Social and clinical determinants of liver transplant evaluation in hepatocellular carcinoma: A multicenter cohort study.
Oct 23, 2026DOI: 10.1097/hc9.0000000000001068Journal ArticleHepatology Communications
Written by: Monica TincopaLana BridiIrine VodkinMignote YilmaKali ZhouAllison KwongJasleen SinghNeil MehtaAlexander KuoRyutaro HiroseSteven A WiselJustin A SteggerdaChris FreiseVeeral Ajmera
INTRODUCTION
Hepatocellular carcinoma (HCC) is the third-leading cause of cancer-related death globally.1 In the United States, HCC is a growing concern with an increasing incidence and mortality over the last 2 decades.2 Liver transplantation (LT) is a curative therapy for HCC that is stage T2 or downstaged and provides excellent long-term survival.3–5 Given that eligibility for LT depends on the burden of HCC, it is essential to receive a timely referral for LT evaluation after HCC diagnosis to decrease the risk of ineligibility due to HCC progression. As HCC is a rapidly growing indication for LT, it is critical to characterize the factors impacting access to LT to optimize clinical outcomes. Prior studies have highlighted differential access to LT by insurance type, race, and other demographics, including sex and neighborhood poverty index.6–11 While disparities in rates of LT among individuals waitlisted for LT are well documented, factors impacting more proximal portions of the LT care cascade, including referral for LT and LT evaluation, remain poorly characterized. Much of the existing literature is comprised of analyses from large secondary databases such as the United Network for Organ Sharing (UNOS) that lack a more detailed assessment of relevant social determinants of health (SDOH) factors or evaluation of single-center data with limited generalizability.11–14 Factors impacting access to LT for patients with HCC have also substantially changed over time, given the changing patient characteristics of individuals in need of LT, with alcohol-associated liver disease (ALD) and metabolic dysfunction–associated steatohepatitis (MASH) now being the primary etiologies for LT. Insight into specific barriers within each step of the LT care cascade among a contemporary cohort of patients is imperative for improving LT access for HCC patients.To address this important knowledge gap, herein we utilize the California Liver Network (CLN), a multicenter research collaborative comprised of 6 large-volume liver transplant centers accounting for over 75% of all LT volume in California. The CLN consortium data includes comprehensive, granular data collection encompassing the time of LT referral to post-transplant outcomes. The aim of this study was to identify social, system, and clinical factors associated with listing, LT, mortality, HCC stage at referral, and time from HCC diagnosis to referral.
METHODS
The CLN is a multicenter consortium of 6 adult LT centers in California. This includes the University of California San Francisco, University of California Los Angeles, University of California San Diego, University of Southern California, Stanford, and Cedars-Sinai academic medical centers. CLN centers are large academic medical centers, but they vary in annual LT volume and waitlist dynamics. Over the study period (2018–2020), centers ranged from higher-volume programs performing ~150–200 liver transplants annually (UCLA, UCSF) to moderate-volume programs performing ~75–130 annually (USC, Cedars-Sinai, Stanford). UCSD had a lower volume (35–50 annually) at the start of the study period, but had substantial growth in volume since study initiation. All centers operate within UNOS Region 5, which is characterized by high organ competition and elevated median MELD at transplant relative to national averages. Any adult (age 18 or older) referred for LT at one of these centers was included. Data collection began in 2018 using a standardized REDcap-based protocol with trained research coordinators under the guidance of a designated transplant hepatologist at each site. For this study, any participant with a diagnosis of HCC was eligible for inclusion. The diagnosis of HCC was based on LIRADS criteria and/or biopsy, with staging according to UNOS criteria using the most recent imaging at the time of LT referral.15 Longer study period duration was prohibited by the substantial cohort size (~10,000 individuals) and extensive data pull for each patient, given the comprehensive detail abstracted for broad categories of variables of interest. A referral was defined as receipt of a formal LT evaluation request at 1 of the 6 CLN centers, including both internal (within the same health system) and external provider referrals. All referrals were logged in the transplant program’s evaluation database. An evaluation was defined as completion of at least the initial hepatology transplant evaluation appointment.
