Application of modified Charlson comorbidity index for predicting outcomes following adult living donor liver transplantation

Abstract

BACKGROUND

The model for end-stage liver disease (MELD) score helps assess the severity of liver disease and can predict survival after liver transplant. The Charlson comorbidity index (CCI) is frequently employed to forecast the 10-year survival probability of patients with multiple health conditions. We employed the CCI to evaluate the impact of comorbid health conditions on patients and assess its predictive capability regarding health complications and mortality following living donor liver transplantation (LDLT).

AIM

To understand the prevalence of extrahepatic comorbidities in our cohort of LDLT patients with modified CCI (mCCI) and to analyze the utility of mCCI as a predictor of morbidity and mortality following LDLT.

METHODS

After obtaining institutional ethics committee approval, a retrospective analysis was conducted on 497 adult patients who underwent LDLT at our institute between January 2021 and December 2023.

RESULTS

Our analysis revealed that the area under the curve (AUC) of the original CCI for predicting 90-day mortality decreased when malignancy was assigned a score of 2 in patients with hepatocellular carcinoma undergoing transplantation. Therefore, we used a mCCI. Both MELD and mCCI scores demonstrated predictive value for 90-day mortality, with AUCs of 0.60 and 0.62, respectively. Using regression coefficients, we developed a composite score defined as: Combined score = [mCCI + (MELD/10)]. This composite metric improved predictive accuracy, yielding an AUC of 0.70 for 90-day mortality prediction. Patients with a CCI > 3 and a MELD > 21 had a significantly higher 90-day mortality rate than others (12.5% vs 5.7%; P = 0.02).

CONCLUSION

The mCCI was independent of decompensation and overall disease severity. Combining MELD and CCI scores enhanced the discriminatory power for predicting morbidity and 90-day mortality.

Key Words: Charlson comorbidity index; Post living donor liver transplantation outcomes; Composite outcome scores in living donor liver transplantation; Charlson comorbidity index vs model for end-stage liver disease; Modified Charlson comorbidity index in liver transplantation; Outcome prediction in living donor liver transplantation

Core Tip: With advances in surgical techniques and perioperative care, the outcomes of liver transplantation have improved significantly, encouraging its use in more elderly patients and those with more comorbidities. In addition to liver disease severity, understanding the impact of extra-hepatic comorbidities on the outcomes will help in prognostication and risk stratification with appropriate resource planning. The Charlson comorbidity index has been validated as a good predictor of long-term survival in varied clinical populations. With limited studies regarding its use in liver transplant patients, we aimed to study its applicability in a cohort of living donor liver transplant patients.

INTRODUCTION

Liver transplantation (LT) is a definitive therapeutic intervention for patients with end-stage liver disease. With ongoing advances in surgical techniques and perioperative care, outcomes after LT have improved significantly over the decades, supporting its use in older patients and those with more substantial comorbidities[1,2]. Nevertheless, LT is frequently associated with several significant complications, such as postoperative infections, renal dysfunction and extended need for organ support[3].

In addition to assessing the severity and waitlist mortality of patients with liver disease, the model for end-stage liver disease (MELD) score also predicts survival after LT[4,5]. The Charlson comorbidity index (CCI) has been commonly used to predict 10-year survival in patients with multiple comorbidities and different clinical populations[6,7]. To expand its utility in LT patients, we recalibrated CCI to generate a modified CCI (mCCI), termed CCI-orthotopic LT, which was found to be a good predictor of post-LT outcomes[8,9]. Here, we analyze the utility of CCI in predicting post-LT outcomes in a cohort of adult living donor liver transplant recipients.

MATERIALS AND METHODS

The aim of this study was to understand the prevalence of extrahepatic comorbidities in a cohort of living donor LT (LDLT) patients with CCI and analyze the utility of mCCI as a predictor of morbidity and mortality following LDLT.

This is a retrospective, single center study of a prospectively collected database. Following approval from the Institutional Ethics Committee, all adult patients who underwent LDLT at our institute between January 2021 and December 2023 were included. As it was a retrospective study, informed consent was waived by the Institutional Ethics Committee (No. ECR/1276/Inst/TN/2019/2025/312). Demographic and clinical data were collected from the electronic database. Exclusion criteria included patients who underwent combined solid organ transplantation or LT for acute liver failure, and pediatric LDLT. This study adhered to the principles of the Declaration of Helsinki and Declaration of Istanbul.

