Learn more: PMC Disclaimer | PMC Copyright Notice
Survival in HIV-positive transplant recipients compared with transplant candidates and with HIV-negative controls
Associated Data
Abstract
Objectives
To evaluate the impact of liver and kidney transplantation on survival in HIV-positive transplant candidates and compare outcomes between HIV-positive and negative recipients.
Design
Observational cohort of HIV-positive transplant candidates and recipients and secondary analysis comparing study recipients to HIV-negative national registry controls.
Methods
We fit proportional hazards models to assess transplantation impact on mortality among recipients and candidates. We compared time to graft failure and death with HIV-negative controls in unmatched, demographic-matched, and risk-adjusted models.
Results
There were 17 (11.3%) and 46 (36.8%) deaths among kidney and liver recipients during a median follow-up of 4.0 and 3.5 years, respectively. Transplantation was associated with survival benefit for HIV-infected liver recipients with model for end-stage liver disease (MELD) greater than or equal 15 [hazard ratio (HR) 0.1; 95% confidence interval (CI) 0.05, 0.01; P <0.0001], but not for MELD less than 15 (HR 0.7; 95% CI 0.3, 1.8; P =0.43) or for kidney recipients (HR 0.6; 95% CI 0.3, 1.4; P =0.23). In HIV-positive kidney recipients, unmatched and risk-matched analyses indicated a marginally significant HR for graft loss [1.3 (P =0.07) and HR 1.4 (P =0.052)]; no significant increase in risk of death was observed. All models demonstrated a higher relative hazard of graft loss or death in HIV-positive liver recipients; the absolute difference in the proportion of deaths was 6.7% in the risk-matched analysis.
Conclusion
Kidney transplantation should be standard of care for well managed HIV-positive patients. Liver transplant in candidates with high MELD confers survival benefit; transplant is a viable option in selected candidates. The increased mortality risk compared with HIV-negative recipients was modest.
Trial Registration
ClinicalTrials.Gov; NCT00074386; http://clinicaltrials.gov/.
Introduction
Almost 2% of people with HIV are estimated to develop kidney failure, and liver failure is an important non-AIDS cause of death [1–4]. In the Solid Organ Transplantation in HIV: Multi-Site Study (HIV-TR), we hypothesized that immunosuppression would not accelerate HIV disease progression nor reduce survival in recipients with a relatively intact immune system and well suppressed viremia, and that outcomes would be similar to other commonly accepted higher-risk transplant patients, such as those over the age of 65 [5–7].
We previously described patient and graft survival in 150 kidney and 125 liver transplant recipients [8–10]. One to three year outcomes in kidney recipients and HIV-hepatitis B virus (HBV) co-infected liver recipients were similar to historical data from transplant registries, whereas outcomes in HIV-hepatitis C virus (HCV) co-infected recipients were poorer [8–10]. To evaluate the impact of transplantation, we now describe graft and patient survival among HIV-infected transplant recipients compared with those who were eligible for but did not receive transplants. To assess the impact of HIV-infection on transplant outcomes, we compare 5-year patient and graft survival in HIV-TR recipients with unmatched, demographic-matched, and risk-matched HIV-negative recipients identified in the national database. We describe predictors of patient and graft survival to aid in the selection and management of HIV-infected transplant candidates. Finally, we describe opportunistic infections, HIV-associated nephropathy (HIVAN), other infections requiring hospitalization, and CD4+ T-cell and HIV-1 RNA trends.
Methods
Study participant selection criteria were described previously [8–10]. Briefly, the CD4+ T-cell count was greater than or equal to 200 cells/ml in kidney recipients and at least 100 cells/ml in liver recipients. Kidney candidates had undetectable HIV-1 RNA whereas liver recipients could have detectable HIV-1 RNA if a fully suppressive antiretroviral regimen was likely to be tolerated post-transplant. Patients with a history of opportunistic infections or cancers without effective therapies were excluded. Endpoints included death and graft failure, defined as the first occurrence of death, re-transplantation or initial return to chronic dialysis in kidney recipients and death or re-transplantation in liver recipients. In un-transplanted candidates, we measured CD4+ T-cell counts and HIV-1 RNA and assessed ongoing study eligibility quarterly. All analyses are stratified by organ.
Candidate–recipient analysis: impact of transplant on survival
Transplant candidates did not receive a transplant due to lack of organ availability, no longer meeting study eligibility requirements, being transplanted off-study, dying before an organ became available, inability to adhere to the study requirements, their own decision, or the study reaching its enrollment cap. We compared mortality between transplant recipients and eligible candidates who did not receive a transplant during the study period via proportional hazards models. Transplantation was examined as a time-dependent covariate to assess its impact of mortality.
