Diarrhea after kidney transplantation: A study of risk factors and outcomesAD Patil, NG Saxena, SB Thakare, AE Pajai, D Bajpai, TE Jamale
Department of Nephrology, Seth GS Medical College and KEM Hospital, Mumbai, Maharashtra, India
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/jpgm.jpgm_601_22
Source of Support: None, Conflict of Interest: None
Keywords: Chronic allograft injury, diarrhea, MMF withdrawal, kidney transplant, tacrolimus toxicity
Diarrhea is a frequent complication after kidney transplantation. It can be associated with a significant decrease in quality of life and if chronic, can lead to several complications like malabsorption, rehospitalization, noncompliance with immunosuppression, and a greater risk of graft loss and death. A comprehensive evaluation is needed for the timely identification of enteric pathogens for eradication with appropriate therapy. Episodes of diarrhea are associated with acute allograft dysfunction due to the ensuing hypovolemia along with an increase in the trough level of tacrolimus at the time of the episode. Reduction or withdrawal of mycophenolate mofetil (MMF) which may be required in cases of chronic diarrhea has been shown to be a risk for rejection and graft loss.,, There is limited data regarding episode-wise analysis of diarrhea in kidney allograft transplant recipients, its etiologies, and its impact on allograft function. We evaluated these clinical outcomes in a cohort of kidney allograft recipients at a tertiary care center.
Study design and data collection
This retrospective cohort study was conducted at a tertiary care referral center in Western India over 1 year from July 2019 to July 2020. One-hundred and ninety-eight kidney allograft recipients, at regular follow-ups in the transplant outpatient clinic, were screened. All recipients with a history of diarrhea and with at least 3 months of follow-up post diarrheal episodes were included in the analysis [Figure 1]. The study was approved by our Institutional ethics committee (EC/OA-19/2020) and written and informed consent was obtained from the participants. Data on demographics, comorbidities, pretransplant dialysis vintage, native kidney disease, donor characteristics, induction, and maintenance immunosuppression, and baseline graft function (eGFR) were recorded from outpatient and inpatient records along with individual case proformas maintained in the clinic. Details of evaluation, time of occurrence post-transplant, etiologies, clinical features, diagnostic modalities, associated acute and chronic allograft dysfunction, reduction of immunosuppression, and withdrawal of MMF were recorded. Diarrhea was categorized into acute (<2 weeks), persistent (≥2 weeks but <4 weeks), and chronic diarrhea (≥ 4 weeks) as per the American College of Gastroenterology (ACG) criteria. Etiologies of the diarrheal episodes were categorized as infectious based on clinical and laboratory criteria: fever, abdominal pain, vomiting, presence of blood or mucous in stool, and/or investigations such as the presence of leucocytosis/leucopenia, detection of pus or blood in stool, stool examination for opportunistic pathogens [conventional and/or multiplex polymerase chain reaction (PCR)] or endoscopic sampling and presence of cytomegalovirus (CMV) infection/disease. Diagnosis of rejection was made based on the Banff 2017 diagnostic criteria on allograft biopsy. Standard evaluation with complete blood counts, kidney function tests, and liver function tests which were done during diarrheal episodes was recorded. All episodes with acute allograft dysfunction were evaluated with our unit-based protocol [Figure 2]. which included extensive evaluation for both infectious and non-infectious etiologies of diarrhea followed by appropriate treatment based on these results. Only after detailed evaluation and failure to respond to prolonged therapy with antiparasitic agents, diarrhea was attributed to MMF. Other potential non-immunosuppressive drugs known to aggravate diarrhea such as metformin, ferrous sulfate, statins, proton pump inhibitors, and bisphosphonates were reduced or discontinued before stopping MMF. In addition, we also asked our patients to withhold milk and milk products during diarrheal episodes considering the possibility of acquired lactose intolerance. In all recipients in whom diarrhea was suspected to be due to MMF, a switch to enteric-coated mycophenolate sodium (EC-MPS) was made. If diarrhea persisted, MMF was given in multiple divided doses, which was subsequently followed by dose reduction if no response. If the above measures failed, MMF was discontinued.