Variables such as patient age, sex, ethnicity, body mass index (BMI), etiological factors, comorbidities, and perioperative complications were assessed and analyzed. A history of admission to the intensive care unit (ICU) at our institution or other hospitals for non-surgical reasons within the 60-day period preceding transplantation was classified as recent ICU admission prior to transplantation. At our center, ICU admission prior to transplantation for patients with nonsurgical liver disease was determined based on the Modified Early Warning Score[10]. The CCI score was calculated using https://www.mdcalc.com/calc/3917/charlson-comorbidity-index-cci. The CCI scoring system is presented in Supplementary Table 1. Because assigning a score for hepatocellular carcinoma (HCC) in the context of underlying liver disease could introduce confounding variables, that component was not included. A liver disease score of 1 was assigned to patients with HCC, and all other patients received a score of 3 for the liver component of CCI.

Intraoperative parameters assessed included cold ischemia time (CIT), warm ischemia time, blood transfusion, graft-to-recipient weight ratio, crystalloids, higher blood transfusion, need for vasopressors for more than 12 hours after arrival to ICU, duration of surgery, and post-reperfusion syndrome (PRS). The PRS definition by Aggarwal et al[11] was used in this study. Higher blood transfusion was classified as administration of ≥ 6 units of whole blood or packed red blood cells (PRBCs) within 24 hours. PRBCs were transfused to achieve a hemoglobin level of 8-9 g/dL. Other blood products were transfused according to clinical situation and thromoboelastometric findings. Crystalloids and colloids were administered according to pulse pressure variations and clinical situation according to the care provider’s decision. A second vasopressor was added when the dose of noradrenaline was > 0.2 µg/kg/minute.

Postoperative morbidity was evaluated using the Clavien-Dindo classification (CDC) and was included as morbidity in the study for grade ≥ 3b[12]. Acute kidney injury was defined according to the Kidney Disease: Improving Global Outcomes guidelines[13]. Non-pulmonary sepsis was diagnosed if there was a ≥ 2 increase in the Sequential Organ Failure Assessment score[14]. Chest infection was diagnosed in patients with new infiltrates on lung imaging, cough, dyspnea, and at least two of the following three clinical features: (1) Fever > 38 °C; (2) Leukocytosis or leukopenia; and (3) Purulent secretions[15]. Early graft dysfunction was defined according to Olthoff et al’s criteria: (1) Bilirubin ≥ 10 mg/dL on postoperative day 7; (2) International normalized ratio ≥ 1.6 on day 7; and (3) Alanine or aspartate aminotransferase > 2000 IU/L within the first 7 days[16]. Prolonged mechanical ventilation was defined as the necessity for invasive mechanical ventilation exceeding 96 hours. Neurological complications included peripheral neuropathy, seizures, altered sensorium, central pontine myelinolysis syndrome, and focal neurological deficits. Mortality rates at the 30-day and 90-day intervals were documented.

Statistical analysis

Continuous variables are presented as median with interquartile range (IQR), whereas categorical variables are expressed as frequencies and percentages. The Mann-Whitney U test was used to compare continuous variables, and the χ² or Fisher’s exact test was used for categorical variables. Correlations between continuous variables were analyzed using Pearson’s correlation coefficient. The predictive capability of mCCI was assessed through receiver operating characteristic (ROC) analysis, from which the optimal point on the ROC curve was determined, facilitating the classification of the study population into two groups: (1) Low mCCI; and (2) High mCCI. Independent predictors of the outcome of interest were identified using a binary logistic regression analysis. Regression coefficients derived from binary regression analysis were utilized to integrate the scores. Variables with P value less than 0.05 were considered statistically significant. All statistical analyses were conducted using MedCalc for Windows, version 23.0.2 (MedCalc Software, Ostend, Belgium).

RESULTS

General characteristics

A total of 497 patients were included in the analysis. Our cohort had a median age of 52 years (IQR: 45, 59) and predominantly male sex [409 (82.3%)]. The median immediate preoperative MELD score was 17 (IQR: 13, 22). Metabolic dysfunction-associated steatotic liver disease (MASLD) was the predominant etiology of liver disease in our study (140 patients; 28.2%). A total of 319 (64.3%) patients were Indian, followed by patients from the Middle-East region [83 (16.7%)].