Predictors of transplant recipient death and graft loss
We examined the following factors in proportional hazards models of death and graft loss: HIV disease stage; potentially hepatotoxic or nephrotoxic antiretrovirals and opportunistic infection prophylaxis; potential drug interactions that may result in toxic or inadequate immunosuppression exposure; and drug interactions that could result in antiretroviral antagonism. HIV-specific predictors in recipient mortality models included enrollment/most recent pretransplant detectable HIV RNA in liver recipients and nadir/enrollment/most recent pre-transplant CD4+ T-cell count. All variables with P <0.1 from univariate models were included in the respective initial multivariate model. Subsequently, variables with P ≥ 0.1 were excluded, the model was re-fit, and interactions were examined.
Complications and other outcomes in HIV-infected transplant recipients
We examined the incidence of opportunistic infections and HIVAN and the incidence, characteristics and predictors of non-opportunistic serious infections resulting in hospitalization. Predictors of within-person CD4+ T-cell trends were examined in univariate and multivariate linear repeated measures models. One- and 3-year Kaplan–Meier estimates for cumulative incidence of loss of HIV RNA control were developed; proportional hazards models were fit by organ to identify predictors of any loss of HIV RNA control. Any loss of HIV RNA control was defined as two or more consecutive detectable measures.
HIV-infected recipient versus HIV-negative control analysis: impact of HIV on patient and graft survival
We compared the outcomes of HIV-TR recipients with controls identified from the Scientific Registry of Transplant Recipients (SRTR) as of 15 March 2013. The SRTR includes data on all transplant recipients in the United States. Identification of HIV-TR recipients in the database was conducted by the SRTR using sex, organ, and birth and transplant dates as identifiers. The control set, restricted to the HIV-TR study period for comparability, excluded HIV-TR recipients, other documented HIV-positive recipients, and re-transplant recipients.
For each of the graft failure and death time-to-event comparisons, four proportional hazards models were fitted to assess the consistency of the findings. These included an unmatched model adjusting for all covariates and using all eligible SRTR recipients as controls; demographic-matched and risk-matched controls examining HIV status; and a demographic-matched control analysis adjusted for risk score. Two matched control selection strategies, demographic-based and risk-based (propensity scores), were conducted with a 1 : 4 study participants to control ratio.
For risk-based matching, individual risk scores were calculated using individual characteristics and parameter estimates from proportional hazards regression models (excluding HIV-positive recipients) evaluating recipient sex, ethnicity, age at transplant, diabetes, hypertension, body mass index (BMI), hepatitis C antibody, hepatitis B core antibody, hepatitis B surface antigen, cytomegalovirus antibody status, work status, education, and primary method of payment; human leukocyte antigen match, cold ischemic time, and time of transplant (10/1/03–12/ 31/05, 1/1/06–12/31/07, 1/1/08–2/7/10); and donor sex, ethnicity, age, diabetes, hypertension, and cause of death. The kidney model was further adjusted for primary cause of kidney failure, time on dialysis, panel reactive antibody, expanded donor criteria [8] and donor relationship to recipient [11]. The liver model was further adjusted for MELD score, multiple organ transplant, donor BMI, smoking, bilirubin and creatinine. The liver model variables were identified in a previous analysis of transplant risk [12].
These 27 and 31 variables were used to calculate kidney and liver recipient risk scores, respectively, with no variable selection. Two events resulted in two risk-matched sets: time to graft failure and to death. The study participants and controls were combined and ranked by their risk score. Two adjacent controls with higher and two with lower risk scores were selected. As risk-based matching already took into account the covariates described, for each of the two time-to-event comparisons, a proportional hazards model was fitted examining only the HIV status.
For the demographic-matched analysis, controls were randomly selected among those matched by time of transplant (within 1 year), age (within 2 years), sex, ethnicity, deceased versus living donor, and receipt of concomitant kidney for liver recipients. For each outcome, an unadjusted demographic-matched proportional hazards model was fitted evaluating the HIV status only. For the adjusted demographic-matched model, a risk factor score was added to account for the variables not adjusted for by demographic-based matching. For the kidney and liver models, the remaining 22 and 25 variables, respectively, were entered into the proportional hazards models, and subsequently the calculated risk score was entered into the final adjusted demographic-match models as a covariate in addition to HIV status.
The same set of 600 and 496 matched controls was used for the two outcomes for the demographic-matched analyses among kidney and liver transplant recipients, respectively. Missing values were incorporated in the models as a new covariate level.
Results
We enrolled 125 liver and 150 kidney transplant recipients and 148 liver and 167 kidney transplant candidates between October 2003 and February 2010 (liver)/June 2009 (kidney) (Table 1) [8–10]. Surviving liver and kidney recipients were followed posttransplant for a median of 3.5 years [interquartile range (IQR) 1.8, 5.0] and 4.0 years (IQR 3.1, 5.0), and a maximum of 6.1 and 5.9 years, respectively. Liver and kidney candidates were followed postenrollment for a median of 1.1 years (IQR 0.1, 2.8) and 1.1 years (IQR 0.5, 2.2), and a maximum of 5.9 and 4.5 years, respectively. Recipient deaths are shown in Table 2.
