Treatment Delays and Survival in Patients With Hepatocellular Carcinoma: A Multicenter Cohort Study

Abstract

INTRODUCTION:

Although treatment delays have been reported for several cancers, fewer studies have described the prevalence, contributing factors, and impact on survival in patients with hepatocellular carcinoma (HCC). We aimed to evaluate these aspects in a multicenter cohort.

METHODS:

We conducted a retrospective cohort study of patients diagnosed with HCC from 2010 to 2018 across 4 US health systems. Treatment delay was defined as the first treatment >90 days after HCC diagnosis. Univariable and multivariable logistic regression analyses were performed to identify factors associated with treatment delays. The association between treatment delay and overall survival was evaluated using multivariable Cox regression analysis.

RESULTS:

Among 1,561 eligible patients (median age 61.7 years, 75.0% men, 65.5% Barcelona Clinic Liver Cancer [BCLC] stage 0/A), treatment delays occurred in 357 (22.9%) patients. Treatment delays were most frequent with BCLC stage 0 (31.1%), followed by stage A (24.3%), stage B (22.6%), and stage C (13.6%). Men (OR 0.68, 95% CI 0.51–0.91) and patients with BCLC stage C HCC (vs stage A: OR 0.40, 95% CI 0.25–0.63) were less likely to experience delays, whereas those with BCLC stage 0 (vs stage A: OR 1.56, 95% CI 1.06–2.29) were more likely to experience delays. In a 6-month landmark analysis, survival did not differ between delayed vs timely treatment (HR 1.09, 95% CI 0.87–1.37). Median survival from the 6-month landmark was 12.1 months (IQR: 5.5–23.3) and 14.6 months (IQR: 5.1–30.0) for the treatment delay and timely treatment groups, respectively.

DISCUSSION:

Treatment delays exceeding 90 days occurred in over one-fifth of patients but were not associated with significant differences in survival.

INTRODUCTION

Hepatocellular carcinoma (HCC) accounts for approximately 80% of primary liver cancers and is the third leading cause of cancer-related mortality worldwide (1). Prognosis varies widely by stage, with 5-year survival rates below 10% for advanced-stage disease compared with over 60% for patients diagnosed at early stages who undergo curative therapy (2–4). This stark contrast in prognosis underscores the importance of semi-annual surveillance in at-risk populations. Although HCC surveillance is associated with earlier detection, higher likelihood of receiving curative treatment, and improved survival outcomes, only in 1 in 4 at-risk patients undergo routine surveillance (5,6).

Failures across the HCC care continuum, including diagnostic delays, treatment delays, and underuse of curative treatments remain a significant concern (7). Indeed, nearly half of patients diagnosed with HCC do not receive any treatment (8). These process failures disproportionately affect Black and Hispanic patients and those of low socioeconomic status, highlighting the multifactorial nature of disparities rooted in demographic, clinical, and social determinants (8–10).

Beyond underuse of surveillance and treatment, delays in cancer diagnosis and treatment are also prevalent (11,12). In other malignancies such as colorectal and breast cancer, even short treatment delays (e.g., 4 weeks) have been associated with increased mortality (13). While fewer studies have evaluated treatment delays specifically in HCC, existing data suggest that such delays are common. Based on tumor doubling times in HCC (14), a delay of 3 months or more may allow for significant tumor progression, potentially reducing treatment effectiveness (15,16). However, data are discordant, with at least 1 prior cohort study reporting no association between treatment delays and overall survival (12). Our study aimed to evaluate the prevalence and contributing factors of treatment delays and to characterize the association with overall survival, using a multicenter US cohort of patients with HCC.

METHODS

Study population

We performed a retrospective multicenter cohort study including patients diagnosed with HCC between January 2010 and August 2018 across 4 large US medical centers: UT Southwestern Medical Center (UTSW) and Parkland Health (Parkland) in Texas and the University of Miami (U Miami) and Jackson Health System (Jackson) in Florida. As previously described (17), UTSW and U Miami are tertiary care referral centers, while Parkland and Jackson serve as safety-net health systems for Dallas County and Miami-Dade County, respectively. All 4 institutions have established multidisciplinary teams and liver tumor boards where HCC cases are reviewed before initiating treatment (18).

