Hepatocellular Carcinoma Surveillance in Patients With Cirrhosis at US Safety-Net Health Systems

Abstract

INTRODUCTION:

Surveillance for hepatocellular carcinoma (HCC) in patients with cirrhosis is associated with improved patient outcomes. We aim to evaluate real-world utilization of HCC surveillance among safety-net populations with cirrhosis.

METHODS:

We performed a retrospective cohort study of adults with cirrhosis across 4 safety-net health systems from March 1, 2017, to February 28, 2022. Receipt of abdominal imaging with ultrasound, computed tomography, or magnetic resonance imaging and the corresponding ICD-9-CM/ICD-10-CM diagnosis codes at 6 months and 12 months were used to assess HCC surveillance.

RESULTS:

Among 14,556 patients with cirrhosis (61.8% male, 73.0% non-White ethnic minorities, 54.4% with Medicaid or indigent care/uninsured), 70.9% and 78.1% received abdominal imaging agnostic to indication within 6 months and 12 months, respectively. When evaluating the receipt of abdominal imaging with a specific indication for HCC surveillance, 29.1% and 34.0% of patients received surveillance within 6 months and 12 months, respectively. On adjusted multivariable regression, greater odds of HCC surveillance were observed in older patients, ethnic minorities, and those with commercial insurance. Lower odds of HCC surveillance were observed in patients with indigent care (vs Medicare: odds ratio [OR] 0.85, 95% confidence interval [CI] 0.72–1.00), drug use (OR 0.63, 95% CI 0.55–0.71), and concurrent mental health/psychiatric diagnoses (OR 0.88, 95% CI 0.80–0.97).

DISCUSSION:

Among a multicenter safety-net cohort of patients with cirrhosis, fewer than 30% received HCC surveillance within 6 months. While greater proportions received abdominal imaging agnostic to indication, the clinical benefit of these examinations for HCC surveillance may be limited because of concerns with abbreviated protocols, quality, and interpretation.

INTRODUCTION

Hepatocellular carcinoma (HCC) remains a leading cause of cancer-related morbidity and mortality (1–3). Delays in HCC diagnosis lead to advanced tumor stage at diagnosis, limiting potentially curative treatment options, and contribute to worse outcomes (4–7). Major society guidelines recommend HCC surveillance in patients with cirrhosis using abdominal ultrasound (US) and alpha fetoprotein every 6 months (8,9). However, existing studies have demonstrated suboptimal utilization of HCC surveillance in patients with cirrhosis (4,10–13). A recent meta-analysis inclusive of 29 studies and a total of 118,799 patients with cirrhosis observed a pooled estimate of 24.0% for HCC surveillance utilization (12). Drivers of low HCC surveillance in patients with cirrhosis are multifactorial and likely reflect patient-, provider-, and system-level challenges (11,14–19). A recent study by our team surveyed primary care providers and gastroenterology/hepatology providers across 5 safety-net health systems in the United States (14). The authors identified important gaps in knowledge regarding HCC surveillance and perceived barriers such as difficulty in identifying patients with cirrhosis and not being up-to-date with HCC surveillance guidelines, especially among primary care providers. Better elucidation of the multifactorial barriers contributing to underutilization of HCC surveillance is critical to guide targeted interventions to improve HCC surveillance gaps in patients with cirrhosis.

A particularly vulnerable population that suffers the highest burden of chronic liver disease in the United States (20–26) is safety-net patients who are underserved and often underinsured. Safety-net health systems care for a large proportion of ethnic minorities, non–English-speaking immigrants, individuals of lower socioeconomic status, and those with active high-risk behaviors, all of which contribute to inherent patient-specific and system-level barriers in access to care, including HCC surveillance (27–33). Hence, better understanding of HCC surveillance utilization and potentially modifiable factors associated with low HCC surveillance among this population can inform health systems to develop targeted interventions to improve cirrhosis care in this vulnerable population. This study aims to evaluate real-world utilization of HCC surveillance among adults with cirrhosis receiving care from 4 geographically distinct safety-net health systems in the United States.