Classifying patients

In our study group, 26 (5.2%) patients had an mCCI score of 1, indicating the presence of compensated liver disease with HCC as an indication for LT. The remaining 68 patients with HCC had decompensations and were assigned a score of 3 in the liver disease domain. Our primary aim was to analyze the prevalence of extrahepatic comorbidities. The distribution of mCCI scores in the study population is presented in Figure 1. Most of these patients fell into the category of mCCI = 3 [240 (48.3%)], followed by 130 (26.2%) patients with mCCI = 4, 53 (10.7%) with a score of 5, 6 (1.2%) with a score of 6, and 7 (1.4%) with a score of ≥ 7. The optimal point from Youden’s index of the ROC curve was used to classify the study population into two groups. Patients with an mCCI score ≤ 3 were classified into the low mCCI group and > 3 into the high mCCI group.

Figure 1 Diagram showing frequency distribution of modified Charlson comorbidity index in our study.

Comparison of preoperative and intraoperative variables

The demographic characteristics and preoperative comorbidities of the patients are shown in Table 1. The median ages of the patients in the low and high mCCI groups were 49 (42, 57) years and 57 (50, 61) years, respectively (P < 0.001). The study cohort consisted predominantly of male patients (82.2%). The distribution of ethnicity between the two groups was comparable. The prevalence of MASLD as an etiology was significantly higher in the high mCCI group (41% vs 19.6%; P < 0.001). Conversely, the low mCCI group had a higher proportion of patients with HCC as the etiology (19.9% vs 17.3%; P < 0.001; Figure 2). The median MELD scores were 16 and 18 in the low and high mCCI groups, respectively (P = 0.01). The prevalence of associated comorbidities such as diabetes mellitus, hypertension, chronic kidney disease, and the incidence of acute kidney injury was higher (88.7%, 33.2%, 10.1%, and 48.9%, respectively) in the high mCCI group than in the low mCCI group (14.2%, 18.6%, 0%, and 33.1%, respectively; all P < 0.001). The occurrence of decompensation, such as spontaneous bacterial peritonitis before transplantation, was comparable between the groups. The requirement for ICU intervention 60 days before LT was 32.2% in the low mCCI group and 38.8% in the high mCCI group (P = 0.16). The correlation between MELD score and mCCI was weak (r = 0.28; P < 0.001).

Figure 2 Distribution of high (orange bars) and low modified Charlson comorbidity index patients (blue bars) across various etiology of liver disease in our study. HCC: Hepatocellular carcinoma; MASH: Metabolic dysfunction-associated steatohepatitis; mCCI: Modified Charlson comorbidity index.

Table 1 Comparison of preoperative characteristics between patients with modified Charlson comorbidity index ≤ 3 and those with modified Charlson comorbidity index > 3, n (%).

Variables		mCCI ≤ 3 (n = 301)	mCCI > 3 (n = 196)	P value
Age (years) (IQR)		49 (42, 57)	57 (50, 61)	< 0.001
Sex, male		252 (83.7)	157 (80.1)	0.30
Body mass index (kg/m2) (IQR)		26 (22.8, 29.7)	27 (23.2, 30)	0.46
Model for end-stage liver disease (IQR)		16 (11, 22)	18 (15, 21)	0.01
Geographical distribution	Indian	191 (63.4)	128 (65.3)	0.58
	Middle East and Egypt	57 (18.9)	26 (13.2)
	Asia (other than India)	26 (8.6)	23 (11.7)
	North Africans	10 (3.3)	6 (3.0)
	Others	17 (5.6)	13 (6.6)
Etiology	Metabolic dysfunction-associated steatohepatitis	59 (19.6)	81 (41.3)	< 0.001
	Ethanol related	53 (17.6)	29 (14.8)
	Viral	18 (6)	15 (7.7)
	Cryptogenic	36 (12)	16 (8.2)
	Autoimmune	27 (9)	10 (5.1)
	Hepatocellular carcinoma	60 (19.9)	34 (17.3)
	Others	48 (15.9)	11 (5.6)
Diabetes mellitus		43 (14.2)	174 (88.7)	< 0.001
Hypertension		56 (18.6)	65 (33.1)	< 0.001
Acute kidney injury		100 (33.2)	96 (48.9)	< 0.001
Chronic kidney disease		0 (0)	20 (10.2)	< 0.001
History of preoperative spontaneous bacterial peritonitis		55 (18.2)	33 (16.8)	0.68
History of preoperative intensive care unit stay with 60 days prior to transplant		97 (32.2)	75 (38.2)	0.16

IQR: Interquartile range; mCCI: Modified Charlson comorbidity index.

