression, display substantially higher levels of viremia than immunocompetent persons [14]. Retrospective studies on transplant recipients as well as virome analyses have provided evidence that the level of alphatorqueviruses may be associated to the degree of immunosuppression after transplantation [15–18]. As shown in a study on lung transplant recipients, the total alphatorquevirus load in blood, including all TTV genogroups, increased markedly in response to immunosuppressive therapy, confirming that these viruses take advantage of a reduction of immunocompetence and that high posttransplant alphatorquevirus levels of >9.2 log₁₀ copies/mL were associated with development of microbial infections [15]. It was further shown that lower levels of alphatorquevirus were significantly associated with episodes of organ rejection [19], and in lung transplant recipients this was especially associated with alphatorquevirus levels <7 log₁₀ copies/mL [20].

Similarly, it was shown in liver transplant recipients that a lower-than-average burden of anelloviruses is indicative of insufficient immunosuppression and associated with acute rejection episodes, even though these patients were subject to the immunosuppressant levels prescribed per protocol [21]. In addition, a cross-sectional study in patients after kidney transplantation [22] indicated that higher alphatorquevirus levels were associated with a decreased risk for antibody-mediated rejection.

The objective of the present study was to assess for the first time in a prospective large cohort of lung transplant recipients, whether levels and kinetics of plasma alphatorquevirus DNA within the first years after lung transplantation are related to the development of acute and chronic rejection and infectious complications and whether alphatorquevirus DNA load is thus a useful marker for identification of the efficacy of immunosuppression after lung transplantation.

MATERIALS AND METHODS

Patients and Methods

All patients transplanted between March 2013 and June 2015 who survived >6 months and had a minimum of 10 blood samples tested for alphatorquevirus were included in this prospective study. A total of 143 of 280 lung transplant recipients fulfilled the inclusion criteria (flowchart 1 inclusion/exclusion, Supplementary Figure). Patient characteristics at time of transplantation are presented in Supplementary Tables 1 and 2. The study was approved by the local ethics committee (EK-Nr: 1710/2014).

Alphatorquevirus load was measured at intervals between 2 weeks and 2 months depending on the time after transplantation and on the frequency of outpatient visits. Overall, 3020 samples were tested, and the mean number of alphatorquevirus analyses was 21 (standard deviation [SD], 7) per patient (range, 10–47).

Induction therapy with alemtuzumab (Campath) was prescribed in 133 patients, 2 received rabbit antithymocyte globulin (ATG) (Fresenius S, Biotech), and 8 had no induction therapy. All patients who received Campath were on dual therapy with calcineurin inhibitors (tacrolimus; target level 8–10 ng/mL first 3 months, 6–8 ng/mL months 4–12) and steroids within the firstposttransplant year; mycophenolate mofetil was added after 12 months (1–2 g/day). All other patients received triple therapy with tacrolimus, mycophenolate mofetil, and steroids (standard protocol).

Patient observation time was 197–1612 days (mean, 937 [SD, 365] days). Clinical data of the patients were obtained from the patients’ charts. Microbial infections were defined as events where, after microbial identification, intravenous antimicrobial treatment and hospital admittance were required. Acute cellular rejection (ACR) was defined following the International Society for Heart and Lung Transplantation (ISHLT) criteria; chronic lung allograft dysfunction (CLAD) was also defined according to ISHLT standards [23].

Determination of Alphatorquevirus Load

Plasma alphatorquevirus DNA was measured by quantitative polymerase chain reaction (PCR) covering the conserved region common to all currently known TTV strains as described previously [15]