METHODS

Adults with cirrhosis who were receiving care at 1 of 4 safety-net health systems in California, Florida, Ohio, and Texas were identified from March 1, 2017, to February 28, 2022. Cirrhosis was identified using established definitions previously used in electronic health record–based studies: ≥1 cirrhosis or cirrhosis-related complications based on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)/International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes on 2 separate clinical encounters (34–36). Patients with HCC (based on ICD-9-CM/ICD-10-CM codes) before or at the time of cirrhosis identification were excluded. Receipt of HCC surveillance was identified using Current Procedural Terminology (CPT) codes corresponding to abdominal imaging with US, computed tomography (CT), or magnetic resonance imaging (MRI) scans (CPT codes: 76700, 76705, 74150, 74160, 74170, 74176, 74177, 74178, 74181, 74182, 74183) as well as the corresponding ICD-9-CM/ICD-10-CM diagnosis codes associated with each imaging examination. While abdominal US is the guideline-recommended HCC surveillance modality, in real-world clinical practice, there may be variations such that providers may order other imaging tests (e.g., CT or MRI of the abdomen) for HCC surveillance, which would provide similar clinical benefits (37). This was our rationale for including the aforementioned imaging modalities in assessing HCC surveillance.

Baseline characteristics were evaluated during the 12 months before the date of cirrhosis identification. Clinical characteristics including risk factors and comorbidities were identified using ICD-9-CM/ICD-10-CM codes and laboratory data (see Supplementary Table 1, Supplementary Digital Content 1, http://links.lww.com/CTG/B324). Characteristics of the cirrhosis cohort were presented as frequencies and proportions for categorical variables and median and interquartile range for continuous variables. Receipt of HCC surveillance within 6 months of cirrhosis identification was evaluated in patients with a follow-up of ≥6 months, and receipt of HCC surveillance within 12 months was evaluated in patients with a follow-up of ≥12 months. We first evaluated the receipt of the aforementioned abdominal imaging regardless of indication. Then, we performed a subsequent analysis evaluating the receipt of abdominal imaging with a specific indication for HCC surveillance based on ICD-10-CM codes (see Supplementary Table 2, Supplementary Digital Content 1, http://links.lww.com/CTG/B324). While we included the receipt of US, CT, or MRI of the abdomen in our analyses, it is possible that some of these patients might have received these imaging modalities in acute care settings (e.g., emergency departments), which may be performed without contrast or performed with limited protocols (e.g., single-phase CT scans). Noncontrast and single-phase CT scans, in particular, may provide limited benefit regarding HCC surveillance. For that reason, we conducted sensitivity analyses excluding CT (i.e., only the receipt of imaging with US or MRI). The proportion of patients who received HCC surveillance was stratified by sociodemographics and relevant clinical characteristics. Comparisons of HCC surveillance between groups were performed using the χ² test. Adjusted multivariable logistic regression models were used to evaluate for predictors of successful HCC surveillance within 6 months and within 12 months of cirrhosis identification. Variables that demonstrated significance in the univariate model (P < 0.10) were selected for inclusion in the multivariable model. Statistical analyses were performed using SAS Studio 3.6 on SAS 9.4 (SAS Institute, Cary, NC). Statistical significance was met with a 2-tailed P value of <0.05. The study was approved by the institutional review boards of each respective institution and health system.

RESULTS

A total of 14,556 patients with cirrhosis were identified, among whom 61.8% were male, with a median age of 58 years (interquartile range 50–64) and a median follow-up time of 40.5 months (interquartile range 15.1–64.0) (Table 1). The cohort was mostly Hispanic (40.9%), followed by non-Hispanic White (27.0%), Black/African Americans (23.6%), and Asians (3.6%). When stratified by primary insurance, 28.2% had indigent care or were uninsured, 26.2% had Medicare, and 26.2% had Medicaid. Overall, 36.8% had diabetes, 14.6% had current drug use, 14.6% had human immunodeficiency virus (HIV) infection, and 5.6% had end-stage renal disease. When evaluating liver disease etiology, 37.3% had alcohol-associated liver disease, 31.5% had hepatitis C virus (HCV), 10.7% had metabolic dysfunction–associated steatotic liver disease/metabolic dysfunction–associated steatohepatitis (MASLD/MASH), and 5.4% had hepatitis B virus. At baseline, 43.4% had at least 1 sign of cirrhosis decompensation, with the most common being ascites.

Table 1.