Table 1
Characteristics of HIV-infected liver and kidney transplant recipients and candidates.
| Kidney
| Liver
| |||
|---|---|---|---|---|
| Recipients (N =150) | Candidates (N =167) | Recipients (N =125) | Candidates (N =148) | |
| Demographic and socioeconomic characteristics | ||||
| Age – median [IQR] years | 46 [40–51] | 45 [39–52] | 48 [43–53] | 49 [44–54] |
| Male sex – no. (%) | 117 (78) | 141 (84) | 97 (78) | 119 (80) |
| Caucasian – no. (%) | 42 (28) | 42 (25) | 86 (69) | 108 (73) |
| African American– no. (%) | 103 (69) | 114 (68) | 31 (25) | 22 (15) |
| Medicare – no. (%) | 80 (53) | 85 (51) | 37 (30) | 43 (29) |
| Medicaid – no. (%) | 20 (13) | 24 (14) | 24 (19) | 28 (19) |
| Private insurance – no. (%) | 46 (31) | 54 (32) | 54 (43) | 66 (45) |
| Employed – no. (%) | 56 (37) | 62 (37) | 29 (23) | 40 (27) |
| Transplantation indications and related characteristics | ||||
| Hepatitis C – no. (%) | 28 (19) | 18 (11) | 89 (71) | 110 (75) |
| Hepatitis B – no. (%) | 5 (3) | 14 (8) | 28 (22) | 24 (16) |
| Hepatocellular carcinoma – no. (%) | N/A | N/A | 45 (36) | 15 (10) |
| MELD at listing – median [IQR] | N/A | N/A | 19 [12–27] | 13 [10–16] |
| MELD at transplant – median [IQR] | N/A | N/A | 20 [15–29] | N/A |
| BMI at listing – median [IQR] | 25 [23–29] | 25 [22–28] | 25 [23–28] | 24 [22–28] |
| BMI at transplant – median [IQR] | 25 [22–29] | N/A | 25 [23–29] | N/A |
| Cause of kidney disease – no. (%) | N/A | N/A | ||
| HIV-associated nephropathya | 36 (24) | 54 (32) | ||
| Hypertension | 38 (25) | 38 (23) | ||
| Diabetic nephropathy | 12 (8) | 18 (11) | ||
| Focal glomerulosclerosis | 9 (6) | 6 (4) | ||
| Unknown or other cause | 55 (37) | 51 (31) | ||
| HIV-specific baseline characteristics | ||||
| CD4+ T-cell (cells/μl)b – median [IQR] | 524 [385–672] | 465 [313–600] | 291 [210–435] | 275 [194–417] |
| Nadir CD4+ T-cell (cells/μl)– median [IQR] | 239 [106–419] | 257 [117–428] | 191 [82–324] | 161 [100–292] |
| HIV RNA undetectableb – no. (%) | 150 (100) | 167 (100) | 110 (88) | 131 (89) |
| Prior opportunistic complication – no. (%) | 37 (25) | 31 (19) | 15 (12) | 29 (20) |
| Pneumocystis jiroveci pneumonia – no. | 21 | 17 | 9 | 9 |
| Kaposi’s sarcoma – no. | 3 | 4 | 0 | 3 |
| Cytomegalovirus retinitis – no. | 6 | 4 | 0 | 2 |
| Cytomegalovirus, other – no. | 1 | 0 | 1 | 2 |
| Mycobacterium avium complex – no. | 6 | 0 | 2 | 2 |
| Prior antiretroviral use – no. (%) | 150 (100) | 167 (100) | 119 (95) | 145 (98) |
| HIV-specific characteristics | ||||
| On antiretrovirals within 1st week posttransplant (%) | 139 (93) | N/A | 104 (83) | N/A |
| Antiretroviral regimen classc – no. (%) | N/A | N/A | ||
| Protease inhibitor | 65 (43) | 64 (51) | ||
| Non-nucleoside (NNRTI) | 62 (41) | 32 (26) | ||
| Protease inhibitor and NNRTI | 15 (10) | 8 (6) | ||
| Nucleosides only | 5 (3) | 15 (12) | ||
| Raltegravir (without protease inhibitor or NNRTI) | 1 (1) | 2 (2) | ||
| None | 2 (1) | 4 (3) | ||
| Specific antiretrovirals of interestc – no. (%) | N/A | N/A | ||
| Ritonavir | 61 (41) | 56 (45) | ||
| Atazanavir | 18 (12) | 12 (10) | ||
| Efavirenz | 56 (37) | 33 (26) | ||
| Nevirapine | 20 (13) | 4 (3) | ||
| Tenofovir | 43 (29) | 74 (59) | ||
| Abacavir or didanosine | 64 (43) | 33 (26) | ||
| Zidovudine or stavudine | 67 (45) | 39 (31) | ||
| Enfurvitide | 0 (0) | 2 (2) | ||
| Raltegravir | 6 (4) | 6 (5) | ||
| Maraviroc | 2 (1) | 0 (0) | ||
| Donor and transplant characteristics | ||||
| Donor age – median [IQR] years | 41 [27–49] | N/A | 40 [25–50] | N/A |
| Male donor sex – no. (%) | 89 (59) | N/A | 72 (58) | N/A |
| Black donor – no. (%) | 36 (24) | N/A | 13 (10) | N/A |
