Social Determinants Are Important Barriers to Completion of the Liver Transplant Pathway and Are Associated with Waitlisting and Mortality in Hepatocellular Carcinoma

Abstract

Background

Hepatocellular carcinoma (HCC) is a serious consequence of chronic liver disease, with liver transplantation (LT) as the most durable curative option. However, significant disparities in access persist. Further, there is a lack of data on social determinants of health (SDOH) and disease-related barriers to HCC–LT pathway completion.

Methods

This retrospective cohort study included adults referred to a single center for LT with HCC from 2017 to 2021. SDOH exposures included race, ethnicity, gender, insurance type, area deprivation index (ADI), and marital status. Data from transplant database notes, including open-text fields documenting reasons for evaluation discontinuation, were used to identify and categorize reasons for failure to complete six predefined steps in the HCC–LT pathway. Multivariable logistic regression and competing risk analysis evaluated factors associated with waitlisting and survival.

Results

Among 495 HCC patients referred for LT, 8.7% were Black and 57.2% were insured by Medicaid, with a mean ADI of 69.0 (± 21.1). Disease-related barriers (39.0%) and social barriers (35.6%) were common reasons for failing to complete the LT pathway. Public insurance (aOR 0.65, 95% CI 0.44–0.95) and being unmarried (aOR 0.60, 95% CI 0.40–0.88) were independently associated with a lower odd of waitlisting, and high ADI (aHR 1.01, 95% CI 1.00–1.02) was associated with mortality.

Conclusions

Insurance type and marital status were associated with a failure to waitlist and ADI with survival. The SDOH were important barriers to completing the HCC–LT pathway. Targeted interventions are needed to support at-risk patients through the HCC–LT process.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10620-025-09278-3.

Introduction

Hepatocellular carcinoma (HCC) ranks among the top ten most prevalent cancers globally, with its incidence in the United States contributing to over 300,000 related deaths in the past two decades [1,2]. Despite the potential of liver transplantation (LT) as a curative option, offering a five-year survival rate of 70–80% [3], less than 30% of eligible patients receive this life-saving treatment [4], underscoring significant underutilization within the healthcare system.

Patients who are referred for LT for HCC face unique challenges compared to their end-stage liver disease counterparts. The pathway to LT for patients with HCC is fraught with additional hurdles, including the need for tumor downstaging and expedited, coordinated care to prevent cancer progression [5]. Furthermore, the challenges extend beyond clinical complexities. The social determinants of health (SDOH) such as insurance coverage, marital status, and neighborhood socioeconomic factors critically hinder access to necessary care, exacerbating disparities in outcomes [6–10]. These systemic barriers are compounded by the emotional and psychological challenges of managing a new cancer diagnosis [11].

Our previous research has identified pronounced racial disparities in HCC–LT accessibility that were not explained by liver disease severity or patient comorbidity [12–14]. A subset analysis suggests more upstream factors may contribute to these disparate outcomes. Moreover, a recent multicenter cancer registry study in Georgia revealed that non-privately insured patients, alongside Black individuals and those residing in high-poverty areas, are significantly less likely to be referred for or complete LT evaluations [15]. However, these studies have not fully explored the obstacles that patients face along the HCC–LT pathway.

There is an urgent need to enhance LT utilization for HCC across all demographic groups. This begins with understanding the intertwined social, systemic, and clinical barriers that obstruct patient progression through the HCC–LT care pathway. Yet, there remains a stark absence of contemporary data detailing the SDOH or disease-related obstacles faced by HCC patients from the point of LT referral to the completion of the LT pathway. By identifying the SDOH barriers linked to each failure point and determinants of waitlist dropout and mortality within the LT care pathway, this study aims to uncover actionable insights that could inform targeted interventions, potentially transforming the landscape of HCC treatment.

Methods

Study Population

This was a retrospective study including all adult patients diagnosed with HCC referred to Indiana University Health Academic Health Center for LT from January 1, 2017, to December 31, 2021. Patients were initially identified through our institution's transplant database. A manual review of each individual patient transplant database file and electronic health record was performed to confirm the diagnosis of HCC. 495 individuals met inclusion criteria. The study was reviewed and approved by the institutional review board at our institution and is compliant with ethical conduct of research. Reasons for exclusion included lack of confirmation of HCC diagnosis during chart review and referral only to establish post-transplant care with a transplant hepatologist in the state of Indiana after being transplanted at another site.

Transplant Database and Abstraction of Reasons for Dropout

The Ottr Organ patient management software, specifically designed for transplant centers, is used to document the clinical course of all patients referred to the Indiana University Academic Health Center. Unlike the standard EHR, this database includes fields for start and stop dates for each step in the LT pathway, as well as open text fields for documenting evaluation end reasons. Additionally, it features a notes section that captures the clinical course outside of clinic visits, including conversations between patients, transplant coordinators, and providers. This comprehensive database, along with provider and social work notes, offers rich documentation of provider-perceived reasons for HCC–LT pathway dropout.

Data from this database were abstracted by three co-authors to identify reasons for failure to complete six predefined steps in the HCC–LT pathway: (1) referral, (2) seen by transplant hepatology, (3) evaluation start, (4) evaluation completion, (5) LT waitlisting, and (6) LT. After abstraction, the reasons for dropout were categorized into themes: comorbidity/too frail/too sick, ongoing alcohol or substance use, adverse SDOH (e.g., lack of social support, unstable housing), patient disinterest and/or fear of LT, loss to follow-up (LTFU) or non-adherence, being too early for LT, cancer progression, opting for resection, choosing alternative therapy, death, or other factors. Death was considered a non-modifiable outcome and analyzed separately from social or disease severity-related reasons for dropout. A conceptual framework is outlined in Fig. 1 and representative quotes in Supplemental Table 1.

Fig. 1.

Conceptual framework

To further differentiate social versus disease-related barriers to dropout, all social reasons—including patient disinterest and fear of LT, LTFU, non-adherence, ongoing substance use, adverse SDOH, and lack of financial approval for evaluation—were grouped together. Disease severity-related barriers, such as cancer progression beyond criteria and comorbidity/being too frail/too sick, were analyzed separately from these social reasons. (Fig. 1).

To ensure the accuracy of our assessment regarding reasons for dropout, fifty charts were re-reviewed by authors (LN, BB, and TA). This process identified only one instance where a case was reclassified from the 'disease severity' domain to the 'social' domain as the reason for dropout.

