Long-term safety and efficacy of antithymocyte globulin induction: use of integrated national registry data to achieve ten-year follow-up of 10-10 Study participants

Antibody induction therapy in kidney transplantation is highly effective in reducing acute rejection [1] and, ultimately, in preserving allograft function [2]. Use of induction agents has increased over recent years such that, by 2013, more than 89 % of kidney transplants in the USA were performed with induction therapy [3]. Notably, while rabbit antithymocyte globulin (rATG) is the most commonly used induction agent in contemporary kidney transplantation [3, 4], it was approved by the US Food and Drug Administration (FDA) in 1998 for the treatment of acute rejection but not for the prevention of rejection (induction) [5]. To date only basiliximab and daclizumab, both antibodies against the inter-leukin-2 receptor (IL2R Abs), have been approved by the FDA for use as induction agents in kidney transplantation, and daclizumab is no longer marketed.

A pivotal component of the evidence base informing current understanding of the effectiveness of alternative induction regimens in kidney transplantation came from the 10-10 Study, which was the first prospective, randomized, international clinical trial comparing rATG and basiliximab among deceased donor transplant recipients perceived to be at increased risk of acute rejection or delayed graft function (defined as the need for dialysis within 7 days of transplantation) [6]. The 278 trial participants included 183 patients enrolled at 17 US centers; participants received cyclosporine, mycophenolate mofetil, and prednisone for maintenance immunosuppression. The 12-month study demonstrated that both induction agents were equally effective (P = 0.34) in preventing the composite quadruple endpoint of acute rejection, delayed graft function, graft loss, or death. However, when analyzed as a more traditional FDA endpoint of acute rejection, graft failure, or death, the differences were statistically significant in favor of rATG (P = 0.02), driven by a lower acute rejection rate in the rATG group (15.6 % versus 25.5 %, P = 0.02) [6].

The duration of follow-up in clinical trials is limited by the willingness of patients to participate for extended periods as well as by the ability of investigators to commit the time and resources necessary to track and monitor participants over a number of years. Extended monitoring beyond an initially determined study period may require additional informed consent and always incurs added costs. Consequently, long-term safety and efficacy data are lacking for many drugs in multiple treatment domains, including transplantation [7]. Thus, there is a need for approaches to assessing long-term outcomes for non-FDA-approved drug uses.

Solid organ transplantation is unique among medical specialties in the universal collection of clinical data in national registries in some countries. In the USA, through the mechanism of the Organ Procurement and Transplantation Network (OPTN), as mandated by the National Organ Transplant Act, transplant centers have been required to submit baseline and follow-up clinical data describing all patients listed for and receiving solid organ transplants since 1987 [8]. The OPTN supplements program-reported outcomes information with data from a national death registry, providing a high level of accuracy for the ascertainment of patient and allograft survival [9]. However, owing to a lack of granularity in the collection of baseline information relevant to eligibility and balanced comparisons as required within a clinical trial framework, it has been difficult to draw unbiased inferences on the long-term efficacy and safety of different immunosuppressive regimens based on registry data alone.

Integration of clinical trial and transplant registry records may circumvent some of the logistical difficulties in conducting long-term clinical studies and the limitations of isolated registry analyses. However, despite the opportunity created by the presence of national transplant registries, examples of the use of this approach in transplantation are limited. We previously linked data from the 10-10 Study with OPTN records to assess 5-year clinical outcomes of US-enrolled participants and found that patients treated with rATG had a lower incidence of a traditional composite endpoint of acute rejection, graft failure or death (37 % versus 51 %, P = 0.04) [10]. Using a similar approach, 15-year follow-up of 133 Australian participants in the Tricontinental Mycophenolate Mofetil Renal Transplantation Study was recently successfully achieved by linking trial records with follow-up reports from the Australia and New Zealand Dialysis and Transplant Registry (ANZDATA) [11].

Ten years have now passed since completion of enrollment in the rATG versus basiliximab induction immunosuppression trial. The goal of the current study was to link records for US participants in the 10-10 Study with current OPTN follow-up records to compare long-term efficacy and safety over 10 years after transplantation.

Results from the first prospective, randomized, multi-center clinical trial comparing the two leading antibody induction agents in renal transplantation, known as the 10-10 Study, provided valuable information on the short-term safety and efficacy of these induction agents [6]. However, information on long-term outcomes associated with initial choice of antibody induction agent is lacking. We integrated clinical trial records for the US participants in the 10-10 Study with current OPTN follow-up records to compare long-term efficacy and safety over 10 years after transplantation.

In the current analysis, we found that patients treated with rATG had similar long-term patient and graft survival at 10 years post-transplant as patients treated with the FDA-approved induction agent basiliximab. Congruent with previously observed superiority of rATG versus basiliximab induction for the FDA-specified composite triple endpoint of allograft rejection, graft failure or patient death used in clinical trials of immunosuppression efficacy observed at 1 year post-transplant [6, 10], superiority with rATG induction for this composite endpoint was confirmed at 5 years using current data for this US-based sub-analyses. At 10 years post-transplant, while freedom from the composite of acute rejection, graft failure or death was not significantly different among US participants in the two trial arms, non-inferiority for rATG compared with FDA-approved basiliximab for 10-year risk of the composite endpoint was demonstrated with a high level of statistical certainty.

