The impact of preexisting and posttransplant diabetes mellitus on outcomes following liver transplantation.

BACKGROUND
Diabetes mellitus (DM) is said to adversely affect transplant outcomes. The aim of this study was to investigate the impact of preexisting and new-onset DM on liver transplant (LT) recipients.


METHOD
A single-center retrospective analysis of prospectively collected data of LT recipients (1990-2015) was undertaken.


RESULTS
Of the 2,209 patients, 13% (n=298) had Pre-DM, 16% (n=362) developed PTDM, 5% (n=118) developed transient hyperglycemia (t-HG) post-LT, and 65% (n=1,431) never developed DM (no DM). Baseline clinical characteristics of patients with PTDM was similar to that of patients with pre-DM. Incidence of PTDM peaked during first-year (87%) and plateaued thereafter. On multivariate analysis (Bonferroni-corrected), nonalcoholic fatty liver disease and the use of Tacrolimus and Sirolimus use were independently associated with PTDM development. Both Pre-DM and PTDM patients had satisfactory and comparable glycaemic control throughout the follow-up period. Those who developed t-HG seems to have a unique characteristic compared to others. Overall, 9%, 5%, and 8% developed end-stage renal disease (ESRD), major cardiovascular event (mCVE), and de novo cancer, respectively. Both Pre-DM and PTDM did not adversely affect patient survival, retransplantation, or de novo cancer. The risks of ESRD and mCVE were significantly higher in patients with Pre-DM followed by P|TDM and no DM.


CONCLUSIONS
In this largest nonregistry study, patients with pre-DM and PTDM share similar baseline clinical characteristics. Pre-DM increases the risk of ESRD and mCVE; however, patient survival was comparable to those with PTDM and without diabetes. Understanding the impact of PTDM would need prolonged follow-up.


INTRODUCTION
Diabetes mellitus (DM) poses an increasing challenge for patients and healthcare systems worldwide. 1 It leads to significant morbidity and mortality through several macrovascular and microvascular complications. 2,3 Furthermore, DM is also associated with an increased risk of several cancers. 4 DM is an important risk factor for the development of nonalcoholic fatty liver disease (NAFLD), which is a leading cause of cirrhosis, liver cancer, and liver transplantation (LT). Furthermore, the presence of DM increases the risk of death in patients with liver disease. 5 LT remains the only treatment option that improves the dismal prognosis of end-stage liver disease. Advances in perioperative management and immunosuppressive treatment has led to excellent long-term outcomes with 1-, 5-, and 10-year post-LT survival rates of >90%, 85%, and 70%, respectively. [6][7][8] However, post-transplant immunosuppression remains associated with a number of side effects including new-onset DM known as post-transplant diabetes mellitus (PTDM). PTDM was first described over 50 years ago 9 but has only more recently gained recognition as a major complication with the potential for serious consequences. 10 The reported incidence of PTDM after all solid organ transplantation ranges from 2% to 53%. 10,11 PTDM has been reported in 7% to 28% of LT recipients LT. 12 Although the pathophysiology of PTDM is not fully understood, pancreatic β-cell dysfunction, impaired insulin secretion (associated with the use calcineurin inhibitors and mammalian target of rapamycin complex 1 inhibitors), postoperative weight gain, and hepatitis C virus infection (associated with insulin resistance) are all thought to play a role. 13,14 The impact of PTDM and preexisting DM on outcomes following LT is limited and equivocal. While some studies demonstrate a negative impact of PTDM on patient and graft survival, [15][16][17][18][19][20] others were not able to confirm this. [21][22][23] Furthermore, these studies have usually not compared the impact of PTDM to that of pre-existing DM, 15,17,18,22 and many had limited follow-up duration potentially leading to underestimating of a potential impact. 16,18,21,22 In addition, most of these studies have been reported from East Asia 15,17,18,21,23 and therefore may not represent the experience in the West. The aim of the present study is to assess the impact of preexisting DM and PTDM on longterm outcomes of LT recipients in the Toronto transplant program, one of the largest in North America.