Statistical Methods

Data are expressed as mean and SD or as median and range. For each patient and all time points for which alphatorquevirus measurements were available and/or relevant events occurred (these events were death, retransplantation, infection requiring hospital admission, CLAD, and ACR), successive records were constructed, with days since lung transplantation as the time indicator. Last record was time of death, retransplantation, or end of follow-up (1 April 2018), whichever occurred first. These snapshot data were converted into time-span data. Variables that were considered constant across all time spans were age at transplantation, sex, type of induction therapy, and baseline alphatorquevirus level (defined as maximum log₁₀ copies/mL during the first 3 months). Time-varying variables were minimum and maximum alphatorquevirus levels during the previous 3 months for each time point (for the time points <3 months after transplantation, the baseline period was used). Furthermore, tacrolimus levels were included for each time point as time-varying variables, if available; these levels were applied to all subsequent time points until the next measurement occurred. The Cox regression model with the Anderson–Gill counting process and robust variance estimation was applied. This model does not stop further evaluation of a patient if an event occurs because it allows multiple events to occur in a patient. However, events occurring within 1 month were counted as 1 event. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated and Breslow survival curves were drawn for levels of maximum or minimum alphatorquevirus viral load during the preceding 3 months (as previously indicated, these levels were chosen as <7, 7–9.5, and >9.5 log₁₀ copies/mL) [20]. We considered age, sex, type of induction therapy, baseline alphatorquevirus levels, and tacrolimus levels in addition to minimum and maximum alphatorquevirus levels during the preceding 3 months that were primary variables of interest. Except the 2 latter variables, all were removed if they were statistically not significant. Events analyzed by this procedure were infection, ACR, and CLAD. Retransplantation occurred in only 3 patients and could not be evaluated. Deaths were analyzed as well; however, as death in these patients typically occurs as a consequence of the other events, we had no specific hypothesis concerning this event. For infections, we hypothesized that high levels of maximum alphatorquevirus and maybe also minimum alphatorquevirus during the preceding 3 months would be associated with an increased hazard, whereas for ACR and CLAD the opposite was conjectured: that high levels should be associated with a reduced hazard.

The analyses were performed using Stata 13.1 (StataCorp) and Statistica 10.0 (StatSoft) software.

RESULTS

Alphatorquevirus Load of the Patients

In 110 of 143 patients, the pretransplant alphatorquevirus levels were available, 19 of whom had a negative alphatorquevirus PCR; 91 were positive with a mean value of 3.6 (SD, 1.9) log₁₀ copies alphatorquevirus DNA/mL plasma. Pretransplant alphatorquevirus levels were not associated with underlying disease (P = .43), sex (P = .36), or age (P = .31). All pretransplant alphatorquevirus-negative patients became TTV positive after transplantation.

After lung transplantation, the viral load strongly increased in all recipients, reaching a mean peak value of 9.5 (SD, 0.56) copies/mL after 134 (SD, 87) days, followed by a plateau value over the later follow-up, generally between 5 and 10 log₁₀ copies/mL with a slow decline afterward (Supplementary Figure 1).

Alphatorquevirus Kinetics and Adverse Events

The 143 patients provided 365 patient-years (PY) of follow-up. During this period, 28 patients (20%) developed an infection requiring hospitalization (7.7% [95% CI, 5.1%–11.1%] PY). These cases included 4 patients with cytomegalovirus (CMV) disease, 7 with bacterial infections (1 Klebsiella, 3 Pseudomonas, 3 other), 11 with fungal infections, 2 with non-CMV viral infections (respiratory syncytial virus and influenza B), and 4 with tuberculosis (2 Mycobacterium tuberculosis, 2 mycobacteria other than tuberculosis) (Supplementary Table 3). Overall, 22 cases (15% of patients) with CLAD were registered. Hence, CLAD occurred with a rate of 6.0% (95% CI, 3.8%–9.1%) PY. ACR was observed in 11 lung transplant recipients (7.7%) (event rate: 3.0% [95% CI, 1.5%–5.4%] PY). Three patients needed retransplantation and 24 patients died (16.8%) (death rate: 6.6% [95% CI, 4.2%–9.8%] PY).

Cox regression for infection events revealed no significant influence of age at transplantation, sex, type of induction therapy, and baseline alphatorquevirus levels (Table 1). These variables were removed for further analysis. The same holds for all other endpoints studied (in these analyses, ATG had to be removed due to variance inflation) (Tables 2 and 3). Only FK506 (tacrolimus) levels had a significant influence, and high levels were associated with increased risk for all events. These levels, in fact, were included as instantaneous values, which probably reflects therapeutic interventions to prevent allograft failure.

Table 1.

Cox Regression Analyses for Infections Occurring During Follow-up After Lung Transplantation

Predictor	HR	(95% CI)	P Value
Age	1.01	(.98–1.03)	.532
Sex, male	0.51	(.23–1.15)	.106
Campath	0.92	(.31–2.69)	.873
ATG	1.20	(.27–5.37)	.814
TTV BL, log10 copies/mL	1.59	(.74–3.40)	.231
FK506 (tacrolimus) level, ng/mL	1.10	(1.03–1.18)	.003
TTV minimuma, log10 copies/mL	0.90	(.66–1.21)	.483
TTV maximuma, log10 copies/mL	4.59	(2.68–7.86)	<.001

Abbreviations: ATG, rabbit antithymocyte globulin; CI, confidence interval; HR, hazard ratio; TTV BL, baseline level of alphatorquevirus.

