Calculated cardiovascular risk after conversion from calcineurin inhibitor to belatacept in kidney transplant recipients: A randomized controlled trial

In renal transplant recipients (RTRs), a belatacept-based immunosuppressive regimen is associated with benecial effects on cardiovascular (CV) risk factors compared with calcineurin inhibitor (CNI)-based regimens. The aim of this randomized, multi-national trial was to compare calculated CV risk between belatacept and CNI (predominantly tacrolimus) treatments using a valdidated model developed for RTRs. From 9 transplant centers, RTRs from 3 to 60 months post-transplantation were recruited to either continue treatment with a CNI-based regimen or switch to belatacept. We compared the change in estimated 7-year risk of major adverse cardiovascular events (MACE) and all-cause mortality after 12 months of treatment. In the 105 RTRs randomized, we found no differences between the treatment groups in predicted risk for MACE or mortality. Diastolic blood pressure was lower after belatacept treatment compared with CNI. The mean changes in traditional CV risk factors, including renal transplant function, were otherwise similar in both treatment groups. The belatacept group had four acute rejection episodes; two were severe rejections, of which one led to graft loss. In conclusion, we found no effects on calculated CV risk by switching to belatacept treatment.


Introduction
The risk of cardiovascular disease (CVD) in patients with renal failure is much higher than in the general population across all age groups 1,2 . While a successful transplant reduces this risk signi cantly, renal transplant recipients (RTRs) still have an annual cardiovascular (CV) event rate of 3.5-5% 3 . Accordingly, CVD remains one of the leading causes of death in RTRs 4,5 . Managing a transplanted patient should therefore include CV risk reduction measures to improve both graft and patient outcomes. Current guidelines for prevention of CVD are based upon data from the general population and from studies speci cally targeting CVD in RTRs 6 . In addition to addressing traditional risk factors for CVD, such as lifestyle choices, hypertension, hyperlipidemia, and diabetes, RTRs present two potentially modi able factors: renal graft function and type of immunosuppressive maintenance regimen.
First, evidence indicates that declining graft function and graft loss are potentially modi able risk factors for CVD and all-cause mortality in this population, which make strategies for optimizing graft function important 7,8 . Second, among immunosuppressive drugs used for transplantation, both steroids and calcineurin inhibitors (CNIs) are associated with adverse CV side effects 9 . Therefore, attempts have been made to minimize or eliminate their use. While these have led to reasonably safe steroid-free regimens [10][11][12] , CNIs are still the cornerstone of immunosuppression in modern solid organ transplantation. Early graft survival improved greatly after the introduction of cyclosporine (CsA) in the early 1980s 13 , and tacrolimus (TAC) has been the CNI of choice since the 1990s 14 . Despite the bene ts of CNIs in the early post-transplant period, they have dose-dependent side effects, including post-transplant diabetes mellitus (PTDM), hypertension, hypercholesterolemia, and nephrotoxicity, leading to progressive decline in renal graft function [15][16][17][18][19] . Therefore, there is an ongoing incentive for development of novel immunosuppressive agents without the side effects of CNIs.
Belatacept, a modi ed form of CTLA4-Ig, binds to CD80 and CD86 on antigen presenting cells, thus blocking CD28 mediated co-stimulation of T-cells. The BENEFIT trials have shown promise for belatacept as an option in designing a more favorable immunosuppressive regimen [20][21][22][23][24] . In brief, despite higher rates of early rejection, the relative risk of death or graft loss after 7 years was reduced by 43% in patients treated with belatacept versus CsA-treated patients, and eGFR in the belatacept-group was on average 22 ml/min/1.73m 2 higher than in the CsA-group. Furthermore, in a meta-analysis comparing belatacept with CNIs, treatment with belatacept was associated with lower blood pressure, lower incidence of diabetes and a more favourable lipid pro le 25 .
However, it is not yet proven whether these ndings translate into overall CVD reduction. Soveri et al. have previously developed a risk calculator for CVD and all-cause mortality for use in RTRs 26 . The group later used the data of the BENEFIT and BENEFIT-EXT trials to calculate the potential bene t associated with belatacept treatment and found a substantial calculated 7-year risk reduction for major adverse cardiac endpoints (MACE) and mortality by converting from CsA to belatacept 27 .
A shortcoming of belatacept that has hindered its implementation in kidney transplantation has been the relatively high rate of early rejection, as well as the lack of studies comparing its e cacy with low-dose TAC, the current standard of care in RTRs. In the present study, our aim was to investigate 1) the effects of conversion from a low-dose CNI-based therapy to belatacept on estimated risk of CVD and all-cause mortality using a previously validated calculator and 2) the changes in traditional markers of cardiovascular health, as well as measures of arterial stiffness.