Exposures of interest included social determinants of health (SDOH) variables, system-level factors, demographics, and clinical characteristics. SDOH variables included documentation status [US citizen, non-US citizen (undocumented), or VISA], primary language, foreign-born status, and need to relocate for transplant. These variables are not available in UNOS/SRTR and represent key novel contributions of the CLN dataset. Insurance type, education level, income, and marital status were also abstracted. The system-level exposure was the LT evaluation site (de-identified). Clinical exposures included MELD score, HCC stage at referral, alpha-fetoprotein (AFP), liver disease etiology, medical comorbidities, prior HCC treatment type, and time from HCC diagnosis to LT referral.
All SDOH variables were abstracted by trained research coordinators using a standardized REDCap protocol from structured psychosocial intake assessments and pre-transplant social work evaluations completed as part of standard care at each center, supplemented by chart review where structured fields were incomplete. Some SDOH variables were not available for all patients, reflecting variation in center-level documentation practices during the study period. Insurance type was categorized using the primary insurer listed at the time of evaluation; patients with dual coverage (eg, Medicare plus Medicaid) were categorized by primary insurer. Variables with more than 5% missing data are noted with sample sizes in Table 1. Outcomes of evaluation, listing, transplantation, and mortality were abstracted. The study was approved by the ethics and/or institutional review committees for each participating center.
Patient characteristics of adults with HCC referred for LT
Note: N included for any variable in which 5% or more of the data are unavailable.
Abbreviations: AFP, alpha-fetoprotein; AIH, autoimmune hepatitis; ALD, alcohol-associated liver disease; BMI, body mass index; CAD, coronary artery disease; DM, diabetes mellitus; ETOH, alcohol use; GED, General Educational Development; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HLD, hyperlipidemia; HTN, hypertension; IT, immunotherapy; MASH, metabolic dysfunction–associated steatohepatitis; MELD, Model for End-Stage Liver Disease; PBC, primary biliary cholangitis; PSC, primary sclerosing cholangitis; PVT, portal vein thrombosis; SBRT, stereotactic body radiation therapy; SDOH, social determinants of health; T1, one lesion <2 cm; T2, within Milan criteria; UNOS, United Network for Organ Sharing; Y90, yttrium-90 radioembolization.
The primary aim of this study was to characterize factors associated with progressions through the post-referral LT care cascade for patients with HCC. Outcomes examined included: (1) listing for LT; (2) receipt of LT; (3) overall mortality; (4) HCC stage at time of LT referral (ordinal); and (5) time from HCC diagnosis to LT referral (continuous in days). Reasons for failure to progress through the LT care cascade were characterized, including reasons for decline for LT listing and reasons for LT waitlist removal. Cause of death and time from referral to death were also captured as part of the overall LT care cascade.
HCC stage was categorized using UNOS criteria: T1 (one lesion <2 cm), T2 (within Milan), UNOS downstaging, beyond UNOS downstaging within all comers criteria, macrovascular invasion, or extrahepatic disease.