Exposures

Exposures of interest included race, gender, ethnicity, insurance type, and neighborhood-level socioeconomic status (SES) indicators. Race and gender were defined by self-report in the transplant database or EHR as Black/African American, White, Asian, American Indian/Alaskan Native, Native Hawaiian/Pacific Islander, other, or unknown, and male or female, respectively. Ethnicity was defined as either Hispanic, Non-Hispanic, or unknown. Individual patient insurance types were collected at the time of referral in the LT database and through manual chart review. Types of insurance were grouped into four categories: Private, Medicare, Medicaid (Medicaid alone or Medicaid and Medicare), VA, uninsured, and other. Additionally, neighborhood SES was measured by utilizing the Area Deprivation Index (ADI) for each patient.

In this study, social determinants of health (SDOH) were defined using a pragmatic, patient-level approach, consistent with prior transplant-focused studies. While the World Health Organization defines SDOH broadly as the conditions in which people are born, grow, live, work, and age, our operationalization focused on factors documented in clinical and transplant evaluations. These included insurance status, marital status, neighborhood-level deprivation (measured by ADI), and documented barriers such as unstable housing, lack of transportation, mental health and substance use disorders, and language barriers. This targeted approach reflects social risk domains most likely to impact the transplant pathway [16].

In the United States, insurance does not offer automatic public funding like other countries for LT and HCC treatment, requiring patients to navigate a complex system to secure coverage. These treatments are mainly funded through a combination of private insurance, Medicare, and Medicaid. Private insurance coverage varies by plan and often requires preauthorization. Medicare covers LT for those aged 65 + , individuals with disabilities, or those with end-stage liver disease at Medicare-approved centers. Medicaid provides coverage for low-income individuals, but eligibility and benefits vary by state [17].

Area Deprivation was obtained by linking patients' addresses to a census tract defined in the 2017 American Community Survey (ACS). The ACS provides data about social, economic, housing, and demographic characteristics for multiple geographic areas down to the census tract level, which are small areas of approximately 1200–8000 (average 4000) persons [18]. The Area Deprivation Index includes 21 ACS variables in the domains of income, education, employment, and housing quality. Scores range from 1 to 100, with 100 representing the highest level of deprivation. These collected scores were grouped into quartiles, Q1: 0–56, Q2: 57–72, Q3: 73–86, and Q4: 87–100 for analysis.

Covariates

Covariates of interest were captured from both the LT database and EMR. Clinical and demographic characteristics include age, BMI, alpha fetoprotein level, underlying liver disease etiology and severity, and cancer stage. Underlying liver disease was defined as alcohol-related liver disease (ALD), metabolic associated steatohepatitis (MASH), hepatitis C virus associated (HCV), hepatitis B virus associated (HBV), autoimmune hepatitis (AIH), cholestatic including primary biliary cholangitis, and primary sclerosing cholangitis (PBC/PSC). All other underlying liver diseases were classified as other.

The Model for End-Stage Liver Disease (MELD)-Na, Milan status, and Barcelona Clinic Liver Cancer (BCLC) stage at the time of referral was obtained for each patient. Marital status was defined as married, not married, separated, widowed, or unknown. Last date of follow-up was defined as last clinical encounter with a provider or the last date of labs or imaging.

Outcomes

We described social and disease-related reasons for failure to complete six predefined steps in the LT care pathway. Demographic, clinical, and SDOH predictors of waitlisting and death were also evaluated.

Statistical Analysis

Subject characteristics were summarized by waitlist status as means and SDs, medians and quartile ranges, and frequencies and percentages, as appropriate. All variables were assessed for normality. Categorical variables were compared to test for heterogeneity using Chi-Square tests, with Fisher’s Exact being used when > 20% of cells had expected cell counts < 5 and using Mantel–Haenszel Chi-Square tests when there were ordinal responses. Continuous variables were compared using Wilcoxon tests. Disease severity-related barriers and social barriers at each step were summarized and compared using the same methods as the subject characteristics.

Time to death was calculated from the date of referral to the date of death. For transplanted subjects, time was calculated from the date of referral to the date of transplant. Alive, non-transplanted subjects were censored at last known follow-up. If the last follow-up date was prior to the referral date, referral date was used as last follow-up. Logistic regression was used to assess the association between the exposures of interest and the primary outcome of waitlisting. Additionally, the effects of demographic, clinical characteristics, and SDOH on survival were analyzed using the Fine and Gray method, treating transplant as a competing risk for death. Age and race were included in all models. Other covariates of interest were considered for inclusion in the multivariable models based on a significance threshold of p < 0.05 from the univariable analyses. Final variables were chosen based upon clinical relevance and multicollinearity concerns.

To assess for potential dropout overinflation due to clinical ineligibility in the primary analysis, which included all patients regardless of referral stage, a sensitivity analysis was performed. We examined progression through key stages of the LT-HCC pathway in two scenarios: (1) those outside of Milan criteria and (2) those with BCLC stage C or BCLC stage D disease who were also outside the Milan criteria. For patients identified as BCLC stage D, those outside the Milan criteria were excluded to ensure that individuals classified as BCLC D were solely due to underlying liver disease severity rather than cancer progression. Both scenarios showed patients with successful progression through key stages: evaluation start (n = 58 and n = 12), waitlisting (n = 20 and n = 3), and transplantation (n = 15 and n = 2) (Supplementary Table 2). Furthermore, there was little change in the reasons for dropout with either scenario (Supplemental Figures, Panel A-B). Consequently, the primary analysis retained all patients. Additionally, a sensitivity analysis compared significant predictors of waitlisting and mortality, excluding patients in both scenarios (Supplementary Tables 3–6).

Data summaries were produced using R Statistical Software (version 4.1.0). All analytic assumptions were verified, and all tests were conducted as two-sided at 0.05 significance level. All analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA).

Results

Patient Characteristics

There were 495 patients with HCC referred for LT during the study period. The majority were men (74.1%), and the median age at referral was 62.0 years (IQR 58.0–67.0), (Table 1). Additionally, 8.7% of the cohort was Black race, 2.8% Asian, and 4.0% Hispanic ethnicity. The most common underlying liver disease etiology was ALD (34.9%), followed by HCV (31.5%) and MASH (21.8%). The mean MELD-Na score was 12.0 (± 5.4), and 73.7% were within the Milan criteria and 91.2% had an AFP <  = 500 at the time of referral. The majority of patients were insured by Medicaid (57.2%), 53.5% of the cohort were married, and the mean ADI was 69.0 (± 21.1) (Table 1).

Table 1.