HCC was diagnosed according to the American Association for the Study of Liver Disease criteria and staged according to the Barcelona Clinic Liver Cancer (BCLC) staging system (3,4). For this analysis, we excluded patients who did not undergo treatment and those with BCLC stage D given limited treatment options outside of liver transplantation (LT) (4).

Data collection

Clinical and demographic information included sex, race, ethnicity, insurance status, socioeconomic status, liver disease etiology and severity, laboratory data, tumor characteristics, HCC treatment modality, time to first treatment, and overall survival.

Race and ethnicity were categorized into mutually exclusive groups: non-Hispanic White (White), non-Hispanic Black (Black), and Hispanic. Owing to the limited number of Asian and American Indian patients, these individuals were grouped into a combined “Asian/other” category for analysis. Insurance status was categorized as Medicare, Medicaid, private insurance, uninsured, other, or unknown. Socioeconomic status (SES) was evaluated using the Yost index (which incorporates neighborhood income, education, occupation, and related variables) and categorized into quintiles (Q1 = lowest SES, Q5 = highest SES) (19). Liver disease etiology was classified as hepatitis B, hepatitis C (HCV postsustained viremic response [SVR] and HCV viremic), alcohol associated liver disease (ALD), metabolic dysfunction associated steatotic liver disease, cryptogenic, or other. Patients with concurrent HCV and ALD were classified as HCV, and those with ALD and metabolic dysfunction–associated steatotic liver disease were classified as ALD. Liver dysfunction was assessed using the Child-Pugh classification, with ascites and hepatic encephalopathy categorized as none, mild or controlled, and severe or uncontrolled.

Tumor characteristics included the number of lesions, maximum tumor diameter, presence of vascular invasion, and/or distant metastasis. Tumor stage at time of diagnosis was classified using the BCLC staging system (stages 0, A, B, C, or D) based on tumor burden, Eastern Cooperative Oncology Group (ECOG) performance status, and liver dysfunction (4). Treatment modalities included LT, surgical resection, local ablation, transarterial chemoembolization (TACE), transarterial radioembolization (TARE), stereotactic body radiation therapy (SBRT), and systemic therapy (chemotherapy) (20). Locoregional therapies included local ablation, TACE, TARE, and SBRT. When patients underwent multiple treatments, they were classified according to their first treatment because it defined the initiation of their HCC treatment course.

Statistical analysis

Patient characteristics were compared using χ² tests for categorical variables. Our primary exposure variable was delayed treatment, and our primary outcome was overall survival. Based on existing literature, treatment delay was defined as receiving the first HCC treatment >90 days after HCC diagnosis (15,16). Date of diagnosis was based on imaging for those with characteristics findings and date of biopsy for those with indeterminate findings that required histological confirmation. We first characterized the proportion of patients with delayed vs timely treatment and then used univariable and multivariable logistic regression models to identify factors associated with treatment delays. We performed a post hoc analysis excluding patients with BCLC stage 0, given the positive predictive value of imaging findings for HCC and potential for physicians to monitor patients for development of T2 lesions that qualify for Model for End-Stage Liver Disease (MELD) exception points if considering LT (3).

We then evaluated the association between treatment delay and overall survival using a landmark analysis to account for immortal time bias. Overall survival was calculated from a landmark date of 180 days, with follow-up censored at last follow-up or transplant. In this analysis, timely treatment was defined as treatment initiated within 90 days of HCC diagnosis, while delayed treatment was defined as receiving treatment between 90 and 180 days. We also performed a sensitivity analysis using a 270-day landmark, with delayed treatment defined as those treated between 90 and 270 days. For both analyses, patients who died before the landmark and those who received HCC treatment after the landmark were excluded. Finally, multivariable Cox regression analysis was performed to identify any factors associated with overall survival, using the 180-day landmark. All statistical analyses were performed using R software (version 4.4.2).