For patient characteristics, summative descriptive statistics were performed for categorical variables expressed as percentages and continuous variables to provide medians with associated interquartile ranges (IQRs). Univariable and multivariable logistic regression were used to calculate odds ratios (ORs) and corresponding 95% confidence intervals (CIs) to identify factors associated with listing for LT, LT, and overall mortality. To assess the association of individual variables with increasing HCC stage, univariate and multivariate ordinal logistic regression were performed. Ordinal logistic regression was selected, given that HCC stages are categorical and represent a natural rank order of related categories that represent monotonic worsening of disease severity and LT eligibility, but the categories are not equally spaced. Ordinal logistic regression allows the model to assess increasing disease severity across the full spectrum of HCC without inflating type I error. Linear regression was used to assess factors associated with time from HCC diagnosis to referral. Our primary analyses were guided by an a priori conceptual framework—specified before model fitting—treating SDOH, clinical severity, and site as distinct domains. Candidate variables were selected for inclusion in multivariable models based on: biologic plausibility, existing literature on determinants of LT access and outcomes in HCC, and degree of statistical significance in univariable analyses accounting for power in outcomes to prevent model overfitting. Variance inflation factors (VIFs) were assessed for primary multivariable models using a linear probability model with identical covariates to each logistic regression model, as STATA’s VIF command is only available following linear regression. This approach is methodologically appropriate as VIF is a property of the predictor matrix and is unaffected by the choice of outcome model. In instances where a variable with substantial missingness was significant on univariate analysis, the primary multivariate analysis was performed excluding this variable, given a reduction in sample size. A separate subanalysis was then provided in Supplemental Materials, https://links.lww.com/HC9/C510. This approach, rather than imputation for missing data, was applied, given the strong assumptions that would be required with imputation for variables with a substantial degree of missingness. A p-value <0.05 was considered statistically significant. All statistical analyses were conducted in STATA software.
Ethical approval for this study was provided by the Institutional Review Board at each participating center. A data use agreement was established and agreed upon between all centers to establish a multicenter registry, with the University of California San Francisco serving as the organizing center. Data were collected using a HIPAA-compliant RedCap system. All research was conducted in accordance with both the Declarations of Helsinki and Istanbul. AI tools were used in the revision process of the manuscript to improve readability and conciseness of the submission.
This study was approved by the Institutional Review Board at each participating center: Cedars-Sinai Medical Center (STUDY00002483), University of Southern California (HS-21-00570), Stanford University (40344), University of California Los Angeles (22-001661), University of California San Francisco (20-32625), and University of California San Diego (201014). A waiver of informed consent was granted by each Institutional Review Board given the retrospective nature of data abstraction from the medical record. A data use agreement was established among all centers to create a multicenter registry, with the University of California San Francisco serving as the organizing center. Data were collected using a HIPAA-compliant REDCap system. All research was conducted in accordance with both the Declarations of Helsinki and Istanbul.
RESULTS
The CLN comprises 9987 adult patients referred for LT evaluation. Of these 9987 patients, a total of 1573 (15.7%) patients had HCC and were thus eligible for inclusion in this study. The baseline characteristics of the 1573 patients with HCC are outlined in Table 1. The median age was 63.3 (IQR 58.4–67.7) years, 72.8% male, 56.1% White, and 40.1% Hispanic. The top etiologies of liver disease include hepatitis C virus (HCV) (34.8%), MASH/cryptogenic (19.1%), and ALD (18.4%). The median biologic Model for End-Stage Liver Disease (MELD) score at time of evaluation was 12.7 (IQR 8–16), and median body mass index (BMI) at evaluation was 27.9 (IQR 24.8–31.1). Insurance was 33.4% private, 39.7% Medicare, and 26.1% Medicaid. English was not the primary language in 31.2% of patients. The median time from HCC diagnosis to referral was 124 days (IQR 32–410). HCC staging distribution at time of referral is outlined in Table 1. HCC treatment before LT referral was received in 55.7% (N=876) patients, with some patients receiving multiple types of HCC treatment before referral. The median AFP at time of referral was 8 ng/mL (IQR 4–33). Other demographic, medical comorbidities, liver disease, and psychosocial characteristics are outlined in Table 1.