The intraoperative parameters are presented in Table 2. There was no significant difference between the groups in any intraoperative parameter except in the number of intraoperative PRBC received, which was more in high mCCI groups (P < 0.001).

Table 2 Comparison of intraoperative parameters between the two groups, n (%)/interquartile range.

Intra operative parameters	mCCI ≤ 3 (n = 301)	mCCI > 3 (n = 196)	P value
Graft-to-recipient weight ratio	0.93 (0.78, 1.08)	0.94 (0.8, 1.12)	0.37
Intra operative peak lactate (mmol/L)	4.1 (2.9, 5.6)	3.6 (2.9, 5.2)	0.85
Shifting lactates (mmol/L)	3.4 (2.1, 5)	3 (2.2, 4.4)	0.61
Crystalloids intra operative (L)	2.5 (1.5, 3)	2.5 (1.5, 3.5)	0.75
Intra operative packed red blood cells	3 (1, 5)	4 (3, 6)	< 0.001
Massive blood transfusion	51 (16.9)	45 (22.9)	0.09
Cold ischemia time (minutes)	98 (75, 135)	103 (75, 129)	0.81
Warm ischemia time (minutes)	43 (36, 51)	43 (36, 53)	0.23
Reperfusion syndrome	19 (6.3)	18 (9.1)	0.23
Surgical duration (minutes)	540 (438, 578)	500 (450, 585)	0.12

mCCI: Modified Charlson comorbidity index.

Postoperative outcomes

The postoperative parameters and outcomes are presented in Table 3. The proportion of patients requiring prolonged vasopressor support following surgery was 42.8% and 37.7% in the low and high mCCI groups, respectively (P = 0.25). The incidence of early allograft dysfunction was 9.6% in the low mCCI group and 12.2% in the high mCCI group (P = 0.35). Postoperative complications, including re-exploration, non-pulmonary sepsis, lung infection, postoperative bleeding, and the duration of ICU and hospital stay, were comparable between the two groups.

Table 3 Postoperative outcomes in the two groups, n (%).

Postoperative outcomes	mCCI ≤ 3 (n = 301)	mCCI > 3 (n = 196)	P value
Prolonged vasopressor support from the end of surgery	129 (42.8)	74 (37.7)	0.25
Early allograft dysfunction	29 (9.6)	24 (12.2)	0.35
Incidence of re-exploration surgery	36 (11.9)	26 (13.2)	0.66
Postoperative renal replacement therapy	11 (3.6)	21 (10.7)	0.002
Incidence of sepsis	53 (17.5)	37 (18.8)	0.71
Incidence of prolonged mechanical ventilation	22 (7.3)	27 (13.7)	0.01
Incidence of lung infections	61 (20.2)	39 (19.8)	0.92
Incidence of cardiac complications	19 (6.3)	22 (11.2)	0.05
Incidence of neurological complications	20 (6.6)	29 (14.7)	0.003
Incidence of wound infection	18 (5.9)	22 (11.2)	0.03
Length of ICU stay (days) (IQR)	6 (5, 8)	6 (5, 10)	0.03
Length of hospital stay (days) (IQR)	16 (13, 20)	17 (14, 22)	0.01
ICU re-admission	29 (9.6)	33 (16.5)	0.01
Morbidity (Clavien-Dindo classification ≥ 3b)	53 (17.6)	57 (29)	0.002
30-day mortality	6 (1.9)	10 (5.1)	0.06
90-day mortality	14 (4.6)	21 (10.7)	0.01

ICU: Intensive care unit; IQR: Interquartile range; mCCI: Modified Charlson comorbidity index.