RESULTS

Patient characteristics

Demographic and clinical characteristics of the 1,561 eligible patients, stratified by timely vs delayed treatment, are outlined in Table 1. In brief, the median age was 61.7 (IQR 56.7–67.7) years, and 1,171 (75.0%) were men. The cohort was diverse regarding race and ethnicity (39.5% White, 31.2% Hispanic, 22.8% Black, 6.5% Asian/other). Patients were equally distributed between Texas (52.9%) and Florida (47.1%). The most common liver disease etiologies were viremic HCV (39.8%) followed by post-SVR HCV (18.7%) and ALD (12.2%). Early-stage HCC, defined as BCLC stage 0 or A, was present in 65.5% of the patients, and 54.0% of the cohort was within the Milan criteria. Insurance coverage included Medicare (33.1%), private insurance (30.5%), Medicaid (15.6%), and uninsured (6.9%). The most common treatments were TACE (37.9%), local ablation (17.7%), surgical resection (16.8%), and systemic therapy (15.5%). Few patients (3.4%) underwent LT as their initial therapy, although an additional 12.1% underwent transplant after bridging therapy.

Table 1.

Patient characteristics, stratified by timely vs delayed treatment

Treatment delay

The median time to treatment was 50 days (IQR 31–85), with most patients (n = 1,204; 77.1%) receiving timely treatment within 90 days of HCC diagnosis; however, over one-fifth (n = 357; 22.9%) of patients experienced a treatment delay. Only 24.3% of patients received treatment within 30 days, whereas 58.0% received treatment within 60 days, 85.1% within 120 days and nearly all (92.7%) received treatment within 180 days.

In univariable analysis, treatment delays were associated with sex, race, liver disease etiology, insurance status, Child-Pugh score, tumor stage, and HCC treatment type. In multivariable analysis, patients with BCLC stage 0 were more likely to experience treatment delays (vs BCLC A: OR 1.56, 95% CI 1.06–2.29). Conversely, men (OR 0.68, 95% CI 0.51–0.91) and patients with BCLC stage C tumors (vs BCLC A: OR 0.40, 95% CI 0.25–0.63) were less likely to experience delays (Table 2). Compared with surgical resection, all other HCC treatment modalities were associated with significantly higher odds of treatment delay (ORs ranging from 3.1–5.5). Treatment delays were observed in 25.5% with BCLC stage 0/A tumors, 22.6% with stage B, and 13.6% of patients with stage C tumors. Similarly, treatment delays occurred in 9.2% of patients undergoing surgical resection, compared with 19.0% receiving systemic therapy and 26.2% receiving locoregional therapy (local ablation, TACE, TARE, or SBRT) (Figure 1). Results were similar in a sensitivity analysis excluding treatment type from the model (data not shown). In an exploratory analysis excluding patients with BCLC stage 0, men (OR 0.67, 95% CI 0.49–0.92) and patients with BCLC stage C (OR 0.39, 95% CI 0.24–0.61) continued to be less likely to experience treatment delays. Results were unchanged in a sensitivity analysis excluding patients who underwent LT.

Table 2.

Factors associated with treatment delays

Figure 1.

Distribution of treatment delays, overall and stratified by BCLC stage and treatment type. BCLC, Barcelona Clinic Liver Cancer.

Most patients who experienced treatment delays had BCLC stage 0/A HCC (73.1%). Within this subgroup, nonsurgical HCC treatment modalities were associated with higher odds of delay, while men were less likely to experience treatment delays (OR 0.68, 95%CI 0.49–0.96). Similarly, among the subgroup within Milan criteria, patients with tumors < 2 cm (OR 1.50, 95%CI 1.05–2.16), ECOG >1 (OR 1.46, 95%CI 1.00–2.12), and type of HCC treatment were independently associated with treatment delays.

Overall survival

Using a 6-month landmark analysis, univariable analysis demonstrated no significant difference in overall survival between patients receiving delayed vs timely treatment (HR 1.09, 95% CI: 0.87–1.37) (Figure 2). Given the crossover observed in the KM curves, proportional hazards assumption was tested and suggested nonproportional hazards (P-value <0.001). Accordingly, we estimated survival at 1, 3, and 5 years. Although 1-year survival was similar between the groups (timely treatment 70.7% vs delayed 75.8%), the difference widened over time, with 5-year survival estimates of 33.8% for timely treatment vs 12.9% for delayed treatment. Overall survival from the 6-month landmark was 14.6 months (IQR: 5.1–30.0) in patients with timely treatment vs 12.1 months (IQR: 5.5–23.3) with delayed treatment. In multivariable analysis, there continued to be no significant association between delayed treatment and overall survival (adjusted HR [aHR] 0.99 95% CI: 0.79–1.24) (Table 3). In addition to tumor stage, Child Pugh class, treatment type, and age older than 65 years were associated with increased mortality.

