Journal Logo

Original Clinical Science—General

Association of More Intensive Induction With Less Acute Rejection Following Intestinal Transplantation: Results of 445 Consecutive Cases From a Single Center

Vianna, Rodrigo MD, PhD1; Farag, Ahmed MD, PhD1,2; Gaynor, Jeffrey J. PhD1; Selvaggi, Gennaro MD1; Tekin, Akin MD1; Garcia, Jennifer MD3; Beduschi, Thiago MD1

Author Information
doi: 10.1097/TP.0000000000003074
  • Free

Abstract

INTRODUCTION

Intestinal transplantation has become one of the best treatment options for patients with irreversible intestinal failure due to refinements in surgical technique, better perioperative care, and development of more effective clinical protocols for immunosuppressive therapy, prophylactic care, and graft surveillance.1-5 While improvements in 1 year patient and graft survival have clearly occurred, concomitant improvements beyond 1 year posttransplant have been less clear.6,7 Achieving long-term graft survival requires minimization of immunosuppressive drug side effects, development of even more effective clinical protocols for early detection and treatment of acute rejection (AR), and greater understanding of the mechanisms of allograft tolerance.8,9

Acute cellular rejection (ACR) still remains a significant challenge to achieving both short- and long-term graft survival. Recent reports from large intestinal transplant centers showed that short-term ACR incidence has been reduced to 30%–40%, due, possibly in part, to the use of more potent/efficacious immunosuppression agents.2-5,7,10-12 Because ACR may occur in the setting of immunosuppression lowering due to development of various potentially life-threatening infections and/or other complications,13 better prophylactic care may clearly help to lower ACR incidence. In addition, determining optimal strategies to prevent/treat the occurrence of preformed and de novo donor-specific antibodies (DSAs) along with antibody-mediated rejection (AMR) has become an important issue, as both DSA types have been associated with AMR-type occurrence and refractory ACR, resulting in chronic rejection and intestinal graft failure.14,15

In terms of immunosuppression, in addition to using tacrolimus (TAC)8,11,16 and corticosteroids (tapered off by 6 mo posttransplant) as maintenance drugs, the polyclonal antibody rabbit antithymocyte globulin (rATG)16 combined with the chimeric anti-CD20 monoclonal antibody rituximab have been used as dual induction therapy at our center since 2013, based on previously reported successful results.17 rATG contains antibodies to a variety of T- and plasma cell/B-cell antigens and may also protect against reperfusion injury.18,19 Rituximab was added with the goal of more efficiently removing B cells, thereby lowering the risk of preformed/de novo DSAs developed/developing against the intestinal graft.

To determine the multivariable influence of baseline demographics, transplant-specific characteristics, and different induction regimens on the incidence rates of developing any ACR, severe ACR, and graft loss due-to-rejection, we analyzed all intestinal transplants performed at our institution since establishment of the intestinal transplant program in 1994. The results of this observational study are presented here.

MATERIALS AND METHODS

Patients and Immunosuppression

Our cohort of 445 consecutive intestinal transplant recipients at the Miami Transplant Institute during 1994–2017 were followed prospectively through March 15, 2019 (last follow-up date). Over the years, the center institutional review board approved each immunosuppression protocol used for these patients; all patients gave written informed consent before enrollment.

Recipients were divided into 5 induction groups. Group 1 (1994–1997) comprised 44 recipients who received no/old induction therapy (high-dose corticosteroids only in 34, anti-CD3 monoclonal antibody OKT3 in 7, and cyclophosphamide in 3). Group 2 (1998–2011) comprised 159 recipients who received an anti-CD25 monoclonal antibody (daclizumab in 156 and basiliximab in 3). Daclizumab (2 mg/kg) was given on postoperative days 0, 7, and 14, and then every 2 weeks during the first 3 months posttransplant; thereafter, daclizumab dose was reduced to 1 mg/kg every 2 weeks for the following 3 months and then stopped. Basiliximab (10 mg) was given on postoperative days 0 and 4, as the 3 recipients were small children (<35 kg). Group 3 (2001–2011) comprised 113 recipients who received alemtuzumab, with the following 2 different schedules being used: 0.3 mg/kg × 4 (preoperatively, immediately posttransplant, and on postoperative days 3 and 7); and 30 mg × 2 (on postoperative days 1 and 4). Group 4 (2006–2012) comprised 34 recipients who were scheduled to receive 3 rATG doses (total planned rATG dose: 5 mg/kg, with 2.0 mg/kg being given on postoperative day 0, and 1.5 mg/kg being given on postoperative days 2 and 4). However, the actual number of rATG doses that these patients received was uneven, as 12 out of 34 received only the first dose, and 3 out of 34 patients received only 2 doses.

Group 5 (2013–2017) comprised 95 recipients who received rATG/rituximab induction therapy, with a total rATG dose of 10 mg/kg divided into 5 equal doses given on postoperative days 0, 2, 4, 6, and 8, and a single dose of rituximab (150 mg/m2) given on postoperative day 1. The rationale in using a total rATG dose of 10 mg/kg was to achieve sufficiently high immunosuppressive protection against rejection risk without concomitantly increasing infection risk.17 In addition, due to concern that patients not receiving a full multivisceral (MV) graft may be at higher ACR risk, basiliximab (40 mg) was additionally given to the subset of group 5 patients who received either an isolated intestine (I) or modified MV (MMV) transplant once every 4 weeks × 3 starting on postoperative day 14.

Maintenance immunosuppression consisted of TAC and corticosteroids (tapered off by 6–9 mo posttransplant) except in patients who received alemtuzumab induction (group 3), where TAC alone was planned to be used. Target TAC trough levels during the first 3 months and beyond 3 months posttransplant were 15–20 ng/mL and 10–15 ng/mL for patients transplanted during 1994–1997, and 12–16 ng/mL and 8–12 ng/mL for patients transplanted during 1998–2012. For patients transplanted during 2013–2017 (group 5), target TAC trough levels during the first 3 months and beyond 3 months posttransplant were 9–15 and 5–9 ng/mL, respectively.

In the attempt to achieve reduced TAC dosing in group 5 while simultaneously avoiding ACR occurrence, a mammalian target of rapamycin (mTOR) inhibitor was combined with TAC as maintenance in 68 of 95 patients, with 25 of 68 receiving TAC/sirolimus, and 43 out of 68 receiving TAC/everolimus. The mTOR inhibitor was not scheduled to start until after 30 days posttransplant to allow for proper wound healing to occur.

Finally, it should be noted that donors received no immunosuppression treatment pretransplant.

Acute Cellular Rejection

Since 1998, frequent surveillance endoscopies and protocol biopsies were used in all patients, being performed twice a week during the first 2 weeks posttransplant, then weekly for 3–8 weeks, and subsequently once a month until stoma closure. Once an ACR was clinically suspected, an immediate endoscopy and biopsy were performed. All ACR episodes were clinically suspected, pathologically diagnosed,20,21 and treated; ACR grade (mild, moderate, or severe) was determined as the maximum pathologic grade observed during that episode.22 High-dose corticosteroids (via intravenous bolus injections) were used to treat mild ACR episodes. Antilymphocyte therapy was used in treating steroid-resistant and moderate-to-severe ACR episodes. Graft dysfunction due to resistant rejection was treated with graft removal and listing for retransplantation.

Of note, in the attempt to reduce the incidence of acute kidney injury due-to-dehydration from occurring in the early posttransplant period, some of the patients transplanted since 2013 received no stoma. Specifically, patients received no stoma when (1) it was technically and anatomically feasible (with appropriate length of native distal colon and rectum) and (2) they were considered to be at lower ACR risk (eg, received an MV transplant and were not highly presensitized). Indication biopsies were therefore only performed in those patients who received no stoma.