Study participants and characteristics
A total of 112 patients from 9 centers signed the patient informed consent form. Of these, one patient was a screen failure (history of rejection) and was never randomized. Of the 111 randomized patients, 6 withdrew consent before any study drug was given, 4 in the belatacept arm and 2 in the CNI arm. Thus, 105 patients were administered study medication: 54 in the belatacept arm and 51 in the CNI arm (de ning our ITT population). In the belatacept-arm 5 patients were withdrawn from the study; 3 due to adverse events (AEs), 1 withdrew consent and 1 moved out of the country. Similarly, there were 2 withdrawals in the CNI arm; 1 due to AE and 1 withdrew consent. The remaining 49 patients in each treatment arm were de ned as the per protocol (PP) population ( Figure 1). As the difference between the PP population and the ITT population was quite small, we did not perform PP analyses to avoid the risk of type I error caused by multiple comparisons. The rst patient was enrolled September 18th, 2014, and the last patient completed the study on September 13th, 2018. Baseline demographic data and clinical characteristics for each group are presented in Table 1.

Estimated risk of MACE and mortality
The primary endpoint was the estimated 7-year risk of MACE and all-cause mortality per the risk calculator developed by Soveri et al. (Figure 2). After 12 months of treatment, there was no statistically signi cant difference between the treatment groups in terms of change in predicted risk, neither for MACE nor for mortality ( Table 2). The difference between interventions in log mean risk prediction for mortality was 0.02 (95% CI: -0.01, 0.05). An overview of the variables used in the risk calculation is presented in Table 3. Data are presented as number (percentage) for categorical data and mean value (standard deviation) for continuous variables. HD = Heart Disease. MACE = major adverse cardiac event. RRT = renal replacement therapy.
Subgroup analysis was also performed to investigate whether time since transplantation in uenced the results in risk calculation. Treatment arms were divided upon the median time after transplantation, thus creating an early and late group (before and after 26 months). There was no difference between belatacept and CNI in calculated risk of MACE (p = 0.33) and mortality (p = 0.56) in the subgroups.

Traditional CVD risk factors
The changes in traditional CV biomarkers from baseline to end of study are presented in Table 4. The mean changes were similar between the treatment groups, except for a signi cant difference in diastolic blood pressure, with lower levels after belatacept treatment compared with CNI. Systolic blood pressure showed a similar reduction, but the difference was not statistically signi cant.

Arterial stiffness
Arterial stiffness was measured at baseline and at end of study using the SphygmoCor® method. Compared with the CNI group, central diastolic pressure in patients of the belatacept group decreased by 6.55 mmHg (95%CI: 1.83, 11.27; p = 0.007) after one year of treatment. For central systolic pressures, the difference of 6.1 mmHg between study groups (95% CI: -0.11, 12.34; p = 0.054) was borderline signi cant. There were no differences between the treatment arms in central pulse pressure, pulse wave velocity and augmentation index (Table 5). Presented as mean (standard deviation). P-value results from two-sample t-tests.

Cardiovascular events and patient survival
During the one-year study period, there were no cardiovascular events observed (including cardiovascular death, non-fatal myocardial infarction, non-fatal stroke, hospitalization due to congestive heart failure or angina pectoris, or coronary intervention) or deaths in the study population.

Safety evaluation
All patients in both study groups reported at least one AE during the duration of the study ( Table 6). The majority of the events were of mild severity and considered unrelated to study drug. More patients in the belatacept-group (53.7% vs 21.6%) reported AEs that were considered possibly or probably related to the intervention. Three patients in the belatacept-group and one patient in the CNI continuation group reported AEs that led to withdrawal from the study. Serious Adverse Events (SAEs) were reported by 29.6% of the patients in the belatacept-group compared with 15.7% in the CNI group. Patients allocated to the belatacept-group had more infections (Table 7). There was 1 case of incident cancer (lung cancer), which occurred in the belatacept-group.  Given as incidence rates (in %) During the study, 8 acute rejection episodes were suspected, and graft biopsies were obtained for further investigation. Acute rejection was con rmed in 4 of the 7 suspected cases in the belatacept-group, and in the single case in the CNI group. Three of the rejection episodes were considered severe (Banff grade IIA or higher): two in the belatacept-group and in the CNI-treated patient. One patient (belatacept) proved refractory despite anti-rejection treatment with methylprednisolone and T-cell depleting antibody. All other rejection episodes recovered upon treatment with corticosteroids or anti-thymocyte globulin as per local practices.

