Detrimental impact of immunosuppressive burden on clinical course in patients with Cytomegalovirus infection after liver transplantation

Cytomegalovirus (CMV)‐infection and reactivation remain a relevant complication after liver transplantation (LT). The recipient and donor serum CMV‐IgG‐status has been established for risk stratification when choosing various pharmaceutical regimens for CMV‐prophylaxis in the last two decades. However, factors influencing course of CMV‐infection in LT remain largely unknown. In this study, the impact of immunosuppressive regimen was examined in a large cohort of patients.

Discussion: CMV-infection remains of clinical importance after LT.Undergone CMVinfection of either recipient or donor requires prophylactic treatment.Additionally, we found a highly significant, dosage-dependent impact of immunosuppression (IS) on long-term outcomes for these patients, underlying the importance of minimization of IS in liver transplant recipients.

INTRODUCTION
Liver transplantation (LT) is the only curative treatment option for selected patients with end-stage liver disease (ESLD). 1,2After LT, life-long immunosuppression (IS) remains standard for prevention of rejection and potential graft loss in majority of cases. 3,4Although substances and regimens have changed over time, calcineurin inhibitors (CNI), mycophenolate mofetil (MMF), mammalian target of rapamycin inhibitors (mTORI), and glucocorticoids (GC) were established as the standard of care in most centers around the globe. 5However, there is consensus, that IS must be highly individualized based on the unique individual risk profile of each patients, including the underlying disease, various comorbidities, immunological factors as well as personal tolerance. 6,77][18] Among these, Cytomegalovirus (CMV) is still one of the most relevant viral pathogens after LT. 19 Before the introduction of CMV prophylaxis, the prevalence of CMV infection after LT was reported to reach 80%.1][22] Extent of CMV infection may vary from asymptomatic viremia to manifest CMV-syndrome or tissue-invasive infections.Acute allograft rejection, bacterial as well as fungal superinfections are associated with CMV-infection, potentially complicating the clinical course. 23fferent risk factors for the development of CMV infection after LT have been identified: A positive serological status of both donor and recipient for a history of CMV infection facilitates CMV reactivation, whereas primary infections are relatively rare. 24CMV reactivation is more likely in the case of intensified IS, which can be present either as induction therapy, as rejection therapy, or as above-average intensive maintenance therapy in the context of autoimmune liver diseases. 23,25rthermore, the risk for CMV infection is reported to be more rel-evant in patients with higher IS after transplantation, for example, patients undergoing small bowel or lung transplantation than LT or kidney transplantation. 26er the course of establishing a personalized immunosuppressive therapy in the last decade, risk-adjusted tapering and individualization of IS have been introduced. 6,27,28The aim of the present study was to answer the question to what extent these changes influence the incidence, severity, and clinical course of CMV infection.

METHODS
All patients who underwent successful LT at our transplant center (Charité Universitaetsmedizin Berlin, Germany) between 2006 and 2018 were considered for the analysis excluding recipients under 18 years and with combined transplantations.LT was conducted in respective of guidelines at the current time.Interdisciplinary conferences were mandatory and international criteria (e.g., MILAN) were upheld. 29However, in individual cases, extended criteria were applied or inclusion in clinical trials was conducted.Allocation was conducted using the MELD-score and was coordinated by Eurotransplant.

Routine follow-up
After successful LT, patients were followed-up periodically at our outpatient clinic in a life-long manner; intervals were dependent on the time after transplantation and ranged between 2×/week to every 12 weeks and included clinical examination and laboratory tests to assess the graft function.Screening for CMV included serological status before and after LT as well as CMV-DNA-PCR for evaluation of viral load and was conducted for the first time as early as within 7 days after LT and repeated once a week upon discharge.Afterward, it was conducted upon clinical or laboratory suspicion and on routine visits (3, 6, 12, 18 months and 3, 5, 7, 10 years after LT) in a life-long manner.