In 2013, we began performing prospective flow cytometry crossmatching as well as routine DSA determinations on all patients both pre and posttransplant. DSA+ was defined to occur when the mean fluorescent intensity count was >1000 (mean fluorescent intensity >3000 defined a high level). In addition, since 2013, all patients with preformed or de novo DSAs received intravenous immunoglobulin (500 mg/kg) weekly ×4 as part of posttransplant treatment, with 4 weekly rituximab doses (375 mg/m2/dose) being additionally added if the recipient received an I or MMV transplant, had persistent DSAs or suspected AMR.

Infection Prophylaxis

Cytomegalovirus (CMV) prophylaxis consisted of a combination of ganciclovir (used during the initial hospital stay), CMV immune globulin, and valganciclovir. For patients transplanted before 2013, valganciclovir was not routinely used; all patients transplanted since 2013 received valganciclovir (900 mg daily) for 12 months. In addition, among CMV donor+/recipient− patients, CMV immune globulin was given for 4 out of 12 months in patients transplanted before/since 2013. For the prevention of fungal infections, before 2013, all patients received nystatin orally for approximately 1 month posttransplant; since 2013, all patients received fluconazole until corticosteroids were discontinued (ie, for approximately 4–6 mo). Finally, in all patients, trimethoprim/sulfamethoxazole was used indefinitely for pneumocystis carinii pneumonia prophylaxis.

Statistics

Frequency distributions were determined for baseline categorical variables; mean and SE were calculated for baseline continuous variables, with geometric means (and corresponding SEs) used for continuous variables with skewed distributions (analyzed using natural logarithmic transformed data). Tests of association among baseline variables were performed using Pearson (uncorrected) chi-squared tests and ordinary (2-sided) t tests.

As of the last follow-up date (March 15, 2019), median follow-up among 59 transplant cases in induction group 5 who were alive with functioning grafts was 43 months (range: 16–68 mo) posttransplant. Since only 10 of 59 of these patients were followed beyond 60 months posttransplant, statistical evaluation of all clinical outcomes was restricted to the first 60 months posttransplant for all transplanted cases in this study. Graft loss was defined as the date of intestinal graft failure (graft removal) or death, whichever occurred first, with the underlying cause of (triggering event leading to) graft loss being determined in each case.23

Stepwise Cox regression was utilized to identify the significant baseline variable predictors of 5 hazard rates during the first 60 months posttransplant: development of any ACR episode, development of a severe ACR episode, and development of graft loss-due-to rejection (AR or chronic rejection), infection, and other causes, respectively. In performing each of these analyses, any competing events occurring other than the cause of interest were treated as censored observations. For 3 baseline variables in which a small subset of patients had a missing value, the observed mean was imputed for missing values in the multivariable analyses.24 Testing the validity of the Cox model proportional hazards assumption was performed by considering the inclusion of time by covariate interaction effects. Finally, the prognostic impact of the development of any ACR and severe ACR on cause-specific graft loss hazard rates were determined, with their occurrences defined as 0–1, time-dependent covariates in the Cox model.

Stepwise linear regression to determine the significant multivariable predictors of the likelihood of being transplanted since 2013 (ie, being in induction group 5) (yes/no) was also performed along with resulting propensity scores.25 Propensity scores are typically used as a way to control for the effects of any unbalanced distributions of other baseline variables existing between 2 study groups (ie, selection bias). In this case, the final Cox models were rerun after controlling for the propensity of being transplanted since 2013.

Finally, because no meaningful differences were observed for any of the clinical outcomes among the 3 less intensive induction therapy groups (ie, groups 1, 2, and 4), for the sake of clarity, these 3 groups were combined in the presentation of all results. Differences in freedom-from-occurrence of each clinical outcome were compared by the log-rank test, with actuarial estimates and time-to-failure curves generated using the Kaplan-Meier method. P ≤ 0.05 was considered to be statistically significant.

RESULTS

Baseline Characteristics

Baseline characteristics are shown in Table 1. Mean age at transplant was 18.8 years, with African-Americans and Hispanics comprising 18.0% (80/445) and 16.6% (74/445), respectively; retransplant cases comprised 12.1% (54/445). The percentage of recipients who received I, liver-intestine (LI), MMV, and full MV allografts were 28.1% (125/445), 8.5% (38/445), 8.8% (39/445), and 54.6% (243/445), respectively.

TABLE 1. - Distributions of selected baseline variables (N = 445)
Mean ± SE (or geometric mean*/SE) if continuous;
Baseline variable Percentage with characteristic if categorical
Recipient age, y 18.8 ± 0.9 (N = 445)
Recipient age
 Child 57.1% (254/445)
 Adult (≥18 y) 42.9% (191/445)
Recipient gender
 Female 49.2% (219/445)
 Male 50.8% (226/445)
Recipient race/ethnicity
 White (nonHispanic) 64.0% (285/445)
 Black (nonHispanic) 18.0% (80/445)
 Hispanic 16.6% (74/445)
 Asian 1.3% (6/445)
CMV status
 D−/R− 31.9% (142/445)
 D−/R+ 19.1% (85/445)
 D+/R− 26.1% (116/445)
 D+/R+ 22.9% (102/445)
Intestinal transplant status
 Primary 87.9% (391/445)
 Retransplant 12.1% (54/445)
Transplant type
 Isolated intestine 28.1% (125/445)
 Liver-intestine 8.5% (38/445)
 Modified multivisceral 8.8% (39/445)
 Multivisceral 54.6% (243/445)
Underwent native splenectomy
 No 36.6% (163/445)
 Yes 63.4% (282/445)
Native pancreaticoduodenal complex removed
 No 36.9% (164/445)
 Yes 63.1% (281/445)
Received a kidney
 No 91.0% (405/445)
 Yes 9.0% (40/445)
Received a large bowel
 No 42.9% (191/445)
 Yes 57.1% (254/445)
Received a liver
 No 36.9% (164/445)
 Yes 63.1% (281/445)
Received a pancreas
 No 31.9% (142/445)
 Yes 68.1% (303/445)
Received a spleen
 No 80.2% (357/445)
 Yes 19.8% (88/445)
Received a stomach
 No 37.3% (166/445)
 Yes 62.7% (279/445)
Cold ischemia time, h 7.16*/1.01 (N = 424)
Warm ischemia time, h 35.90*/1.01 (N = 420)
In hospital (vs at home) before transplant
 No 60.5% (256/423)
 Yes 39.5% (167/423)
Induction type
 Received no/old induction 9.9% (44/445)
 Received anti-CD25 35.7% (159/445)
 Received alemtuzumab 25.4% (113/445)
 Received rATG (pre-2013) 7.6% (34/445)
 Received rATG/rituximab (since 2013) 21.4% (95/445)
*/, multiplied and divided by; anti-CD25, antiinterleukin-2 receptor alpha chain (daclizumab or basiliximab); CMV, cytomegalovirus; D, donor; R, recipient; rATG, rabbit antithymocyte globulin (thymoglobulin).

Crosstabulations of transplant type with the removal of native organs/receiving donor organs are shown in Table 2. Native splenectomy was performed in few I/LI versus most MMV/MV recipients. For 2 I cases showing that a native splenectomy was performed, these 2 cases were retransplants of previously failed MV grafts. Similarly, the native pancreaticoduodenal complex was removed in none of I/LI versus all but 1 of MMV/MV cases. Finally, the donor spleen was transplanted into no I/LI cases versus 31.2% (88/282) of MMV/MV cases.