The overall HCC care cascade for patients referred for LT evaluation with HCC is displayed in Figure 1. Of the 1573 patients with HCC referred for LT, 1565 (99%) were evaluated, with 1552 (98.6%) completing their evaluation. The median time from referral to evaluation was 36 days (IQR 19–61). The median time from referral to selection was 110 days (IQR 50–241.5). Of all referred (N=1573), 892 (56.7%) were listed, 510 (32.4%) were transplanted, and 420 (26.7%) died. Of those who completed the evaluation, 892 (57%) were listed for LT. The top reasons for decline were medical/surgical contraindication or too sick (including advanced HCC stage) (N=266, 43.9%), substance use or psychosocial contraindications (N=199, 32.8%), and patient choice (N=72, 11.9%). The median time from referral to listing was 104 days (IQR 52–198). The median MELD at listing was 10 (IQR 8–14), with 526 (65.7%) receiving MELD exception points with a median MELD exception score of 29 (IQR 27–30) and time to MELD exception points of 283 days (IQR 196–443) (Table 2). Of those listed for LT, 510 (57.1%) were transplanted, with 321 (35.9%) removed from the LT waitlist, and the remaining 61 patients were active on the LT waitlist at the end of the study period. Unadjusted listing and transplantation rates by LT site are shown in Figure 2. The primary indications for waitlist removal were medical/surgical contraindication or too sick (N=183, 57%), and loss to follow-up or other/unknown (N=79, 24.6%). The median time from referral to waitlist removal was 522 days (IQR 273–927) and median time from referral to LT was 457 days (IQR 304–675) (Figure 1). In the overall HCC cohort, 420 (N=26.8%) died with a median time from referral to mortality of 526 days (254–928) and a median time to last follow-up of 1189 days (IQR 385–1737) (Table 2). Leading causes of death are outlined in Table 2.
Referral cascade and outcomes for patients with hepatocellular carcinoma referred for liver transplant. Time to event is presented as median (IQR) in days. Abbreviations: IQR, interquartile range; LT, liver transplant.
Liver transplant evaluation and clinical outcomes for HCC patients
Abbreviations: CAD, coronary artery disease; CVA, cerebrovascular accident; DDLT, deceased donor liver transplantation; ESLD, end-stage liver disease; LDLT, living donor liver transplantation; MELD, Model for End-Stage Liver Disease; MOF, multiorgan failure; MVA, motor vehicle accident; SLK, simultaneous liver-kidney transplant.
Unadjusted listing and transplantation rates by liver transplant evaluation site.
Factors associated with listing for LT on univariable analysis included male sex, race, MELD score at evaluation, liver disease etiology, history of hyperlipidemia (HLD) and coronary artery disease (CAD), AFP at evaluation, HCC stage at time of referral, history of prior HCC treatment, type of prior HCC treatment received, LT evaluation site, insurance type, documentation status and marital status (Supplemental Table S1, https://links.lww.com/HC9/C510). On multivariable analysis, AFP at evaluation (OR 0.99, 95% CI 0.99–0.99, p<0.001), HCC stage, LT evaluation site, Medicaid insurance (OR 0.55, 95% CI 0.39–0.77, p=0.001), and being married/having a domestic partner (OR 1.90, 95% CI 1.32–2.71, p<0.001) were independently associated with odds of listing for LT (Figure 3A). Due to the restricted number of patients with documentation status data, multivariate analysis, including those with documentation status available, is provided in Supplemental Table S2, https://links.lww.com/HC9/C510. Factors associated with receipt of LT on univariable analysis included history of diabetes, HCC staging, LT evaluation site, and marital status (Supplemental Table S1, https://links.lww.com/HC9/C510). On multivariable analysis, only more advanced HCC stage (OR 0.42, 95% CI 0.20–0.87, p=0.02) and LT evaluation site (site 1 OR 0.46, 95% CI 0.31–0.69, p<0.001; site 4 OR 1.78, 95% CI 1.01–3.14, p=0.04) were independently associated with odds of receipt of LT (Figure 3B). Overall mortality was associated with race, Hispanic ethnicity, MELD score at evaluation, liver disease etiology, history of diabetes, hypertension (HTN) and chronic dialysis, presence of ascites, hepatic encephalopathy (HE), portal vein thrombosis (PVT), history of prior resection for HCC, insurance type, formal education level, and income (Supplemental Table S1, https://links.lww.com/HC9/C510). Factors independently associated with odds of overall mortality included MELD score at evaluation (OR 1.02, 95% CI 1.01–1.05, p=0.008), history of HTN (OR 1.44, 95% CI 1.08–1.91, p=0.01), chronic dialysis (OR 2.48, 95% CI 1.09–5.64, p=0.03), HE (OR 1.54, 95% CI 1.12–2.11, p=0.007), and PVT (OR 1.44, 95% CI 1.01–2.06, p=0.04) (Figure 3C).