| Deceased donor – no. (%) | 102 (68) | N/A | 123 (98) | N/A |
| Six-antigen-matched kidney – no. (%) | 21 (14) | N/A | N/A | N/A |
| High infectious risk donor – no. (%) | 32 (31) | N/A | 10 (8) | N/A |
| Hepatitis C positive donor – no. (%) | 10 (7) | N/A | 12 (10) | N/A |
| Hepatitis B core antibody positive donor – no. (%) | 17 (11) | N/A | 16 (13) | N/A |
| Combined liver-kidney transplant – no. (%) | N/A | N/A | 9 (7) | N/A |
| Post-transplant characteristics | ||||
| Follow-up posttransplant – median [IQR] years | 4.0 [3.1–5.0] | N/A | 3.5 [1.8–5.0] | N/A |
| Initial calcineurin inhibitor – no. (%) | N/A | N/A | ||
| Cyclosporine | 34 (23) | 42 (34) | ||
| Tacrolimus | 102 (68) | 76 (61) | ||
| None | 14 (9) | 7 (6) | ||
| Initial mycofenolate – no. (%) | 132 (88) | N/A | 89 (71) | N/A |
| With abacavir or didanosine – no. (%) | 63 (42) | 22 (18) | ||
| With zidovudine or stavudine – no. (%) | 52 (35) | 16 (13) | ||
| Any thymoglobulin – no. (%) | 56 (37) | N/A | 2 (2) | N/A |
Table 2
Graft failure and death among HIV-positive transplant recipients and HIV-uninfected registry control groupsa.
| Graft failure | Death | |
|---|---|---|
| Kidney recipients | ||
| HIV-TR recipients (N =150) | 46 (30.7%) | 17 (11.3%) |
| Unmatched controls (N =85 153) | 20 757 (24.4%) | 10 933 (12.8%) |
| Demographic-match controls (N =600) | 174 (29.0%) | 76 (12.7%) |
| Risk-match controlsa (N =600) | 162 (27.0%) | 71 (11.8%) |
| Liver recipients | ||
| HIV-TR recipients (N =124) | 57 (46.0%) | 45 (36.3%) |
| Unmatched controls (N =32 399) | 10 342 (31.9%) | 8475 (26.2%) |
| Demographic-match controls (N =496) | 170 (34.3%) | 133 (26.8%) |
| Risk-match controlsa (N =496) | 161 (32.5%) | 147 (29.6%) |
Candidate–recipient analysis: impact of transplant on survival
In multivariate models, transplantation was associated with a significant survival benefit for liver recipients with MELD at least 15 [hazard ratio (HR) 0.1; 95% confidence interval (CI) 0.05, 0.1; P <0.0001], but not for MELD less than 15 (HR 0.7; 95% CI 0.3, 1.8; P =0.43) or for kidney recipients (HR 0.6; 95% CI 0.3, 1.4; P =0.23) (Tables 3 and and44).
Table 3
Impact of liver transplantation on mortality comparing HIV-infected transplant candidates and recipients.
| Univariate predictor | HR (95% confidence interval) | P value |
|---|---|---|
| Age (by decade) | 1.4 (1.1, 1.8) | 0.02 |
| Sex (female) | 0.8 (0.4, 1.4) | 0.41 |
| Race (white) | 0.7 (0.4, 1.1) | 0.11 |
| Nadir CD4+ T-cell count (per 50 cells/μl) | 1.0 (0.9, 1.1) | 0.96 |
| CD4+ T-cell at enrollment (per 50 cells/μl) | 1.0 (0.9, 1.0) | 0.35 |
| Detectable HIV RNA at enrollment | 2.1 (1.2, 3.8) | 0.01 |
| BMI at enrollment (<21) | 1.5 (0.9, 2.6) | 0.11 |
| MELD at enrollment | 1.1 (1.0, 1.1) | <0.0001 |
| Dual organ transplant | 2.9 (1.4, 6.0) | 0.005 |
| HCV-infected | 1.6 (0.9, 2.9) | 0.08 |
| Most recent MELD pretransplanta | 1.1 (1.1, 1.1) | <0.0001 |
| Most recent CD4+ T-cell pretransplanta (per 50 cells/μl) | 0.9 (0.8, 1.0) | 0.03 |
| Detectable HIV RNA (most recent pretransplant)a | 1.5 (0.6, 3.6) | 0.37 |
| Transplantation with MELD ≥ 15a | 1.0 (0.6, 1.6) | 0.92 |
| Transplantation with MELD <15a | 0.6 (0.2, 1.6) | 0.33 |
|
| ||
| Multivariate predictors | HR (95% confidence interval) | P value |
|
| ||
| Transplantation with MELD ≥ 15a | 0.1 (0.05, 0.1) | <0.0001 |
| Transplantation with MELD <15a | 0.7 (0.3, 1.8) | 0.43 |
| Most recent MELD pretransplanta | 1.1 (1.1, 1.2) | <0.0001 |
| HCV-infected | 3.7 (2.0, 6.9) | <0.0001 |
| Age (by decade) | 1.4 (1.1, 1.9) | 0.02 |
MELD, model for end-stage liver disease.