TABLE 2. - Cross-tabulations of transplant type with removal of native organs (no/yes) and receiving donor organs (no/yes)
Transplant type
Organ-specific surgery I LI MMV MV
Native splenectomy 1.6% (2/125) 5.3% (2/38) 92.3% (36/39) 99.6% (242/243)
Native PC removed 0.0% (0/125) 0.0% (0/38) 97.4% (38/39) 100.0% (243/243)
Received a kidney 4.0% (5/125) 2.6% (1/38) 10.3% (4/39) 12.3% (30/243)
Received a large bowel 48.0% (60/125) 21.1% (8/38) 56.4% (22/39) 67.5% (164/243)
Received a liver 0.0% (0/125) 100.0% (38/38) 0.0% (0/39) 100.0% (243/243)
Received a pancreas 0.8% (1/125) 52.6% (20/38) 100.0% (39/39) 100.0% (243/243)
Received a spleen 0.0% (0/125) 0.0% (0/38) 35.9% (14/39) 30.5% (74/243)
Received a stomach 0.0% (0/125) 0.0% (0/38) 100.0% (39/39) 98.8% (240/243)
I, isolated intestine; LI, liver-intestine; MMV, modified multivisceral; MV, multivisceral; PC, pancreaticoduodenal complex.

Propensity to Be Transplanted Since 2013

The following 6 baseline variables were associated in stepwise linear regression with a significantly higher propensity to be transplanted since 2013 (Table 3) (listed in order of selection): lower log{warm ischemia time} (P < 0.000001), received a large bowel (P < 0.000001), did not receive a spleen (P < 0.000001), recipient age ≥50 years (P = 0.00003), Hispanic recipient (P = 0.0003), and recipient not being in the hospital pretransplant (P = 0.005). Once these 6 variables were controlled, no other variables offered additional prognostic value (P > 0.05).

TABLE 3. - Linear regression model (via stepwise regression) for the propensity to receive rATG/rituximab (since 2013) (overall, 95/445 received rATG/rituximab)
Selected linear model a,b
Variable c,d P Coefficient ± SE
Log {WIT} <0.000001 −0.518 ± 0.051
Received a large bowel <0.000001 0.281 ± 0.031
Received a spleen <0.000001 −0.295 ± 0.037
Recipient age ≥50 y 0.00003 0.224 ± 0.054
Hispanic recipient 0.0003 0.141 ± 0.038
In-hospital pretransplant 0.005 −0.086 ± 0.030
aThe 6 selected variables are listed in order of selection into the linear regression model.
bThe R2 (coefficient of determination) for this 6 variable model was 0.47.
cVariables selected in the linear regression model were defined as follows: log {WIT} (continuous variable), received a large bowel = {1 if recipient received a large bowel, 0 otherwise}, received a spleen = {1 if recipient received a spleen, 0 otherwise}, recipient age ≥50 y = {1 if recipient age ≥50 y, 0 otherwise}, Hispanic recipient = {1 if recipient was Hispanic, 0 otherwise}, and in-hospital pretransplant = {1 if recipient was in the hospital pretransplant, 0 if the recipient was at home pretransplant}.
dThe intercept term ±SE for this 6 variable model was 1.959 ± 0.188 (P < 0.000001).
Log, natural logarithm; rATG, rabbit antithymocyte globulin; WIT, warm ischemia time.

Of note, the donor spleen was transplanted into 0.0% (0/66) of MMV/MV recipients since 2013 versus 40.7% (88/216) of previous MMV/MV recipients (P < 0.000001); in addition, donor large bowel was transplanted into nearly all recipients since 2013, 94.7% (90/95), versus 46.9% (164/350) of previous recipients (P < 0.000001).

Any ACR

ACR was observed in 61.3% (273/445) of transplanted cases during the first 60 months posttransplant, with the first ACR grade being mild or moderate in 75.8% (207/273) and severe in 24.2% (66/273) of cases. Among the first ACR episodes, 92.7% (253/273) occurred during the first 18 months posttransplant. Figure 1A shows that during the first 18 months posttransplant, MMV and MV transplant recipients had a significantly more favorable freedom-from-ACR in comparison with I and LI transplant recipients (P = 0.002), with LI recipients experiencing the poorest freedom-from-ACR. No significant difference was observed between MMV and MV transplant recipients (P = 0.61). Figure 1B shows highly significant differences in freedom-from-ACR by induction group (P = 0.000002), with group 5 (rATG/rituximab) having the most favorable outcome overall followed by group 3 (alemtuzumab). In fact, Figure 1C shows that during the first 24 days (0.8 mo) posttransplant, the time period in which the hazard rate of developing a first ACR (ie, negative slope of the Kaplan-Meier curve on the natural logarithm scale) was greatest, clear favorable effects of rATG/rituximab and alemtuzumab induction were seen (P < 0.000001). Of note, 52.0% (142/273) of first ACR episodes had occurred by 24 days posttransplant. Figure 1C also shows that much smaller effects of induction group were seen beyond 24 days posttransplant. Actuarial estimates of ACR-free survival by induction group (Table 4) shows that at 1 month posttransplant, the rATG/rituximab and alemtuzumab groups had more favorable values, 84% and 76%, compared with 49% for the 3 other (less intensive) induction groups combined; however, outcome for the alemtuzumab group became similarly less favorable by 60 months posttransplant.

TABLE 4. - Actuarial estimates of ACR-free survival by induction group
Months posttransplant Other/no induction (%) Alemtuzumab induction (%) rATG/rituximab induction (%)
1 49 76 84
3 39 51 70
6 35 45 67
12 33 38 55
24 28 30 50
60 23 23 44
ACR, acute cellular rejection.

FIGURE 1.
FIGURE 1.:
Kaplan-Meier freedom-from-ACR comparisons. A, Kaplan-Meier freedom-from-ACR by 4 transplant types (I, LI, MMV, and MV) during the first 18 mo posttransplant. B, Kaplan-Meier freedom-from-ACR by 3 induction therapy groups (other vs alemtuzumab vs rATG/rituximab) during the first 18 mo posttransplant. C, Kaplan-Meier freedom-from-ACR by 3 induction therapy groups (other vs alemtuzumab vs rATG/rituximab) during the first 3 mo posttransplant. D, Kaplan-Meier freedom-from-ACR by transplant type (I/LI vs MMV/MV) and induction therapy (other vs alemtuzumab vs rATG/rituximab) during the first 18 mo posttransplant. ACR, acute cellular rejection; I, isolated intestine; LI, liver-intestine; MMV, modified multivisceral; MV, multivisceral; rATG, rabbit antithymocyte globulin.

Three multivariable predictors were selected into the Cox model indicating a significantly lower hazard rate of developing a first ACR (Table 5A) (shown by order of selection): received rATG/rituximab induction (P < 0.000001), transplant type MMV or MV (P = 0.0009), and received alemtuzumab induction (P = 0.004). Once these 3 variables were controlled, no other variables offered additional prognostic value (P > 0.10).