Baseline characteristics of the study cohort

	Frequency (N)	Proportion (%)
Total	14,556	100.0
Follow-up time, mo, median (IQR)	40.5	15.1–64.0
Sex
Male	8,997	61.8
Female	5,559	38.2
Age, yr, median (IQR)	58	50–64
Race-ethnicity
Non-Hispanic White	3,927	27.0
Black/African American	3,437	23.6
Hispanic	5,952	40.9
Asian	522	3.6
Other/unknown	718	4.9
Language
English	8,958	61.5
Non-English	5,437	37.4
Insurance status
Medicare	3,810	26.2
Medicaid	3,808	26.2
Commercial	1,347	9.3
Other	1,489	10.2
None/indigent care	4,102	28.2
Risk factors and comorbidities
Drug use	2,128	14.6
HIV	2,128	14.6
Cardiovascular disease	11,952	82.1
Diabetes	5,351	36.8
End-stage renal disease	815	5.6
Mental health/psychiatric	9,190	63.1
Non-HCC cancers	3,462	23.8
Liver disease etiologies
HBV	786	5.4
HCV	4,587	31.5
MASLD/MASH	1,554	10.7
ALD	5,431	37.3
Cirrhosis decompensation at baseline
Ascites	5,972	41.0
Hepatic encephalopathy	240	1.7
Variceal bleeding	717	4.9
Hepatorenal syndrome	313	2.2
≥1 cirrhosis-related decompensation	6,311	43.4

ALD, alcohol-associated liver disease; HBV, hepatitis B virus; HCV, hepatitis C virus; IQR, interquartile range; MASH, metabolic dysfunction–associated steatohepatitis; MASLD, metabolic dysfunction–associated steatotic liver disease.

Receipt of abdominal imaging

Overall, 70.9% of patients received abdominal imaging agnostic to indication within 6 months of cirrhosis identification. Receipt of abdominal imaging within 6 months was similar between men and women, and when stratified by age, patients age 49–49 years had the highest proportion of receiving abdominal imaging (Figure 1). When stratified by race/ethnicity, the highest proportion of abdominal imaging within 6 months was observed in Hispanics. Non–English-speaking patients had significantly lower proportion of receiving abdominal imaging than English-speaking patients (68.3% vs 72.3%, P < 0.001). When stratified by insurance status, the lowest proportion receiving abdominal imaging was among patients with cirrhosis with Medicare insurance (Figure 1). The proportion receiving abdominal imaging was similar across liver disease etiologies, but patients with decompensated cirrhosis were significantly more likely to receive abdominal imaging vs patients with compensated cirrhosis (82.8% vs 62.9%, P < 0.001). On adjusted multivariable regression, greater odds of completing abdominal imaging within 6 months were observed among racial and ethnic minorities compared with non-Hispanic Whites (Black/African Americans: odds ratio [OR] 1.14, 95% confidence interval [CI] 1.01–1.28; Hispanics: OR 1.25, 95% CI 1.11–1.42; Asians: OR 1.70, 95% CI 1.32–2.17) and those with Medicaid or indigent care vs Medicare (Medicaid: OR 1.17, 95% CI 1.04–1.32; indigent care: OR 1.30, 95% CI 1.11–1.52) (Table 2). Significantly greater odds of abdominal imaging within 6 months were also observed in patients who were HIV-positive and had underlying MASLD/MASH, concurrent non-HCC cancer, cardiovascular disease, or evidence of decompensated cirrhosis (Table 2).

Figure 1.

Receipt of abdominal imaging within 6 months and 12 months of cirrhosis diagnosis. ALD, alcohol-associated liver disease; HBV, hepatitis B virus; HCV, hepatitis C virus; MASH, metabolic dysfunction–associated steatohepatitis; MASLD, metabolic dysfunction–associated steatotic liver disease.

Table 2.

