Silent acute rejection during prolonged delayed graft function reduces kidney allograft survival

Abstract

Background.  

The relationship between the effects of early, silent, acute rejection (AR) and delayed graft function (DGF) on kidney allograft survival remain controversial, and the role of protocol biopsies during DGF is unclear. We hypothesized that protocol biopsies during DGF would reveal a high incidence of silent AR that may adversely affect long-term allograft survival.

Methods.  

We routinely carried out protocol biopsies in patients requiring dialysis 7 to 10 days posttransplant. We retrospectively examined the extent to which silent AR, diagnosed by protocol biopsies during prolonged DGF, may mediate the adverse effects of DGF on graft survival in 410 consecutive transplants using Cox proportional hazards analysis.

Results.  

By 40 days posttransplant, the cumulative incidence of AR was 57.2% among 65 patients who had a protocol biopsies during DGF, while it was only 15.1% among the 345 who did not need a protocol biopsy. Mild DGF (n=30) requiring one or two dialysis treatments had no effect on graft survival, but the unadjusted risk ratio (and 95% confidence interval) associated with more prolonged DGF (n=104) was 3.08 (2.09–4.52, P <0.0001). The risk for graft failure from AR detected on protocol biopsy was 2.91 (1.60–5.27, P =0.0004) and was similar to the risk from early AR in patients without DGF, 2.95 (1.72–5.07, P <0.0001). After taking the effects of AR into account, the risk of graft failure attributable to prolonged DGF was reduced to 1.76 (1.06–2.94, P =0.0294), suggesting that much of the risk of DGF was because of the risk of AR.

Conclusions.  

Silent AR is common during DGF. Prolonged DGF is associated with reduced graft survival after kidney transplantation, and much of this association can be explained by silent AR. In the absence of data from randomized trials, protocol biopsies and treatment of silent AR during prolonged DGF appear to be warranted.

Studies of large transplant registries suggest that delayed graft function (DGF), which is usually defined as the need for dialysis during the first week after transplantation, has an adverse effect on kidney allograft survival (^1,2). Some single-center studies have also found that DGF has an adverse effect on graft survival, (^3–5) while others have not (^6–8). There may be many reasons for this heterogeneity in study results. One possibility is that patients with DGF may have a higher incidence of early acute rejection (AR), (^2,6,8) and it is possible that the detection and treatment of AR could influence the effect of DGF on graft survival. Previous studies have examined the incidence of AR in patients undergoing protocol biopsies during DGF (^9–11). However, it is unclear from these studies how much of the adverse effects of DGF on graft survival can be explained by the effects of clinically silent AR. We therefore tested the hypothesis that silent AR is common during DGF, and that silent AR during DGF is associated with reduced long-term graft survival.

METHODS

Patient Population

We examined 410 consecutive kidney transplants performed between January 1, 1995 and March 21, 2000. In 1995, there were 42 patients treated with a sequential, antithymocyte globulin (ATG, Atgam, Upjohn, Kalamazoo, MI) induction protocol. These patients were treated for up to 14 days with ATG, or until urine output was established and the serum creatinine was declining. When ATG was discontinued, cyclosporine A (CsA, Sandimmune, Novartis Pharmaceuticals Corporation, East Hanover, NJ) was started. This protocol was abandoned after a randomized, controlled trial failed to substantiate any clinically relevant differences between ATG and CsA induction (¹²). Thereafter, patients who did not receive ATG received CsA induction, i.e., CsA was started immediately after transplantation. Triple therapy with CsA was used for maintenance immunosuppression in all but 9 (haploidentical, living related) transplants; 305 (74.4%) were treated with CsA microemulsion (Neoral, Novartis Pharmaceuticals Corporation). All patients received corticosteroids, 100 (24.4%) received azathioprine, and 310 (75.6%) received mycophenolate mofetil.

Biopsy Protocol

During the period of study, patients with DGF, defined by the need for dialysis after transplantation, were monitored with protocol biopsies. Specifically, patients who still required dialysis and did not have increasing urine output after 7 to 10 days underwent biopsy. Biopsies were repeated every 7 to 10 days until allograft function improved. AR was diagnosed using Banff criteria.

Analysis

We used logistic regression analysis to examine independent clinical correlates for DGF. In this analysis, we included recipient and donor age, race, ethnicity, and gender, as well as recipient body mass index, the number of major histocompatibility mismatches, the panel reactive antibody titer (peak and at the time of transplantation), the primary cause of kidney failure, the number of years of treatment for kidney failure before transplantation, and the type of prophylactic immunosuppression used (ATG or CsA induction protocols described above). We used Cox proportional hazards analysis to examine the independent effects of DGF and AR on graft failure. ARs were analyzed as time-dependent covariates. Analyses were carried out using the statistical software package SAS version 8.1 (SAS Institute, Inc., Cary, NC). Results were considered statistically significant for P <0.05.