Forest plot of multivariate predictors of (A) listing; (B) transplantation; and (C) overall mortality. Reference categories: White race, MASH/cryptogenic; T1 HCC stage; private insurance; <high school education; no income; single/widowed. AFP, alpha-fetoprotein; AIH, autoimmune hepatitis; ALD, alcohol-associated liver disease; CAD, coronary artery disease; CI, confidence interval; GED, General Educational Development; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; LT, liver transplantation; MASH, metabolic dysfunction–associated steatohepatitis; MELD, Model for End-Stage Liver Disease; PBC, primary biliary cholangitis; PSC, primary sclerosing cholangitis; T1, one lesion <2 cm; T2, within Milan criteria; Y90, yttrium-90 radioembolization.
Formal collinearity diagnostics demonstrated no evidence of meaningful multicollinearity in the multivariable models. Mean VIF values were 1.59 (listing), 1.37 (transplantation), and 1.60 (mortality), with no individual covariate exceeding a VIF of 3.18 across any model, well below conventional thresholds of concern (VIF >5). SDOH variables demonstrated uniformly low VIF values across models: insurance type (VIF 1.44–1.50), marital status (VIF 1.94–2.10), education level (VIF 2.18–2.46), and income source (VIF 2.10–2.87), confirming that intercorrelations among SDOH variables did not materially influence model estimates or the interpretation of individual variable contributions.
Factors associated with an increasing HCC stage are detailed in Table 3. On univariable analysis, medical factors associated with more advanced stages of HCC at time of LT referral included: no prior history of HE, higher AFP at time of evaluation, and type of prior HCC therapy at time of evaluation. Sociodemographic factors associated with more advanced stages of HCC at time of LT referral on univariable analysis included: male sex, Medicaid and other insurance, those with non-US citizen/resident VISA status, language other than English as a primary language, and less than high school education. On multivariable analysis, male sex (OR 1.55, 95% CI 1.18–2.05, p=0.002), higher AFP (OR 1.01, 95% CI 1.01–1.01, p <0.001), type of prior HCC treatment, Medicaid insurance (OR 1.40, 95% CI 1.02–1.95, p=0.04), and English as primary language (OR 0.70, 95% CI 0.52–0.95, p=0.02) were independently associated with more advanced HCC stage at time of LT referral (Table 3). Factors associated with time from HCC diagnosis to referral for LT are shown in Table 4. Liver disease etiology, receipt of prior HCC treatment, LT evaluation site, and documentation status were associated with time to referral for LT on univariable analysis. On multivariable analysis, having dual HCV/ALD etiology (coefficient 309.7, p=0.002), and type of prior HCC treatment were independently associated with longer duration from HCC diagnosis to referral for LT (Table 4). Multivariate analysis for more advanced HCC stage and time from diagnosis to referral, including those with documentation status available, is provided in Supplemental Tables S3 and S4, https://links.lww.com/HC9/C510.
Predictors of more advanced HCC stage at time of LT referral
Abbreviations: AFP, alpha-fetoprotein; AIH, autoimmune hepatitis; ALD, alcohol-associated liver disease; BMI, body mass index; CAD, coronary artery disease; CI, confidence interval; DM, diabetes mellitus; ETOH, alcohol use; GED, General Educational Development; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HLD, hyperlipidemia; HS, high school; HTN, hypertension; IT, immunotherapy; MELD, Model for End-Stage Liver Disease; PBC, primary biliary cholangitis; PSC, primary sclerosing cholangitis; PVT, portal vein thrombosis; ref, reference; SBRT, stereotactic body radiation therapy; SDOH, social determinants of health.
