Introduction

Type 2 Diabetes mellitus (T2DM) is associated with increased risk of fibrosis progression and metabolic complications in patients with non-cirrhotic chronic liver disease, particularly in patients with metabolic dysfunction-associated steatotic liver disease (MASLD) [1,2,3,4]. However, the published literature evaluating the impact of T2DM in patients with cirrhosis is limited and has produced mixed results [5,6,7]. The projected increase in the prevalence of both T2DM [8] and cirrhosis [9] over the next decade underscores the importance of better understanding the interaction between them as they impact both liver-associated clinical events (LACE) and non-LACE in this patient population. Furthermore, as more patients develop cirrhosis and progress to end-stage liver disease (i.e., decompensated cirrhosis), it is imperative to know how T2DM may affect clinical outcomes in these patients as they await liver transplantation (LT) [10].

The published literature evaluating the impact of diabetes in patients with cirrhosis is limited by relatively small sample size [11], and mostly constitutes of patients with viral hepatitis as the predominant etiology of cirrhosis which is not reflective of current trends that favors alcohol and MAFLD as the leading causes of cirrhosis [12]. Furthermore, the existing literature lacks heterogeneity in patient population that is inclusive of both patients with compensated and decompensated cirrhosis [13], thereby, potentially diffusing the effect size as the risk of decompensation is relatively low in patients with compensated cirrhosis [14]. There is also a lack of studies that have evaluated the impact of diabetes on morbidity and mortality in patients awaiting LT to better define how T2DM affects key transplant-related parameters such as waitlist mortality, likelihood of receiving LT and impact on metabolic complications such as cardiovascular disease and renal function.

As the impact of T2DM in patients with end-stage liver disease on LT waitlist is not known, we conducted the current study to evaluate the relationship between T2DM and clinical outcomes among patients with end-stage liver disease, undergoing LT evaluation. We hypothesized that the presence of T2DM will be associated with increased LACEs, metabolic complications, and reduced likelihood of receiving a LT. This hypothesis was tested in a large retrospective cohort of LT waitlist registrants using patient-level data.

Materials and Methods

The current study is a retrospective analysis of patients who were registered for LT waitlist at Virginia Commonwealth University (VCU). The study was approved by the institutional review board. All authors reviewed and approved the manuscript prior to submission.

Study Cohort

All adult patients, aged 18 years or older, who were registered on LT waitlist at Virginia Commonwealth University (VCU) between 1/2010 and 1/2017 were included in the analysis. Patients who were evaluated but not registered on the LT waitlist were excluded since the main objective of the current study was to evaluate the impact of T2DM on waitlist complications in LT candidates. Patients listed for dual organ transplants (i.e., liver-kidney or liver-heart) were also excluded. As part of the LT evaluation, all patients underwent detailed cardiovascular evaluation which included 2D-echocardiography and coronary artery disease assessment, which could include coronary angiography [15,16,17].

Follow-Up and Clinical Events

All patients were followed at 3-month intervals where routine bio-clinical data were collected, such as medical history, medication changes, laboratory, and vital signs. Patients also had their medical charts reviewed for any clinical events, which were characterized as LACE or non-LACE. LACE events included variceal hemorrhage, hepatic encephalopathy, and ascites. Additional complications of cirrhosis included development of varices, hepatic hydrothorax, hepatopulmonary syndrome, portopulmonary hypertension, and spontaneous bacterial peritonitis (SBP) [18]. Development of hepatocellular carcinoma (HCC) was not evaluated as a clinical outcome as HCC is a significant reason for LT waitlist registration and development of HCC, while waiting for LT is relatively uncommon, thereby, precluding significant statistical analysis [19]. Non-LACE parameters evaluated included hospitalization-related parameters that included number of hospitalizations, intensive care unit (ICU) level care, need for mechanical ventilation, and initiation of renal replacement therapy. The cause of hospitalization was enumerated as hepatic decompensation, infection, renal failure, and cardiovascular events including myocardial infarction, stroke, and cardiovascular death.

Definitions

Patients were considered to have T2DM if they had a medical diagnosis of T2DM, were taking exogenous insulin, hypoglycemic agents, or had HbA1c > 6.5%. Similarly, diagnosis of hepatic decompensation was extracted from the chart and included ascites, hepatic hydrothorax, esophageal varices, acute variceal hemorrhage (AVH), spontaneous bacterial peritonitis (SBP), and hepatic encephalopathy (HE).