^aMinimum and maximum alphatorquevirus level during 3 months before each time point.

Table 2.

Cox Regression Analyses for Acute Cellular Rejection Occurring During Follow-up After Lung Transplantation

Predictor	HR	(95% CI)	P Value
Age	1.03	(.98–1.08)	.238
Sex, male	1.31	(.35–4.92)	.692
Campath	0.72	(.10–5.47)	.755
TTV BL, log10 copies/mL	0.86	(.50–1.48)	.593
FK506 (tacrolimus) level, ng/mL	1.20	(1.06–1.35)	.004
TTV minimuma, log10 copies/mL	0.51	(.27–.96)	.037
TTV maximuma, log10 copies/mL	1.50	(.87–2.59)	.149

Abbreviations: ATG, rabbit antithymocyte globulin; CI, confidence interval; HR, hazard ratio; TTV BL, baseline level of alphatorquevirus.

^aMinimum and maximum alphatorquevirus level during 3 months before each time point.

Table 3.

Cox Regression Analyses for Chronic Lung Allograft Dysfunction Occurring During Follow-up After Lung Transplantation

Predictor	HR	(95% CI)	P Value
Age	1.00	(.97–1.03)	.902
Sex, male	1.11	(.46–2.69)	.822
Campath	0.82	(.19–3.50)	.788
TTV BL, log10 copies/mL	0.81	(.65–1.00)	.052
FK506 (tacrolimus) level, ng/mL	1.05	(.86–1.28)	.645
TTV minimuma, log10 copies/mL	0.71	(.54–.94)	.015
TTV maximuma, log10 copies/mL	0.97	(.71–1.33)	.850

Abbreviations: ATG, rabbit antithymocyte globulin; CI, confidence interval; HR, hazard ratio; TTV BL, baseline level of alphatorquevirus.

^aMinimum and maximum alphatorquevirus level during 3 months before each time point.

High maximum levels of alphatorquevirus viral load during the 3-month time window strongly increased risk of infection (HR, 5.05 [95% CI, 2.94–8.67]). A significant impact (P = .004) was also detected for tacrolimus with an 8% increased hazard for infection from an increase of the FK506 level by 1 ng/mL (Table 4). Whereas in patients with the maximum level staying below 7 log₁₀ copies/mL no infection was estimated, the cumulative frequency of occurrence of an infection in those with a 3-month window level >9.5 log₁₀ copies/mL increased to >40% (Figure 1).

Table 4.

Cox Regression Analyses for the Different Endpoints Occurring During Follow-up After Lung Transplantation

Endpoint	Predictor	HR	(95% CI)	P Value
Infection	FK506	1.08	(1.02–1.14)	.004
	TTV minimum	0.93	(.70–1.25)	.631
	TTV maximum	5.05	(2.94–8.67)	<.001
ACR	FK506	1.15	(1.04–1.27)	.008
	TTV minimum	0.48	(.26–.88)	.018
	TTV maximum	1.47	(.90–2.42)	.126
CLAD	FK506	1.06	(.90–1.24)	.506
	TTV minimum	0.71	(.54–.93)	.013
	TTV maximum	1.01	(.77–1.32)	.945
Death	FK506	1.11	1.02–1.21)	.013
	TTV minimum	2.30	(.86–6.14)	.097
	TTV maximum	0.54	(.20–1.46)	.227

Analysis restricted to tacrolimus levels (FK506 ng/mL) and maximum as well as minimum levels of torque teno virus (log₁₀ copies/mL) during the 3 months before each time point.

Abbreviations: ACR, acute cellular rejection; ATG, rabbit antithymocyte globulin; CI, confidence interval; CLAD, chronic lung allograft dysfunction; FK506, tacrolimus; HR, hazard ratio; TTV BL, baseline level of alphatorquevirus.

^aMinimum and maximum alphatorquevirus level during 3 months before each time point.

Figure 1.

Breslow estimate of cumulative fraction without infection by range of maximum alphatorquevirus (TTV) levels (copies/mL) 3 months before each time point.

For ACR the minimum 3-month alphatorquevirus level was a significant predictor (HR, 0.48 [95% CI, .26–.88]). For each increase by 1 log₁₀ copies/mL, the risk of ACR reduced to about one-half (Table 4). Also for ACR, tacrolimus plasma levels were associated with an increased risk (HR, 1.15 [95% CI, 1.04–1.27]). As mentioned above, this could reflect consequences of therapeutic interventions, as high FK506 should reflect efficient immunosuppression. Cumulative fraction with ACR was increased in those with minimum alphatorquevirus levels <7 log₁₀ copies/mL during the preceding 3 months (Figure 2).