Discussion
In this randomized study, where stable renal transplant patients were converted from a CNI-based maintenance immunosuppressive regimen to belatacept, no difference in calculated 7-year risk of MACE or all-cause mortality could be demonstrated after 1 year of follow-up. We were unable to nd a signi cant effect on any of the three modi able cardiovascular risk factors which were used as inputvariables in the risk calculator (serum LDL-cholesterol, diabetes-prevalence, and serum creatinine). The belatacept arm had signi cantly lower diastolic blood pressure, measured both centrally (SphygmoCor® method) and peripherally. We found a similar improvement for systolic pressure (Table 4, 5), but this difference was not statistically signi cant.
Of the three modi able risk factors in the calculator, the expectations regarding effect on lipid pro le were limited. While CsA has been implicated in dyslipidemia 28 , TAC seems to be less detrimental to lipid status. In our study, 94% of the participants were on TAC before randomization. Ferguson et al. 29 compared three steroid-avoiding regimens of immunosuppression: belatacept with MMF vs belatacept with sirolimus vs TAC with MMF. Both belatacept-arms had lower LDL (23.9 mg/ml and 25.0 mg/ml vs 34.0 mg/ml for TAC with MMF) after one year, but the difference was non-signi cant, possibly related to the limited sample-size of the study. Another observational study focusing on the metabolic effects of conversion from TAC to belatacept found improvement in GFR and acid-base status, but not in blood lipids 30 . Our ndings are in line with these reports, as we found no effect on LDL-cholesterol (Table 3).
However, we are surprised by the lack of effect on GFR, which is in contrast to the BENEFIT studies, as well as other conversion studies reported in the literature 24,31−33 . In those studies, there was a consistent improvement in graft function by converting to belatacept. One possible explanation for this was the predominant use of TAC by our study participants with relatively low trough levels ( Table 1) at baseline.
In the Symphony trial 34 , the low-dose TAC group had an average trough-level of 6.7 ng/ml 1 year after transplantation and achieved an eGFR on average 5.7 ml/min/1.73m 2 higher than the other 3 comparator groups. A belatacept conversion study by Grinyo et al. 35 [39][40][41][42] . Thus, we expected to improve glycemic metabolism in converting from TAC to belatacept. However, no subject in our study reversed diabetes mellitus or developed PTDM in either study arm (Table 3). Also, triglycerides, serum ApoB, and serum ApoA1 did not improve (Table 4), which is of interest, since all three of these parameters are mentioned as risk factor for developing PTDM. 43,44 Beside trough levels, we also need to consider another bias. All patients were already treated with CNI for a median of 26 months since transplantation. Serious negative side effects of CNI-treatment could be less likely found in the control group, as patients suffering from these side effects could have been converted to alternative immunosuppression earlier on and thus not be eligible for this study.
The only positive effect that we found for belatacept was a signi cant improvement in diastolic blood pressure, measured both centrally (SphygmoCor® method) and peripherally. For systolic pressure, a similar improvement was found (Table 4), but it was not statistically signi cant (p = 0.09), most likely due to the relatively small sample size of this study. Although not included in the calculator, blood pressure is of course an established risk factor for cardiovascular disease. Moreover, high blood pressure is strongly associated with risk of graft failure and nding an improvement in this parameter could still indicate an advantage for belatacept-treatment 45 .
Regarding safety, AEs occurred in both groups, but SAEs were reported almost twice as often in belatacept-treated patients (29.6% vs 15.7%), and the latter were more likely (5.6% vs 2.0%) to discontinue their study treatment than patients treated with CNI. Rejection was seen more often in the belataceptpatients. Four episodes of biopsy-proven acute rejections occurred in the belatacept-group vs one single episode in the CNI-group (7.4% vs 2.0%). Three patients showed signs of vascular in ammation in the biopsy, corresponding to Banff grade II, two of which were in the belatacept-group. All three patients were treated according to local protocol with high-dose steroids (4) and T-cell depleting antibodies (1), despite which one belatacept-patient suffered graft loss and re-initiated dialysis treatment. The other two recovered with treatment.
The rate of rejection in this study is in line with earlier reports. For example, in the trial by Grinyo et al. 35 , 7.1% of belatacept-patients experienced rejection versus none in the CNI-group. In another trial by Adams et al. 46 , 1-year rejection rates were around 50% when belatacept was used right after transplantation, declining to 33% when TAC was tapered off 3-5 months after transplantation. When TAC was tapered after 11 months, the rejection rates between TAC-and belatacept-treated patients were similar, around 16%. Other reports have described varying (0-11%) rates of rejection, but these are data from nonrandomized 'rescue'-settings after even longer time post-transplantation and are therefore not comparable with our results 47,48 .
Beside rejection, urinary tract infections (UTIs), nasopharyngitis and other respiratory tract infections (RTIs) were more often seen in the belatacept arm ( Table 7). The present study's planned visits could have led to a bias in the reporting of uncomplicated infections, since a study visit was planned every month for belatacept-patients, instead of every 3 months for the CNI-continuation group.
Not a single case of pneumocystis-jirovecii pneumonia, cytomegalovirus-(CMV), polyoma-or EBVassociated disease was seen in the belatacept-patients. Three cases of CMV-infection were seen in CNIpatients. Previous reports have been inconclusive on opportunistic infections (OPIs) in belatacepttreatment. The follow-up study to the rst belatacept-conversion trial noted a slightly higher incidence of viral infection (11% vs 14%) 35 . In a recent study by Bertrand e.a. 50 OPI's were noted in 453 patients treated with belatacept (9.8%) 49 . In a multivariate analysis of that study, the authors concluded that patients with low GFR (<25 ml/min) and patients converted early after transplantation (within six months) were more likely to develop OPIs.
There was one case of lung cancer in the belatacept group in the present study. Previous studies have not indicated a higher risk of malignancy in belatacept beyond post-transplant-lymphoproliferative disorder 24,35 .
A major strength of the current study is the international multicenter-approach, making it representative for European transplantation practice. However, this study also has important limitations which must be taken into account. The study duration of 1 year was most likely too short to reveal a signi cant difference in renal function between the two study groups. We have overestimated the potential reduction in MACE and mortality for patients that use low-dose TAC instead of CsA. Another limitation was the heterogeneous time from transplantation to trial enrollment, and the small number of patients on CsA and the relatively large span of eGFR also contributed to the heterogeneity. Patients with severely diminished graft function were less likely to bene t from conversion.
In conclusion, we have shown no effect on calculated cardiovascular risk or renal function in this study comparing late conversion to belatacept with continuation of CNI-based immunosuppression. We did show a signi cant difference in diastolic blood pressure. We re-con rmed the increased chance of rejection when converting to belatacept. After more than 10 years of clinical experience, the place of belatacept in kidney transplantation is still not fully established, but it may be an attractive option when  Figure 3). Patients randomized to the control group with continuation of CNI treatment were to maintain trough levels of CsA between 75 and 200 ng/ml and TAC between 5 and 10 ng/ml. Both groups were to continue their underlying immunosuppressive regimen, consisting of mycophenolate mofetil (MMF) or mammalian target of rapamycin inhibitor and corticosteroids. Any other concomitant medication necessary to maintain the patients' baseline condition or to treat a coexisting disease was permitted.