Demographic and Clinical Characteristics of Patients with Hepatocellular Carcinoma Waitlisted for Liver Transplantation

	Not Waitlisted (N = 330)	Waitlisted (N = 165)	Total (N = 495)	p value
Age at Referral				0.346
Mean (SD)	61.3 (8.7)	62.2 (6.3)	61.6 (8.0)
Median (Q1, Q3)	62.0 (57.0, 67.0)	63.0 (59.0, 67.0)	62.0 (58.0, 67.0)
Range	22.0–84.0	41.0–75.0	22.0–84.0
Gender				0.059
Women	94 (28.5%)	34 (20.6%)	128 (25.9%)
Men	236 (71.5%)	131 (79.4%)	367 (74.1%)
Race				0.030a*
American Indian/Alaskan Native	1 (0.3%)	1 (0.6%)	2 (0.4%)
Asian	8 (2.4%)	6 (3.6%)	14 (2.8%)
Black/African American	36 (10.9%)	7 (4.2%)	43 (8.7%)
Native Hawaiian/Pacific Islander	2 (0.6%)	1 (0.6%)	3 (0.6%)
White	256 (77.6%)	144 (87.3%)	400 (80.8%)
Other	1 (0.3%)	0 (0.0%)	1 (0.2%)
Unknown	26 (7.9%)	6 (3.6%)	32 (6.5%)
Ethnicity				0.010*
Hispanic or Latino	11 (3.3%)	9 (5.5%)	20 (4.0%)
Non-Hispanic	291 (88.2%)	153 (92.7%)	444 (89.7%)
Unknown	28 (8.5%)	3 (1.8%)	31 (6.3%)
Marital Status				0.010a*
Married	157 (47.6%)	108 (65.5%)	265 (53.5%)
Not Married	70 (21.2%)	21 (12.7%)	91 (18.4%)
Separated	50 (15.2%)	24 (14.5%)	74 (14.9%)
Widowed	19 (5.8%)	12 (7.3%)	31 (6.3%)
Unknown	34 (10.3%)	0 (0.0%)	34 (6.9%)
Insurance				0.025a*
Medicaid	199 (60.3%)	84 (50.9%)	283 (57.2%)
Private	119 (36.1%)	77 (46.7%)	196 (39.6%)
VA	1 (0.3%)	2 (1.2%)	3 (0.6%)
Uninsured	8 (2.4%)	1 (0.6%)	9 (1.8%)
Other	2 (0.6%)	1 (0.6%)	3 (0.6%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.2%)
ADI				0.008*
Mean (SD)	70.8 (20.8)	65.6 (21.3)	69.0 (21.1)
Median (Q1, Q3)	74.0 (58.0, 88.0)	69.0 (52.0, 82.0)	72.0 (56.0, 86.0)
Range	9.0–100.0	8.0–100.0	8.0–100.0
N-Miss	5	0	5
ADI Quartile				0.003*
Q1	76 (23.4%)	51 (30.9%)	127 (25.9%)
Q2	76 (23.4%)	48 (29.1%)	124 (25.3%)
Q3	89 (27.4%)	41 (24.8%)	130 (26.5%)
Q4	84 (25.8%)	25 (15.2%)	109 (22.2%)
N-Miss	5	0	5
Etiology				0.046a*
Alcohol	120 (36.4%)	53 (32.1%)	173 (34.9%)
MASH	62 (18.8%)	46 (27.9%)	108 (21.8%)
HCV	111 (33.6%)	45 (27.3%)	156 (31.5%)
AIH	1 (0.3%)	4 (2.4%)	5 (1.0%)
Cholestatic (PBC/PSC)	2 (0.6%)	4 (2.4%)	6 (1.2%)
HBV	7 (2.1%)	8 (4.8%)	15 (3.0%)
Other	12 (3.6%)	4 (2.4%)	16 (3.2%)
Unknown	15 (4.5%)	1 (0.6%)	16 (3.2%)
Within Milan?				< 0.001*
No	98 (29.7%)	20 (12.1%)	118 (23.8%)
Yes	222 (67.3%)	143 (86.7%)	365 (73.7%)
Unknown	10 (3.0%)	2 (1.2%)	12 (2.4%)
BCLC stage				0.006*
0	8 (2.6%)	7 (4.3%)	15 (3.2%)
A	176 (57.3%)	112 (69.6%)	288 (61.5%)
B	68 (22.1%)	26 (16.1%)	94 (20.1%)
C	16 (5.2%)	2 (1.2%)	18 (3.8%)
D	39 (12.7%)	14 (8.7%)	53 (11.3%)
N-Miss	23	4	27
MELD-Na				0.039*
Mean (SD)	11.7 (5.3)	12.6 (5.7)	12.0 (5.4)
Median (Q1, Q3)	10.0 (8.0, 14.0)	11.0 (8.0, 16.0)	10.0 (8.0, 15.0)
Range	6.0–30.0	6.0–32.0	6.0–32.0
N-Miss	25	0	25
BMI				0.597
Mean (SD)	29.8 (6.4)	29.4 (5.6)	29.7 (6.1)
Median (Q1, Q3)	29.1 (25.2, 33.2)	28.4 (25.3, 32.4)	28.9 (25.2, 33.0)
Range	16.4–48.5	9.0–43.2	9.0–48.5
N-Miss	40	1	41
AFP at Referral				<0.001*
Mean (SD)	1089.6 (7028.9)	42.1 (173.3)	730.9 (5718.3)
Median (Q1, Q3)	8.7 (4.3, 49.9)	5.3 (3.3, 12.2)	7.0 (3.8, 31.0)
Range	0.7–99,209.0	0.9–1541.0	0.7–99,209.0
N-Miss	40	14	54

*p < 0.05

Black vs. White vs. All Other

Married vs. Not Married

Private vs. All Other

Alcohol vs. MASH vs. HCV

Significant demographic differences were observed between those who were waitlisted and those who were not, including race, ethnicity, marital status, insurance, ADI, etiology, Milan criteria, BCLC stage, AFP, and MELD-Na. Among Black patients referred for LT, 16.3% were waitlisted compared to 36.0% of White patients (p = 0.030). The waitlisted cohort included 9 (5.5%) Hispanic patients, 153 (92.7%) in the non-Hispanic waitlisted group, and 3 (1.8%) of unknown ethnicity (p = 0.010). Additionally, a significantly higher proportion of the waitlisted patients were married compared to non-waitlisted patients (65.5% vs. 47.6%, p = 0.010), and waitlisted subjects were more commonly to be privately insured than non-waitlisted subjects (46.7% vs. 36.1%, p = 0.025). Those waitlisted had a significantly lower ADI (mean 65.6 ± 21.3 vs. 70.8 ± 20.8, p = 0.008). Waitlisted patients were more often within Milan criteria (86.7% vs. 67.3%, p < 0.001), had a BCLC stage of 0 or A (73.9% vs. 59.9%, p = 0.006), had higher MELD-Na values (mean 11.7 ± 5.3 vs. 12.6 ± 5.7 p = 0.039), and higher AFP levels (median 8.7 [IQR 4.3–49.9] vs. 5.3 [IQR 3.3–12.2], p < 0.001), (Table 1).