Management of CMV
CMV-infection was classified as following: (i) CMV-viremia in patients without clinical symptoms but de novo CMV infection with proof of CMV-DNA in blood specimens, (ii) CMV-syndrome in patients with clinical symptoms corresponding de novo CMV-infection, and (iii) tissue-invasive CMV-infection in patients with histological proof of CMV in tissue samples.
Center protocol for CMV-prophylaxis was conducted in respect of recipient/donor seroconstellation.Thus, R+/D+, R+/D−, and R−/D+ patients were routinely evaluated for prophylactic treatment in respect to IgG status at LT.However, if comorbidities/expected side effects mandated, prophylaxis was not administered in an individual manner.For prophylactic therapy for CMV-infection and other herpes viruses, acyclovir and valganciclovir (VGCV) were administered in nontherapeutic dosages depending on renal function as the standard of care, respectively. 25Switch from aciclovir to VGCV as standard agent was conducted based on scientific data.Preemptive therapy was conducted in patients with new CMV-viremia.

Immunosuppression
Routine IS consisted of CNI, MMF, MTORI, and GC in an individualized manner.Induction therapy, defined as additional treatment with antibodies/IS except those mentioned above, was considered individually in younger patients, known autoimmune-diseases, impaired renal function, and absence of malignancies.Choice of induction regimen (e.g., antithymocyte-globuline (ATG) and basiliximab) was based on individual factors, physicians' expertise, and current scientific data.
To investigate the impact of IS, dosages and trough levels (CNI, MMF, and mTORI) at time of CMV-infection or at last follow-up in patients without CMV-infection were analyzed as well as mode of induction for therapy at LT.For stratification, IS dosages were scored using a scale introduced by Vasudev et al. 30 Here, each of the following dosages per day were calculated with one unit for: 5 mg prednisone; 100 mg azathioprine; 100 mg cyclosporine; 2 mg tacrolimus; 500 mg MMF; 2 mg mTORI and added in case of combination therapy.
Tacrolimus trough levels were recorded at time of LT (median of all available values 7 days after LT) and after 90 days (median of all available values 7 days prior and 7 days after 90 days-mark), allowing for calculation of cumulative 90 days exposure.This time interval was chosen, as the vast majority of CMV-infections occurred within 90 days.