Outcomes and definitions
The outcomes of our study were chronic allograft injury (CAI) and MMF withdrawal. In addition, the impact of MMF withdrawal on the development of CAI was studied. Acute allograft dysfunction during the episode was defined as an increase in creatinine by 25% from the baseline and persisting for <3 months. Chronic allograft injury was defined as creatinine ≥1.5 mg/dl for ≥3 months and in recipients with baseline abnormal eGFR, we defined it as an increase in creatinine by 25% from baseline persisting for ≥3 months. Graft failure was defined as the return of the patient to hemodialysis for ≥3 months post-transplant. Socioeconomic status was classified using the Modified Kuppuswamy Scale. A marginal donor was defined as a living donor with an age ≥60 years and/or GFR ≤60 ml/min/1.73 m2 [Diethylenetriamine pentaacetate (DTPA) by double plasma sampling). Baseline allograft function at the onset of diarrhea was assessed by eGFR using CKD-EPI (chronic kidney disease epidemiology collaboration) formula. Cryptosporidiosis was defined as any diarrheal episode with isolation of cryptosporidium either by stool examination and/or D2 biopsy (second part of duodenum) by upper gastrointestinal (GI) endoscopy. We defined severe diarrhea as per the definition used by the DIarrhea Diagnosis Aid and Clinical Treatment (DIDACT) study as ≥3 stools/day or ≥7 consecutive days accompanied by weight loss ≥2 kg and/or fever and/or mucous and/or blood in the stool and/or hospitalization. Tacrolimus toxicity was defined as either supratherapeutic whole blood trough tacrolimus levels with acute allograft dysfunction or biopsy-proven changes of acute or chronic tacrolimus toxicity. Immediate response to treatment was defined as improvement in symptoms within 48 hours of starting antibiotics.
We expressed categorical variables as numbers with percentages. Mean with standard deviation was calculated for continuous variables which were normally distributed and median with interquartile range (IQR) for those which were not. Recipients were divided into two groups, those with chronic allograft injury vs no chronic allograft injury and MMF withdrawal vs no MMF withdrawal. For the comparison of categorical variables between these groups (patient-wise), we used Fisher's exact test when observations were less than 5, and Pearson's Chi-square test when they were not. Independent samples t-test was used for continuous variables that were normally distributed and the Mann–Whitney test for those not normally distributed. The strength of association was measured using odds ratios (OR) with 95% confidence intervals (CI). Predictors of the outcomes of MMF withdrawal and allograft injury were analyzed by multivariate binary logistic regression analysis. We assessed event-free survival analysis for assessing the difference in the risk of CAI in two groups, one with and the other without MMF withdrawal. Kaplan-Meier survival curves were generated for comparing time to event (CAI) in these two groups and this comparison was done using the log-rank test. The significance level was fixed at P < 0.05 and all tests were two-tailed.
140 diarrheal episodes were analyzed in 85/198 (42.9%) kidney allograft recipients. The median (IQR) time to onset of diarrhea was 25 (7–60) months post-transplant [Table 1] and [Table 2]. Seventy-two (84.7%) were males and the mean age was 38 ± 12 years. Fifty of these recipients (58.8%) belonged to a lower socioeconomic class and had a median (IQR) dialysis vintage of 12 months (6–24) before transplant. 82 (96.5%) received kidneys from a living donor and 3 (3.5%) from a deceased donor. Sixty-nine (81%) living donors were related and had a first-degree relationship with their respective recipients whereas the remaining were from unrelated donors. Forty-two (49.5%) received antibody induction: anti-thymocyte globulin in 39 (45.9%), basiliximab in 2 (2.4%), and daclizumab in 1 (1.2%). The most common immunosuppression regimen at the time of transplant was tacrolimus (TAC), MMF, and steroid in 80 (94.1%). Hypertension (HTN) and new-onset diabetes after transplant (NODAT) were the most common comorbidities seen in 33 (38.9%) and 32 (37.6%) respectively with a median (IQR) duration of NODAT being 5 years (2–7).