Table 4
Impact of kidney transplantation on mortality comparing HIV-infected transplant candidates and recipients.
| Univariate predictor | HR (95% confidence interval) | P value |
|---|---|---|
| Age (by decade) | 1.6 (1.2, 2.2) | 0.001 |
| Sex (female) | 0.3 (0.1, 1.2) | 0.09 |
| Race (white) | 1.2 (0.5, 2.5) | 0.69 |
| Nadir CD4+ T-cell count (per 50 cells/μl) | 1.0 (0.9, 1.1) | 0.62 |
| CD4+ T-cell at enrollment (per 50 cells/μl) | 1.0 (1.0, 1.1) | 0.49 |
| HCV infected | 1.8 (0.8, 4.1) | 0.19 |
| BMI at enrollment (<21) | 2.3 (1.0, 5.1) | 0.04 |
| Most recent CD4+ T-cell pretransplanta (per 50 cells/μl) | 1.0 (0.9, 1.1) | 0.67 |
| Transplantationa | 0.7 (0.3, 1.5) | 0.36 |
|
| ||
| Multivariate predictors | HR (95% confidence interval) | P value |
|
| ||
| Transplantationa | 0.6 (0.3, 1.4) | 0.23 |
| Age (by decade) | 1.7 (1.2, 2.3) | 0.001 |
| BMI at enrollment (<21) | 2.6 (1.2, 5.7) | 0.02 |
Predictors of transplant recipient death and graft loss
In the multivariate model, dual organ transplant [HR 3.8 (95% CI 1.6, 8.8), P =0.002], pretransplant BMI less than 21[HR 2.2 (95% CI 1.1, 4.4), P =0.03], higher donor age [HR 1.3 (95% CI 1.1, 1.6) per decade, P =0.01] and HCV-infection [HR 2.1 (95% CI 1.0, 4.6), P =0.06] were associated with increased mortality in liver transplant recipients (Supplementary Table 1, http://links.lww.com/QAD/A808). The same factors appeared in the graft loss model, but with HCV infection being significant (HR 2.4, P =0.02) and BMI being marginally significant (HR 1.9, P =0.06) (Supplementary Table 2, http://links.lww.com/QAD/A808). These HIV-specific factors were not associated with mortality in unadjusted models: nadir, enrollment or pretransplant CD4+ T-cell count or detectable HIV-1 RNA. Liver graft loss was not associated with any factors examined to address potential antiretroviral-immunosuppressant drug interactions, including prior 3-month protease inhibitor or efavirenz use, or use of mycophenolate mofetil (MMF) with abacavir or didanosine.
Thymoglobulin use within first week following transplant [HR 3.5 (95% CI 1.3, 9.1), P =0.01] and higher age [HR 1.7 (95% CI 1.1, 2.6) per decade, P =0.01] were associated with kidney recipient mortality (Supplementary Table 1, http://links.lww.com/QAD/A808). This longer term analysis confirmed previously reported associations [8] of treated rejection (HR 4.0, P =<0.0001) and of initial thymoglobulin use (HR 1.8, P =0.048) with graft loss; MMF use in the prior 3 months was protective [HR 0.5 (95% CI 0.3, 0.99), P =0.046] (Supplementary Table 2, http://links.lww.-com/QAD/A808). None of the previously described CD4+ T-cell count or antiretroviral variables, posttransplant HIV-1 RNA, or tenofovir were associated with kidney transplant recipient mortality or graft loss (data not shown).
Complications and other outcomes in HIV-infected transplant recipients
There were no recurrent opportunistic infections in recipients with a pretransplant history, and opportunistic infection history was not associated with survival. There were four cases of cutaneous Kaposi’s sarcoma, five esophageal and one bronchial candidiasis, two Pneumocystis jirovecii pneumonia (PCP) and one cryptosporidiosis. Four liver recipients who had candidiasis or PCP died due to multisystem organ failure, cerebrovascular accident or recurrent hepatitis C.
There were three cases of recurrent HIVAN. CD4+ T-cell counts at the time of HIVAN diagnosis were 0, 97 and 770 per cubic millimeter. Graft losses occurred at 31, 22 and 14 months after the diagnosis, respectively.