Predictors of abdominal imaging within 6 months of cirrhosis diagnosis

	OR	95% LL	95% UL	P value
Female	1.00	Reference
Male	0.97	0.89	1.06	0.529
Age (yr)
<40	1.00	Reference
40–49	1.17	0.98	1.40	0.100
50–59	1.10	0.93	1.29	0.268
60–69	1.19	1.00	1.41	0.053
70 and over	1.10	0.89	1.35	0.401
Race/ethnicity
Non-Hispanic White	1.00	Reference
Black/African American	1.14	1.01	1.28	0.031
Hispanic	1.25	1.11	1.42	<0.001
Asian	1.70	1.32	2.17	<0.001
Other/unknown	1.05	0.85	1.30	0.661
Insurance status
Medicare	1.00	Reference
Medicaid	1.17	1.04	1.32	0.011
Commercial	1.18	1.00	1.41	0.055
Indigent care	1.30	1.11	1.52	0.001
Other	1.55	1.37	1.76	<0.001
Language
English	1.00	Reference
Non-English	1.01	0.90	1.14	0.816
Drug use	0.95	0.84	1.07	0.375
HIV-positive (vs HIV-negative)	1.22	1.08	1.38	0.002
MASLD/MASH (vs no MASLD/MASH)	1.54	1.34	1.78	<0.001
Mental health/psychiatric diagnosis	1.07	0.97	1.18	0.172
Non-HCC cancer	1.57	1.41	1.75	<0.001
Cardiovascular disease	1.48	1.29	1.71	<0.001
Cirrhosis decompensation	2.48	2.26	2.73	<0.001

Model adjusted for sex, age, race/ethnicity, insurance, language, drug use, HIV, MASLD/MASH, mental health/psychiatric diagnoses, non-HCC cancers, cardiovascular disease, cirrhosis decompensation, and study site.

HCC, hepatocellular carcinoma; LL, lower limit; MASH, metabolic dysfunction–associated steatohepatitis; MASLD, metabolic dysfunction–associated steatotic liver disease; OR, odds ratio; UL, upper limit.

Overall, 78.1% of patients received abdominal imaging within 12 months of cirrhosis identification (Figure 1). On adjusted multivariable regression, greater odds of abdominal imaging were observed in patients age 40–49 years compared with those age <40 years (OR 1.43, 95% CI 1.16–1.74); racial and ethnic minorities compared with non-Hispanic Whites (Black/African Americans: OR 1.17, 95% CI 1.02–1.34), Hispanics (OR 1.32, 95% CI 1.15–1.52), and Asians (OR 1.75, 95% CI 1.33–2.31); and those with indigent care (vs Medicare: OR 1.25, 95% CI 1.05–1.49) (see Supplementary Table 3, Supplementary Digital Content 1, http://links.lww.com/CTG/B324). Greater odds of imaging within 12 months were also observed in patients who were HIV-positive and had MASLD/MASH, non-HCC cancer, cardiovascular disease, and evidence of decompensated cirrhosis.

To address potential for screening overestimation based on suboptimal CT (e.g., single phase), on sensitivity analysis excluding CT, 54.2% of patients with cirrhosis received imaging within 6 months (see Supplementary Figure 1, Supplementary Digital Content 1, http://links.lww.com/CTG/B324). Receipt of imaging was similar between men and women and similar between English vs non–English-speaking patients. When stratified by age, cirrhosis patients age 40–49 years had the highest proportion of abdominal imaging, and when stratified by race and ethnicity, the highest proportion of abdominal imaging was observed among Black/African Americans. When stratified by insurance, the highest proportion receiving abdominal imaging was observed among patients who were uninsured or had indigent care and those with Medicaid, whereas the lowest proportion was among those with commercial insurance. On adjusted multivariable regression, greater odds of abdominal imaging with US or MRI within 6 months was observed among Black/African American and Asian patients (vs non-Hispanic Whites), patients insured by Medicaid (vs Medicare), and non–English-speaking patients (vs English-speaking) (see Supplementary Table 4, Supplementary Digital Content 1, http://links.lww.com/CTG/B324). Patients with cirrhosis with commercial insurance had lower odds of abdominal imaging compared with Medicare-insured patients.

Completion of abdominal imaging with HCC surveillance indication

When evaluating the receipt of abdominal imaging with a specific indication for HCC surveillance, 29.1% received surveillance within 6 months (Figure 2). Receipt of HCC surveillance was similar between men and women, whereas when stratified by age, patients aged 50–69 years had the highest proportion of surveillance. No significant race/ethnicity-specific differences were observed. When stratified by insurance status, the greatest proportion receiving HCC surveillance was observed for indigent care or uninsured patients and those with Medicaid, and the lowest proportion was observed in those with commercial insurance (Figure 2). The proportion receiving HCC surveillance was similar across etiologies and depends on whether patients were compensated or decompensated. On adjusted multivariable regression, greater odds of HCC surveillance within 6 months were observed in older patients (vs age <40 years: OR 1.67, 95% CI 1.34–2.09), racial and ethnic minorities compared with non-Hispanic Whites (Black/African American: OR 1.14, 95% CI 1.01–1.29; Hispanics: OR 1.46, 95% CI 1.29–1.66; Asians: OR 1.55, 95% CI 1.20–2.01), and those with commercial insurance (vs Medicare: OR 1.21, 95% CI 1.01–1.46). Significantly lower odds of HCC surveillance were observed in patients with indigent care vs Medicare, drug use, concurrent mental health/psychiatric diagnoses, and end-stage renal disease (Table 3).