Because rATG is a more potent induction agent than IL2R Abs, there is concern over increased risks of death and malignancy in the long-term. We found that there were no significant differences in these outcomes at 10 years. Although our observations contradict prior reports raising concern for more common post-transplant malignancies with rATG [16, 17], there were numerically higher point estimates of lymphoproliferative disorder and non-melanoma skin cancer, but not other non-skin cancer, in the rATG arm, and the lack of significance may be impacted by the low study power. Continued assessment of the impact of induction immunosuppression on cancer risk is warranted. Importantly, the close alignment of the patient survival curves and very similar magnitude of the 10-year survival fractions supports truly equivalent long-term survival. A strength of the current analysis is that we studied a carefully characterized patient sample with confirmed induction exposure, and linked to long-term transplant registry data for outcomes ascertainment, including comprehensive national death records. We did not assess early complications aside from those captured as study outcomes in the current analysis. Common concerns with the use of rATG early after transplantation include thrombocytopenia, leucopenia, pyrexia, and hypotension. We previously reported associations of rATG with transient thrombocytopenia and leucopenia (driven by lymphopenia) that resolved by post-operative day 14 [6].

Maintenance immunosuppression warrants consideration in any study of induction therapies. Cyclosporine, mycophenolate, and low-dose prednisone were used in the 10-10 Study. While the most common maintenance immunosuppression regimen in current US practice is tacrolimus, mycophenolate, and low-dose prednisone, cyclosporine continues to be used as the primary calcineurin inhibitor among many patients internationally. A subsequent randomized comparison of rATG versus the IL2R Ab daclizumab in the context of maintenance tacrolimus, mycophenolate, and low-dose prednisone among high immunologic-risk patients showed that rATG induction was associated with lower rates of delayed graft function (32 % versus 45 %, P = 0.04), acute rejection (15 % versus 27 %, P = 0.02), and need for anti-lymphocyte treatment of acute rejection (2.7 % versus 14.9 %, P = 0.002) at 1 year post-transplant [18]. This benefit of rATG in lowering acute rejection compared with an IL2R Ab was nearly identical to the benefit seen in the 10-10 Study, despite the use of tacrolimus as the maintenance calcineurin inhibitor. Furthermore, a recent open-label trial comparing rATG versus dacluzimab or basiliximab, also in the context of maintenance tacrolimus, mycophenolate, and corticosteroids, found that only rATG was protective against 6-month and 1-year acute rejection in African-American recipients [19]. Finally, our analysis considered maintenance therapy from an intention-to-treat perspective, but it is possible that maintenance regimens changed over time.

There are other limitations to our study. We were unable to capture follow-up data for the European patients enrolled in the original trial, owing to use of a US-based transplant registry. No similar comprehensive registry existed in the participating European countries at the time of the trial. As a result, 34 % of the initial sample was not included in the current analysis, compromising statistical power. Although 96 % of trial and OPTN records matched exactly on all linkage parameters or with ±1 day extension of transplant date (to account for differences in recorded anesthesia induction versus transplant date in the two sources), 4 % of matches required broadening of agreement windows for transplant or birthdates, and thus indicate recording errors for these dates in one of the sources; such recording errors cannot be resolved with the current design based on extracted information without direct access to center records. Observed baseline characteristics were well balanced across the included subjects in the two treatment arms, but imbalances in unmeasured characteristics are possible. While the intervention of interest, induction immunosuppression, is solely an early exposure, loss of blinding after the completion of the trial and subsequent unmeasured impacts on other aspects of care is also a potential limitation of the study design.

The long-term data presented were also limited by the fields captured in the OPTN registry. For example, histological assessments of the type and severity of acute rejection episodes are not collected by the OPTN. Although center-reported death records were supplemented with the SSDMF and thus considered to be complete, outcomes (acute rejection, graft failure) that relied solely on center reports could be incompletely captured. Because the OPTN registry incorporates events attributable to prior periods if newly reported by centers after an annual survey, we also re-analyzed 5-year outcomes in the recently extracted data and found slightly higher event rates compared with the prior 2007 extraction [10] but with very consistent patterns. It is possible that re-extraction after a longer latency could also impact estimates of 10-year event frequencies if some events close to the end of the study period were recognized and reported after our data extraction.

Conducting clinical trials with long-term follow-up poses many challenges, from financial burdens to the unwillingness of patients, physicians, and commercial and academic sponsors to participate in studies that require years of involvement. Consequently, long-term safety and efficacy data are lacking for most drugs [20]. Although post-marketing surveillance for adverse effects is mandatory by the FDA, the success of such programs relies heavily on spontaneous reporting by healthcare providers. In the field of transplantation in particular, although the majority of contemporary immunosuppressive agents have been used in clinical practice for decades, there are few long-term, prospective clinical trials evaluating safety and efficacy [21‐25]; hence, the optimal immunosuppressive regimen remains a subject of debate. Here, we describe an efficient, non-obtrusive method of monitoring long-term treatment outcomes using the combination of a properly designed clinical trial and registry outcomes data. Our data integration approach could be particularly useful in addressing other uncertainties for immunosuppression-related outcomes in transplantation when clinical trial and registry data are available. Despite the opportunity created by the presence of national transplant registries, we are aware of only two prior demonstrations of this approach in transplantation: a 5-year follow-up of the 10-10 Study by our group [10], and linkage of records from a maintenance immunosuppression trial with ANZDATA records [11]. We believe this methodology has potential for more common application in the field of transplantation, as well as broader implications for other fields with access to both trial and registry data.

In summary, integrating records from a clinical trial of induction immunosuppression therapies with national transplant registry data enabled ascertainment of long-term follow-up information for 9 years beyond the duration of the original comparative trial. We found that, compared with an FDA-approved induction agent, the benefits of polyclonal induction with rATG for an FDA-defined composite endpoint of acute rejection, graft failure, or death were sustained through 5 years post-transplant. At 10 years post-transplant, rATG induction had comparable efficacy and safety to FDA-approved basiliximab. Our methodology demonstrates an efficient strategy for monitoring the long-term safety and efficacy of therapies examined in clinical trials when registry data for participants are also available. This approach may be extended to clinical trials outside of transplantation.