Completion of Steps in the HCC Liver Transplantation Pathway

There were 495 patients referred for LT. Their progression through six steps in the LT pathway is described in Fig. 2. Of those referred for LT, 452 (91.3%) were seen by transplant hepatology to be considered for evaluation, 313 (63.2%) started LT evaluation, 182 (36.8%) completed LT evaluation, 165 (33.3%) were waitlisted, and 133 (26.9%) underwent LT.

Fig. 2.

Reasons for dropout from each step in the HCC–LT pathway

Among the 313 patients with evaluation start dates, the median time from referral to evaluation start was 36 days (mean 64.6 days). For the 165 waitlisted patients, the median time from referral to waitlisting was 188 days (mean 250.6), and the median time from evaluation start to waitlisting was 133 days (mean 181.7). For the 133 transplanted patients, the median time from waitlisting to transplant was 140 days (mean 168.6). Time from evaluation start was not statistically associated with waitlisting or transplant, nor with being categorized as having social or clinical barriers.

Overall, 129 patients failed to complete the HCC–LT pathway due to social reasons (Fig. 3). A full comparison of how these patients compare to the rest of the cohort is described in Table 2. Those patients who failed to complete the HCC–LT pathway for social reasons were more likely to be younger (59.4 ± 9.5 vs. 62.4 ± 7.3, p = 0.013) and unmarried (65.0% vs 43.4%, p = 0.034) than the rest of the cohort (Table 2).

Fig. 3.

Composite social and disease-related reasons for dropout in the HCC–LT pathway

Table 2.

Patients with social barriers to completing the HCC–LT pathway

	No social barriers (N = 366)	Social barriers (N = 129)	Total (N = 495)	p value
Age at Referral				0.013*
Mean (SD)	62.4 (7.3)	59.4 (9.5)	61.6 (8.0)
Median (Q1, Q3)	63.0 (58.0, 67.0)	61.0 (56.0, 66.0)	62.0 (58.0, 67.0)
Range	23.0–84.0	22.0–73.0	22.0–84.0
Gender				0.701
Women	93 (25.4%)	35 (27.1%)	128 (25.9%)
Men	273 (74.6%)	94 (72.9%)	367 (74.1%)
Race				0.141a
American Indian/Alaskan Native	1 (0.3%)	1 (0.8%)	2 (0.4%)
Asian	9 (2.5%)	5 (3.9%)	14 (2.8%)
Black/African American	27 (7.4%)	16 (12.4%)	43 (8.7%)
Native Hawaiian/Pacific Islander	2 (0.5%)	1 (0.8%)	3 (0.6%)
White	301 (82.2%)	99 (76.7%)	400 (80.8%)
Other	1 (0.3%)	0 (0.0%)	1 (0.2%)
Unknown	25 (6.8%)	7 (5.4%)	32 (6.5%)
Ethnicity				0.718
Hispanic or Latino	15 (4.1%)	5 (3.9%)	20 (4.0%)
Non-Hispanic	330 (90.2%)	114 (88.4%)	444 (89.7%)
Unknown	21 (5.7%)	10 (7.8%)	31 (6.3%)
Marital status				0.034a*
Married	207 (56.6%)	58 (45.0%)	265 (53.5%)
Not Married	62 (16.9%)	29 (22.5%)	91 (18.4%)
Separated	52 (14.2%)	22 (17.1%)	74 (14.9%)
Widowed	22 (6.0%)	9 (7.0%)	31 (6.3%)
Unknown	23 (6.3%)	11 (8.5%)	34 (6.9%)
Insurance				0.970a
Medicaid	210 (57.4%)	73 (56.6%)	283 (57.2%)
Private	145 (39.6%)	51 (39.5%)	196 (39.6%)
VA	2 (0.5%)	1 (0.8%)	3 (0.6%)
Uninsured	6 (1.6%)	3 (2.3%)	9 (1.8%)
Other	2 (0.5%)	1 (0.8%)	3 (0.6%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.2%)
ADI				0.071
Mean (SD)	68.1 (21.2)	71.7 (20.8)	69.0 (21.1)
Median (Q1, Q3)	71.0 (55.0, 86.0)	75.0 (59.8, 88.0)	72.0 (56.0, 86.0)
Range	8.0–100.0	10.0–99.0	8.0–100.0
N-Miss	4	1	5
ADI Quartile				0.092
Q1	98 (27.1%)	29 (22.7%)	127 (25.9%)
Q2	96 (26.5%)	28 (21.9%)	124 (25.3%)
Q3	92 (25.4%)	38 (29.7%)	130 (26.5%)
Q4	76 (21.0%)	33 (25.8%)	109 (22.2%)
N-Miss	4	1	5
Etiology				0.060a
Alcohol	123 (33.6%)	50 (38.8%)	173 (34.9%)
MASH	89 (24.3%)	19 (14.7%)	108 (21.8%)
HCV	110 (30.1%)	46 (35.7%)	156 (31.5%)
AIH	4 (1.1%)	1 (0.8%)	5 (1.0%)
Cholestatic (PBC/PSC)	6 (1.6%)	0 (0.0%)	6 (1.2%)
HBV	11 (3.0%)	4 (3.1%)	15 (3.0%)
Other	10 (2.7%)	6 (4.7%)	16 (3.2%)
Unknown	13 (3.6%)	3 (2.3%)	16 (3.2%)
Milan				0.596
No	84 (23.0%)	34 (26.4%)	118 (23.8%)
Yes	274 (74.9%)	91 (70.5%)	365 (73.7%)
Unknown	8 (2.2%)	4 (3.1%)	12 (2.4%)
BCLC				0.830
0	14 (4.0%)	1 (0.8%)	15 (3.2%)
A	209 (59.7%)	79 (66.9%)	288 (61.5%)
B	72 (20.6%)	22 (18.6%)	94 (20.1%)
C	16 (4.6%)	2 (1.7%)	18 (3.8%)
D	39 (11.1%)	14 (11.9%)	53 (11.3%)
N-Miss	16	11	27
MELD-Na				0.035*
Mean (SD)	12.3 (5.6)	11.1 (4.9)	12.0 (5.4)
Median (Q1, Q3)	11.0 (8.0, 16.0)	9.0 (8.0, 12.0)	10.0 (8.0, 15.0)
Range	6.0–32.0	6.0–29.0	6.0–32.0
N-Miss	15	10	25
BMI				0.770
Mean (SD)	11.8 (4.9)	10.7 (4.5)	11.5 (4.8)
Median (Q1, Q3)	11.0 (8.0, 15.0)	9.0 (8.0, 12.0)	10.0 (8.0, 14.0)
Range	6.0–30.0	6.0–30.0	6.0–30.0
N-Miss	13	8	21
AFP at Referral				0.175
Mean (SD)	758.4 (6038.3)	653.8 (4730.1)	730.9 (5718.3)
Median (Q1, Q3)	7.1 (3.8, 40.2)	6.5 (3.8, 16.4)	7.0 (3.8, 31.0)
Range	0.9–99,209.0	0.7–49,886.0	0.7–99,209.0
N-Miss	41	13	54