DISCUSSION
In this diverse, multicenter consortium of 6 adult LT centers encompassing nearly 10,000 individuals with detailed data on psychosocial and SDOH factors, we characterized predictors independently associated with progression through each phase of the LT care cascade for adults with HCC starting at the time of referral. Among the 1573 patients with HCC, we highlight the impact of both clinical (more advanced HCC, higher AFP, disease etiology), socioeconomic (Medicaid insurance, non-English primary language), psychosocial factors (marital status), along with system factors (LT evaluation site) on progression through the post-referral LT care cascade. We identified that the main drop-off in the HCC care cascade occurred at selection for waitlisting, with just over half of referrals approved for listing. Our work further outlines potential targets for interventions to improve access to LT for HCC with a focus on mechanisms to optimize timely referral, given that the primary reason for declining listing was being too ill/advanced stage. Dual HCV/ALD etiology was associated with delayed referral, underscoring the potential for ancillary social and substance use interventions to facilitate LT access.
Our findings build on existing literature by examining a robust, large, multicenter consortium with detailed data from the time of LT referral through post-LT outcomes. In particular, our work was able to examine crucial SDOH factors that are missing from large secondary databases (documentation status, primary language, foreign-born status, and need to relocate). The diverse nature of our centers also provides an opportunity to examine heterogeneous patient populations that overcome limitations of existing single-center studies. One such study of individuals in Georgia found a lower likelihood of LT referral and completion of LT evaluation among those with Medicaid, though this study was limited to individuals with early-stage HCC.16 Evaluation of data of referral for surgical treatment of HCC in Pennsylvania similarly found a negative association of Medicaid insurance and likelihood of referral.17 A single-center study of 705 patients waitlisted for LT with HCC found patients with public insurance had lower rates of LT and higher rates of waitlist drop out.18 A contemporary single-center study similarly identified public insurance and unmarried status as independent barriers to waitlisting.19 In our study, Medicaid insurance was independently associated with lower odds of listing (OR 0.55) and more advanced HCC stage at referral (OR 1.40). Our findings are consistent with prior studies and suggest that insurance-related barriers influence both disease burden at presentation and likelihood of listing once referred.
These data elucidate key areas to target to optimize access across the HCC care cascade, in particular, the clinical implications of late referral for LT for individuals with more advanced HCC and those with more complex medical and surgical comorbidities. These characteristics account for the highest percentages of reasons for LT decline and LT waitlist removal. Beyond markers of HCC stage and severity, Medicaid insurance, LT site, and marital status were independently associated with odds of listing, underscoring the role of both socioeconomic and system-level factors in determining cascade progression independent of clinical severity.20 Marital status in the transplant candidacy context likely captures a composite construct encompassing social support availability. While social support is a clinically relevant and guideline-endorsed consideration for transplant candidacy, such criteria warrant careful scrutiny to ensure they do not disproportionately disadvantage patients from non-traditional households. Prior work has shown that racial and ethnic minority patients are disproportionately declined for waitlisting for psychosocial reasons independent of medical contraindications, underscoring the need for equitable and transparent evaluation criteria.21 The reason for differences by LT evaluation site remains unclear, as the multivariable model accounted for demographics, MELD, disease etiology, medical comorbidities, HCC stage, AFP, insurance, prior HCC treatment, and other psychosocial factors to evaluate odds of listing. Variable listing practices may reflect different risk thresholds by center as well as differential psychosocial evaluations, though further analyses to better understand these differences are needed. From a patient perspective, these findings emphasize the potential benefit of being evaluated at multiple LT centers when possible, given the impact of site variation. Several system-level interventions could address site variation without placing the burden on patients: (1) OPTN/UNOS requirement for transparent public reporting of HCC-specific listing rates and decline reasons by center; (2) development of standardized minimum criteria for HCC LT candidacy evaluation to reduce threshold variation across programs; (3) education of referring hepatologists and oncologists about center-specific listing practices to inform referral decisions; and (4) regional multidisciplinary HCC tumor boards to promote standardized candidacy evaluation. Of note, the LT site was also independently predictive of odds of transplantation, which is likely a result of differences in odds of listing. Factors associated with more advanced HCC at the time of referral and longer time from HCC diagnosis to referral underscore the impact of prior HCC care on perceived need for or eligibility for LT by referring providers. Individuals with prior locoregional therapy had more advanced HCC at the time of referral and longer time to referral, indicating that referring providers may wait to refer until initial treatments fail or delay referral until after HCC treatment has already been initiated. Finding the optimal balance between initiating locoregional therapy and referring for LT consideration for HCC for referring providers represents an area for future research to ensure timely treatment is initiated without losing a potential window for LT when needed. Notably, Medicaid insurance and non-English primary language were independently associated with more advanced HCC stage at referral, suggesting that socioeconomic and language barriers may delay HCC diagnosis or reduce engagement with surveillance programs prior to referral. These findings identify specific high-risk groups who may benefit from targeted outreach and navigation support to facilitate earlier referral.