Seventy-three (86%) recipients were on MMF at the onset of diarrhea, and 10 were on AZA, and 2 on CNI plus a steroid.
Clinical features of diarrheal episodes
Out of the 140 diarrheal episodes, 106 (75.7%) occurred 6 or more months post-transplant whereas 34 (24.3%) episodes occurred less than 6 months post-transplant [Table 3]. Hospitalization was required in 67/140 (47.8%) episodes, and 33/140 (23.6%) episodes were associated with hemodynamic instability. Eighty-three episodes (59.2%) were categorized as severe diarrhea.
Ninety (64.3%) episodes were categorized as infectious episodes based on clinical presentation and investigations [Table 3]. Microbiological diagnosis by stool examination or endoscopic sampling (either a single or more than one pathogen) was obtained in 28/90 (31%) episodes.
Of these, patients the diagnosis was established by a stool multiplex PCR in 16 of 28 (57%), conventional stool assays in 10/28 (35.7%), and endoscopy with D2 (duodenal) biopsy in 2/28 (7.1%).
The most common pathogens were parasites in 22/28 (78.6%) with giardia and cryptosporidium seen in 11/28 (39%) and 6/28 (21.4%) episodes, respectively. Entamoeba histolytica was seen in 4/28 (14.3%) episodes and hymenolepis nana in 1 episode (3.5%). Among bacterial causes, escherichia coli were detected in 10/28 (35.7%) episodes. Eight out of 28 (28.5%) episodes had a viral etiology with norovirus being most common in 5/28 (17.8%) episodes. Fifty (35.7%) episodes were categorized as non-infectious. Upper GI endoscopy and colonoscopy were required for diagnosis in 15 (10.7%) and 7 (5%) episodes respectively [Supplementary Table 1][Additional file 1].
The most commonly used empirical/specific antibiotic therapy was nitazoxanide in 125 (89.2%) episodes followed by metronidazole 15 (10.7%), azithromycin 9 (6.4%), ciprofloxacin 8 (5.7%), doxycycline 9 (6.4%) and rifaximin in 0.7% (1). Intravenous antibiotics were required in 10 (7.1%) episodes. The median (IQR) duration of treatment was 7 days (7–14). Most episodes [125 (89.2%)] immediately responded to treatment. Non-immunosuppressive medications known to aggravate diarrhea such as metformin, ferrous sulfate, statins, proton pump inhibitors, and bisphosphonates were either reduced or stopped in 67 (47.9%) episodes.
Seventy-eight recipients (91.7%) had acute allograft dysfunction during diarrhea and 20 (25.6%) of these did not recover to baseline after the resolution of diarrhea. An allograft biopsy was done in 13/20 (65%) of these recipients in whom the acute allograft dysfunction did not recover to baseline. [Supplementary Table 1]. 12 (14.1%) developed rejection.
Chronic allograft injury
Forty-one recipients (48.2%) developed chronic allograft injury out of which two developed graft failure. CAI was more likely with a later onset of diarrhea (36 months vs 20 months, P = 0.01), with cryptosporidiosis (12.9% vs 0, OR = 1.171, 95% CI (1.032–1.330, P = 0.021) and in recipients who received kidneys from the marginal donors (19.5% vs 4.5%, P = 0.032) [Table 4]. Recipients with CAI also had a lower baseline eGFR at the onset of diarrhea although it was not statistically significant (59.58 vs 66.13, P = 0.127).
Seventy-three (86%) recipients were on MMF at the onset of diarrhea and had 128 diarrheal episodes. A dose reduction of MMF was required in 15/73 (20.5%) recipients. Thirty-five out of 73 (48%) recipients required withdrawal of MMF and subsequent change to AZA (Azathioprine). Comparison of variables between recipients requiring MMF withdrawal vs no MMF withdrawal as summarized in [Table 5]. Baseline mean eGFR at the onset of diarrhea and, the time to onset of diarrhea post-transplant was similar between both groups. Eighteen out of the 35 recipients (51.4%) that required MMF withdrawal developed CAI. Log-rank test [Figure 3] showed a higher probability of CAI in recipients requiring MMF withdrawal (P = 0.026) with a hazard ratio (HR) of 0.46 (0.22–0.94).