*p < 0.05

Black vs. White vs. All Other

Married vs. Not Married

Private vs. All Other

Alcohol vs. MASH vs. HCV

Additionally, 141 patients failed to complete the HCC–LT care pathway due to disease severity-related barriers (Fig. 3). A full comparison of how these patients compare to the rest of the cohort is described in Table 3. Those patients who failed to complete the HCC–LT pathway for disease severity-related reasons were more commonly women (34.0% vs 22.6%, p = 0.009) and insured by Medicaid (65.2% vs 54.0%, p = 0.015) than the rest of the cohort (Table 3). Notably, among women referred, 37.5% of women had disease-related barriers compared to 25.3% of men referred (p = 0.009).

Table 3.

Patients with disease-related barriers to completing the HCC–LT pathway

	No disease-related barriers (N = 354)	Disease-related barriers (N = 141)	Total (N = 495)	p value
Age at Referral				0.468
Mean (SD)	61.4 (8.3)	62.1 (7.3)	61.6 (8.0)
Median (Q1, Q3)	62.0 (57.2, 67.0)	63.0 (59.0, 66.0)	62.0 (58.0, 67.0)
Range	22.0–84.0	37.0–80.0	22.0–84.0
Gender				0.009*
Women	80 (22.6%)	48 (34.0%)	128 (25.9%)
Men	274 (77.4%)	93 (66.0%)	367 (74.1%)
Race				0.151a
American Indian/Alaskan Native	2 (0.6%)	0 (0.0%)	2 (0.4%)
Asian	11 (3.1%)	3 (2.1%)	14 (2.8%)
Black/African American	25 (7.1%)	18 (12.8%)	43 (8.7%)
Native Hawaiian/Pacific Islander	1 (0.3%)	2 (1.4%)	3 (0.6%)
White	288 (81.4%)	112 (79.4%)	400 (80.8%)
Other	1 (0.3%)	0 (0.0%)	1 (0.2%)
Unknown	26 (7.3%)	6 (4.3%)	32 (6.5%)
Ethnicity				0.333
Hispanic or Latino	16 (4.5%)	4 (2.8%)	20 (4.0%)
Non-Hispanic	313 (88.4%)	131 (92.9%)	444 (89.7%)
Unknown	25 (7.1%)	6 (4.3%)	31 (6.3%)
Marital status				0.121a
Married	196 (55.4%)	69 (48.9%)	265 (53.5%)
Not Married	55 (15.5%)	36 (25.5%)	91 (18.4%)
Separated	56 (15.8%)	18 (12.8%)	74 (14.9%)
Widowed	21 (5.9%)	10 (7.1%)	31 (6.3%)
Unknown	26 (7.3%)	8 (5.7%)	34 (6.9%)
Insurance				0.015a*
Medicaid	191 (54.0%)	92 (65.2%)	283 (57.2%)
Private	152 (42.9%)	44 (31.2%)	196 (39.6%)
VA	3 (0.8%)	0 (0.0%)	3 (0.6%)
Uninsured	6 (1.7%)	3 (2.1%)	9 (1.8%)
Other	1 (0.3%)	2 (1.4%)	3 (0.6%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.2%)
ADI				0.951
Mean (SD)	69.3 (20.3)	68.4 (23.0)	69.0 (21.1)
Median (Q1, Q3)	72.0 (57.2, 86.0)	71.0 (51.8, 89.0)	72.0 (56.0, 86.0)
Range	8.0—100.0	8.0—99.0	8.0—100.0
N-Miss	4	1	5
ADI Quartile				0.980
Q1	84 (24.0%)	43 (30.7%)	127 (25.9%)
Q2	94 (26.9%)	30 (21.4%)	124 (25.3%)
Q3	103 (29.4%)	27 (19.3%)	130 (26.5%)
Q4	69 (19.7%)	40 (28.6%)	109 (22.2%)
N-Miss	4	1	5
Etiology				0.530a
Alcohol	127 (35.9%)	46 (32.6%)	173 (34.9%)
MASH	78 (22.0%)	30 (21.3%)	108 (21.8%)
HCV	106 (29.9%)	50 (35.5%)	156 (31.5%)
AIH	4 (1.1%)	1 (0.7%)	5 (1.0%)
Cholestatic (PBC/PSC)	4 (1.1%)	2 (1.4%)	6 (1.2%)
HBV	11 (3.1%)	4 (2.8%)	15 (3.0%)
Other	12 (3.4%)	4 (2.8%)	16 (3.2%)
Unknown	12 (3.4%)	4 (2.8%)	16 (3.2%)
MELD-Na				0.850
Mean (SD)	12.1 (5.6)	11.8 (5.1)	12.0 (5.4)
Median (Q1, Q3)	10.0 (8.0, 15.2)	10.0 (8.0, 14.0)	10.0 (8.0, 15.0)
Range	6.0—32.0	6.0—31.0	6.0—32.0
N-Miss	22	3	25
BMI				0.945
Mean (SD)	29.7 (5.9)	29.7 (6.5)	29.7 (6.1)
Median (Q1, Q3)	29.0 (25.2, 32.9)	28.7 (25.2, 33.0)	28.9 (25.2, 33.0)
Range	17.5—48.5	9.0—48.5	9.0—48.5
N-Miss	36	5	41
AFP at Referral				<0.001*
Mean (SD)	775.2 (6657.6)	628.3 (2427.6)	730.9 (5718.3)
Median (Q1, Q3)	6.1 (3.7, 18.4)	11.9 (4.4, 93.6)	7.0 (3.8, 31.0)
Range	0.7–99,209.0	0.9–19493.0	0.7–99,209.0
N-Miss	46	8	54

*p < 0.05

Black vs. White vs. All Other

Married vs. Not Married

Private vs. All Other

Alcohol vs. MASH vs. HCV

After referral, 35 died and 57 patients had other reasons for dropout (Fig. 2).