Patient Selection and Data Collection
This was a retrospective analysis of prospectively collected data from a single, high-volume liver transplant center. All adult patients who received a liver transplant in the Toronto liver transplant program for any indication between January 1990 and December 2015 were eligible for inclusion. The following patients were excluded from the study: (a) patients who underwent combined solid organ transplantation and (b) patients who were transplanted in other centers and subsequently followed up in the Toronto program.
Demographic and clinical data were extracted from electronic transplant database (OTTR: Transplant Care Platform 6, OTTR Chronic Care Solutions, Omaha, NE) and supplemented with manual chart review.
For the purposes of this study, patients were categorized into 4 groups based on diabetic status: those without DM before and after transplantation (no DM); those with pretransplant DM (Pre-DM); those who developed PTDM; and those who developed transient hyperglycemia (t-HG). Main analysis included no DM, Pre-DM, and PTDM groups. The t-HG group was compared with the other 3 groups separately but was not included in the main analysis.
The study was approved by the Toronto University Health Network Research Ethics Board (REB 16-5023).

Diagnosis of PTDM
PTDM was diagnosed based on the 2003 International Consensus Guidelines. 24 LT recipients who were not diabetic before transplantation but developed symptoms of DM with an elevated random plasma glucose (≥11.1 mmol/L) or an elevated fasting plasma glucose (≥7.0 mmol/L) or an elevated 2-hour plasma glucose (≥11.1 mmol/L) during an oral glucose tolerance test were diagnosed to have PTDM. In addition, those who were initiated on hypoglycemic agents for more than 14 continuous days post-transplant were also considered to have PTDM. Glucose levels are monitored daily (pre and postmeal) during the immediate post-transplant period. Following discharge, random glucose level is regularly checked in LT recipients at their outpatient clinic appointments and those with elevated glucose levels are referred to their family physicians for further investigations.
Those patients who were diagnosed with DM range hyperglycemia based on the above criteria but became euglycemic without the need for hypoglycemic agents within 6 months of LT were categorized as t-HG.

Immunosuppression Regimen
All patients received intravenous methylprednisolone followed by an oral course of prednisone and tacrolimus immediately following LT; basiliximab (before that thymoglobulin) induction was used to delay introduction of calcineurin inhibitors in some cases (eg, patients with renal failure). Oral prednisone dose followed a standard tapering regimen and was stopped at 3 months after LT. Tacrolimus was substituted with cyclosporine in case of tacrolimus intolerance or neurotoxicity. Mycophenolate mofetil was added if calcineurin inhibitor (tacrolimus or cyclosporine) dose reduction was required. Long-term dual immunosuppressant regimen (calcineurin inhibitor or sirolimus and mycophenolate mofetil) was the standard of care for patients transplanted for autoimmune liver diseases, acute liver failure, and inborn metabolic disorders. Long-term low-dose prednisone was part of the immunosuppressant regimen for patients with a history of recurrent acute cellular rejection or chronic rejection. Furthermore, the target trough levels of cyclosporine, tacrolimus, and sirolimus remained unchanged over the study period. However, the use of immunosuppressants evolved during the 25-year study period. In brief, these include introduction of Tacrolimus as the primary immunosuppressant in 2001/2002, before which cyclosporine remained the primary immunosuppressant and the use of sirolimus since 2005 as the primary immunosuppressant in those who were intolerant to calcineurin inhibitors.
For the purposes of this study, only those patients who were taking an immunosuppressant continuously for a minimum of 6 months were considered to have a positive exposure to that immunosuppressant.

Endpoints
Patient survival was the primary endpoint. Patient survival was defined as the time from LT to death from any cause. Patients who were lost to follow-up, or moved to another center for ongoing care, were censored at the time of their last clinic visit. In the event of more than one LT, cases were censored at the time of the second LT.
Re-LT and development of end-stage renal disease (ESRD), major cardiovascular events (mCVEs), and de novo cancer were secondary endpoints. ESRD was defined as GFR <15 mL/min/1.73 m 2 for 6 months or more, requirement of renal replacement therapy, or renal transplantation. Myocardial infarction and stroke were considered as mCVE.