RESULTS

Protocol Biopsies

Patients were divided into 5 groups (Table 1) according to whether or not they had DGF (groups 1–4) or no DGF (group 5). Patients with DGF were further subdivided into those who underwent protocol biopsy (group 1), those who for no clear reason did not have a protocol biopsy (group 2), those who did not have a protocol biopsy because urine output was increasing (group 3), and those who did not have a protocol biopsy because they had regained function (group 4). Two patients in group 2 did not undergo biopsy because they were too ill (both died). Three patients in group 4 had a biopsy, even though function appeared to be improving (none of the three had AR).

Table 1: Patient characteristics and outcomes (unadjusted)^a

Delayed Graft Function

Of 410 consecutive transplants, 134 (32.7%) required dialysis within the first week. Independent risk factors for DGF included (Table 2): cadaver donor source (12 of 176 living donor transplants had DGF vs. 122 of 234 cadaver donor transplants), a history of a previous transplant (26 of 51 vs. 108 of 359 without a previous transplant), donor age less than or equal to 30 years (29 of 122 vs. 105 of 288 age >30 years), recipient age less than or equal to 30 years (6 of 41 vs. 128 of 369 age >30 years), and white race (97 of 318 vs. 37 of 92 of other races). The number of years of treatment for end-stage kidney disease before transplantation was also associated with DGF in univariate (data not shown), but not multivariate, analysis (Table 2).

Table 2: Independent risk factors for delayed graft function

Of the 134 patients with DGF, 30 (22.4%) required only one or two dialysis treatments, while 104 (77.6%) required more than two dialysis treatments. Only three patients with one or two dialysis treatments underwent biopsy. Of those requiring more than two dialysis treatments, 65 of 104 (62.5%) underwent biopsy, while 39 of 104 (37.5%) did not. In 18 of 39 of these latter cases, a biopsy was not performed because the urine output was increasing. In 2 of 39 cases, the biopsy was not performed because the patient was too ill, but in 19 of 39, there was no explanation for why a biopsy was not performed, i.e., protocol violation. Those who underwent biopsy ultimately required 10.5 (8.5–12.4) dialysis treatments (mean and 95% confidence interval) compared with 5.5 (4.5–6.5) for those who did not undergo biopsy (P <0.0001).

Among the 410 transplants, 5 had primary nonfunction. The allograft from one of these patients was removed shortly after transplantation due to thrombosis. The other four underwent protocol biopsy while receiving dialysis; one showed AR and three showed ischemia–thrombosis without rejection.

Acute Rejection

Patients with DGF underwent a protocol biopsy every 7 to 10 days until kidney function recovered or the graft was deemed lost. Overall, AR was detected in 33 of 65 (50.8%) of patients who had at least one protocol biopsy for prolonged DGF. AR was detected in the first protocol biopsy in 24, in the second biopsy in 8 and not until a third biopsy in 1 (Fig. 1). In the face of persistent DGF, 3/4 of the repeat biopsies showed ongoing AR. The time from transplantation to the first AR diagnosed by protocol biopsy (n=33) was 16.6±9.6 days (mean±SD, range 3.0–39.0 days). During this same time period, the time to the first clinical AR diagnosed in patients without DGF requiring protocol biopsies (n=54) was 17.2±9.9 days (range 4.0–37.0 days). At 40 days posttransplant, the cumulative (Kaplan-Meier) incidence of AR was 57.2% among the 65 patients who had a protocol biopsy because of prolonged DGF, while it was only 15.1% among the 345 who did not have a protocol biopsy.

Figure 1: Results of multiple biopsies obtained in 65 patients with persistent delayed graft function (needing >2 dialysis treatments). About 3/4 of follow-up biopsies obtained for persistent delayed graft function showed on-going acute rejection (AR).

Graft Failure

The 134 patients with DGF were 2.86 (1.94–4.20) times more likely to have graft failure (P <0.0001) in unadjusted Cox proportional hazards analysis. However, for the 30 patients requiring less than or equal to two dialysis treatments, there was no adverse effect of DGF on graft survival (unadjusted P =0.7992), while for the 104 patients requiring more than two dialysis treatments, the unadjusted risk of graft failure was 3.08 (2.09–4.52) times higher compared with patients who required none or only one or two dialysis treatments (P <0.0001). The DGF requiring more than two dialysis treatments was an independent risk factor for graft failure after adjusting for multiple other variables (Table 3). However, much of the risk of prolonged DGF was attributable to AR (Table 3). If fact, the risk of graft failure from DGF requiring more than two dialysis treatments was reduced from more than 3-fold to 1.76 (1.06–2.94, P =0.0294) after taking the effects of AR into account (Table 3). AR detected by protocol biopsy during DGF increased the risk of graft failure almost three-fold, and was similar to the risk of early AR in patients who did not have DGF (Table 3).