Figure 2.

Imaging within 6 months and 12 months of cirrhosis diagnosis with specific imaging indication for HCC surveillance. ALD, alcohol-associated liver disease; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MASH, metabolic dysfunction–associated steatohepatitis; MASLD, metabolic dysfunction–associated steatotic liver disease.

Table 3.

Predictors of imaging within 6 months of cirrhosis diagnosis with specific imaging indication for HCC surveillance

	OR	95% LL	95% UL	P value
Female	1.00	Reference
Male	0.94	0.86	1.02	0.138
Age (yr)
<40	1.00	Reference
40–49	1.41	1.16	1.71	0.001
50–59	1.61	1.35	1.92	<0.001
60–69	1.71	1.42	2.07	<0.001
70 and over	1.67	1.34	2.09	<0.001
Race/ethnicity
Non-Hispanic White	1.00	Reference
Black/African American	1.14	1.01	1.29	0.034
Hispanic	1.46	1.29	1.66	<0.001
Asian	1.55	1.20	2.01	0.001
Other/unknown	1.12	0.89	1.40	0.335
Insurance status
Medicare	1.00	Reference
Medicaid	1.13	1.00	1.28	0.052
Commercial	1.21	1.01	1.46	0.040
Indigent care	0.85	0.72	1.00	0.050
Other	1.43	1.27	1.62	<0.001
Language
English	1.00	Reference
Non-English	0.92	0.82	1.04	0.170
Drug use	0.63	0.55	0.71	<0.001
HIV-positive (vs HIV-negative)	1.16	1.01	1.33	0.034
Mental health/psychiatric diagnosis	0.88	0.80	0.97	0.012
Non-HCC cancer	1.02	0.93	1.13	0.641
Cardiovascular disease	1.16	1.04	1.30	0.010
End-stage renal disease	0.39	0.30	0.50	<0.001
Cirrhosis decompensation	1.08	0.99	1.18	0.098

Model adjusted for sex, age, race/ethnicity, insurance, language, drug use, HIV, end-stage renal disease, mental health/psychiatric diagnoses, non-HCC cancers, cardiovascular disease, cirrhosis decompensation, and study site.

HCC, hepatocellular carcinoma; LL, lower limit; OR, odds ratio; UL, upper limit.

Overall, 34.0% of patients received HCC surveillance within 12 months (Figure 2). On multivariable regression, similar predictors of receiving HCC surveillance were seen as those observed for the 6-month interval of HCC surveillance (see Supplementary Table 5, Supplementary Digital Content 1, http://links.lww.com/CTG/B324).

On sensitivity analyses excluding CT imaging, 25.2% of patients with cirrhosis received abdominal imaging with US or MRI for an indication of HCC surveillance within 6 months (see Supplementary Figure 2, Supplementary Digital Content 1, http://links.lww.com/CTG/B324). Receipt of HCC surveillance was similar by sex, race/ethnicity, and primary language spoken. Cirrhosis patients age 50–59 years and 60–69 years had the highest proportion of HCC surveillance. When stratified by insurance, the highest proportion receiving HCC surveillance was among patients with indigent care or uninsured patients and those with Medicaid. No differences in HCC surveillance were observed by liver disease etiology or the presence of cirrhosis decompensation. On adjusted multivariable regression, greater odds of HCC surveillance with US or MRI were observed among older patients (vs age <40 years), racial and ethnic minorities (vs non-Hispanic Whites), patients insured by Medicaid (vs Medicare), and non–English-speaking patients (vs English-speaking) (see Supplementary Table 6, Supplementary Digital Content 1, http://links.lww.com/CTG/B324). Lower odds of HCC surveillance with US or MRI were observed in patients with commercial insurance (vs Medicare) and those with concurrent drug use.