The high mCCI group exhibited a significantly higher incidence of prolonged mechanical ventilation, wound infection, cardiac complications, neurological complications, and the requirement for postoperative renal replacement therapy than the low mCCI group. The ICU readmission rate was higher in the high mCCI group (16.5%) than in the low mCCI group (9.6%; P = 0.01).

The overall 90-day mortality rate was 7.8% and the morbidity rate was 22.1%. Overall morbidity > grade 3b was higher in the high mCCI group (29%) than in the low mCCI group (17.6%; P = 0.002). The mortality rate at 90 days was higher in the high mCCI group (10.7% vs 4.6%; P = 0.01). Mortality and morbidity for each mCCI score are displayed in Figure 3.

Figure 3 Morbidity and 90-day mortality in different classes of modified Charlson comorbidity index. A: Morbidity; B: 90-day mortality. Blue bars: Relative frequency of patients with morbidity and mortality. Orange bars: Non-morbid patients.

Combining the mCCI and preoperative MELD

We analyzed the MELD score to determine the severity of liver disease and the mCCI to assess the severity of comorbid conditions and to predict outcomes. We generated a composite score based on the regression coefficients, which was mCCI + (MELD/10). This composite score was used to predict the 90-day mortality and morbidity rates.

ROC curve analysis

The overall 90-day mortality rate was 7.8% in our study population, and the morbidity rate was 22.1%. The MELD, mCCI, and composite score predicted morbidity with area under the curve (AUC) values of 0.51, 0.57, and 0.59, respectively. The MELD and mCCI scores predicted 90-day mortality with AUC values of 0.60 and 0.62, respectively (Table 4). The composite score with integration of MELD and mCCI scores enhanced the predictive capability, yielding an AUC of 0.70 to predict 90-day mortality. Patients with both CCI > 3 and MELD > 21 exhibited a higher 90-day mortality rate (12.5%) than other patients (5.7%; P = 0.02; Figure 4).

Figure 4 Mortality in different groups of preoperative models for end-stage liver disease and modified Charlson comorbidity index category. mCCI: Modified Charlson comorbidity index; MELD: Model for end-stage liver disease.

Table 4 Area under the curve for the predictors of outcome after living donor liver transplantation.

Preoperative parameter	Postoperative outcome	Area under the curve	Cut-off	Sensitivity (%)	Specificity (%)
MELD	CDC ≥ 3b	0.51	> 21	40	73
	90-day mortality	0.60		42	74
mCCI	CDC ≥ 3b	0.57	> 3	52	65
	90-day mortality	0.62		63	49
Combination score [mCCI + (MELD/10)]	CDC ≥ 3b	0.59	> 5	54	65
	90-day mortality	0.70		40	87

CDC: Clavien-Dindo classification; mCCI: Modified Charlson comorbidity index; MELD: Model for end-stage liver disease.

A combined score of > 5 predicted 90-day mortality with a sensitivity of 40% and specificity of 87%. A model is considered effective if the goodness of fit is > 0.5, which is the lower bound of the 95%CI. A goodness of fit below 0.5 indicates that the model is no better than random prediction. The mCCI had a lower bound of 0.50, and the MELD had 0.46 in predicting 90-day mortality, suggesting poor model performance. However, the combination score had a lower bound of 0.64, indicating that the model is effective. As a continuous variable, the composite score was elevated in patients with morbid conditions and in those who died within 90 days (Figure 5).

Figure 5 Box and whisker plots for the combined score [modified Charlson comorbidity index + (preoperative model for end-stage liver disease/10)] in patients with morbidity and 90-day mortality. A: Morbidity; B: 90-day mortality. mCCI: Modified Charlson comorbidity index; MELD: Model for end-stage liver disease.

Regression analysis

Owing to the limited number of patients experiencing 90-day mortality, we shifted our focus to morbidity as the primary outcome for regression analysis. Binary logistic regression analysis for morbidity in the univariate context identified several significant factors: (1) Sex (female); (2) mCCI > 3; (3) Diabetes mellitus; (4) Preoperative history of renal replacement therapy; (5) Preoperative ICU stay within 60 days; (6) Recurrent large-volume paracentesis; and (7) higher intraoperative blood transfusion. In the multivariate analysis, the following were determined to be independent predictors of mortality: (1) Sex (female) [odds ratio (OR) (95%CI): 1.99 (1.17-3.40)]; (2) mCCI > 3 [OR (95%CI): 1.66 (1.06-2.60)]; (3) Preoperative history of renal replacement therapy [OR (95%CI): 6.31 (2.04-19.47)]; (4) Recurrent large-volume paracentesis [OR (95%CI): 1.66 (1.03-2.67)]; and (5) Higher intraoperative blood transfusion [OR (95%CI): 2.33 (1.40-3.88)] (Table 5).