Rejection
Acute cellular rejection (ACR)-corresponding to a T-cell mediated rejection-was diagnosed based on histological specimen by an experi-enced pathologist after ultrasound-guided biopsy of the graft at suspicion and was scored using the classification established Banff Working Group based on portal and biliary inflammation, and endothelialitis. 31nce, a score of 0-2 was classified as "no rejection," 3 as "borderline," a score of 4-5 as "grade 1/mild," 6-7 as "grade 2/moderate," and 8-9 as "grade 3/severe" ACR.All patients with ACR diagnosed and included in the present analysis were cases of biopsy-proven acute rejection.
General clinical characteristics and follow-up information were acquired by inhospital data, our outpatient center, and reports or consultations from the community-based setting (e.g., primary care physi-
Two hundred fifty patients received induction regimen of immunosuppressive therapy and in about half of the cases (n = 136%/54.4%),the interleukin-two-receptor-antibody basiliximab was administered.
Histologically proven ACR was diagnosed in 333 (38.4%).Grade one rejections were most common (n = 159%/47.7%)and GC-pulse therapy was administered in 291 (87.4%) of these patients with an additional following therapy with ATG in 34 ( patients.An overview of patient characteristics is presented in Table 1. CMV-infections were diagnosed in 325 (37.5%) patients after a median time of 28 (0-3102) days after LT.Most CMV-infections (n = 270%/83.1%)occurred within 90 days after LT.For comparative analysis, the group with CMV infection post LT (group CMV-positive; n = 325) was matched with the group without CMV infection (group CMV-negative; n = 542).Groups' characteristics are shown in Table 1.
CMV-syndrome appeared in 84 (25.8%) of these patients and tissue invasion was detected in 4 (1.2% CMV-infections were registered in the cohort under a prophylactic regimen and 165 (44.5%) infections occurred in the patient population not receiving treatment after LT (p < .001).Although this effect was seen throughout all risk-constellations, it was most prominent in the high-risk groups, as seen in Figure 1.
We further investigated correlation of occurrence of ACR and CMV between groups as they both represent major IS-related events with effects immediately after LT. months, respectively (Log rank = 0.004; Figure 2).The subgroup analysis revealed most favorable outcome for patients with at least one episode of acute rejection but without CMV-infection after LT (Log rank = 0.004; Figure 3).
Overall, immunosuppressive regimen did not differ between groups and was mainly CNI based with tacrolimus as the mainstay of therapy.In concordance, IS-intensity score did not show significant differences in IS-burden with a median score of 2.5 (0-11) units in the CMVpositive group and 2.3 (0-12) units in the CMV-negative group (p = .6).
Recorded tacrolimus trough levels 90 days after LT were comparable in patients suffering from CMV-infection with 8.1 (±2.6) ng/mL and 8.4 (±2.5) ng/mL in patients without CMV infection (p = .09).Interestingly, comparing the tacrolimus trough level for patients with CMV-infection using the record closest to date of CMV-infection revealed a mean of 9.2 (±9.2) ng/mL but did not show statistical significant difference compared to the 90 day trough level of CMV-negative patients (p = .11).
No association of dichotomized IS-intensity score on occurrence of CMV-infection was found (p = .65).MTORI therapy with its reported CMV-protective effect was found more often in the CMV-negative group (n = 66%/12.2%)than in the CMV-positive group (n = 26%/8.0%)as shown by Chi squared test with a statistical significance (p = .05).

Subgroup analysis on patients
Of all clinically relevant parameters to assess relevance of impact on development of CMV-infection after LT, CART analysis revealed pretransplant risk constellation according to serostatus of donor and recipient the most important.Furthermore, prophylactic CMVand induction therapy revealed relevance in a subset of patients.No impact was found for extent of IS or regimen of IS.For details see Grouped IS-intensity score showed statistically significant differences with worst overall survival in patients with >4 units (Log rank < .001).
We further analyzed impact of tacrolimus trough levels on survival in our cohort and found a striking dependency on cumulative 90 days trough levels, with longer survival of patients undergoing "(aggressive) minimization" and minimal levels (<320 ng/mL and 321579 ng/mL) and impaired survival in patients with "high exposure" (>840 ng/mL, Log rank < .001).Median survival ranged from 80. tacrolimus dosage-dependency (Log rank < 0.02).For details, see Figure 8.
Multivariate Cox-analysis using clinically relevant parameters with putative impact on survival showed statistically significant impaired overall survival for patients who did receive induction therapy for LT (HR = 1.99,CI: 1.05-3.8;p = .036)and who had to undergo retransplantation (HR = 3.64, CI: 1.39-9.52;p = .009).In concordance with the univariate analysis, higher tacrolimus trough level were associated with a significantly shorter survival in a dosage-dependent manner (p = .017).Analyzing the cohort of CMV-positive patients after infection, only impact of cumulative tacrolimus trough level on survival prevailed with statistical significance (p = .002).For details of multivariate analysis, see Table 2.
F I G U R E 6 Extent of cytomegalovirus (CMV)-infection and survival after liver transplantation (LT).

F I G U R E 7
Influence of mammalian target of rapamycin inhibitors (mTORI) on course after liver transplantation (LT).