A key strength of this study is the detailed assessment of system-level, SDOH, and clinical factors captured from the time of referral across 6 diverse LT centers, including variables unavailable in other large secondary databases. The demographic diversity of this cohort, including 40% Hispanic and 18% Asian patients, reflects the California LT catchment area and represents populations underrepresented in prior LT disparities research. Several limitations should be considered when interpreting these findings, however. The demographic composition of this cohort differs from national LT populations, particularly with respect to Black patient representation (3.3%). Findings, therefore, may not be directly generalizable to transplant centers serving populations with different demographic profiles. It is important to acknowledge transplantation rates and median MELD scores at transplant differences in California compared with other geographic regions. California is in UNOS Region 5, which historically has had lower LT rates per patient per year and higher median MELD at transplant, reflecting a high-acuity, competitive organ allocation environment. Transplantation rates may therefore underestimate progression through the HCC care cascade in lower-MELD regions. These differences impact rates of LT but not patterns in LT listing. The 99% evaluation completion rate in this study likely reflects the characteristics of this specific HCC cohort, wherein 56% had received prior HCC treatment, reflecting active engagement in specialty care. Additional limitations to note are the potential for incomplete data capture as part of retrospective chart abstraction. The absence of area-level deprivation metrics such as the Area Deprivation Index is also of note, as it precludes assessment of neighborhood-level structural factors. Importantly, this cohort does not capture pre-referral barriers or characterize patients with HCC never referred for LT.
Lastly, analyses of SDOH introduce potential for collinearity of variables. To address this, we performed formal VIF diagnostics, which demonstrated that multicollinearity did not account for the attenuation of SDOH associations observed in multivariable models. The attenuation of certain SDOH associations in the full multivariable model may reflect associations with clinical disease severity; however, formal mediation or sequential hierarchical modeling analyses were not performed, and the mechanism of attenuation cannot be determined from these analyses alone.
CONCLUSIONS
In this large, diverse multicenter cohort, progression through the LT care cascade for HCC was shaped by an interplay of clinical severity, prior HCC management, socioeconomic, psychosocial factors, and center-level variation. Delayed referral and referral at more advanced HCC stage were common and strongly associated with prior HCC treatment and disease characteristics, suggesting that transplant consideration is often deferred until later in the disease course. Medicaid insurance and non-English primary language were independently associated with more advanced HCC stage at referral, identifying socioeconomically vulnerable and linguistically isolated patients as high-priority targets for earlier engagement and navigation support. While clinical factors primarily drove mortality, Medicaid insurance, psychosocial characteristics, and transplant center were key determinants of listing and transplantation, highlighting potentially modifiable socioeconomic and system-level influences on access to potentially curative therapy. Together, these findings underscore the importance of timely referral, coordinated multilingual care pathways, and standardized center-level evaluation practices that integrate transplant evaluation earlier in HCC management.