E cacy assessment and procedures
The primary endpoint of this trial was estimated cardiovascular risk after 12 months, using a prediction model developed for RTRs by Soveri et al. 26 . The estimated 7-year risk of MACE and mortality in the two treatment groups were calculated as a linear combination of the following variables: age, previous coronary heart disease, previous smoker, current smoker, creatinine, diabetes mellitus, low-density lipoprotein (LDL), number of transplants and total time on renal replacement therapy ( Figure 2

Sample size and randomization
We performed a power calculation hypothesizing that the intervention arm would decrease the risk of MACE by 30%. We came to that estimate by extrapolation of the reduction in calculated risk in the previously mentioned paper by Soveri e.a. 27 ; the calculated risk of MACE for BENEFIT-patients decreased by 31.2% (from 14.3-10.9%), and for mortality by 40% (17.5-12.5%). The corresponding risk reduction for BENEFIT-EXT-patients was 27.8% (22.5 to17.6%) and 22.6% (30.9-25.2%). For a two-sample t-test on a two-sided signi cance level of 0.05, assuming a standard deviation of 0.64 (on the natural logarithmic scale), a sample size of 51 per group was required to obtain a power of 0.8 (80%) to detect a 30% calculated risk reduction in MACE. The ANCOVA model was expected to have slightly greater power than the two-sample t-test, and therefore a sample size of 102 patients was seen as su cient for this study. To account for 8% drop-out, a total of 110 patients, 55 per treatment arm were included in the study.
Randomization to treatment arm was performed using a computerized procedure, strati ed by center, in a 1:1 ratio.

Statistical analysis
The primary endpoint was a comparison of the estimated CV risk between treatment groups (CNI-vs. belatacept-based immunosuppression) at one year. For patients who discontinued the study before one year, the last available estimate of CV risk was used in the analysis of the intention-to-treat (ITT) population. Due to a skewed distribution, estimated CV risk was log-transformed. The primary analysis was performed using analysis of covariance (ANCOVA) with treatment as a group variable and baseline log CV risk for MACE and center as covariates. All other comparisons on primary and secondary endpoints were based on ITT comparisons of treatment groups using two-sample t-test, or ANCOVA with correction for baseline variables and/or center. A two-sided P value of <0.05 was considered statistically signi cant. Analyses were performed using SAS® version 9.4 (SAS Institute, Cary, NC). The data used to support the ndings are included in the manuscript. Any additional (raw) data are available from the author (OWB) and co-author (KTS) upon reasonable request. Restrictions to availability may apply due to privacy or ethical reasons.

Figure 1
Study ow chart. Conversion and dosing scheme.