DISCUSSION
In With 60%, the rate of CMV-infections in our study population prior to LT was representative for its prevalence in western countries. 32IS mainly consisted of CNI, MMF, and mTORI with GC during the initial phase after LT and induction therapy based upon the individual's risk profiles according to current standards and recommendations. 4,28V-infection with measurable viral load via PCR in recipient blood samples occurred in about 40% of patients in our overall cohort, mostly within 90 days after LT, as has been reported before. 335][36][37] Preemptive CMV-therapy upon diagnosis of CMV-viremia-requiring high-dosage of, for example, VGCV-has been reported to effectively reduce occurrence of CMV-syndrome further but is associated with increased logistic and financial burden as well as toxicity (e.g., leukopenia).4][35] The prophylactic regimen in this study mainly consisted of acyclovir and VGCV in recent years.Current guidelines on prophylaxis and treatment of CMV post LT define ganciclovir-based therapy as standard, whereas VGCV possesses a favorable bioavailability in oral regimens. 33,34,38Although our current standard regimen upholds these recommendations, this cohort reflects the change of regimen in the last decades.More experimental regimens based on novel drugs have already been explored and described but have not been implemented in clinical routine yet. 39surprisingly, CMV-infection after LT was significantly associated with higher donor/recipient risk constellations in our cohort, as recommendations for prophylaxis specifically focus on this aspect.We used CART-analysis including various clinical variables to assess parameters with potential impact on CMV-infection.2][43] Here, it confirmed that serostatus at LT still remains the most important clinical risk factor for developing CMVinfection requiring prophylactic treatment especially of patients with R+ seroconstellation.Of note, in our findings, "moderate" (R+/D+ and R+/D−) as well as "high" (R−/D+) risk constellation have com-parable infection rates and seem to profit equally from preventive measures.5][46] Although indication for induction therapy remains controversial, Bittermann et al. found similar rates of 27% in the US. 47,48We did not find a statistical significant impact of IS dosage-dependence on CMV-occurrence, although a trend was observed.Thus, we estimate the results of this study not as contradictory, because ISand its dosage-is undoubtedly causally connected to CMV-infection after transplantation, but other factors as mentioned above (e.g., CMV-seroconstellation) contribute with stronger influence under the fundament of a immunosuppressive regimen.
Acute rejection has been linked to CMV-infections in kidney and LT, potentially increasing the risk of graft loss or death in affected patients. 23,49,50In this study, ACR was significantly associated with CMV-infection as well but occurred more frequently after CMV- Still, the role of rejection in graft-and patient outcomes remains unclear. 51,524][55][56] Favorable courses for subsets of patients with ACR have been reported, but the mechanism is not yet understood. 57Of importance, rejection frequently requires GC pulse therapy with increased cost and side effects. 56,58Thus, regarding ACR and the desire to minimize IS in LT patients, a "fine balance" is required. 59 found a striking adverse impact of CMV-infection on survival, reflecting the importance of thorough adherence to prophylaxis or even preemptive therapy recommendations and screening post-LT.
Other studies have reported similar observations, but interestingly, CMV-infection itself can only rarely be linked directly to graft failure or death. 24,60,61However, we found a dosage-dependent effect of IS on extent of CMV-infection.
More infection-related deaths were observed in the cohort of CMV+ patients, potentially indicating the weakened immune status in these patients.
Moreover, we analyzed the influence of IS and revealed inverse dosage-dependent survival differences in tacrolimus trough levels, with the best outcome for patients with the lowest cumulative serum levels.This effect was highly significant in the subgroup of patients suffering from CMV-infection after LT.It has become an important focus of research in LT aftercare to individually assess IS in long-term clinical course to prevent adverse events of administered substances, ranging from nephrotoxicity to metabolic syndrome, edema, thromboembolism to de novo malignancies. 7,10,62,63Recently, dosage-dependent effects of IS from the subacute post-LT setting on de novo malignancies have been reported. 148][69] Although we did not find direct impact of IS on CMV-infection-possibly due to the fact that induction therapy with t-cell depletion and preoperative CMV-serostatus is far more relevant-the strong correlation on IS-dependent survival differences urges for amendment of even early IS-regimens with the aim of going "as low as possible."[72][73] The use of mTORi is associated with a lower rate of CMV-infection, so this substance-adapted to comorbidity-seems to be helpful at least in this situation.These results are supported by previously published work. 74,75However, it remains unclear whether the use of mTORi alone generates the benefit, or whether the reduction of CNI contributes to it.
Certain limitations of this study have to be mentioned.The retrospective character is associated with a potential risk of bias with the possibility that underlying confounding variables might have not been analyzed.Further, therapeutic approaches regarding prophylaxis and (preemptive) treatment have evolved and specialized and our long-term observational analysis incorporated these changes with, for example, switch from acyclovir to VGCV in routine prophylaxis for CMV.Also, although assessment of IS at time of CMV-infection gives a meaningful insight for this cohort, IS-burden for the CMV-negative cohort had no comparable "time-stamp" and thus, IS at 90 days after LT in these patients has been evaluated.This might have obscured analysis on differences in IS-burden and impact on CMV-infection.Impact of maintenance IS in long-term outcome was analyzed, but still, significant changes in the regimen in individual cases cannot be ruled out completely and in these rare cases, impact of IS might be unprecise.
Further the IS-intensity score might not be able to reflect IS-burden in the given cohort, where, for example, mTORi increase the scale but are deemed rather beneficial, therefore not contributing to the risk of CMV-infection or its complicated course afterward.While assessment of IS-burden using this scale remains a helpful tool, it may not be suitable for all IS-related investigations in organ transplantation.Similar, stratification of tacrolimus exposure as utilized in this study has not been reported in the exact setting of CMV infection.Still, it enables a rather practical approximation of dosage-dependent effects in IS after LT, as Rodríguez-Perálvarez et al showed. 14