DISCUSSION

Among a large multicenter cohort of patients with cirrhosis receiving care at safety-net health systems in the United States, fewer than 30% of patients received guideline-recommended HCC surveillance within 6 months. This observation is consistent with existing studies demonstrating low utilization of HCC surveillance in patients with cirrhosis (4,10–13,36). However, when evaluating the receipt of abdominal imaging agnostic to indication, over 70% of patients with cirrhosis received abdominal imaging with US, CT, or MRI within 6 months. When evaluating only US or MRI, 54.2% of patients with cirrhosis received abdominal imaging within 6 months. This is an important clinical observation because while US is the recommend HCC surveillance modality, other cross-sectional abdominal imaging, particularly MRI, may provide similar benefit for HCC surveillance (38,39). While contrast-enhanced abdominal CT or MRI generally is at least as sensitive and specific (and likely more so) as US for HCC diagnosis, limited data exist evaluating the role of these alternative imaging modalities from a surveillance perspective. Pocha et al evaluated 163 adults with compensated cirrhosis randomized to biannual ultrasonography or yearly CT and observed similar sensitivity and specificity with both approaches (40). A prospective cohort study from South Korea observed greater sensitivity and specificity for detection of early-stage HCC using MRI- vs US-based surveillance approaches (41). In a meta-analysis evaluating abbreviated MRI for HCC surveillance, abbreviated MRI both with or without contrast had a pooled per-patient sensitivity and specificity superior to US alone (42). An ongoing clinical trial—PREMIUM Study (CSP #2023 PREventing liver cancer Mortality through Imaging with Ultrasound vs MRI)—is evaluating abbreviated MRI vs US for HCC surveillance among Veterans with cirrhosis.

Another important consideration is that abdominal imaging performed in acute inpatient or emergency settings for indications other than HCC surveillance may be performed without contrast or limited single-phase protocols, limiting the benefit for HCC surveillance. We attempted to address this concern by performing sensitivity analyses excluding CT and observed that 54.2% of patients with cirrhosis received US or MRI abdominal imaging within 6 months. When limited to imaging with a specific indication for HCC surveillance, the findings were similar whether CT imaging was included.

We observed greater odds of HCC surveillance in racial and ethnic minorities. While previous studies have observed high rates of HCC surveillance in Asians, our findings are in contrast to what has been previously reported on lower rates of HCC surveillance among Black/African Americans (30,31,43–45). For example, a single-center safety-net study of 904 patients with cirrhosis from 2008 to 2011 observed that African Americans were 39% less likely to receive HCC surveillance compared with non-Hispanic Whites (31). Data from the Veterans Affairs HCV Clinical Case Registry study of 13,002 patients with HCV-related cirrhosis observed lower rates of HCC surveillance in African Americans vs Whites (15.5% vs 68.1%; OR 0.60, 95% CI 0.45–0.81) (46). The reasons underlying these discordant observations are not clear. It is possible that all the safety-net health systems included in our study are affiliated with teaching hospitals with various levels of medical trainees providing care and thus may be more aware and up to date with HCC surveillance recommendations. It is possible that teaching hospital affiliations and incorporation of medical trainees lead to providers being more cognizant of healthcare inequities among ethnic minorities, or there may be targeted outreach programs focused on underserved ethnic minorities that might have mitigated some of the previous disparities in HCC surveillance observed. While these disparities are worth noting, the bigger picture of overall low HCC surveillance utilization across all race and ethnic groups should still be emphasized and deserves greater attention.

We observed significant differences in odds of HCC surveillance by insurance. Compared with patients with Medicare, greater odds of surveillance at 6 months were observed among both commercial and Medicaid insured patients. However, it is interesting that patients with indigent care or uninsured patients had greater odds of imaging overall, but when focusing on imaging with a specific indication of HCC surveillance, the direction of the effect changed, such that indigent care or uninsured patients with cirrhosis had lower odds of HCC surveillance. This may be explained by observations that uninsured patients in particular are more likely to use acute care or emergency services and less likely to be consistently engaged in ambulatory care (47–49), given that HCC surveillance is almost always performed in elective settings. While lower HCC surveillance was also observed for patients with drug use, mental health/psychiatric diagnoses, or end-stage renal disease, when focusing on imaging with HCC surveillance indications, these differences were no longer significant. These observations likely reflect similar aforementioned considerations of acute care vs ambulatory imaging. Patients with active drug use or concurrent mental health/psychiatric diagnoses may be less engaged or more prone to not follow through with elective imaging requests.