Figure 2.

Breslow estimate of cumulative fraction without acute cellular rejection (ACR) by range of minimum alphatorquevirus (TTV) levels (copies/mL) 3 months before each time point.

For CLAD, the only significant predictor was the minimum alphatorquevirus plasma level during the 3-month time window, with higher levels reducing the risk of dysfunction (HR, 0.71 [95% CI, .54–.93]). For each increase by 1 log₁₀ copies/mL, the risk of CLAD reduced to about 70% (Table 4). In those with plasma levels <7 log₁₀ copies/mL, the cumulative fraction developing CLAD increased to about 30% (Figure 3).

Figure 3.

Breslow estimate of cumulative fraction without chronic lung allograft dysfunction (CLAD) by range of minimum alphatorquevirus (TTV) levels (copies/mL) 3 months before each time point.

Mortality could not be predicted by any of the variables considered and only the minimum plasma alphatorquevirus level indicated some relationship (Table 4).

A formal receiver operating characteristic (ROC) analysis based on the moving window data revealed an upper threshold of 9.2 log₁₀ copies/mL and a lower one of 8.1 log₁₀ copies/mL for infections and CLAD, respectively, with a sensitivity of 87% for infections and 95% for CLAD, and a specificity of 71% for infections and 55% for CLAD.

Association Between Alphatorquevirus Load and Other Parameters

A small but statistically significant correlation of alphatorquevirus plasma levels and drug levels of tacrolimus was found (Pearson r = 0.167). Testing lagged correlations revealed a maximum for the tacrolimus level 5 weeks before (r = 0.232, P < .001). No correlation between sex (P = .069) or age at transplantation (P = .91) and pretransplant alphatorquevirus load was found. In addition, baseline alphatorquevirus levels showed no correlation to subsequent alphatorquevirus peak levels (r = 0.005, P = .962) or time to reach the peak (Spearman R = –0.049, P = .603).

DISCUSSION

One of the most important goals in lung and other solid organ transplantation (SOT) is to tailor immunosuppressive therapy to the individual needs of the patient, avoiding both rejection episodes caused by insufficient immunosuppression, and opportunistic infections and malignancies, which are consequences of overimmunosuppression and remain a significant cause of death after transplantation. While monitoring of immunosuppression is currently performed mainly on the basis of pharmacokinetic characteristics, which do not necessarily predict clinical outcome in the individual patient, it was hypothesized, based on retrospective analyses, that alphatorquevirus load may reflect the level of immunosuppression in patients after lung transplantation [15].

In this first prospective study of a cohort of 143 lung transplant recipients, from whom >3000 plasma samples were analyzed for alphatorquevirus DNA and who were followed up for up to 3 years posttransplantation, we could provide clear evidence for alphatorquevirus DNA load in lung transplant recipients being significantly associated with the development of clinical complications; hence, this can be conjectured as a basis for monitoring individual level of immunosuppression.

A number of smaller and/or retrospective studies on lung transplant recipients have already suggested that alphatorquevirus load may be a marker reflecting the level of immunosuppression [20]. Also in other patient groups as liver and renal transplantation first overall associations, especially between lower alphatorquevirus load and organ rejection, were shown [20, 22, 25].

All of these studies suggested that alphatorquevirus load may be useful as a new specific biomarker reflecting the biologic effect of immunosuppressive drugs after SOT [15, 19, 20].

However, the usefulness of alphatorquevirus load for clinical management is highly dependent on the definition of threshold levels, either suggestive of a high risk for infections or malignancies or of a high risk for development of rejection episodes. We have therefore, based on our previous evaluations, defined patient groups according to specific threshold levels. An alphatorquevirus level of >9.5 log₁₀ copies/mL was taken as critical and possibly reflecting too high a level of immunosuppression [15], and a level of <7 log₁₀ copies/mL was considered to reflect a potentially too low immunosuppression and bearing a high risk for development of rejection, as shown by a recent case-control study [20].