TABLE 5. - Cox model for the hazard rate of developing a first ACR during the first 60 mo posttransplant (273 events)
(A) Selected Cox model a (via stepwise regression)
Variable b P Coefficient ± SE
Received rATG/rituximab induction <0.000001 −0.830 ± 0.169
Transplant type MMV or MV 0.0009 −0.412 ± 0.125
Received alemtuzumab induction 0.004 −0.421 ± 0.148
(B) Selected Cox model including the propensity to receive rATG/rituximab Induction
Variable b P Coefficient ± SE
Received rATG/rituximab induction 0.0008 −0.731 ± 0.219
Transplant type MMV or MV 0.0007 −0.420 ± 0.125
Received alemtuzumab induction 0.005 −0.416 ± 0.148
Propensity to receive rATG/rituximab 0.48 −0.211 ± 0.298
(C) Final Cox model including the significant time × covariate interaction effects c
Variable b P Coefficient ± SE
Received rATG/rituximab (during first 24 d) <0.000001 −1.664 ± 0.317
Received rATG/rituximab (beyond 24 d) 0.47 −0.159 ± 0.221
Transplant type MMV or MV 0.001 −0.406 ± 0.125
Received alemtuzumab (during first 24 d) 0.000005 −1.135 ± 0.239
Received alemtuzumab (beyond 24 d) 0.17 0.284 ± 0.206
aThe 3 selected variables are listed in order of selection into the Cox model.
bVariables included in the Cox model were defined as follows: received rATG/rituximab induction = {1 if recipient received rATG/rituximab induction, 0 otherwise}, transplant type MMV or MV = {1 if transplant type = MMV or MV, 0 otherwise}, received alemtuzumab = {1 if recipient received alemtuzumab induction, 0 otherwise}, propensity to receive rATG/rituximab (continuous variable), received rATG/rituximab (during first 24 d) = {1 if recipient received rATG/rituximab induction and time ≤24 d posttransplant, 0 otherwise}, received rATG/rituximab (beyond 24 d) = {1 if recipient received rATG/rituximab induction and time >24 d posttransplant, 0 otherwise}, received alemtuzumab (during first 24 d) = {1 if recipient received alemtuzumab induction and time ≤24 d posttransplant, 0 otherwise}, and received alemtuzumab (beyond 24 d) = {1 if recipient received alemtuzumab induction and time >24 d posttransplant, 0 otherwise}.
cLikelihood ratio tests of the interaction of received rATG/rituximab with time (during vs beyond the first 24 d posttransplant), transplant type MMV or MV with time (during vs beyond the first 24 d posttransplant), and received alemtuzumab with time (during vs beyond the first 24 d posttransplant) yielded P = 0.00003, 0.31, and 0.000002, respectively.
ACR, acute cellular rejection; MMV, modified multivisceral; MV, multivisceral; rATG, rabbit antithymocyte globulin.

Of note, there were no significant effects of recipient age in either univariable or multivariable analysis, and the Cox model result in Table 5A, remained reasonably consistent when analyzed separately for children and adults (results not shown). Other factors such as receiving a donor spleen were also not significant in either univariable or multivariable analysis (results not shown). In addition, Table 5B, shows that the Cox model results remained unchanged after controlling for the propensity to be in induction group 5.

Table 5C, shows (as reflected in Figure 1C) that the protective effects of receiving rATG/rituximab and alemtuzumab induction both disappeared over time. Specifically, the magnitudes of protective effects (Cox model coefficients) of receiving rATG/rituximab and alemtuzumab induction during the first 24 days posttransplant (P < 0.000001 and P < 0.000005, respectively) were similar (−1.664 and −1.135, respectively). While a negligible protective effect of rATG/rituximab remained beyond 24 days posttransplant (ie, negative Cox model coefficient, but P = 0.47), a nonsignificant, detrimental effect of alemtuzumab was observed beyond 24 days posttransplant (positive Cox model coefficient; P = 0.17). Likelihood ratio tests of the interaction of receiving rATG/rituximab and alemtuzumab induction with time yielded P = 0.00003 and P = 0.000002, respectively. The likelihood ratio test of an interaction of transplant type MMV/MV with time was nonsignificant (P = 0.31), indicating that a consistent protective effect of MMV/MV transplant existed over time.

Finally, Kaplan-Meier freedom-from-ACR curves (Figure 1D) by transplant type (I/LI versus MMV/MV) and induction group (other versus alemtuzumab versus rATG/rituximab) show reasonable consistency with the Cox model results, as the ACR rate was lower for MMV/MV (versus I/LI) across induction group, and lower for rATG/rituximab followed by alemtuzumab across transplant type.

Severe ACR

Severe (grade 3) ACR developed in 22.2% (99/445) of transplanted cases during the first 60 months posttransplant, with 66.7% (66/99) of these episodes occurring at the time of first ACR and 33.3% (33/99) occurring after the first ACR episode. Among the first severe ACR episodes, 84.8% (84/99) occurred during the first 18 months posttransplant. Figure 2A shows that during the first 18 months posttransplant, MMV and MV transplant recipients had a significantly more favorable freedom-from-severe ACR in comparison with I and LI transplant recipients (P = 0.000004), with LI recipients experiencing the poorest freedom-from-severe ACR (although negligibly different from I recipients). No significant difference was observed between MMV and MV recipients (P = 0.34). Figure 2B shows significant differences during the first 18 months posttransplant in freedom-from-severe ACR by induction group (P = 0.05), with group 5 (rATG/rituximab) having the most favorable outcome followed by group 3 (alemtuzumab). In fact, Figure 2C shows that during the first 24 days (0.8 mo) posttransplant, the time period in which the hazard rate of developing a severe ACR (ie, negative slope of the Kaplan-Meier curve on the natural logarithm scale) was greatest, clear favorable effects of rATG/rituximab and alemtuzumab induction were seen (P = 0.0003). Of note, 39.4% (39/99) of severe ACR episodes had occurred by 24 days posttransplant. Figure 2C also shows that much smaller effects of induction group were seen beyond 24 days posttransplant. Actuarial estimates of severe ACR-free survival by induction group (Table 6) shows that at 1 month posttransplant, the rATG/rituximab and alemtuzumab groups had more favorable values, 96% and 97%, compared with 85% for the 3 other (less intensive) induction groups combined; however, outcome for the alemtuzumab group became similarly less favorable by 60 months posttransplant.

TABLE 6. - Actuarial estimates of severe ACR-free survival by induction group
Months posttransplant Other/no induction (%) Alemtuzumab induction (%) rATG/rituximab induction (%)
1 85 97 96
3 81 90 92
6 78 90 91
12 75 81 88
24 73 75 84
60 66 71 81
ACR, acute cellular rejection; rATG, rabbit antithymocyte globulin.

FIGURE 2.
FIGURE 2.:
Kaplan-Meier freedom-from-severe ACR comparisons. A, Kaplan-Meier freedom-from-severe ACR by 4 transplant types (I, LI, MMV, and MV) during the first 18 mo posttransplant. B, Kaplan-Meier freedom-from-severe ACR by 3 induction therapy groups (other vs alemtuzumab vs rATG/rituximab) during the first 18 mo posttransplant. C, Kaplan-Meier freedom-from-severe ACR by 3 induction therapy groups (other vs alemtuzumab vs rATG/rituximab) during the first 3 mo posttransplant. D, Kaplan-Meier freedom-from-severe ACR by transplant type (I/LI vs MMV/MV) and induction therapy (other vs alemtuzumab vs rATG/rituximab) during the first 18 mo posttransplant. ACR, acute cellular rejection; I, isolated intestine; LI, liver-intestine; MMV, modified multivisceral; MV, multivisceral; rATG, rabbit antithymocyte globulin.

Three multivariable predictors were selected into the Cox model indicating a significantly lower hazard rate of developing a severe ACR (Table 7A) (shown by order of selection) (note: a type I error of 0.10 was used here, as a significant deviation from the proportional hazards model was expected): transplant type MMV or MV (P < 0.000001), received rATG/rituximab induction (P = 0.01), and received alemtuzumab induction (P = 0.07). Once these 3 variables were controlled, no other variables offered additional prognostic value (P > 0.10). Of note, the Cox model result in Table 7A, remained reasonably consistent when analyzed separately for children and adults (results not shown). In addition, Table 7B, shows that the Cox model results remained unchanged after controlling for the propensity to be in induction group 5.