CONCLUSION
CMV infection still plays a negative role for recipients and represents an important burden after LT.All constellations with either donor or recipient seropositivity indicating undergone CMV-infection prior to LT require prophylaxis equally.In cases of a manifest CMV-infection it must be properly addressed.Although CMV-infection might not be IS-dosage dependent, it is clearly IS-induced.Survival is dramatically affected by the extent of CNI-based IS, so it must always be adjusted according to the principle "as little as possible, as much as necessary" to achieve holistic success after LT.

F I G U R E 2 F I G U R E 3
with elevated risk of CMV infection based on seroconstellation (R+/D+ and R+/D− and R−/D+) did not show statistically significant differences in tacrolimus trough levels at time of CMV-infection, cumulative 90-day exposure or IS-score (p = .12,p = .94,and p = .67,respectively).Analyzing severity of CMV-infection, 90 days grouped cumulative tacrolimus dosage did not show significant impact on extent of CMVinfection (p = .22).ANOVA test comparing exact tacrolimus trough F I G U R E 1 Impact of recipient and donor cytomegalovirus (CMV)-seroconstellation and prophylaxis on CMV-infection after liver transplantation (LT).D, donor; PPX, prophylaxis; R, recipient.Impact of acute rejection on overall survival after liver transplantation (LT).BIPAR, biopsy proven acute rejection.Impact of acute rejection and cytomegalovirus (CMV)-infection on overall survival after liver transplantation (LT).ACR, acute cellular rejection.levels at time of CMV-infection showed statistically significant elevated levels of the immunosuppressive agent in patients with manifest CMV-syndrome with 11.5 (±16.6)ng/mL compared CMV-viremia only with 8.3 (±3.1) ng/mL (p = .001).