Although we did not directly use these suggested threshold levels within data analysis, as it is not clear how they can be applied prospectively, we defined a 3-month monitoring window and demonstrated that highest alphatorquevirus levels during this period are strongly related to risk of infection with a >5-fold increased risk for each 1 log₁₀ increase of alphatorquevirus load. Considering that the infection rate was 7.7% PY in our patients, occurring at a mean maximum 3-month window level of 9.7 log₁₀ copies/mL, reducing this level to 8 log₁₀ copies/mL is expected to reduce infection rate to about 0.9% PY, and to 7 log₁₀ copies/mL to 0.6% PY. On the other hand, too low a level, reflecting low immunosuppression, increased the risk of rejection and chronic allograft dysfunction. CLAD occurred with a rate of 6.0% PY. Three-month moving window minimum levels of alphatorquevirus were 6.5 log₁₀ copies/mL for CLAD as well as for ACR. Risk of CLAD increased about 40% and of ACR about 2-fold for each 1 log₁₀ decrease of alphatorquevirus levels. The estimated rate of CLAD by increasing the minimum alphatorquevirus level from 6.5 to 8 log₁₀ copies/mL is 2.8% PY. Likewise, increasing the log₁₀ TTV level to 8 copies/mL would be expected to reduce the ACR rate from 3.0% PY to about 1% PY. Because there is always a tradeoff between reducing the risk of infection and increasing the risk of rejection, the proposal to intervene as soon as monitoring of alphatorquevirus reveals maximum levels in excess of 9.5 log₁₀ copies/mL and minimum levels fall below 7 log₁₀ copies/mL gain support from the present investigation. A formal ROC analysis based on the moving window data revealed that the upper threshold of 9.2 log₁₀ copies/mL was associated with a high risk for infections, which is similar to what was previously found [19, 25]. However, instead of the previously defined lower cutoff of 7 log₁₀ copies/mL [20], present investigations found that an alphatorquevirus level of <8 log₁₀ copies/mL was already associated with increased risk of CLAD.

Thus, rather than 7–9.5 log₁₀ copies/mL, a range of 8–9.2 log₁₀ copies/mL should be maintained to minimize the risk of both infections and rejections/allograft dysfunction.

The alphatorquevirus load values presented in the present study are of course based on the specific alphatorquevirus PCR system used. The development of an international standard is needed to allow interlaboratory comparisons and general application of the proposed threshold levels.

The overall comparison of alphatorquevirus load with tacrolimus levels in all samples showed that there is a certain correlation between higher alphatorquevirus load and higher drug tacrolimus levels in blood, and this is especially true for tacrolimus levels a few weeks earlier. This is in agreement with data from virome analyses [18]. As instantaneous FK506 plasma levels also showed a correlation (independent of alphatorquevirus virus load) with infections as well as rejections/CLAD, and in both cases increased levels were associated with increased risk, this might, in the latter case, reflect dose adjustments in close temporal relation to the event rather than a true causal relationship.

First studies have been performed in other SOT recipients; however, so far it is not yet clear which thresholds of alphatorquevirus load could be used in these patients. Further analyses will be needed to assess whether these levels are similar to the threshold levels proposed for lung transplant patients. In a previous study on liver transplant recipients, there was a correlation between pretransplant alphatorquevirus load and posttransplant clinical course [25], which was not the case in our lung transplant cohort.

There are some limitations in this study. The use of alphatorquevirus as a marker for immunosuppression seems not to be useful within the first 1–2 months postoperatively, because in this period the alphatorquevirus kinetics differs between the individual patients without correlating with clinical parameters nor with induction therapy as shown previously [19, 21]. Furthermore, different types of immunosuppressive protocols (ATG, alemtuzumab, or no induction) were used in our patients; although we found no evidence that alphatorquevirus kinetics and risk of subsequent events were affected by these protocols, it cannot be completely excluded, as only 2 patients received ATG, that this may exert an influence to a certain degree. Another limitation is the somehow variable spacing of alphatorquevirus measurements with the possibility to miss time points with higher or lower alphatorquevirus activity. Finally, thus far little is known about the possible pathogenic role of alphatorquevirus itself. Although it is generally accepted that alphatorquevirus is not pathogenic, it is possible that at extremely high levels, as occasionally found in some patients, alphatorquevirus may have some negative influence on the immune system.

In conclusion, the present study provides evidence that alphatorquevirus load may serve as a useful marker to assess the level of immunosuppression in lung transplant recipients and to predict clinical complications. Thus, measurement of alphatorquevirus load may provide a tool for tailoring immunosuppressive therapy to the patients’ individual needs. Further multicenter prospective studies comparing different immunosuppressive regimens are needed to validate the efficacy of such optimizing personalized posttransplant therapeutic protocol.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.