TABLE 7. - Cox model for the hazard rate of developing a severe ACR during the first 60 mo posttransplant (99 events)
(A) Selected Cox model a (via stepwise regression)
Variable b P Coefficient ± SE
Transplant type MMV or MV <0.000001 −0.985 ± 0.204
Received rATG/rituximab induction 0.01 −0.727 ± 0.296
Received alemtuzumab induction 0.07 −0.440 ± 0.249
(B) Selected Cox model including the propensity to receive rATG/rituximab induction
Variable b P Coefficient ± SE
Transplant type MMV or MV <0.000001 −0.973 ± 0.205
Received rATG/rituximab induction 0.02 −0.905 ± 0.385
Received alemtuzumab induction 0.07 −0.449 ± 0.249
Propensity to receive rATG/rituximab 0.47 0.378 ± 0.521
(C) Final Cox model including the significant time × covariate interaction effects c
Variable b P Coefficient ± SE
Transplant type MMV or MV 0.000001 −0.985 ± 0.204
Received rATG/rituximab (during first 24 d) 0.01 −1.530 ± 0.603
Received rATG/rituximab (beyond 24 days) 0.43 −0.281 ± 0.354
Received alemtuzumab (during first 24 d) 0.003 −1.766 ± 0.603
Received alemtuzumab (beyond 24 d) 0.63 0.143 ± 0.294
aThe 3 selected variables are listed in order of selection into the Cox model.
bVariables included in the Cox model were defined as follows: transplant type MMV or MV = {1 if transplant type = MMV or MV, 0 otherwise}, received rATG/rituximab induction = {1 if recipient received rATG/rituximab induction, 0 otherwise}, received alemtuzumab = {1 if recipient received alemtuzumab induction, 0 otherwise}, propensity to receive rATG/rituximab (continuous variable), received rATG/rituximab (during first 24 d) = {1 if recipient received rATG/rituximab induction and time ≤24 d posttransplant, 0 otherwise}, received rATG/rituximab (beyond 24 d) = {1 if recipient received rATG/rituximab induction and time >24 d posttransplant, 0 otherwise}, received alemtuzumab (during first 24 d) = {1 if recipient received alemtuzumab induction and time ≤24 d posttransplant, 0 otherwise}, and received alemtuzumab (beyond 24 d) = {1 if recipient received alemtuzumab induction and time >24 d posttransplant, 0 otherwise}.
cLikelihood ratio tests of the interaction of transplant type MMV or MV with time (during vs beyond the first 24 d posttransplant), received rATG/rituximab with time (during vs beyond the first 24 d posttransplant), and received alemtuzumab with time (during vs beyond the first 24 d posttransplant) yielded P = 0.89, 0.05, and 0.001, respectively.
ACR, acute cellular rejection; MMV, modified multivisceral; MV, multivisceral; rATG, rabbit antithymocyte globulin.

Table 7C, shows (as reflected in Figure 2C) that the protective effects of receiving rATG/rituximab and alemtuzumab induction both disappeared over time. Specifically, the magnitudes of protective effects (Cox model coefficients) of receiving rATG/rituximab and alemtuzumab induction during the first 24 days posttransplant (P = 0.01 and 0.003, respectively) were similar (−1.530 and −1.766, respectively). While a negligible protective effect of rATG/rituximab remained beyond 24 days posttransplant (ie, negative Cox model coefficient, but P = 0.43), a nonsignificant, detrimental effect of alemtuzumab was observed beyond 24 days posttransplant (positive Cox model coefficient, P = 0.63). Likelihood ratio tests of the interaction of receiving rATG/rituximab and alemtuzumab induction with time yielded P = 0.05 and 0.001, respectively. The likelihood ratio test of an interaction of transplant type MMV/MV with time was nonsignificant (P = 0.89), indicating that a consistent protective effect of MMV/MV transplant existed over time.

Finally, Kaplan-Meier freedom-from-severe ACR curves (Figure 2D) by transplant type (I/LI versus MMV/MV) and induction group (other versus alemtuzumab versus rATG/rituximab) show reasonable consistency with the Cox model results, as the severe ACR rate was lower for MMV/MV (versus I/LI) across induction group, and lower for rATG/rituximab followed by alemtuzumab across transplant type.

Overall and Cause-specific Graft Survival

The observed incidence of graft loss due-to-any cause was 58.7% (261/445) during the first 60 months posttransplant, with the underlying cause of graft loss being due to rejection, infection, and other causes in 15.5% (69/445), 16.6% (74/445), and 26.5% (118/445) of cases, respectively. Specific graft losses due-to-rejection included AR (N = 53) and chronic rejection (N = 16). Specific graft losses due-to-infection included: sepsis (N = 53), pneumonia (N = 17), and other infections (N = 4). Specific graft losses due-to-other causes included cardiovascular event (N = 21), graft-versus-host diseases (GVHD) (N = 13), posttransplant lymphoproliferative disorder (PTLD) (N = 11, all types), ruptured pseudoaneurysm (N = 10), etc.

Observed percentages of patients with graft loss due-to-AR and graft loss due-to-chronic rejection who previously experienced a severe ACR episode were 94.3% (50/53) and 50.0% (8/16), respectively. In fact, log-rank test of the prognostic effect of developing a severe ACR on the hazard rate of developing graft loss due-to-rejection was highly significant (P < 0.000001).

Figure 3A shows that MMV and MV transplant recipients had a significantly more favorable freedom-from-graft loss due-to-rejection in comparison with I and LI transplant recipients (P < 0.000001). Figure 3B shows that during the first 6 months posttransplant, the time period in which the hazard rate of developing graft loss due-to-rejection (ie, negative slope of the Kaplan-Meier curve on the natural logarithm scale) was greatest, clear favorable effects of rATG/rituximab and alemtuzumab induction were seen (P = 0.008). Of note, 52.2%(36/69) of the rejection-specific graft losses had occurred by 6 months posttransplant. Figure 3B also shows that much smaller effects of induction group were seen beyond 6 months posttransplant.

FIGURE 3.
FIGURE 3.:
Kaplan-Meier freedom-from-graft loss due to rejection comparisons. A, Kaplan-Meier freedom-from-graft loss due-to-rejection by 4 transplant types (I, LI, MMV, and MV) during the first 60 mo posttransplant. B, Kaplan-Meier freedom-from-graft loss due-to-rejection by 3 induction therapy groups (other vs alemtuzumab vs rATG/rituximab) during the first 60 mo posttransplant. I, isolated intestine; LI, liver-intestine; MMV, modified multivisceral; MV, multivisceral; rATG, rabbit antithymocyte globulin.

The 3 significant multivariable predictors for the hazard rate of developing a severe ACR were also fitted in a Cox model for the hazard rate of developing graft loss due-to-rejection (Table 8A). As expected, the most significant protective factor for graft loss due-to-rejection was transplant type MMV or MV (P < 0.000001). Table 8A, also shows that under the proportional hazards assumption (ie, the prognostic variables’ effects remaining constant over time), the protective effects of rATG/rituximab and alemtuzumab induction were not statistically significant at the 0.05 level.