Sixty-nine (55%) liver and 75 (50%) kidney recipients experienced 215 and 197 non-opportunistic infections requiring hospitalization, respectively. Half occurred within 6 months post-transplant. Among cultured liver recipient infections, 81% were bacterial, 7% fungal, 5% viral and 1% protozoal. Infection sites were blood (20%), respiratory tract (13%), and genitourinary tract (13%). Among cultured kidney recipient infections, 70% were bacterial, 9% fungal and 12% viral and sites of infection were genitourinary tract (26%), blood (18%), and respiratory tract (17%). Hepatitis C was associated with increased serious infection risk for both liver (HR 2.8; 95% CI 1.4, 5.4; P-value 0.003) and kidney (HR 2.1; 95% CI 1.2, 3.7; P-value 0.006) recipients. MELD, nonwhite race and enrollment BMI less than 21 were also associated with serious infections for liver recipients; early thymoglobulin use and lower nadir CD4+ T-cell count were associated for kidney recipients (data not shown).
Liver transplantation, HCV infection and higher MELD score were associated with decreased CD4+ T-cell count; higher nadir and enrollment CD4+ T-cell count and time post-enrollment were associated with increased CD4+ T-cell count, suggesting post-transplant recovery of CD4+ T-cells. In the kidney model, transplantation and thymoglobulin were associated with decreased CD4+ T-cell count, whereas higher nadir and enrollment CD4+ T-cell count were associated with increased CD4+ T-cell count. Time posttransplant was not protective (Supplementary Table 3, http://links.lww.com/QAD/A808).
Among 15 liver recipients with detectable HIV RNA at transplant, 13 were undetectable within 3 months, and the others after 5 and 9 months post-transplant. Of 110 liver recipients with undetectable HIV RNA at baseline, 18 (16%) experienced any loss of HIV RNA control. Of 150 kidney recipients, 20 (13%) had any loss of HIV RNA control. One- and 3-year Kaplan–Meier estimates (95% CI) for cumulative incidence of initial loss of HIV RNA control were 12% (7, 20) and 20% (13, 31) in liver recipients and 8% (4, 14) and 16% (11, 24) in kidney recipients. Only three liver and five kidney recipients with initial loss of HIV RNA control failed to subsequently suppress HIV RNA. In both liver and kidney models, being off antiretroviral therapy was significantly associated with increased risk of loss of HIV RNA control. Higher nadir CD4+ T-cell count was marginally protective in the liver model (data not shown). BMI, dual organ transplantation, MELD score, and antiretroviral and immunosuppression regimens were not associated with loss of HIV RNA control. Antiretroviral-related drug toxicity was never indicated as a contributing factor in graft failure or death reports.
HIV-infected recipient versus HIV-negative control analysis: impact of HIV on patient and graft survival
The median time posttransplant for kidney and liver transplant survivors was 4.1 years (IQR 3.1–5.1) and 4.3 years (IQR 2.9–6.0), respectively. All HIV-TR kidney recipients, and 124 of 125 liver recipients, were identified by the SRTR. These and 227 additional HIV-positive kidney and 68 liver recipients were excluded as potential controls. Median follow-up time among the study participants and three control groups were similar: 4.0 to 4.1 years in kidney and 3.9 to 4.0 years in liver. Death and graft failure outcomes are shown in Table 2. In kidney recipients, the unmatched and risk-matched analyses showed a marginally significant hazard ratio for graft loss [HR 1.3 (P =0.07) and HR 1.4 (P =0.052)], but no statistically significant increase in risk was observed for HIV-positive transplant recipients for the other models (Table 5). All the models show that HIV-positive liver recipients had a higher relative hazard (1.5–1.6) of experiencing graft failure than the HIV-negative controls. HIV status was not significant in any of the death censored graft failure models. Although the total event number was small, the HRs were also smaller for this event. The relative hazard of death was not significantly higher among HIV-positive liver transplant recipients in the risk-matched analysis, but was in the other three models (HRs 1.4–1.6). The absolute difference in the proportion of deaths was 6.7% in the risk-matched control analysis.
Table 5
Proportional hazards regression results for HIV status by outcome and match model, stratified by organ.