Data Analysis
Continuous variables were presented as median and interquartile range (IQR). Categorical variables were presented as number and percentage. Normality of distributions was assessed using histograms and confirmed using Shapiro-Wilk test. Kruskal-Wallis, Mann-Whitney, and χ 2 test were used as appropriate to compare basic demographics of the 4 study groups. A P value of <0.05 was considered significant.
Predictors of PTDM were investigated using univariate and multivariate analyses. Univariate analysis was performed using Mann-Whitney test and χ 2 for continuous and categorical variables, respectively. Variables with a P < 0.10 were included in the multivariate logistic regression model. Variables were considered to have independent association only if the P value reached Bonferronicorrected level of significance.
Patient survival analysis for all-cause mortality (primary outcome) was performed using the Kaplan-Meier method. The probability of re-LT and developing ESRD, mCVE, and de novo cancer were investigated using a competing risk model. Death was considered as a competing event and the cumulative incidence of re-LT, ESRD, mCVE, and de novo cancer were calculated using Gray's test.
Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 20.0 (Armonk, NY) for univariate and multivariate prediction of PTDM, GraphPad prism 5 (San Diego, CA) for Kaplan-Meier analysis, and visualization and R statistical software, version 3.4.1 (http://cran.r-project.org/) for competing risk analysis. The R-package "cmprsk" version 2.2-7 was used to compare the cumulative incidence and conduct a regression analysis of competing risks.

Demographics and Clinical Characteristics
The study included 2209 patients, who underwent LT between January 1990 and December 2015 and fulfilled the eligibility criteria. The median age at LT was 54 years (IQR 47 to 60) and 67% (n = 1480) were male patients ( Table 1). The median duration of post-LT follow-up was 6.7 years (IQR 2.9 to 11.3).   In comparison to the no DM group, the following parameters were associated with the development of PTDM (Table 2): on univariate analysis, age, male sex, body mass index (BMI), NAFLD as the cause of liver disease, pre-LT dyslipidemia, cold ischemia time, Roux-en-Y biliary anastomosis, and the use of cyclosporin, sirolimus, and prednisone for immunosuppression. In addition to the above parameters, pre-LT hypertension (P = 0.07) and the use of tacrolimus (P = 0.06) and mycophenolate (P = 0.07) were included in the multivariate analysis (total of 13 parameters); the Bonferroni-corrected level of significance in the multivariate analysis was P < 0.0038. On multivariate (Bonferroni-corrected) analysis, NAFLD as the cause of liver disease (odds ratio [OR] 2.03, 95% CI 1.31-3.16; P = 0.002) and the use of tacrolimus (OR 2.76, 95% CI 1.78-4.30; P < 0.001) and sirolimus (OR 2.72, 95% CI 1.67-4.42, P < 0.001) as immunosuppression were independently associated with the development of PTDM.
The clinical characteristics of PTDM patients at LT were similar to Pre-DM patients but were different to that of no DM patients. Compared with the no DM group, PTDM patients were older (P < 0.0001), predominantly male (P = 0.001), with higher BMI (P < 0.0001), and were likely to have NAFLD as the cause of liver disease (P < 0.0001). In contrast, PTDM patients had marked similarities with Pre-DM patients at LT-age (P = 0.40), sex (P = 0.09), BMI (P = 0.99), and NAFLD etiology (P = 0.54). Interestingly, hepatitis C etiology did not differ between PTDM and the no DM (P = 0.21) or Pre-DM (P = 0.23) groups.

Primary Outcome (Patient Survival)
Over a third (n = 782, 35%) of the total study population died following LT. The 1-, 5-, 10-, and 20-year patient survival rates were 90%, 79%, 64%, and 23% in the Pre-DM group, 97%, 88%, 74%, and 22% in the PTDM group, and 90%, 79%, 68%, and 34% in the no-DM group, respectively. Overall, there was no difference in Only those who were not diabetic at the time of transplantation (ie, not including those with pretransplantation diabetes) were included in this analysis. Parameters with a P < 0.10 on univariate analysis were included in the multivariate analysis and these parameters are indicated by an asterisk (*). The Bonferroni-corrected level of significance in this analysis was P < 0.0038, indicated in bold. BMI, body mass index; ΔBMI, change in BMI; CI, confidence interval; HCV, hepatitis C; LT, liver transplantation; NAFLD, nonalcoholic fatty liver disease.
patient survival between the 3 study groups (P = 0.30, Figure 2). A summary of the major causes of death are shown in Table S1 (SDC, http://links.lww.com/TP/B734).