24/78 (30.7%) required tacrolimus dose reduction due to supratherapeutic whole blood trough levels and acute allograft dysfunction.
Multivariate binary logistic regression [Supplementary Table 2][Additional file 2] showed that chronic diarrhea (P = 0.001) was an independent predictor of MMF withdrawal.
Our results show that diarrhea is a frequent complication in kidney transplant recipients with a high risk of allograft injury. A significant number of these cases need MMF withdrawal for chronic diarrhea. MMF withdrawal increases the probability of chronic allograft injury.
The prevalence of post-transplant diarrhea was 42.9%, which is significantly higher than that reported by the United States Renal Data System (USRDS) registry (22%). The prevalence reported from various Indian studies were 8.98%, 10%, and 47%. A higher prevalence can be related to greater environmental exposure to opportunistic GI pathogens in our population (a large majority of them being from a lower socioeconomic class) than in the West. Another possibility is the higher risk of MMF-induced diarrhea due to a lower body mass index in the Indian population. The former may also explain the higher number of episodes being due to an infectious etiology in our study (64.3%) in comparison to that reported from the USRDS data (16.7%). The higher prevalence of infections cause (compared to the reported Indian studies) may have to do with the detailed protocol-based evaluation of each episode and the number of infectious episodes was similar to another Indian study which reported more than half of episodes being due to infectious causes.
We had a higher rate of detection of pathogens by stool examination (31%) with the most common etiology being parasitic (78.6%) followed by bacteria (35.7%) and viruses (28.5%) in comparison to the findings of the DIDACT study in which only 20.3% of recipients had a microbiological diagnosis with a higher prevalence of bacterial infections (15.7%) whereas only 3.6% were parasitic infections and 0.9% viral infections. It was also higher than that reported by Singh et al. (opportunistic parasites in 13%). The most common parasite was giardia (39%) followed by cryptosporidium (21.4%) which was similar to that reported in another study from India (cryptosporidium was the cause in 28.9%). More than one-third (35.7%) of these diagnoses were from conventional stool assays as compared to the study by Coste et al. who reported only 23% of them had a microbiological diagnosis using conventional stool examinations. Molecular methods such as stool multiplex PCR was required for diagnosis in 57%. This was higher than that reported by Sonambekar et al. (biofire PCR assay in 24% of episodes). But it was only slightly lesser as compared to a French study where 72% of diagnoses were done using multiplex PCR assays. Higher microbiological yield in our cohort is likely related to a unit-based protocolized evaluation: repeated stool examinations (minimum 3), use of modified ZN (Ziehl-Neelsen) stain, use of multiplex PCR, and lower threshold for endoscopic examination with sampling. This was carried out by a multidisciplinary team involving transplant physicians, gastroenterologists, and microbiologists. The incidence of CMV disease and infection (4.4%) was less than that reported in the DIDACT study (7.4%) and this may be related to the lower risk of CMV in our recipients who were predominantly the recipients of live-related donor transplants where antibody induction is used infrequently.
We categorized 35.7% of episodes as noninfectious which were lesser compared to the 80% episodes reported by the DIDACT study but similar to the studies from India (40% due to MMF) and 17% due to immunosuppressive medications.
Majority of the recipients in our study had acute allograft dysfunction (91.7%) and 74.4% of them recovered to baseline. Volume depletion and tacrolimus toxicity during the episodes could be the reasons for the higher rates of acute allograft dysfunction in our study. Diminished metabolism of tacrolimus by the enzyme Cytochrome P450 3A4 (CYP3A4) in the intestinal mucosa due to diarrhea leads to an increase in tacrolimus blood levels during the episode which contributes to acute allograft dysfunction due to acute vasoconstriction of the afferent arterioles as well as chronic allograft dysfunction as a result of increased production of stimulators of extracellular matrix production such as transforming growth factor (TGF)-beta, osteopontin (a potent macrophage chemoattractant) and chemokines. In 30.7% of the diarrheal episodes that were associated with acute allograft dysfunction, evidence of tacrolimus toxicity was observed. This is similar to the previous report by Bhadauria et al. (33.3%).