Detailed reasons for failing to complete each step are described below:

Dropouts Between Step 1 (Referral for Evaluation) and Step 2 (Being Seen by Transplant Hepatology)

Of the 495 patients referred for LT for HCC, 452 (91.3%) were seen by transplant hepatology. Reasons for not seeing transplant hepatology after referral included insurance denial (n = 14) and referring provider’s intention being tumor board review instead of LT (n = 29).

Dropouts Between Step 2 (Seen by Transplant Hepatology) and Step 3 (Evaluation Started)

Of the 452 patients who were seen by transplant hepatology, 139 (30.8%) did not have an evaluation started. The most common reason for not having an evaluation started was patients' disinterest and fear (n = 36, 25.9%) followed by cancer being outside of criteria (n = 24, 17.3%) (Supplemental Fig. 1a–d). Failure to have an evaluation start for additional social reasons to patient disinterest and fear included ongoing alcohol or substance use (n = 13, 9.4%), loss to follow-up and non-adherence (n = 13, 9.4%), and adverse SDOH (n = 5, 3.6%) amounting in total to 48.2% of those who did not have an evaluation started (Supplemental Fig. 1a–d).

Dropouts Between Step 3 (Evaluation Start) and Step 4 (Evaluation Completion)

Of the 313 patients who began evaluation, 131 (41.9%) were unable to complete the process. This represented the largest dropout step in the pathway. The most common reasons for inability to complete this step were cancer progression outside of criteria (n = 36, 27.5%), becoming too sick or frail (n = 36, 27.5%), and patient disinterest and fear (n = 20, 15.3%) (Supplemental Fig. 1a–d). Inability to complete an evaluation for social reasons, including ongoing alcohol of substance use (n = 12, 9.2%), loss to follow-up and non-adherence (n = 6, 4.6%), adverse SDOH (n = 7, 5.3%), and disinterest and fear (n = 20, 15.3%), amounted in total to 34.4% of those who did not complete evaluation (Supplemental Fig. 1a-d).

Dropouts Between Step 4 (Evaluation Completion) and Step 5 (Waitlisting)

Of the 182 patients who completed evaluation, 165 (90.7%) were waitlisted for LT. The most common reason for not being waitlisted after evaluation completion was cancer progression outside of criteria (n = 5, 29.4%) and becoming too sick or frail (n = 5, 29.4%). There was one patient not waitlisted after evaluation completion due to disinterest/fear (n = 1, 5.9%) (Supplemental Fig.1a–d).

Dropouts Between Step 5 (Waitlisting) and Step 6 (Transplant)

Of the 165 patients who were waitlisted, 133 underwent LT. The most common reasons for waitlist dropout were cancer progression outside of criteria (n = 15, 46.9%) and becoming too sick or frail (n = 9, 28.1%). There was one patient who failed to complete this step due to non-adherence (n = 1, 3.13%) and one who was disinterested/fearful (n = 1, 3.13%) (Supplemental Fig. 1a–d).

Predictors of Waitlisting and Mortality

From univariable analysis, Black vs White race, underlying etiology, Milan status, BCLC stage, marital status, insurance type, AFP, and ADI were significant predictors of waitlisting (Table 4). From multivariable analysis, being within Milan (aOR 3.44, 95% CI 1.98–6.00, p =  < 0.001), having private insurance (aOR 1.74, 95% CI 1.10–2.73, p = 0.017), and being married (aOR 1.62, 95% CI 1.04–2.52, p = 0.033) remained significant predictors of being waitlisted (Table 4).

Table 4.

Demographic, clinical, and SDOH associated with odds of waitlisting

	Univariable				Multivariable
Predictor	Estimated odds ratio	95% confidence limits		p value	Estimated odds ratio	95% confidence limits		p value
Age at Referral	1.02	0.99	1.04	0.202	1.02	0.99	1.05	0.327
Gender (Women vs. Men)	0.65	0.42	1.02	0.060
Race (Overall p value)	–	–	–	0.039*	–	–	–	0.295
Race (Black vs. White)	0.35	0.15	0.80	0.013*	0.49	0.20	1.20	0.118
Race (Other vs. White)	1.19	0.47	2.97	0.717	0.98	0.36	2.65	0.972
Ethnicity (Overall p value)	–	–	–	0.020*
Ethnicity (Non-Hispanic vs Hispanic)	0.64	0.26	1.58	0.337
Ethnicity (Unknown vs. Hispanic)	0.13	0.03	0.58	0.007*
Married (Yes vs. No)	1.68	1.13	2.49	0.010*	1.62	1.04	2.52	0.033*
Insurance (Private vs. Non-Private)	1.54	1.06	2.26	0.025*	1.74	1.10	2.73	0.017*
Etiology (Alcohol vs. Not Alcohol)	0.78	0.52	1.16	0.212
Etiology (MASH vs. Not MASH)	1.59	1.03	2.47	0.039*
Etiology (HCV vs. Not HCV)	0.70	0.46	1.05	0.085
Etiology (AIH/PBC/PSC vs. Not AIH/PBC/PSC)	5.33	1.40	20.38	0.014*
Within Milan? (Overall p value)	–	–	–	< 0.001*	–	–	–	< 0.001*
Within Milan? (Unknown vs No)	0.98	0.20	4.82	0.980	0.64	0.07	5.63	0.689
Within Milan? (Yes vs No)	3.16	1.87	5.33	< 0.001*	3.44	1.98	6.00	< 0.001*
BCLC (C-D vs. 0-B)	0.51	0.28	0.92	0.024*
MELD-Na	1.03	1.00	1.07	0.061
BMI	0.99	0.96	1.02	0.454
AFP	0.99	1.00	1.00	0.021*
ADI (Continuous)	0.99	0.98	1.00	0.011*
ADI Quartile (Overall p value)	–	–	–	0.023*	–	–	–	0.232
ADI (Q2 vs. Q1)	0.94	0.57	1.56	0.815	0.95	0.54	1.68	0.854
ADI (Q3 vs. Q1)	0.70	0.41	1.18	0.181	0.75	0.42	1.33	0.321
ADI (Q4 vs. Q1)	0.45	0.26	0.79	0.005*	0.56	0.30	1.03	0.061

^*p < 0.05

Predictors of death during LT evaluation for HCC from univariable analysis included MELD-Na score, AFP, Milan status, BCLC stage, ADI, marital status, and insurance type (Table 5). From multivariable analysis, Milan status (aHR 0.59, 95% CI 0.40–0.87, p = 0.008), MELD-Na score (aHR 1.04, 95% CI 1.01–1.07, p = 0.008), and ADI quartile 4 (aHR 1.72, 95% CI 1.08–2.75, p = 0.023) remained significant predictors of death (Table 5).