Figure 2.

Overall survival using 6-month landmark, stratified by timely vs delayed treatment.

Table 3.

Factors associated with overall survival using a 6-month landmark

In subgroup analyses of patients with early-stage HCC, delayed treatment was not associated with mortality when early stage was defined by BCLC 0/A (HR 1.31, 95%CI 0.99–1.75) (see Supplementary Figure 1A, Supplementary Digital Content, http://links.lww.com/AJG/D869), although it was associated with increased mortality in those within Milan criteria (HR 1.44, 95%CI 1.05–1.98) (see Supplementary Figure 1B, Supplementary Digital Content, http://links.lww.com/AJG/D869). Similarly, delayed treatment was not significantly associated with mortality in patients who underwent curative treatment (HR 1.32, 95%CI 0.81–2.14) or noncurative treatment (HR 0.84, 95%CI 0.65–1.09), as well as younger than 65 years (HR 1.13, 95%CI 0.85–1.51) or older (HR 1.01, 95%CI 0.69–1.48) patients. In subgroup analyses by treatment type, delayed treatment was not associated with survival among patients undergoing locoregional therapy (HR 0.95, 95%CI 0.74–1.22).

A sensitivity analysis excluding patients who underwent LT demonstrated results consistent with the primary analysis (Table 4). Similarly, overall survival was not significantly associated with treatment delay using alternative definitions of 60 days (HR 0.87, 95%CI 0.70–1.08) or 120 days (HR 0.95, 95%CI 0.76–1.18).

Table 4.

Factors associated with overall survival using a 6-month landmark approach, among patients who did not undergo liver transplantation

DISCUSSION

In this large and diverse cohort of patients with HCC, over 1 in 5 patients experienced treatment delays. Delays were associated with both tumor stage and treatment modality, with delays observed in over one-fourth of patients with BCLC stage 0/A HCC and those undergoing locoregional therapy. However, we failed to observe a significant association between treatment delays and overall survival in either the 6-month or 9-month landmark analyses.

The prevalence of treatment delays observed in our study aligns with existing literature among patients with HCC. Treatment delays are also common across other cancer types such as colorectal, breast, and lung cancer, with several studies reporting associations with worse survival (21–23). Previous studies have attributed treatment delays to a combination of provider, patient, and system-level factors (11,12). In other cancers, delays have been also attributed to financial barriers, limited education levels, and lack of awareness (24,25). Indeed, treatment delays can result from the complexity of HCC management, in which treatment decisions are an interplay of liver dysfunction, patient performance status, and tumor burden (4). The care continuum across multiple specialties including transplant hepatology, surgery, interventional radiology, medical oncology, and radiation oncology might further contribute to potential treatment delays. A meta-analysis underscored the importance of multidisciplinary approach for patients with HCC, demonstrating improved overall survival (26). Additional reported benefits include improved communication among providers, increased guideline-concordant care, and faster time to treatment initiation (18).

The lack of association between treatment delays and overall survival in our study is consistent with existing literature among patients with HCC (27). Indeed, similar findings have also been reported in other cancer types as well (12,27,28). Although we found this finding persisted in multivariable analysis adjusting to tumor stage and type of HCC treatment, the lack of association may still be due to selection bias, in which patients with indolent tumors are more likely to be monitored and those with concerns of more aggressive tumor biology or worse prognosis are expedited for treatment (14,29). On an individual patient basis, treatment delays can have notable impact on treatment eligibility and prognosis. For example, complete response rates after local ablation are highly correlated with tumor size, with the best responses observed in tumors smaller than 3 cm (30). Similarly, risk of microvascular invasion and recurrence after surgical resection increases with larger tumor sizes (31). Furthermore, there may be other negative consequences of treatment delays including prolonged psychological distress associated with the new HCC diagnosis (32,33). Future research should continue to comprehensively evaluate the consequences of treatment delays to inform optimal care delivery models that are patient centered. Structured care navigation programs and early referral pathways may facilitate earlier evaluations, treatment initiation, and improved survival (34–36).