TABLE 8. - Cox model for the hazard rate of developing graft loss-due-to (acute or chronic) rejection (69 events), considering the same 3 variables selected for the hazard rate of developing a severe ACR
(A) Three variable Cox model
Variable a P Coefficient ± SE
Transplant type MMV or MV <0.000001 −1.273 ± 0.254
Received rATG/rituximab induction 0.06 −0.656 ± 0.352
Received alemtuzumab induction 0.14 −0.435 ± 0.300
(B) Final Cox model including the significant time × covariate interaction effects b
Variable a P Coefficient ± SE
Transplant type MMV or MV <0.000001 −1.268 ± 0.254
Received rATG/rituximab (during first 6 mo) 0.02 −1.419 ± 0.608
Received rATG/rituximab (beyond 6 mo) 0.96 0.023 ± 0.458
Received alemtuzumab (during first 6 mo) 0.03 −1.069 ± 0.486
Received alemtuzumab (beyond 6 mo) 0.69 0.161 ± 0.404
(C) Final Cox model including the significant time × covariate interaction effects and the propensity to receive rATG/rituximab induction
Variable a P Coefficient ± SE
Transplant type MMV or MV <0.000001 −1.267 ± 0.255
Received rATG/rituximab (during first 6 mo) 0.03 −1.444 ± 0.675
Received rATG/rituximab (beyond 6 mo) 0.99 −0.001 ± 0.537
Received alemtuzumab (during first 6 mo) 0.03 −1.071 ± 0.487
Received alemtuzumab (beyond 6 mo) 0.69 0.160 ± 0.405
Propensity to receive rATG/rituximab 0.93 0.055 ± 0.633
aVariables included in the Cox model were defined as follows: transplant type MMV or MV = {1 if transplant type = MMV or MV, 0 otherwise}, received rATG/rituximab induction = {1 if recipient received rATG/rituximab induction, 0 otherwise}, received alemtuzumab induction = {1 if recipient received alemtuzumab induction, 0 otherwise}, propensity to receive rATG/rituximab (continuous variable), received rATG/rituximab (during first 6 mo) = {1 if recipient received rATG/rituximab induction and time ≤6 mo posttransplant, 0 otherwise}, received rATG/rituximab (beyond 6 mo) = {1 if recipient received rATG/rituximab induction and time >6 mo posttransplant, 0 otherwise}, received alemtuzumab (during first 6 mo) = {1 if recipient received alemtuzumab induction and time ≤6 mo posttransplant, 0 otherwise}, and received alemtuzumab (beyond 6 mo) = {1 if recipient received alemtuzumab induction and time >6 mo posttransplant, 0 otherwise}.
bLikelihood ratio tests of the interaction of transplant type MMV or MV with time (during vs beyond the first 6 mo posttransplant), received rATG/rituximab with time (during vs beyond the first 6 mo posttransplant), and received alemtuzumab with time (during vs beyond the first 6 mo posttransplant) yielded P = 0.87, 0.05, and 0.05, respectively.
ACR, acute cellular rejection; MMV, modified multivisceral; MV, multivisceral; rATG, rabbit antithymocyte globulin.

However, Table 8B, shows (as reflected in Figure 3B) that the protective effects of receiving rATG/rituximab and alemtuzumab induction were significant during the first 6 months posttransplant (P = 0.02 and 0.03, respectively), with each of these protective effects disappearing over time (ie, nonproportional hazards). The magnitudes of protective effects (Cox model coefficients) of receiving rATG/rituximab and alemtuzumab induction during the first 6 months posttransplant were similar (−1.419 and −1.069, respectively). While the effect of rATG/rituximab was negligible beyond 6 months posttransplant (ie, Cox model coefficient close to 0, P = 0.96), a nonsignificant, detrimental effect of alemtuzumab was observed beyond 6 months posttransplant (positive Cox model coefficient; P = 0.69). Table 8C, shows that the Cox model results remained unchanged after controlling for the propensity to be in induction group 5.

In terms of graft loss due-to-infection, Figure 4 shows that freedom-from-graft loss due-to-infection was consistently more favorable for rATG/rituximab induction in comparison with the other induction groups over time (P = 0.0005), with actuarial freedom-from-graft loss due-to-infection at 60 month posttransplant for the rATG/rituximab, alemtuzumab, and other induction groups combined being 94%, 72%, and 72%, respectively.

FIGURE 4.
FIGURE 4.:
Kaplan-Meier freedom-from-graft loss due-to-infection by 3 induction therapy groups (other vs alemtuzumab vs rATG/rituximab) during the first 60 mo posttransplant. rATG, rabbit antithymocyte globulin.

The following 3 multivariable predictors were selected into the Cox model indicating a significantly lower hazard rate of developing graft loss due-to-infection (Table 9A) (shown by order of selection): recipient age ≥1 year (P = 0.001), primary transplant (versus retransplant) (P = 0.0002), and received rATG/rituximab induction (P = 0.003). Once these 3 variables were controlled, no other variables offered additional prognostic value (P > 0.05). Tables 9B and C, show that these Cox model results remained unchanged after controlling for (1) the propensity to receive rATG/rituximab, and (2) the unfavorable effect of developing an ACR episode, respectively. Thus, the significantly favorable protection offered by the rATG/rituximab group against graft loss due-to-infection appeared to be greater than the protection solely offered by reduced first ACR risk.

TABLE 9. - Cox model for the hazard rate of developing graft loss-due-to infection (74 events)
(A) Selected Cox model a via stepwise regression
Variable b P Coefficient ± SE
Recipient age <1 y 0.001 0.845 ± 0.267
Retransplant 0.0002 1.090 ± 0.312
Received rATG/rituximab induction 0.003 −1.316 ± 0.471
(B) Selected Cox model, controlling for the propensity to receive rATG/rituximab
Variable b P Coefficient ± SE
Recipient age <1 y 0.001 0.843 ± 0.271
Retransplant 0.0002 1.091 ± 0.312
Received rATG/rituximab induction 0.01 −1.300 ± 0.536
Propensity to receive rATG/rituximab 0.95 −0.036 ± 0.592
(C) Final Cox model, including the time-dependent covariate showing the prognostic impact of developing an ACR
Variable b P Coefficient ± SE
Recipient age <1 y 0.001 0.880 ± 0.268
Retransplant 0.0004 1.114 ± 0.312
Received rATG/rituximab induction 0.01 −1.194 ± 0.474
Any ACR {t} 0.02 0.589 ± 0.256
aThe 3 selected variables are listed in order of selection into the Cox model.
bVariables included in the Cox model were defined as follows: recipient age <1 y = {1 if recipient age <1 y, 0 otherwise}, retransplant = {1 if the transplant is a retransplant case, 0 otherwise}, received rATG/rituximab induction = {1 if recipient received rATG/rituximab induction, 0 otherwise}, propensity to receive rATG/rituximab (continuous variable), and any ACR {t} = {1 if the patient experienced an ACR episode at or before time t, 0 otherwise}.
ACR, acute cellular rejection; rATG, rabbit antithymocyte globulin.

Only 1 predictor, that being whether the patient was at home (versus in the hospital) pretransplant, was selected into the Cox model indicating a significantly lower hazard rate of developing graft loss due-to-other causes (P = 0.0004; table not shown); once this variable was controlled, no other variables offered additional prognostic value (P > 0.05). Receiving rATG/rituximab induction was not associated with the hazard rate of developing graft loss due-to-other causes in either univariable (P = 0.13) or multivariable (P = 0.31) analysis.

Finally, actuarial estimates of overall graft survival by induction group (Table 10) show a significant protective effect of group 5 (rATG/rituximab) that was maintained over time (P = 0.0002; log-rank test); differences between the alemtuzumab and other induction groups were negligible.

TABLE 10. - Actuarial estimates of graft survival by induction groupa
Months posttransplant Other/no induction (%) Alemtuzumab induction (%) rATG/rituximab induction (%)
1 86 81 94
3 72 71 86
6 65 65 82
12 56 56 77
24 47 43 68
60 37 34 58
aThe log-rank test comparing the 3 induction groups (2 degrees of freedom) for freedom-from-graft loss due to any cause yielded P = 0.0007. The log-rank test comparing rATG/rituximab vs all other induction groups combined yielded P = 0.0002.
rATG, rabbit antithymocyte globulin.