Table 5 Logistic regression analysis of factors associated with postoperative morbidity.

Characteristics of patients	Univariable analysis		Multivariable analysis
	OR	95%CI	OR	95%CI
Age	0.93	0.57-1.53	-	-
Sex	2.12	1.28-3.52b	1.99	1.17-3.40a
Body mass index	0.98	0.94-1.02	-	-
Smoking	1.27	0.80-2.01	-	-
MELD	1.02	0.66-1.55	-	-
Diabetes mellitus	1.75	1.15-2.69b	0.7	0.4-1.5
Insulin dependent diabetes mellitus	1.85	1.14-3.02b	0.8	0.4-1.6
Coronary artery disease	1.33	0.82-2.18	-	-
Hypertension	1.01	0.62-1.65	-	-
Chronic kidney disease	2.45	0.98-6.16	-	-
Etiology metabolic dysfunction-associated steatohepatitis	0.94	0.49-1.8	-	-
Preoperative intensive care unit stay	1.99	1.29-3.07b	0.7	0.5-1.1
Recurrent large volume paracentesis	2.10	1.34-3.29b	1.66	1.03-2.67a
Preoperative need of dialysis	7.64	2.56-22.85b	6.31	2.04-19.47b
Intraoperative massive blood transfusion	2.34	1.44-3.82b	2.33	1.40-3.88b
Graft-to-recipient weight ratio	0.99	0.94-1.04	-	-
mCCI > 3	1.92	1.25-2.94b	1.66	1.06-2.60a
mCCI + MELD/10	2.05	1.32-3.18b	0.99	0.96-1.01

^aP < 0.05.

^bP < 0.01.

mCCI: Modified Charlson comorbidity index; MELD: Model for end-stage liver disease; OR: Odds ratio.

DISCUSSION

In the current era, LT has shown promising results in patients with end-stage liver disease. The implementation of the MELD score for liver allocation in the United States began in 2002, aiding in prioritizing organ allocation for the sickest patients on the waitlist[17,18]. Recently, to reduce the waitlist duration for transplants, there has been an increased reliance on extended criteria donor liver grafts[19-21]. Survival following a transplant is influenced by factors related to both the donor and the recipient. In addition to predicting the severity of the condition and mortality on the waitlist, the MELD score forecasts survival after liver transplant[22,23]. In addition to examining hepatic markers, we aimed to assess extrahepatic comorbidities independent of LT and their impact on outcomes following LT. The American Society of Anesthesiologists (ASA) grading system is the most widely used method for evaluating patients based on their comorbidities. However, another tool, the CCI, comprises 19 items that correspond to various comorbid medical conditions, each with a distinct clinical weight based on the adjusted risk of 1-year mortality[7]. The total CCI score was used to predict mortality due to comorbidities over a span of 10 years. While the CCI was not originally intended to forecast perioperative mortality in surgical cohorts, Laor et al[24] demonstrated that the CCI was more strongly correlated with the risk of perioperative death in elderly patients than age alone. In another study by Varady et al[25], CCI was more accurate than ASA score in predicting 1-year mortality following hip surgery. We selected CCI rather than the ASA score because most patients presenting for LT are classified as grade 3 or higher, limiting the score’s discriminatory ability. Accordingly, we used the mCCI, which we anticipated would provide better stratification across the patient population.