Statistical Analysis

Categorical variables are reported as frequency and percentages, whiles continuous variables were reported as mean ± standard deviation (SD) if normally distributed and median with IQR if non-normal distribution. As this patient population consists mainly of patients with end-stage liver disease, most of the patients had already experienced a hepatic decompensating event. Thus, we first evaluated the relationship between likelihood of developing incident portal hypertension-related complications (i.e., incident ascites in patients without ascites at time of LT evaluation) via Kaplan-Meyer analysis and Cox regression in patients with and without T2DM. Regression analysis was adjusted for age, gender [male], etiology of liver disease, and baseline model for end-stage liver disease (MELD) score.

In the next analysis, the impact of T2DM on (1) individual cirrhosis-associated complications (2) hospitalizations (3) waitlist mortality and (4) likelihood of receiving LT was evaluated. To gain a deeper understanding on how the presence of T2DM might affect MELD score, the components of MELD score (creatinine, INR, bilirubin) were plotted over time, stratified according to T2DM status. In this analysis, patients who received a LT were excluded after the LT. Analysis of variance (ANOVA) and Student’s t test to test for differences in normally distributed continuous variables, Mann–Whitney U test for non-normally distributed continuous variables, and Chi-square or Fisher’s exact test for categorical variables to compare groups. A nominal p value of < 0.05 was considered statistically significant. The analysis was performed with the SPSS (IBM Inc., Cary, NC).

Results

Baseline Characteristics of Study Cohort

A total of 593 patients met inclusion criteria and were included in our final analysis (Table 1). The mean age of the cohort was 53 ± 10 years and consisted mainly of men (n = 382, 64%) and non-Hispanic whites (n = 430, 73%). The most common etiology of cirrhosis was Hepatitis C (37%), followed by alcohol-induced (29%) and MASH (17%). At baseline, T2DM was present in 192 (32%) of patients and of these, 73 patients were on exogenous insulin. The baseline mean HbA1c was 6.0 ± 1.84 and the fasting blood glucose was 126 ± 57.74 mg/dL. The mean MELD score at time of LT waitlist registration was 15.06 ± 6.95. The prevalence of ascites, hepatic encephalopathy, and history of variceal hemorrhage at baseline was 64%, 53%, and 17%, respectively.

Table 1 Baseline characteristics of the entire cohort
Full size table

Patients with T2DM were older (57 ± 7 vs. 52 ± 10 years, p < 0.001) and more likely to be males (70% vs 61%) (Supplemental Table 1). The BMI was 31 ± 6.6 kg/m2 in patients with T2DM compared to 29.2 ± 6.8 kg/m2 patients without T2DM (p = 0.004). Significant differences in prevalence of etiologies of end-stage liver disease requiring LT were noted in patients T2DM, when compared to patients without T2DM (Supplemental Table 1). Metabolic comorbidities, such as coronary artery disease, hypertension, dyslipidemia, and obesity, were also more common in patients with T2DM.

Relationship Between T2DM and Portal Hypertension at Baseline

Patients with T2DM were more likely to have esophageal varices (61% vs. 47%, p = 0.002) and history of acute variceal hemorrhage (23% vs. 16%, p = 0.03) (Fig. 1). The presence of other complications of cirrhosis including hepatic encephalopathy, ascites, SBP, and hepatic hydrothorax was similar in patients with and without T2DM at baseline. Baseline MELD scores were similar between the two cohorts; however, patients with T2DM had lower total bilirubin when compared to patients without T2DM (3.01 ± 3.59 vs. 4.45 ± 5.24 mg/dL; p < 0.001). No other differences in laboratory values between the two cohorts were noted.

Fig. 1
figure 1

Baseline burden of portal hypertension stratified according to the presence of type 2 diabetes

Full size image

In unadjusted analysis, the presence of T2DM was significantly associated with the presence of varices (OR 1.77, 95% CI 1.23, 2.53, p = 0.002) and history of acute variceal hemorrhage (OR 1.57, 95% CI 1.02, 2.43, p = 0.04). An association between hepatic hydrothorax and diabetes was noted; however, this did not reach the significance threshold (OR 2.03, 95% CI 0.98, 4.20, p = 0.05). In adjusted analysis, the relationship between T2DM and portal hypertension-related complications of the presence of varices (OR 1.51, 95% CI 1.03, 2.20, p = 0.03) and hepatic hydrothorax (OR 2.53, 95% CI 1.154, 5.54, p = 0.02) remained significant. The association between T2DM and HPS was noted; however, this did not reach significance (p = 0.08).

Impact of T2DM on Developing Hepatic Clinical Outcomes

The impact of T2DM on incident risk of hepatic decompensation was investigated via Cox regression. Patients with T2DM were more likely to develop incident ascites after 12 months of follow-up when compared to patients without T2DM (HR 1.91, 95% CI 1.11, 3.28, p = 0.019) (Fig. 2A). No statistically significant association between the presence of T2DM and likelihood of developing new esophageal varices, acute variceal hemorrhage, HE, SBP, and hepatic hydrothorax was noted (Fig. 2B–F).