Table 5.

Demographic, clinical, and SDOH associated with odds of death

	Univariable				Multivariable
Predictor	Estimated hazard ratio	95% confidence limits		p value	Estimated hazard ratio	95% confidence limits		p value
Age at referral	1.00	0.98	1.02	0.921	1.00	0.97	1.02	0.817
Gender (Women vs. Men)	1.20	0.84	1.73	0.324
Race (Overall p value)	–	–	–	0.674	–	–	–	0.803
Race (Black vs. White)	1.24	0.70	2.18	0.458	1.14	0.63	2.08	0.661
Race (Other vs. White)	0.82	0.34	2.00	0.660	0.80	0.32	1.98	0.627
Ethnicity (Overall p value)	–	–	–	0.150
Ethnicity (Non-Hispanic vs Hispanic)	1.36	0.63	2.93	0.436
Ethnicity (Unknown vs. Hispanic)	2.58	0.93	7.17	0.070
Married (Yes vs. No)	0.71	0.51	0.99	0.041*	0.77	0.52	1.12	0.168
Insurance (Private vs. Non-Private)	0.70	0.50	0.98	0.040*	0.68	0.46	1.02	0.060
Etiology (Alcohol vs. Not Alcohol)	1.23	0.89	1.71	0.218
Etiology (MASH vs. Not MASH)	1.04	0.72	1.50	0.849
Etiology (HCV vs. Not HCV)	1.02	0.74	1.41	0.894
Etiology (AIH/PBC/PSC vs. Not AIH/PBC/PSC)	1.34	0.57	3.18	0.507
Within Milan? (Overall p value)	–	–	–	0.019*	–	–	–	0.024*
Within Milan? (Unknown vs No)	2.30	0.75	7.06	0.147	1.27	0.24	6.69	0.780
Within Milan? (Yes vs No)	0.69	0.48	0.99	0.044*	0.59	0.40	0.87	0.008*
BCLC (C-D vs. 0-B)	2.35	1.54	3.58	< 0.001*
MELD-Na	1.04	1.01	1.07	0.008*	1.04	1.01	1.07	0.008*
BMI	1.00	0.97	1.03	0.885
AFP	1.00	1.00	1.00	<0.001*
ADI (Continuous)	1.01	1.00	1.02	0.009*
ADI Quartile (Overall p value)	–	–	–	0.007*	–	–	–	0.090
ADI (Q2 vs. Q1)	1.14	0.70	1.84	0.601	1.10	0.65	1.86	0.717
ADI (Q3 vs. Q1)	1.18	0.72	1.95	0.519	1.14	0.67	1.95	0.622
ADI (Q4 vs. Q1)	1.98	1.28	3.04	0.002*	1.72	1.08	2.75	0.023*

^*p < 0.05

To assess the robustness of our findings regarding the association between exposure and outcome and account for dropout overinflation due to LT ineligibility due to clinical reasons, we conducted a series of sensitivity analyses. The direction and magnitude of the association between exposure and outcome remained fairly consistent with our primary analysis with some notable changes and additional significant predictors of waitlisting and death.

In multivariable analysis excluding patients outside of Milan criteria (n = 377), insurance type remained a significant predictor of waitlisting. Notably, ADI quartile 3 vs 1 (aOR 0.51, 95% CI 0.27–0.97, p = 0.040) and ADI quartile 4 vs 1 (aOR 0.42, 95% CI 0.21–0.84, p = 0.014) were newly significant predictors of waitlisting (Supplementary Table 3). ADI quartile 4 vs 1 remained a significant predictor of death. Additionally, marital status (aHR 0.64, 95% CI 0.42–1.00, p = 0.048) emerged as a new significant predictor of death in this scenario (Supplementary Table 4).

In multivariable analysis excluding patients with BCLC C or D status outside of Milan criteria (n = 465), marital status and insurance type remained significant predictors of waitlisting. ADI quartile 4 vs 1 (aOR 0.52, 95% CI 0.28–0.97, p = 0.040) emerged as newly significant predictors of waitlisting (Supplementary Table 5). MELD-Na and ADI quartile 4 (aHR 1.82, 95% CI 1.12–2.95, p = 0.016) remained significant predictors of death (Supplementary Table 6).

We also conducted an exploratory analysis to examine the effect of distance on waitlisting status. Using the US Census Bureau geocoder, we calculated the distance in kilometers from patient address to transplant center for 452 of the 495 participants. Limiting the analysis to participants within Indiana, Kentucky, Ohio, or Illinois (N = 437), we found that distance to the transplant center was not significantly associated with waitlisting status (OR 1.00, 95% CI 1.00–1.00, p = 0.938).

Discussion

Disparities in LT for HCC have been identified in the literature but less is known about the determinants of these disparities or the barriers patients face in completing the complex HCC–LT pathway. Patients with HCC face unique challenges compared to those with end-stage liver diseases seeking LT. These include the need to expedite and coordinate care to prevent cancer progression while completing evaluation testing, managing psychological and emotional stressors related to the cancer diagnosis, and contending with their own SDOH. In addition to cancer progression and becoming too sick for LT, we identified potentially modifiable social risk factors that hinder progression through the care pathway. A quarter of the cohort failed to complete the pathway due to social reasons. Additionally, SDOH were associated with failure to waitlist or death during LT evaluation. Several predictors of waitlisting and death, such as MELD score, AFP levels, and address, are routinely captured in the EHR at the time of referral, highlighting an opportunity for future research to develop validated clinical risk stratification tools to identify high-risk patients earlier in the HCC–LT pathway.

In our multivariable models, marital status, insurance type, and area-level deprivation were significantly associated with lower odds of waitlisting and higher risk of death during LT evaluation. Furthermore, in our sensitivity analyses, which excluded sicker patients, area deprivation and marital status became more prominent predictors of failure to waitlist and mortality. Race, insurance type, and area-level poverty were previously associated with referral and completion of LT evaluation in a cohort of patients with HCC in Georgia [15]. Furthermore, we found similar associations with failure to complete evaluation and death in our cohort exploring both ESLD and HCC patients referred for LT. [7] This suggests that insurance type and area-level deprivation are important attributes of high-risk phenotypes that may transcend region.