We evaluated 497 patients in total. Among them, 94 had HCC; 68 of these patients had decompensated liver disease and HCC and were assigned a score of 3. The remaining 26 patients (5.2%), for whom HCC was the indication for LT, were assigned a score of 1 in the liver component of the CCI. Extrahepatic comorbidities significantly affected approximately half of the transplant population, with uncomplicated diabetes being the most common, present in about 40% of patients. Most individuals had an mCCI score of 3, with progressively fewer patients in higher score categories. We evaluated CCI performance using two methods: (1) Applying the original CCI method, which assigns score of 2 for malignancy; and (2) Omitting the additional malignancy score, given that malignancy in LT recipients is associated with better short-term post-transplant outcomes. We refer to this modified version as the mCCI. We found that the AUC of the original CCI was lower when a score of 2 was assigned to malignancy. Thus, we conclude that carcinoma/malignancy, considered a poor-prognosis component in the original CCI score, does not confer the same adverse impact in patients undergoing LT and therefore should not be weighted as such in this population. Consequently, we chose to use the mCCI. We then used the optimal cut-off from Youden’s index of ROC to distinguish patients with higher morbidity and mortality, which was > 3 for mCCI and > 21 for preoperative MELD. We compared the preoperative, intraoperative, and postoperative variables between the patients by dividing them into two groups: (1) mCCI > 3 and (2) mCCI ≤ 3.

In our study, patients with a high mCCI were generally older and had a higher incidence of MASLD. This may be attributed to the well-known association between diabetes, a component of metabolic syndrome, and MASLD[26-28]. In our study, the high mCCI group had a higher median MELD score, along with increased incidences of diabetes and acute kidney injury, as well as a greater prevalence of chronic kidney disease. Postoperatively, prolonged mechanical ventilation was more frequent in the high mCCI group, potentially due to older patient age. This observation aligns with existing literature, which links poor preoperative condition, high ASA, and elevated MELD scores to prolonged duration of mechanical ventilation after surgery[29-32].

Postoperative cardiac and neurological complications were more prevalent, which we attributed to the higher preoperative incidence of diabetes and the advanced patient age in the high mCCI group. Additionally, the postoperative wound infection rate was elevated in patients with high mCCI, likely due to the increased incidence of preoperative diabetes in this group. Diabetes is recognized as a risk factor not only for traditional complications, such as stroke, coronary heart disease, peripheral neuropathy, retinopathy, and nephropathy, but also for complications such as cancer and infections[33]. In a systematic review and meta-analysis of diabetes and surgical site infections (SSI), Martin et al[34] found a significant association between diabetes and SSI, which remained consistent across various types of surgeries, even after adjusting for BMI. They also confirmed a link between both preoperative and postoperative hyperglycemia and SSI, as well as a history of diabetes[34].

Owing to the low number of patients experiencing 90-day mortality in our study, we shifted our focus to morbidity as the outcome for regression analysis. We defined morbidity as CDC ≥ 3b, which became our outcome of interest, and conducted binary logistic regression analysis. This analysis identified several independent predictors of morbidity: (1) Sex (female); (2) Preoperative events; (3) Recurrent large-volume paracentesis and the need for dialysis; (4) mCCI > 3; and (5) Higher intraoperative blood transfusion.

We aimed to develop a score that would assign importance to both the hepatic and extrahepatic components. Although MELD was not significant in the univariate binary logistic regression analysis, we included it because the existing literature suggests its association with postoperative morbidity. Preoperative MELD was used as an indicator of liver disease severity and mCCI was used as a marker of extrahepatic comorbidities. To integrate both, we applied a conversion factor derived from the regression coefficients of these two variables, resulting in a combined composite score of mCCI + (MELD/10). However, the mCCI remained a significant independent risk factor in the regression analysis, whereas the combination score did not. This suggests that the preoperative MELD score contributes little to postoperative comorbidity.

Our study was limited by the retrospective nature of the analysis of prospectively collected data. Another limitation was the number of preoperative variables available for analysis. Further research, including comparisons with various comorbidities by propensity score matching, is needed in a prospective manner. Whether the equation of adding 1/10^th of the MELD to the mCCI holds should be confirmed with external validation. Despite these limitations, the dataset provided a large representative sample for predicting morbidity and 90-day mortality in patients undergoing LT. This may also be useful in anticipating the risk of perioperative complications and could aid in risk stratification, prognostication and appropriate resource allocation.

CONCLUSION

Carcinoma, which is a component with a poor outcome in the original CCI score, cannot be extrapolated to liver transplant patients and therefore, an mCCI score is needed. The mCCI was independent of liver decompensation and disease severity, and 60% of our patients had an mCCI ≤ 3. Combining the MELD and CCI scores improved the discriminatory power to predict morbidity and 90-day mortality.