Figure 4 .Figure 5 )
Figure 4. Treatment of CMV-infection was mainly conducted intravenously with ganciclovir (n = 122%/37.5%), in 88 (27.1%) patients, oral therapy with ganciclovir/VGCV was initiated.Of note, in 26 (8.0%) patients, the reduction of IS was sufficient to treat CMV infection and in 89 (27.4%) patients, no therapy was recorded.There was no significant difference in survival after CMV-infection in regard to therapeutic regimen of CMV-infection (p = .29).At the end of follow-up, 128 patients in the CMV-positive group and 199 patients in the CMV-negative group had deceased.Although the total number of deaths did not differ between groups (p = .43),subgroup analysis of causes of death revealed significant differences, with more infectious-disease related deaths in the CMV-positive group and more malignancies in the CMV-negative group (p = .003),as presented in Table1.Survival analysis showed significantly impaired overall long-term results and from time of CMV-infection for patients with recorded CMV-infection with median survival of 44.0 (0-141.8)months and 66.5 (0-142.1)months, respectively (Log rank = 0.03; Figure5).Additionally, in the Kaplan-Meier analysis, a statistically significant survival benefit was found for patients with sole viremia when compared to those with CMV-syndrome or tissue-invasive infection with a median survival of 41.3 (0-141.8),67.7 (0.4-139.9), and 7.6 (0.8-25.0) months, respectively (Log rank < .001,Figure6).MTORI therapy as part of IS was also associated with improved overall survival in our cohort in survival analysis.Median survival for patients with mTORI was 47.6 (6.1-137.4)months and 62.3 (0.03-142.2) months, respectively (Log rank = 0.04, see Figure7).The MTORI associated survival benefit could not be shown in the subgroup analysis for patients with or without CMV-infection (p = .15and p = .21,respectively).
3 (57.5-103.0)months in patients with cumulative exposure of <320 ng/mL to 47.5 (0.3-142.1) months in patients with >840 ng/mL cumulative tacrolimus trough levels.This dosage-dependent observation was highlighted in the survival analysis for the subgroup of patients with CMV-infection after LT (Log rank < .001).Here, median survival after CMV-infection was 103.0 (103) months in patients with aggressive minimization and 31.5 (0.1-138.5) months in the "high exposure" cohort.Survival analysis in CMV-negative patients showed similar statistical significant F I G U R E 4 Classification and regression tree (CART)-analysis on impact of clinical variables on occurence of cytomegalovirus (CMV)-infection after liver transplantation (LT).CART method was used to investigate the relevance of specific clinical variables with putative influence on CMV infection.The regression tree classified the 867 patients in this study for their associated variables in regard of occurrence of CMV infection.The nodes indicate relevant parameters that improve correct classification with significance and thus decreasing impurity toward the terminal (upper) nodes.Although seven variables were included in this analysis (see below), only recipient/donor risk constellation, a prophylactic regimen for CMV-infection and induction showed significant impact in improvement of correct classification for occurrence of CMV infection.Variables included in analysis: cumulative 90 days tacrolimus trough level (<320; 321-579; 580-839; >840)) ng/mL; immunosuppression intensity score; induction therapy (yes/no); rejection (yes/no); mammalian target of rapamycin inhibitors (mTORI) administration (yes/no); cytomegalovirus (CMV)-prophylaxis (yes/no); risk constellation (R+/D+, R+/D−, R−/D+, R−/D−).F I G U R E 5 Impact of cytomegalovirus (CMV)-infection on overall survival after liver transplantation (LT).
this study, we investigated clinical relevance of CMV-infection and analyzed various risk-factors with potential impact on the clinical course of CMV-infection after LT under current treatment standards.Our patient cohort reflected the situation of the last one and a half decades and the beginning shift of indications for transplantation away from hepatitis virus induced ESLD in western countries toward other indications such as cirrhosis and carcinoma, although we did not evaluate nonalcoholic steatohepatitis specifically, one of the rapidly increasing indication for LT.

F I G U R E 8
Dependence of survival after liver transplantation (LT) from tacrolimus trough level: (A) overall patient cohort; (B) patients with cytomegalovirus (CMV)-infection after LT; (C) patients without CMV-infection after LT.
Overall patient characteristics and constellation of groups.
Multivariate regression analysis for impact on overall survival (a) and after cytomegalovirus (CMV)-infection in CMV+ patients (b).