| Outcome | Model | Kidney
| Liver
| ||
|---|---|---|---|---|---|
| Hazard ratio (95% CI) | P value | Hazard ratio (95% CI) | P value | ||
| Graft failure | Unmatched, adjusted for covariates | 1.311 (0.974–1.765) | 0.07 | 1.518 (1.167–1.974) | 0.002 |
| Demographic-matched unadjusted | 1.081 (0.759–1.540) | 0.67 | 1.601 (1.161–2.209) | 0.004 | |
| Demographic-match adjusted for risk score | 1.115 (0.780–1.593) | 0.55 | 1.607 (1.159–2.227) | 0.004 | |
| Risk-matched | 1.418 (0.997–2.017) | 0.052 | 1.530 (1.106–2.115) | 0.01 | |
| Death | Unmatched, adjusted for covariates | 1.145 (0.700–1.873) | 0.59 | 1.518 (1.129–2.040) | 0.01 |
| Demographic-matched unadjusted | 0.995 (0.574–1.725) | 0.99 | 1.612 (1.120–2.321) | 0.01 | |
| Demographic-match adjusted for risk score | 1.089 (0.617–1.920) | 0.77 | 1.644 (1.134–2.381) | 0.01 | |
| Risk-matched | 1.172 (0.669–2.055) | 0.58 | 1.354 (0.941–1.949) | 0.10 | |
Discussion
In one of the largest prospective cohorts of HIV-infected transplant recipients, transplantation conferred survival benefit in liver recipients with pre-transplant MELD greater than 15, consistent with HIV-uninfected recipient outcomes [13–15]. HIV-positive liver recipients had a higher relative hazard of both graft failure and death in most control analyses; however, the absolute difference in the proportion of deaths was only 6.7% in the risk-matched analysis. HIV-TR liver recipient survival was 63.7%, comparable to 62.2% 5-year survival among 693 liver transplant recipients greater than age 65 transplanted between 1997 and 2000 in the United States and 254 liver transplant recipients with malignant neoplasms, 237 with ‘other liver disease,’ 299 with ‘other’ conditions, and 821 repeat transplant recipients with survival rates of 54.1, 63.6, 64.4 and 54.4%, respectively. [5] Taken all together, the modest increase in risk compared with HIV-negative recipients, as well as in the absolute proportion of those who died, and comparability with other transplant populations, support liver transplant as a viable option in carefully selected and managed recipients [16].
HIV-HBV co-infected liver transplant recipients have been effectively managed with regular use of hepatitis B immune globulin and nucleos(t)ide analogues, with excellent survival [10]. In previous analyses comparing HIV-HCV co-infected and HCV-mono-infected liver recipients, patient and graft survival rates were similar among patients who did not have combined kidney–liver transplantation, a BMI less than 21 kg/m2 [2], or an HCV-positive donor, highlighting the potential for improved outcomes with careful patient selection and management [9]. Strategies to control HCV using interferon-free regimens and direct acting antivirals will revolutionize the treatment of recurrent HCV, the major contributor to the poorer outcome in co-infected participants [9,16,17]. Resources to assist with patient and donor selection, prevention of acute rejection, management of drug–drug interactions, and treatment of recurrent viral hepatitis are increasingly available.
Although we did not detect a survival benefit with kidney transplantation, this may be attributed to the small number of deaths limiting our power to demonstrate a survival difference [18–22]. HIV-positive kidney recipients experienced similar outcomes as their HIV-negative counterparts approximately 5 years post-transplant in all control analyses. There was, however, a marginally significant HR in two of the graft loss analyses which is likely the result of high rates of rejection in HIV-infected kidney recipients [8]. To improve kidney graft survival, better immunosuppressive agents and/or strategies are needed [23,24].
Of particular importance is the very low incidence of de-novo opportunistic infections/neoplasms. Similarly low rates have been reported in other centers [25–30]. The lack of association of history of pre-transplant opportunistic infection and nadir CD4+ cell count with mortality indicates that current immune status and degree of viral control is a more important consideration than HIV clinical history in pre-transplant evaluation. In contrast, non-opportunistic serious infections were an important challenge, occurring in approximately half of patients. We found no association of HIV-related factors with these infections. HCV infection was associated with risk of infection in both liver and kidney recipients. High rates of infectious complications have been reported in HIV-HCV co-infected recipients in several centers [28,30–34]. The disproportionate burden of serious infections within the first 6 months implicates initial immunosuppression, but the mechanisms by which HCV infection increases risk of serious infections, particularly sepsis, are not well understood.
Severity of HIV infection, measured by CD4+ T lymphocyte count nadir and CD4+ cell count at baseline and transplantation, did not affect graft loss or mortality, validating the HIV-related eligibility criteria. These have been adopted by most transplant centers in the United States and Europe. That HIV was well controlled may explain why detectable HIV RNA was not associated with mortality as it was in other reports [35]. The only significant predictor of loss of virologic control was being off antiretroviral therapy (ART).
Post-transplant management of HIV-infected recipients is challenging due to interactions between immunosuppressant and antiretroviral medications. ART regimens employed in this cohort reflect agents most commonly prescribed during the enrollment period, namely protease inhibitor and NNRTI-based regimens, which have inhibitory and inductive effects on P450-CYP3A enzyme systems, respectively, and mixed effects when employed together. Nontherapeutic immunosuppressant drug exposure resulting from drug–drug interactions and the alternate dosing intervals employed to maintain target trough levels has been hypothesized as a cause of the high rejection rates in this cohort [8]. In kidney recipients, cyclosporine use was associated with an increased rate of graft loss [8]. MMF use emerged as protective against graft loss. This finding, along with the lack of an effect of combining MMF with specific NRTIs on HIV control, suggests limited clinical relevance of the in-vitro interactions among these agents [36]. We found no impact of antiretroviral drug class on patient or graft survival, or on organ rejection rates, and no indication that antiretroviral toxicity contributed to graft failure or death. Thus, drug–drug interactions appeared to have been manageable. The relative contribution of antiretroviral–immunosuppressant medication interactions to high rejection rates may be better discernable in the future with increased reliance on integrase inhibitor-based antiretroviral regimens, which lack most of these drug interactions.