Other Analyses

Among patients transplanted since 2013, the percentage of I, MMV, and MV recipients who received no stoma was 0.0% (0/29), 20.0% (1/5), and 42.6% (26/61), respectively. As the use of no stoma led to fewer biopsies being performed, we compared the first ACR rate between patients who received versus did not receive a stoma (among those transplanted since 2013). While the use of stoma was associated in univariable analysis with a higher ACR rate (P = 0.05), it was not significant once the prognostic effect of transplant type (I versus MMV/MV) was controlled (P = 0.20).

Among patients transplanted since 2013, the use of an mTOR inhibitor (in 68/95 patients) as a time-dependent covariate was not associated with reduced ACR risk (P = 0.64 for first ACR; P = 0.26 for severe ACR). Furthermore, in comparing the hazard ratio of first ACR for I versus MV (and MMV versus MV) transplants between patients transplanted before and since 2013, no reduction in the hazard ratio was observed for patients transplanted since 2013, suggesting that there was little (if any) benefit in using 3 basiliximab doses (starting at day 14 posttransplant) among I and MMV recipients.

Finally, there was no significant difference in GVHD incidence between patients transplanted before versus since 2013 (P = 0.15, 26/350 versus 13/95), nor in PTLD incidence between patients transplanted before versus since 2013 (P = 0.50, 35/350 versus 8/95).

DISCUSSION

This observational study of the complete intestinal transplant experience at the Miami Transplant Institute (445 cases transplanted over a 24-y period) is one of the largest series of such cases reported to date. Our Cox model analysis found the same following 3 factors to be associated with a significantly more favorable freedom-from-ACR and freedom-from-severe ACR: MMV or MV transplant, rATG/rituximab induction (group 5), and alemtuzumab induction (group 3). However, the significant protective effects of these 2 more intensive induction regimens existed only during the first 24 days posttransplant (ie, while the hazard rate of first ACR development was at its peak). As expected, development of a severe ACR episode was a highly significant predictor of the hazard rate of developing graft loss due-to-rejection,22 and the significantly more favorable severe ACR rates observed during the first 24 days posttransplant in the more intensive induction groups 3 and 5 translated into a significantly lower graft loss due-to-rejection rate as well (but only during the first 6 mo posttransplant). Our Cox model analyses also showed that a significant lowering of the graft loss due-to-infection rate occurred in induction group 5, even after controlling for other significant predictors, including the development of a first ACR. Thus, the protective effect of induction group 5 for graft loss due-to-infection extended beyond its lowering of first ACR risk, suggesting that the longer use of viral and fungal prophylactic medications in patients transplanted since 2013 may have directly translated into fewer graft losses due-to-infection. While distinct improvements in overall patient and graft survival among patients transplanted since 2013 were observed, we are nonetheless constrained by this study’s observational nature in being able to make definitive cause and effect inferences.

In terms of using rATG and alemtuzumab as induction agents, as stated earlier, rATG may lessen ACR risk via its demonstrated protection against reperfusion injury.18,19 rATG has also been shown to promote an increase in FoxP3+ regulatory T cells,26,27 while results for alemtuzumab are contradictory, with 1 study28 showing versus 2 other studies26,29 not showing this type of increase. One study27 reported that rATG favorably alters the balance of regulatory T-to-effector memory T-cells posttransplant, while another study29 reported that alemtuzumab unfavorably alters this balance. Effector memory T cells have been reported to be uniquely prevalent at rejection.30 In terms of B cells, rATG has strong anti-B-cell activity,31 while alemtuzumab promotes expansion of regulatory B cells.32,33 However, 2 studies34,35 reported that the type of B-cell depletion/reconstitution seen following alemtuzumab use actually promotes de novo DSA development and higher AMR risk. It is currently unknown how these basic science differences translate into observed ACR/AMR outcomes. It is also unknown how one might utilize these induction agents more effectively in reducing longer-term ACR/AMR risk.

MMV and MV allografts were associated with a significantly lower incidence of developing ACR and severe ACR, as previously published,11,22,23 while liver inclusion (in LI and MV grafts) offered no noticeable protection against ACR development. To allow sufficient space in transplanting a MMV or MV allograft, extensive removal of native lymphoid tissue (ie, spleen, mesenteric lymph nodes, and intestinal mucosal lymphoid tissue) may offer a protective role, which was shown in a cardiac allograft animal model with indefinite immunological tolerance after removal of secondary lymphoid organs.36 Although results from the University of Pittsburgh show a clear protective effect of liver inclusion against the development of chronic rejection,2,3 with most of their reported graft losses due to “rejection” being due to chronic (rather than acute) rejection, we are not aware of any specific comparison made by that group showing a protective effect of liver inclusion on ACR development. In addition, while the recent Intestinal Transplant Registry report shows a significant protective effect of liver inclusion for overall graft survival, no distinction between predictors of graft loss due-to-acute versus chronic rejection was made.7 Abu-Elmagd et al15 did show that liver inclusion offered a protective effect against the occurrence of both persistent preformed and de novo DSAs, each of which were associated with a significantly higher hazard rate of developing chronic (but not acute) rejection. A more definitive analysis at our center of graft loss due-to-chronic rejection would require much longer follow-up than the 60-month period used in this study. While organ graft tolerance due to liver inclusion was clearly shown in the laboratory over 25 years ago,37-39 its applicability to clinical outcomes following intestinal transplantation is still not fully understood.

Numerous study limitations did exist. First and foremost is the fact that we are reporting the results of an observational study—the gold standard for reporting comparisons of immunosuppressive regimens would be a randomized clinical trial. In spite of our best attempt to control for selection bias via a propensity score analysis, the possibility still remains that some other uncontrolled variable (or combination of variables) would more accurately explain the graft survival improvement seen at our center among patients transplanted since 2013. The fact that graft loss due to all causes was statistically more favorable among patients transplanted since 2013 may simply reflect more experience in all fields of intestinal transplant management. A statistical determination as to whether the improvements seen were due an era effect rather than specific changes made in immunosuppression/prophylactic care could be attempted via analysis of the Intestinal Transplant Registry data, whereby different immunosuppression protocols would be compared within a given era, and transplant center would be considered as a confounder.

Second, DSA and humoral rejection data were not available in most patients transplanted before 2013; thus, no attempt to analyze such results were made here. Third, infection-specific data with dates of occurrence and infection types were not available in patients transplanted before 2013, nor were adverse event data such as development of persistent neutropenia, etc. Forth, while target TAC levels have been continually declining over time since our intestinal transplant program began in 1994, we did not have serial TAC trough levels available in patients transplanted before 2013. Thus, an analysis of the prognostic impact of serial TAC levels on ACR development was not possible.

In practice, while we are continuing to use the rATG/rituximab induction regimen as described in this report, we are simultaneously considering the inclusion of other potentially useful agents. In addition, we are continuing to perform MV transplants without an ostomy in selected cases.40

In summary, a distinct improvement in 5-year patient and graft survival was observed among patients transplanted in the most recent era (ie, since 2013), where the more intensive induction regimen consisting of rATG (total dose: 10 mg/kg)/rituximab was used in combination with longer prophylactic care. While I and MMV transplant recipients (considered to be at higher ACR risk in comparison with MV recipients) also received 3 standard basiliximab doses (starting on day 14 posttransplant) as part of an extended induction, it is our belief that this more intensive/effective induction regimen used at our center since 2013 has clinical relevance in terms of yielding fewer rejection episodes over a total period of 60 months posttransplant, which concomitantly led to a lowering of the total amount of immunosuppression required in these patients during that period. It is our hope that the results presented in this rather large observational study will provide some guidance in terms of how to further develop an immunosuppressive strategy which leads to even greater improvements in clinical outcomes and long-term patient care following intestinal transplantation.