Transient Hyperglycemia
One hundred and eighteen patients fulfilled the diagnostic criteria of t-HG. All achieved a state of euglycemia without the need for antidiabetic medications within 6 months of LT.
t-HG patients were unique in their baseline clinical characteristics. The age of t-HG patients at LT was no different from that of patients with Pre-DM (P = 0.06), PTDM (P = 0.25), and no DM (P = 0.11). However, BMI of t-HG patients at LT was significantly higher than that of no DM patients (P = 0.03) but was similar to that of Pre-DM (P = 0.50) and PTDM (P = 0.54) patients. Similarly, t-HG patients were predominantly males compared with no DM patients (P = 0.01) but were not different from Pre-DM (P = 0.47) or PTDM (P = 0.60) patients. Yet, the prevalence of NAFLD being the underlying etiology of liver disease was similar between t-HG and no DM (P = 0.36) patients but was significantly lower compared with that of Pre-DM (P = 0.0006) and PTDM (P = 0.002) patients. More interestingly, the prevalence of hepatitis C being the underlying etiology of liver disease was significantly higher in t-HG patients compared with no DM (P = 0.002) and PTDM (P = 0.04) patients but similar to that of Pre-DM (P = 0.25) patients.
Outcomes of patients with t-HG were not analyzed due to a small number of events.