Our findings must be interpreted in the context of the study's limitations. First, given the retrospective nature of our study, there is the potential for measurement error and unmeasured confounders. As discussed above, a risk of channeling bias for which patients may be monitored vs treated expeditiously can affect the analyses examining overall survival. Unfortunately, reasons for treatment delay were not captured and can be difficult to determine retrospectively. Second, we evaluated landmarks of 6 and 9 months based on tumor doubling time, but there may be an alternative landmark that is more clinically relevant. Third, we excluded patients who did not undergo treatment, which may be interpreted as the most extreme form of treatment delay. In a post hoc analysis using the 6-month landmark, those with delayed or no treatment had significantly worse survival than those with timely treatment. Finally, our study was conducted across 2 tertiary care centers and 2 safety-net hospitals in distinct geographical areas, but the results may not be generalizable to other areas or health systems. We feel that these limitations are balanced by our study's notable strengths including its evaluation of a clinically important question, large and diverse patient population, and its rigorous analytic approach including landmark analyses to account for immortal time bias.

In conclusion, treatment delays exceeding 3 months were common but were not associated with significant differences in overall survival. Continued efforts are needed to define metrics for optimal care delivery models that improve clinical outcomes for patients with HCC.

CONFLICTS OF INTEREST

Guarantor of the article: Amit G. Singal, MD, MS.

Specific author contributions: Conceptualization: A.Y. and A.G.S. Data curation: all authors. Formal analysis: M.J. Funding acquisition: A.Y. and A.G.S. Investigation and methodology: M.J. and A.G.S. Project administration: A.G.S. Resources: A.G.S. Supervision and Validation: A.G.S. Writing–original draft: M.J. Writing–review and editing: all authors.

Financial support: This study was conducted with support from National Institute of Health (NIH) U01 CA283935, R01 MD012565, and R01 CA256977. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The funding agencies had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation of the manuscript.

Potential competing interests: Amit G. Singal has served as a consultant or on advisory boards for Genentech, AstraZeneca, Eisai, Bayer, Exelixis, Merck, Elevar, Boston Scientific, Sirtex, FujiFilm Medical Sciences, Exact Sciences, Helio Genomics, Roche, Glycotest, Abbott, DELFI, IMCare, and Universal Dx. Nicole E. Rich has served as consultant or on advisory boards for AstraZeneca, Eisai, Exelixis, Genentech and Madrigal. None of the other authors have any relevant conflicts of interest to disclose.

Study Highlights.

WHAT IS KNOWN

• ✓ The care continuum for patients with hepatocellular carcinoma (HCC) is prone to gaps, such as treatment delays.

• ✓ Multidisciplinary care approach has the potential to streamline care and reduce treatment delays in HCC management.

WHAT IS NEW HERE

• ✓ Treatment delays exceeding 90 days were common, occurring in more than 1 in 5 patients with HCC.

• ✓ Patients with early-stage HCC were more likely to experience treatment delays compared with those with more advanced tumor burden.

• ✓ Treatment delays were not significantly associated with overall survival.

Supplementary Material

acg-121-2473-s001.pdf ^{(179.4KB, pdf)}

ABBREVIATIONS:

ALD

alcohol associated liver disease

BCLC

Barcelona Clinic Liver Cancer

Black

non-Hispanic Black

chemotherapy

systemic therapy

ECOG

Eastern Cooperative Oncology Group

HCC

hepatocellular carcinoma

HCV

hepatitis C

Jackson

Jackson Health System

LT

liver transplantation

MELD

Model for End-Stage Liver Disease

Parkland

Parkland Health

SBRT

stereotactic body radiation therapy

SES

socioeconomic status

SVR

postsustained viremic response

TACE

transarterial chemoembolization

TARE

transarterial radioembolization

U Miami

University of Miami

UTSW

UT Southwestern Medical Center

White

non-Hispanic White