Fig. 2
figure 2

Impact of type 2 diabetes mellitus (T2DM) on incidence of developing portal hypertension-related outcomes. Patients with T2DM were more likely to develop ascites (A); however, T2DM did not affect likelihood of developing varices, acute variceal hemorrhage (AVH), hepatic encephalopathy (HE), spontaneous bacterial peritonitis (SBP) or hepatic hydrothorax (B–F, respectively)

Full size image

Patients with T2DM were more likely to require hospitalizations (56% vs. 49%) over 12 months of follow-up; however, this did not reach statistical significance (p = 0.06) (Fig. 3A). The mean number hospitalization between patients with and without T2DM was similar (2.0 ± 1.2 vs. 1.8 ± 1.1; p = 0.21) over the 12-month period. However, patients with T2DM were more likely to require renal replacement therapy during their hospitalization compared to patients without T2DM (Fig. 3B). Patients with diabetes were more likely to be hospitalized with portal hypertension-related complications (22% vs. 14%; p = 0.026). During these hospitalizations, diabetes did not impact the probability of Intensive care unit (ICU) admission or mechanical ventilation (Fig. 3C, D).

Fig. 3
figure 3

Impact of type 2 diabetes mellitus (T2DM) on likelihood of hospitalization and associated complications. While T2DM status did not affect likelihood of hospitalization (A), patients with T2DM were more likely to initiated hemodialysis during their hospitalization (B). No association between intensive care unit (ICU) admission (C) or need mechanical ventilation (D) and T2DM were noted

Full size image

Impact of T2DM on Transplant Related Outcomes

Patients with T2DM were more likely to die (35% vs. 27%; p = 0.03); however, on survival analysis, survival rates between patients with and without T2DM were similar (Fig. 4A). Patients with T2DM were less likely to receive a LT (37% vs. 45%; p = 0.04) when compared to cirrhosis patients without T2DM (Fig. 4B). When components of MELD score (INR, bilirubin, and creatinine) were plotted over time, patients with T2DM were noted to have significantly lower serum bilirubin (Fig. 4C). No differences in INR and creatine over time were noted over time (Fig. 4D, E). Finally, liver-related causes of mortality were similar in patient with and without diabetes (Fig. 4F).

Fig. 4
figure 4

Interplay between type 2 diabetes mellitus (T2DM) and liver transplant (LT) metrics. While overall survival was similar between patients with and without T2DM (A), patients with T2DM were less likely to receive a liver transplantation (B). Trend in international normalized ratio (INR; C), total bilirubin (D), and creatinine (E) revealed that patients with T2DM have lower serum bilirubin than patients without T2DM. Causes of mortality are enumerated in F

Full size image

Discussion

The overall prevalence of diabetes is increasing rapidly [20], particularly among patients with chronic liver disease and cirrhosis, underscoring the importance of better understanding the interaction between them as driver of hepatic decompensation and clinical outcomes [21]. The present study expands on the published literature by demonstrating that patients with T2DM and end-stage liver disease have higher morbidity.

Study Findings in the Context of Published Literature

The published literature evaluating the impact of T2DM on liver-associated clinical events (LACE) is limited, particularly in patients with end-stage liver disease. Moreover, the literature is limited by relatively small sample size, exclusion of patient with known history of decompensated cirrhosis, and lack of association between T2DM and liver transplantation [13, 14]. The present study designed to overcome these limitations demonstrates significantly higher burden of portal hypertension in patients with end-stage liver disease and T2DM. The study corroborates prior reported data from studies with smaller sample size that also demonstrated higher portal hypertension-related burden in patients with T2DM [10, 22,23,24]. While the association between compensated cirrhosis and T2DM has been reported previously [25, 26], the current study expands on this by demonstrating the patients with T2DM and end-stage liver disease at significantly higher risk for liver-related morbidity and mortality. Furthermore, patients with T2DM were more likely to develop incident complications of portal hypertension, notably ascites, acute variceal hemorrhage, or varices over the study period. While historically, T2DM is over-represented in MASH (vs. other etiologies of cirrhosis). However, in the current cohort of patients with T2DM, the proportion of patients with MASH cirrhosis is slightly higher, but similar to the proportion of patients with HCV, the two most common etiologies of cirrhosis in the cohort. This provides preliminary data on how T2DM might drive clinical outcomes irrespective of the etiology of chronic liver disease. Furthermore, as MASH is the leading indication for LT in women and elderly and poised to be the leading indication among all potential LT recipients, these findings are generalizable to what a typical LT population will be in the near future.