Table 3: Independent risk factors for graft failure

DISCUSSION

The incidence of AR detected with protocol biopsies during DGF could vary depending on the type of immunosuppression used, the duration and severity of DGF, and the indications for biopsy. We conducted a Medline search and a hand search of bibliographies from key articles on DGF and found three studies reporting the incidence of AR on protocol biopsies during DGF (Table 4) (^9–11). In two of these studies, the incidence of DGF was 40.8 to 46.7%, i.e., similar to the crude incidence of 48.5%, and the cumulative incidence of 51.4%, seen in the present study (^9,10). However, in the third study, the incidence of AR associated with DGF was only 18.2% (¹¹).

Table 4: Incidence of acute rejection on protocol biopsies obtained during delayed graft function

In two of the previous studies, (^9,10) as in our study, biopsies were only obtained when DGF was prolonged or severe. This selection of higher-risk patients would be more expected to increase the incidence of AR detected by protocol biopsies than if biopsies were obtained on all patients with DGF. For example, in the study by Thistlethwaite and coworkers (⁹), 16 of 61 patients with DGF did not undergo biopsy because “the duration of graft dysfunction was too short”. In the study of Gaber and colleagues (¹⁰), a slightly greater proportion of patients with DGF underwent biopsy, but the incidence of AR was nevertheless similar. In a more recent study, the incidence of AR on protocol biopsy during DGF was only 18.2% (¹¹). However, this study excluded 21 of 109 patients who required a diagnostic biopsy before 1 week, and the incidence of AR in these patients was not indicated (¹¹). In all of these studies, the definition of DGF requiring dialysis posttransplant is subjective, and if physicians are more or less likely to order dialysis after transplantation, then the incidence of AR found on protocol biopsies in patients with DGF could also vary.

A study of Howard and coworkers (¹³) reported a 48% incidence of AR in patients with DGF. However, these authors did not specify exactly how many patients underwent protocol biopsies and how many of the episodes of AR were detected by protocol biopsies. As a result, it is difficult to determine the true incidence of AR on protocol biopsy in their study and to compare their results with those of our study or of others.

Altogether, it appears that the incidence of AR during prolonged or severe DGF is 40 to 50%. This high incidence seems to be independent of the type of immunosuppression used, specifically whether or not antibody induction was used. However, only a randomized trial with protocol biopsies in patients with DGF treated with and without antibody induction could determine whether antibody induction affects the incidence of silent AR during DGF.

Whether DGF has an adverse effect on graft survival, and whether an effect of DGF on graft survival can be attributed to AR, are controversial. Some, (^2,6,8) but not all, (^4,14) studies suggest that DGF is associated with a higher incidence of AR. Because AR is associated with an increased risk of graft failure, (²) an increased incidence of AR during DGF could be expected to result in a higher rate of graft failure. A number of studies suggest that DGF is associated with graft failure and that this association is independent of early ARs (^1–5). However, it is likely that not all AR episodes during DGF were detected in these studies, especially if protocol biopsies were not obtained routinely in all patients with DGF. Thus, it is still plausible that most, if not all, of the adverse effects of DGF on graft survival could be caused by the effects of silent AR on graft survival.

In our study, AR detected by protocol biopsy during DGF had a similar adverse effect on graft survival as AR detected by a rise in creatinine in patients who did not have DGF (Table 3). In addition, it appeared that much, but not all, of the adverse effects of DGF on graft survival could be explained by AR during DGF (Table 3). In all of these patients, the AR was treated with additional immunosuppression. Nevertheless, it is impossible to determine in our retrospective study whether the outcome would have been any different had we not obtained protocol biopsies during DGF, and had these rejection episodes gone undetected. In addition to the important long-term effects of silent AR, DGF without rejection was still associated with a significant reduction in long-term graft function.

There are no data from randomized controlled trials to determine whether the detection and treatment of clinically silent AR improves graft survival. In a small, randomized trial, treatment of AR detected on protocol biopsies during the first 6 months after transplantation was associated with a lower serum creatinine at 2 years (¹⁵). AR during DGF may be more severe than “silent” AR in patients with graft function because by definition all ARs during DGF are “silent.”

In summary, the results of this study suggest that the incidence of AR during prolonged DGF may be as high as 50%. Furthermore, AR detected by protocol biopsy during DGF is associated with reduced graft survival. Therefore, in the absence of data from randomized trials, protocol biopsies in patients with prolonged DGF appear to be warranted, and AR detected on protocol biopsies in this setting appears to warrant treatment.