DISCUSSION
This study represents the largest reported nonregistry long-term experience on DM (Pre-DM and PTDM) following LT to date. Furthermore, it represents a multiethnic Western population, whereas previous literature predominantly involves studies of monoethnic cohorts from East Asia. Additionally, this study explores the risk factors of PTDM and the impact of Pre-DM and PTDM on renal, cardiovascular, and de novo cancer outcomes. Neither the Pre-DM nor the PTDM appear to impact patient survival, retransplantation, or development of de novo cancer. However, not surprisingly, the risk of developing ESRD and mCVE was significantly higher in patients with Pre-DM. Although not statistically significant, the impact of PTDM on ESRD and mCVE was numerically between that of Pre-DM and no DM.
A quarter of the patients who were not diabetic at the time of LT developed PTDM, which is on par with the previously reported 19% to 35%. 15,17,18,[25][26][27][28] A few previous studies, however, have reported a lower incidence rate (6% to 8%) 21,29 and the reason/s for this substantial difference is not apparent. Furthermore, the occurrence of PTDM was considerably high during the early post-transplant period in this study and previous studies. 18,21,26,27 In addition, the demographic and clinical characteristics at the time of LT of patients with Pre-DM and patients who later developed PTDM were similar. This remarkable similarity and the onset of PTDM immediately following LT raises important questions including whether PTDM is merely the clinical manifestation of a previously subclinical prediabetic state and whether those at risk of developing PTDM may potentially be identified pretransplant with rigorous investigations such as oral glucose tolerance test or glucose clamp studies. The above notions are perhaps supported by the findings of a retrospective study, 30 which suggest a 2-fold increase in the development of PTDM with every 10 mg/dL increase in pretransplant fasting plasma glucose level. However, prospective studies are needed for the validation of these hypotheses.
A wide variety of recipient and donor factors, cause and severity of liver disease, and surgical and post-transplant management factors have been suggested to play a role in the development of PTDM, 15,17,18,21,[25][26][27][28][29][30][31][32][33] with limited consistency in common independent risk factors. The possible explanations for such discrepancy between studies include difference in cohort characteristics, cohort size, follow-up duration, and factors included in the analysis. A recent meta-analysis involving over 4500 patients from 19 retrospective studies identified male sex, body mass index, hepatitis C etiology, pre-LT impaired fasting glucose, and use of tacrolimus as independent risk factors of PTDM. 34 However, this meta-analysis failed to include recipient age, NAFLD etiology, and donor factors in the analysis, which have been shown to increase the risk of PTDM development. Although we included a range of factors that were shown to be associated with the development of PTDM in our analysis, Bonferroni correction was undertaken to mitigate the problem of multiple comparisons. In our study, NAFLD etiology and the use of sirolimus and tacrolimus were the only independent factors associated with the development of PTDM. Sirolimus inhibits insulinmediated inactivation of hepatic glycogenolysis, 35 causes a reduction in human pancreatic ductal cells, and inhibits glucose-stimulated insulin secretion, 36 which potentially explains its role in the development of PTDM. Similarly, reduced insulin secretion and increased insulin resistance have been proposed as potential mechanisms of tacrolimus-related PTDM. 37 On the contrary, the independent association between NAFLD etiology and PTDM may be due to the subclinical prediabetic status of these individuals (as discussed earlier).
Interestingly, neither the Pre-DM nor the PTDM adversely impacted patient survival and retransplantation. However, despite the optimal HbA1c control in the post-LT period, Pre-DM adversely affected renal and cardiovascular outcomes. The possible reasons for this are the presence of micro and macrovascular end-organ damage even before the formal diagnosis of DM 38 and the questionable reliability of HbA1c as a marker of glycemic control. 39 Moreover, the impact of PTDM on ESRD and mCVE was in between that of Pre-DM and no DM. Given that the atherosclerosis that leads to both micro and macrovascular complications require prolonged duration of DM, we suspect that PTDM would have an adverse impact on renal and cardiovascular outcomes that is comparable to that of Pre-DM in the long term. These findings enhance our understanding of Pre-DM and PTDM and their effect on the quality and quantity of life after LT, which is helpful especially given the paucity of such data to date. However, given that the use of tacrolimus was an independent risk factor for PTDM, the development of ESRD especially in patients with PTDM may have been confounded by the use of tacrolimus. Furthermore, as we enter an era where recurrent hepatitis C after LT becomes obsolete and de novo/recurrent NAFLD increase in incidence, we are likely to see changes in the patterns of incidence and impact of PTDM.
The literature on the impact of both Pre-DM and PTDM on outcomes in other solid organ transplantation has been limited and contradictory. In a South Korean study of 176 kidney transplant recipients without diabetes at transplantation, PTDM was associated with the increased incidence of cardiovascular disease but did not compromise graft or patient survival. 40 In a single-center study from Portugal involving 648 kidney transplant recipients, PTDM was shown to have no impact on renal dysfunction, cardiovascular events, graft, and patient survival. 41 Similarly, in a study from South Africa involving 111 kidney transplant recipients, PTDM did not impact graft or patient survival. 42 However, in a multicenter study involving over 2000 kidney transplant recipients from 5 North American and 2 European centers, PTDM was independently associated with graft survival; this study failed to investigate the impact of PTDM on patient survival or other outcomes. 43 In a registry study involving more than 9000 lung transplant recipients, both Pre-DM and PTDM were shown to impact patient survival adversely. 44 Literature on the impact of Pre-DM and PTDM on outcomes of heart transplant recipients were not available.
Interestingly, those who developed t-HG had unique demographic and clinical characteristics, with hepatitis C being the most common underlying etiology of liver disease in these patients; prevalence of NAFLD was at its lowest despite increased BMI. Whether hepatitis C played a crucial role in the development of transient hyperglycemia during the immediate post-LT period and why the hyperglycemic state did not continue beyond the first 6 months remain open to speculation.
Our study has its own strengths and limitations. It is the largest nonregistry study to date. Unlike the national registry studies, single-center studies, such as ours, have the advantage of consistency with respect to the use of immunosuppression and the management of DM. Our study also has the advantage of representing the longest duration of post-LT follow-up, with up to 20 years after transplantation; previous studies have been restricted with follow-up and have been limited by relatively small patient numbers. Furthermore, ours is the first nonregistry study to investigate the impact on the development ESRD, mCVE, and de novo cancer in liver transplant recipients with Pre-DM, PTDM, and no DM, in the long term. On the contrary, the retrospective nature of our study is a natural limitation, due to the risk of bias inherent to study design when compared with a prospective trial. One such limitation is the lack of data on the duration of DM in Pre-DM patients, which has limited our ability to distinguish those who developed DM secondary to cirrhosis from those with DM that is unrelated to cirrhosis. Interestingly, neither the BMI at listing nor the change in BMI during the first year of transplantation were associated with development of PTDM. This, however, needs to be interpreted with caution because of the limitations in the accuracy of BMI measurement in the pretransplant population (eg, due to fluid retention). Furthermore, though our study represents one of the longest experiences, it may still not be adequate to capture the "true" impact of PTDM on outcomes.
In conclusion, this is the largest nonregistry study to investigate the impact of Pre-DM and PTDM on LT outcomes. The baseline (at LT) clinical characteristics of patients who later developed PTDM were similar to that of patients with Pre-DM, suggesting that the genetic susceptibility may be similar between the 2 clinical entities. Pre-DM seems to pose the highest risk toward the development of ESRD and mCVE following LT, despite satisfactory HbA1c control. This emphasizes the urgent need for better investigation/s of glycemic control to minimize the long-term impact of DM. Although PTDM appears to have a lesser impact than that of Pre-DM, this needs to be interpreted with caution as the pathogenesis of macrovascular and microvascular complications requires prolonged duration of DM.