No significant differences with regard to hospitalizations, ICU stay, or survival were noted in patients with and without T2DM. As the baseline risk of decompensation, hospitalization, and death is higher in the study cohort, it is not surprising that no significant differences in these outcomes were noted at study end. Furthermore, the causes of mortality were similar in patients with and without T2DM and expectedly, hepatic causes were the leading cause of mortality. Patients with T2DM were more likely to require renal replacement therapy during their hospitalizations. This might potentially be due to the presence of underlying chronic kidney disease that patients with T2DM are higher risk for primed them to developing renal injury that required renal replacement therapy [27]. Interestingly, serum creatinine at baseline was similar between the two cohorts suggesting the renal susceptibility may not be fully captured by commonly used renal parameters such as serum creatinine. Moreover, creatinine is known to be affected by muscle mass and sarcopenic patients have a reduced muscle mass. Given that sarcopenia is more prevalent in patients with T2DM, the use of serum creatinine and creatinine-based eGFR in this population may overestimate renal function. These findings underscore the importance of developing and validating tools that better evaluate the interaction between muscle mass to provide a better risk adjustment model. While some studies have reported reduced survival in patients with T2DM, other studies do not corroborate these findings [5, 10, 11, 26].

Impact of Study Findings on Clinical Practice

Patients with T2DM were less likely to receive LT over study duration when compared to patients without T2DM. As the observed survival and morbidity were similar in the two cohorts, a possible explanation for these findings is patients with DM may develop co-morbid conditions that lead to inactivation such as infections therefore disadvantaging them on the LT waitlist. While patients with T2DM were noted to have lower serum bilirubin, which persisted over the duration of the study cohort. However, other contributors of MELD score including INR and serum creatinine were not significantly different between the two cohorts. As the prevalence of chronic kidney disease is higher in patients with T2DM, lack of differences in renal function-related parameters (creatinine, blood urea nitrogen, eGFR) in the study cohort can potentially be explained by other competing factors such as diuretic use, HRS, and dehydration, that can lead to similar serum creatinine levels in patients with and without T2DM. The combined observations demonstrating higher baseline prevalence and incident risk of portal hypertension-related complications and lower probability of LT suggest that the MELD score alone underestimates the morbidity in patients with T2DM. Changes to MELD score calculation have been proposed over the years (i.e., MELDNa, MELD 3.0) with the overarching goal of improving organ allocations; however, how these scores might perform in patients with T2DM is not as well established. Given the study findings, it might be prudent to evaluate the relationship between T2DM and MELD score to improve outcomes in LT waitlisted patients. While the findings of this study are preliminary, it is important to confirm them in well-designed prospective studies to allow for better risk stratification of patients with T2DM undergoing LT evaluation. Recognizing this increased risk is fundamental to developing risk mitigation strategies to not only improve clinical outcomes of patients awaiting LT.

Study Strengths and Limitations

The current study is the largest study to date evaluating impact of T2DM on patients with end-stage liver disease undergoing LT evaluation. The clinical outcomes were carefully adjudicated and the study design evaluated outcomes important in clinical management including both hepatic and non-hepatic complications. The study is retrospective in design; however, as there are no data on T2DM among potential LT waitlist registrants, this study provides foundational data for study design and sample size determination for future prospective studies. The impact of diabetes treatment was not evaluated but will be difficult to do so in a retrospective study as hospitalizations, decompensation (i.e., renal failure) has the potential to not only change the type of hypoglycemic agents but also the intensity. Thus, a well-designed prospective study that incorporates the complexity of this patient population is necessary to better evaluate the relationship between type and intensity of hypoglycemic agents in patients with end-stage liver disease. The impact of glycemic control on clinical outcomes was not evaluated. As patients with end-stage liver disease are at higher risk for blood loss (bleeding, hemolysis) and may require blood transfusions, HbA1c may not be reliable. Serum fructosamine levels may provide more insight [28]; however, these are not collected as part of standard of care and thus could not be evaluated in this retrospective study. Given the limitations of classic measures of insulin resistance (i.e., fasting glucose and HbA1c) in this population, the potential impact of glycemic control and insulin resistance on clinical outcomes requires well-designed prospective studies with glucose tolerance test (clamp vs. frequent sampling) and a priori defined clinical outcomes are necessary. Moreover, such an approach can also lead to validation and development of better point of care testing (i.e., HOMA-IR, OGTT, etc.) to assess insulin resistance in this challenging patient population. Finally, while an association between RRT and T2DM was noted, the current study due to its retrospective study design is not able to provide a deeper understanding of these findings. Well-designed prospective studies are necessary to fully understand such an association.

In summary, T2DM in patients with end-stage liver disease is associated with increased morbidity and may negatively affect likelihood of receiving a LT. Additional studies are necessary to confirm these study findings and develop clinical strategies to improve outcomes in this patient population.