The impact of SES factors on healthcare outcomes may differ in countries with a publicly funded healthcare system, where cancer diagnoses are recorded within the national healthcare registry. For instance, in contrast to our findings, a European study examining the relationship between the European Deprivation Index (EDI), hepatocellular carcinoma (HCC) characteristics, and liver transplantation (LT) outcomes found that EDI did not significantly influence survival following LT for HCC. [19]

Our study is the first to describe reasons for failure to complete each step of the HCC–LT pathway. Here, 26% of the cohort failed at some step in the LT care pathway due to potentially modifiable social factors, including disinterest and fear. Pursuing LT for HCC, which is often asymptomatic, may be different than seeking LT for ESLD. Fear and anxiety are well-documented emotions associated with other cancer types [20]. Interventions to mitigate this fear have included providing patients with detailed information about their treatment, what to expect, how to manage side effects early in their disease course, and counseling and support groups to overcome stress and fear associated with treatment. [21]

Additional social reasons, including adverse SDOH, non-adherence, and active substance use, were also prominent reasons for early dropout. There is a growing body of literature on social need interventions in medicine that contribute to improved adherence, treatment uptake, and outcomes [22]. The evidence on health-related social need interventions addressing food insecurity and transportation challenges among patients with cancer is the most robust, with clinical trial evidence of food voucher plus pantry being an effective intervention to improve treatment completion [23,24]. Transplant patients who receive adequate social support often experience better adherence to treatment protocols and improved overall outcomes. This includes increased quality of life and better management of post-transplant care [25]. Building on this body of literature, specific social need interventions for patients with HCC pursuing LT should be tested and evaluated.

Importantly, the largest portion of the cohort dropped out between evaluation start and evaluation completion. This dropout was largely driven by social reasons (34%), which was the second most common cause. Social reasons were also the largest contributor to dropout between referral and evaluation start. These findings suggest that interventions to address social drivers of attrition may be most impactful when implemented early—at the time of diagnosis or referral.

Notably, disease severity factors, such as cancer progression and becoming too sick for LT, accounted for 39% of the reasons patients failed to complete the HCC–LT care pathway. Individuals most likely to be in this group were predominantly women and those with public insurance. These data suggest a potential relationship between these dropout reasons and underlying social determinants. The socioecological model of disease, along with the framework presented here, suggests that upstream social factors likely overlap and contribute to delayed diagnosis [7,26]. In other datasets, insurance type has been associated with late-stage diagnosis in multiple cancers, including hepatocellular carcinoma (HCC). Identifying this at-risk group early may help expedite and coordinate cancer treatments for these patients. Additionally, earlier disease identification, coupled with more robust screening uptake, is crucial in uninsured populations [27]. Finally, the impact of frailty on women with liver disease is increasingly recognized. As data on interventions to improve frailty continue to emerge, incorporating nutritional and physical exercise regimens into this cohort's care plan should be considered to reduce the risk of frailty. [28,29]

To our knowledge, this is one of the first studies to evaluate reasons for failure to complete each step in the LT care pathway for HCC, building on previous work in the field exploring SDOH as risk factors for failure of waitlisting and death during LT evaluation. The study identifies factors associated with failed waitlisting and death that might be detected using routinely available clinical data, even in the absence of detailed SDOH or social needs screening. Importantly, the largest portion of the cohort dropped out between evaluation start and evaluation completion, with social reasons being the second most common cause, accounting for 34%. We identify social and disease-related factors associated with failure to progress through the LT care pathway that may be promising targets for future interventions. This was a single-center study, and thus, findings may be influenced by center-specific factors such as referral patterns, evaluation protocols, or institutional policies. Other centers may face different structural or systemic barriers, and validation in diverse settings is warranted. However, we believe our study provides a robust framework for studying these barriers and identifies many social and disease severity-related factors that are likely transferable. In addition, future work should aim to replicate and expand these findings in multi-center or national datasets. While national transplant registries such as UNOS currently capture data beginning only at the time of waitlisting, emerging data infrastructures such as the CHART (Consortium for Organ Transplant Research) initiative may offer new opportunities to capture longitudinal, pre-waitlist data.³⁰ In addition, our study population does not reflect the racial demographics of our catchment area; racial and ethnic minorities have been shown in other cancer types to face unique barriers to accessing cancer treatment. Therefore, more diverse cohorts may identify additional barriers to HCC–LT pathway completion. Notably, where our cohort lacks racial diversity, it provides a unique opportunity to explore these factors in a predominantly rural, Medicaid-insured population within a single transplant center. Finally, our categorization of dropout reasons relied on provider documentation. While detailed conversations were often recorded, it is possible that transplant coordinators may have mischaracterized patients' barriers. Qualitative work to understand these challenges from the patient perspective is warranted.

In conclusion, our study highlights the multifaceted barriers that patients with HCC face in completing the LT care pathway, including both social and disease severity-related factors. By identifying social risk factors and disease severity-related factors, we underscore the need for targeted interventions. Our findings suggest that early identification of high-risk phenotypes using available EHR data could facilitate timely and personalized interventions. Interventions to reduce fear, anxiety, and support adverse SDOH are warranted. Future research should focus on implementing and evaluating specific interventions to address these identified challenges and improve equitable access to LT for HCC.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

HCC

Hepatocellular carcinoma

LT

Liver transplantation

ADI

Area deprivation index

SDOH

Social determinants of health

MELD

Model for end-stage liver disease

ACS

American community survey

LTFU

Lost to follow-up

SES

Socioeconomic status

ALD

Alcohol-related liver disease

MASLD

Metabolic dysfunction-associated steatotic liver disease

HCV

Hepatitis C virus associated

HBV

Hepatitis B virus associated

AIH

Autoimmune hepatitis

PBC

Primary biliary cholangitis

PSC

Primary sclerosing cholangitis

BCLC

Barcelona clinic liver cancer

Author Contributions

B.B. wrote the main manuscript text, assisted with data collection, and with table/figure preparation. T. A. assisted in writing parts of the manuscript, data collection, and with figure preparation. M. P. assisted with data collection. A.C. performed statistical analysis of the data and assisted in table/figure preparation for publication. C. S. assisted with data collection. L.N. was the main PI for this study and wrote the main manuscript text, assisted with data collection, and with table/figure preparation. All authors reviewed the manuscript.

Funding

This study was funded by the National Institute on Minority Health and Health Disparities, NMIHD 1K23MD018090-01.

Data Availability

No datasets were generated or analyzed during the current study.

Declarations

Conflict of interest

The authors declare no competing interests.