The principal limitations of this study are lack of randomization and prospective enrollment of a control group. Multiple subgroups of patients and heterogeneity of both antiretroviral and immunosuppressant treatment regimens limit our ability to detect differential effects of specific therapies or drug classes on outcomes. Limited information on co-morbidities such as hypertension, hyperlipidemia, cardiovascular disease, and bone demineralization, and concomitant medications used to treat these, did not allow us to ascertain confounding effects of other disease processes on outcomes. A limitation of the registry analysis is the potential misclassification bias in the selection of the control population. Twenty-four percent of recipients identified in the SRTR registry reported an unknown HIV status. However, as HIV infection is rare in renal and liver transplant cohorts, the rate of any misclassification and corresponding impact is expected to be very small. A fourth limitation is the relatively small cohort of HIV-positive transplant recipients evaluated. With relatively small numbers of events and a large number of factors evaluated in regression models of registry controls, over-fitting is a concern. However, the consistency of the results with the various models and outcomes used demonstrate the robustness of our conclusions. Nevertheless, outcomes should continue to be evaluated in the HIV-positive recipients as transplant becomes more common in this population and new treatments evolve.
In conclusion, these analyses support kidney and liver transplantation as an option for carefully selected people with HIV infection. Refinement of post-transplant management in the era of integrase inhibitor-based ART and direct acting agents for pre-, peri-, and post-transplant treatment of HCV co-infection will be necessary. It is likely that effective HCV therapy will lead to liver transplantation outcomes that are similar to those in HCV-mono-infected patients, as effective HBV treatment has resulted in similar outcomes for HIV-HBV co-infected and HBV mono-infected recipients.
Acknowledgments
This work was supported by the Solid Organ Transplantation in HIV: Multi-Site Study (AI052748) funded by the National Institute of Allergy and Infectious Diseases. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Allergy and Infectious Diseases or the National Institutes of Health.
We would like to thank the following investigators from the HIV-TR Study Team for their contributions to the final manuscript: Laurie Carlson, Norah A. Terrault, Rodney L. Rogers (University of California, San Francisco); Thomas D. Schiano (Icahn School of Medicine at Mount Sinai); Michael Wong (Beth Israel Deaconess Medical Center); Robert Redfield, Charles Davis (University of Maryland); Michael Yin, Lorna Dove (Columbia University); Shmuel Shoham, Aruna Subramanian (Johns Hopkins University); Jimmy A. Light (Washington Hospital Center); Margaret Ragni (University of Pittsburgh); Joseph Timpone (Georgetown University); Valentina Stosor (Northwestern University); Dushyantha T. Jayaweera (University of Miami); Fred Poordad, Nicholas Nissen (UCLA and Cedars Sinai Medical Center); Brian Wispelway (University of Virginia); Kenneth Pursell (University of Chicago); David Mushatt (Tulane University); Thomas C. Pearson (Emory University); Judith Feinberg, Kenneth E. Sherman (University of Cincinnati); Alan J. Taege (Cleveland Clinic), Jeffrey M. Jacobson (Drexel University); Brigid Betz-Stablein, Don Stablein (EMMES Corporation), Lawrence Fox, Jonah Odim (National Institute of Allergy and Infectious Diseases).
We especially thank all of the Solid Organ Transplantation in HIV: Multi-Site Study investigators and coordinators (HIV-TR Study Team) for their hard work and dedication to the study and participants.
Funding/Support: This work was supported by the Solid Organ Transplantation in HIV: Multi-Site Study (AI052748) funded by the National Institute of Allergy and Infectious Diseases. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Allergy and Infectious Diseases or the National Institutes of Health.
Footnotes
Conflicts of interest
The authors of this manuscript have no conflicts of interest to disclose.
Authors’ contributions: Conceived, designed and drafted the article: M.R., B.B., P.S., G.B., D.H.; clinical care of the participants and acquisition of data – M.R., S.H., B.M., D.H., E.B., K.O., D.S., W.H., G.B., P.S.; study design and clinical execution – all; data analysis – B.B.; data interpretation – all; approval of the final version of the manuscript – all.
Institutional review board approval: The Committee on Human Research at the University of California, San Francisco approved the study protocol, as did the Human Subjects Review Board from each center. Each participant provided written informed consent.
Data analysis: B.B. and Dr Stablein have full access to all of data in the study and take responsibility for the integrity and accuracy of the data and analysis. The SRTR data reported were supplied by the Minneapolis Medical Research Foundation (MMRF) as the contractor for the SRTR. The interpretation and reporting of these data are the responsibility of the authors and are not an official policy of or interpretation by the SRTR or the US Government.