REFERENCES

1. Abu-Elmagd K, Bond G, Reyes J, et al. Intestinal transplantation: a coming of age. Adv Surg. 2002; 36:65–101
2. Abu-Elmagd K, Reyes J, Bond G, et al. Clinical intestinal transplantation: a decade of experience at a single center. Ann Surg. 2001; 234:404–416
3. Abu-Elmagd KM, Costa G, Bond GJ, et al. Five hundred intestinal and multivisceral transplantations at a single center: major advances with new challenges. Ann Surg. 2009; 250:567–581
4. Grant D, Abu-Elmagd K, Reyes J, et al.; Intestine Transplant Registry. 2003 report of the intestine transplant registry: a new era has dawned. Ann Surg. 2005; 241:607–613
5. Fishbien TM. Intestinal transplantation. N Engl J Med. 2009; 361:998–1008
6. Sudan D. The current state of intestine transplantation: indications, techniques, outcomes and challenges. Am J Transplant. 2014; 14:1976–1984
7. Grant D, Abu-Elmagd K, Mazariegos G, et al.; Intestinal Transplant Association. Intestinal transplant registry report: global activity and trends. Am J Transplant. 2015; 15:210–219
8. Abu-Elmagd KM, Costa G, Bond GJ, et al. Evolution of the immunosuppressive strategies for the intestinal and multivisceral recipients with special reference to allograft immunity and achievement of partial tolerance. Transplant Int. 2009; 22:96–109
9. Abu-Elmagd KM, Kosmach-Park B, Costa G, et al. Long-term survival, nutritional autonomy, and quality of life after intestinal and multivisceral transplantation. Ann Surg. 2012; 256:494–508
10. Reyes J, Mazariegos GV, Bond G, et al. Pediatric intestinal transplantation: historical notes, principles and controversies. Pediatr Transplant. 2002; 6:193–207
11. Tzakis AG, Kato T, Levi DM, et al. 100 multivisceral transplants at a single center. Ann Surg. 2005; 242:480–490
12. Smith JM, Skeans MA, Horslen SP, et al. OPTN/SRTR 2013 Annual Data Report: Intestine. Am J Transplant. 2015; 15Suppl 21–16
13. Sudan DL, Kaufman SS, Shaw BW Jr, et al. Isolated intestinal transplantation for intestinal failure. Am J Gastroenterol. 2000; 95:1506–1515
14. Kato T, Mizutani K, Terasaki P, et al. Association of emergence of HLA antibody and acute rejection in intestinal transplant recipients: a possible evidence of acute humoral sensitization. Transplant Proc. 2006; 38:1735–1737
15. Abu-Elmagd KM, Wu G, Costa G, et al. Preformed and de novo donor specific antibodies in visceral transplantation: long-term outcome with special reference to the liver. Am J Transplant. 2012; 12:3047–3060
16. Reyes J, Mazariegos GV, Abu-Elmagd K, et al. Intestinal transplantation under tacrolimus monotherapy after perioperative lymphoid depletion with rabbit anti-thymocyte globulin (thymoglobulin). Am J Transplant. 2005; 5:1430–1436
17. Vianna RM, Mangus RS, Fridell JA, et al. Induction immunosuppression with thymoglobulin and rituximab in intestinal and multivisceral transplantation. Transplantation. 2008; 85:1290–1293
18. Beiras-Fernandez A, Chappell D, Hammer C, et al. Influence of polyclonal anti-thymocyte globulins upon ischemia-reperfusion injury in a non-human primate model. Transpl Immunol. 2006; 15:273–279
19. Bogetti D, Jarzembowski TM, Sankary HN, et al. Hepatic ischemia/reperfusion injury can be modulated with thymoglobulin induction therapy. Transplant Proc. 2005; 37:404–406
20. Lee RG, Nakamura K, Tsamandas AC, et al. Pathology of human intestinal transplantation. Gastroenterology. 1996; 110:1820–1834
21. Ruiz P, Bagni A, Brown R, et al. Histological criteria for the identification of acute cellular rejection in human small bowel allografts: results of the pathology workshop at the VIII international small bowel transplant symposium. Transplant Proc. 2004; 36:335–337
22. Selvaggi G, Gaynor JJ, Moon J, et al. Analysis of acute cellular rejection episodes in recipients of primary intestinal transplantation: a single center, 11-year experience. Am J Transplant. 2007; 7:1249–1257
23. Kato T, Gaynor JJ, Selvaggi G, et al. Intestinal transplantation in children: a summary of clinical outcomes and prognostic factors in 108 patients from a single center. J Gastrointest Surg. 2005; 9:75–89
24. Afifi AA, Elashoff RM. Missing observations in multivariate statistics II. Point estimation in simple linear regression. J Am Statist Assoc. 1967; 62:10–29
25. Stukel TA, Fisher ES, Wennberg DE, et al. Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods. JAMA. 2007; 297:278–285
26. Lopez M, Clarkson MR, Albin M, et al. A novel mechanism of action for anti-thymocyte globulin: induction of CD4+CD25+foxp3+ regulatory T cells. J Am Soc Nephrol. 2006; 17:2844–2853
27. Gurkan S, Luan Y, Dhillon N, et al. Immune reconstitution following rabbit antithymocyte globulin. Am J Transplant. 2010; 10:2132–2141
28. Bloom DD, Chang Z, Fechner JH, et al. CD4+ CD25+ FOXP3+ regulatory T cells increase de novo in kidney transplant patients after immunodepletion with campath-1H. Am J Transplant. 2008; 8:793–802
29. Macedo C, Walters JT, Orkis EA, et al. Long-term effects of alemtuzumab on regulatory and memory T-cell subsets in kidney transplantation. Transplantation. 2012; 93:813–821
30. Pearl JP, Parris J, Hale DA, et al. Immunocompetent T-cells with a memory-like phenotype are the dominant cell type following antibody-mediated T-cell depletion. Am J Transplant. 2005; 5:465–474
31. Zand MS, Vo T, Huggins J, et al. Polyclonal rabbit antithymocyte globulin triggers B-cell and plasma cell apoptosis by multiple pathways. Transplantation. 2005; 79:1507–1515
32. Cherukuri A, Salama AD, Carter C, et al. An analysis of lymphocyte phenotype after steroid avoidance with either alemtuzumab or basiliximab induction in renal transplantation. Am J Transplant. 2012; 12:919–931
33. Heidt S, Hester J, Shankar S, et al. B cell repopulation after alemtuzumab induction-transient increase in transitional B cells and long-term dominance of naïve B cells. Am J Transplant. 2012; 12:1784–1792
34. Bloom D, Chang Z, Pauly K, et al. BAFF is increased in renal transplant patients following treatment with alemtuzumab. Am J Transplant. 2009; 9:1835–1845
35. Todeschini M, Cortinovis M, Perico N, et al. In kidney transplant patients, alemtuzumab but not basiliximab/low-dose rabbit anti-thymocyte globulin induces B cell depletion and regeneration, which associates with a high incidence of de novo donor-specific anti-HLA antibody development. J Immunol. 2013; 191:2818–2828
36. Lakkis FG, Arakelov A, Konieczny BT, et al. Immunologic ‘ignorance’ of vascularized organ transplants in the absence of secondary lymphoid tissue. Nat Med. 2000; 6:686–688
37. Davies HFS, Kamada N, Roser BJ. Mechanisms of donor-specific unresponsiveness induced by liver grafting. Transplant Proc. 1983; 15:831–835
38. Starzl TE, Demetris AJ, Murase N, et al. Cell migration, chimerism, and graft acceptance. Lancet. 1992; 339:1579–1582
39. Calne R, Davies H. Organ graft tolerance: the liver effect. Lancet. 1994; 343:67–68
40. Garcia J, Beduschi T, Pfeiffer B, et al. Multivisceral transplant without an ostomy – 5-year experience. Transplantation. 2019; 103Suppl 2 7S2S6–S7
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.