Nearly half of our recipients (48.2%) had CAI (chronic allograft injury)-the higher prevalence of which could be explained by the use of a sensitive definition of CAI which is not yet reported in the available literature. Recipients with CAI were more likely to have a later onset of diarrhea which might have to do with a decline in baseline graft function over a period of time due to multiple factors.
Reduction in immunosuppression included a reduction of MMF (20.5%), reduction of MMF and tacrolimus (25.3%), and reduction of tacrolimus only (30.7%). This was lower than that reported from the DIDACT study in which 60% of recipients required changes in immunosuppression. In our cohort, dose reduction of both MMF and tacrolimus was required in about one-fourth of the episodes. The available literature suggests that regimens based on MMF and tacrolimus have been associated with a greater incidence of infectious and unspecified non-infectious diarrhea but an exact prevalence is not reported yet. Recipients who required withdrawal of MMF and subsequent change to AZA (48%) were higher than that reported by the DIDACT study in which MMF reduction or stoppage was done in 31.4% of recipients, reflecting lower tolerability to MMF in our population. The probability of CAI was significantly higher in recipients requiring MMF withdrawal which is consistent with the USRDS data which reported that reduction or withdrawal of MMF is a risk factor for rejection and graft loss,,, and the risk of graft failure increases with MMF doses reduction ≥50%.
Categorization of diarrhea in kidney transplant recipients based on duration (acute, persistent, and chronic) is not available to date. From our analysis, chronic diarrhea significantly predicted MMF withdrawal. This emphasizes the need for timely diagnosis, management of infectious episodes with empirical or specific antibiotics in appropriate dosages and duration, and correction of hypovolemia.
As a part of an evaluation of each diarrheal episode, we subjected recipients to endoscopic evaluation which included both upper gastrointestinal endoscopy and colonoscopy in 15.7% of episodes showed varied findings of gastric antral vascular ectasias, pathogens such as giardia and cryptosporidium on D2 biopsy, gastric erosions, non-erosive colonic lesions, and colitis with cryptic distortion. This was similar to that of the DIDACT study in which 17.5% of recipients had to undergo a colonoscopy for diagnosis of diarrhea revealing macroscopic changes such as erosions, ulcerations, and microscopic findings of villous atrophy.
Our study has several strengths. First, to our knowledge, this is the largest data from South Asia on the evaluation and outcomes of diarrhea post-kidney transplant where a protocol-based evaluation of each episode was undertaken by a multidisciplinary team. Second, detailed evaluation including multiplex PCR led to higher microbiological yield, which can guide the therapy. Third, a detailed patient assessment for each diarrheal episode, their clinical features, etiologies, acute allograft dysfunction, its recovery to baseline, and careful assessment of pharmacokinetics of tacrolimus during the episodes could be obtained from a dedicated transplant clinic for analysis.
Limitations of our study include its retrospective nature and the lack of a control group, results of Chi-square test not supporting the association of MMF withdrawal with CAI. and potential unmeasured confounders and underlying biases as this is an observational study.
To conclude, diarrhea post-kidney transplant is associated with significant morbidity, MMF withdrawal was necessitated in 35/73 (48%) and CAI was observed in 41 of 85, i.e., 48.2% of the recipients. A thorough and systematic evaluation of each episode can help identify various infectious etiologies in a significant number of recipients before dose reduction or withdrawal of MMF as this is a risk factor for chronic allograft injury.
We acknowledge the departments of Microbiology and Gastroenterology, Seth GS Medical College, Mumbai for the help in the evaluation of patients. We acknowledge the help of Mr. Kishor Vardam in data collection for this study.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]