Incorporating a diverse cohort of patients with cirrhosis across 4 safety-net health systems allowed comprehensive assessments of real-world HCC surveillance utilization among underserved populations. However, as is typical for observational studies incorporating ICD-9-CM/ICD-10-CM codes, our analyses may be subject to misclassification bias. While we incorporated CPT codes and corresponding ICD-9-CM/ICD-10-CM codes to assess HCC surveillance, we did not have data to determine types of providers ordering imaging or hospital locations from which the orders originated. Furthermore, while we captured completion of HCC surveillance, we were not able to capture data on imaging examinations that were ordered but not completed. Hence, it was difficult to determine with certainty whether the lack of surveillance was due to providers not ordering abdominal imaging or patients not completing imaging that was ordered. While a comprehensive data pull ensured complete extraction of relevant data to assess HCC surveillance, it is possible that patients received care outside of study sites. However, we believe that this would primarily affect imaging that was agnostic to indication and would not significantly affect imaging with specific indications for HCC surveillance. Because of insurance limitations, most safety-net populations would likely not receive elective HCC surveillance outside of safety-net settings given that HCC surveillance does not constitute acute or emergency care (29,50,51). While our study evaluated various factors and their association with timely receipt of HCC surveillance, we acknowledge that successful implementation of effective HCC surveillance is complex, and gaps and delays in surveillance likely reflect multiple patient-, provider-, and system-level factors. The observational nature of our study design limits the more granular assessment of these specific factors in affecting HCC surveillance. System-level factors such as limited radiology department capacity, lack of integrated electronic health record–based reminders, patient navigators, and patient financial and logistical barriers may be important factors posing challenges to effective implementation of HCC surveillance (11,16,52). Indeed, broader system barriers to clinical care result in many patients not being engaged in routine care before HCC presentation (53). Finally, it should be noted that HCC surveillance is only 1 step in the cancer care continuum and studies have also demonstrated downstream failures including diagnostic delays and underuse of curative treatment (7,54,55). Future research that more clearly elucidates these potentially modifiable system-level factors followed by design of targeted interventions to address these barriers will be critical to address gaps in HCC surveillance, particularly among safety-net populations.

In conclusion, among a large cohort of patients with cirrhosis receiving care at US safety-net health systems, fewer than 30% received HCC surveillance within 6 months. While a greater proportion received abdominal imaging when indication was not considered, the clinical benefit of these examinations for HCC surveillance must be interpreted with caution given potential concerns with abbreviated protocols, quality, and interpretation of these examinations. Continuous efforts to bridge existing gaps in HCC surveillance are critically needed, given the importance of surveillance in early detection of HCC leading to improved treatment and patient outcomes.

CONFLICTS OF INTEREST

Guarantor of the article: Robert J. Wong, MD, MS.

Specific author contributions: All authors: study concept and design, acquisition of data, and analysis and interpretation of the data. M.T., O.K., and Y.Z.: statistical analyses. R.J.W.: drafting of the manuscript. All authors: Critical revision of the manuscript for important intellectual content. R.J.W.: study supervision. R.J.W. had full access to the data in the study and takes responsibility for the integrity of the data and accuracy of the data analysis.

Financial support: This study was supported by the National Institute on Minority Health and Health Disparities (NIMHD) 5R01MD017063. A.G.S. was also supported by NCI U01 CA271887, R01 MD012565, and R01 CA256977. M.K. was also supported by NIAAA K24AA022523, R01AA029312, and NIMHD U24MD017250. P.D.J. was also supported by NIMHD R01MD012565, NIDDK U01DK130185, NCI K08CA255413, U01CA288421, and The V Foundation for Cancer Research Grand ID# DEC2022-015.

Potential competing interests: R.J.W. has received research grants (to his institution) from Gilead Sciences, Exact Sciences, Theratechnologies, and Durect Corporation and has served as a consultant (without compensation) for Gilead Sciences, Mallinckrodt Pharmaceuticals, and Salix Pharmaceuticals. P.J. has received research grants (to her institution) from Gilead Sciences. B.N. and P.P. have no disclosures. M.T. has received research grants (to her institution) from Gilead Sciences. O.K. has received research grants (to his institution) from Gilead Sciences. Y.Z. has received research grants (to her institution) from Gilead Sciences. R.F. has no disclosures. G.T. has received grants (to his institution) from Gilead Sciences and Novo-Nordisk and serves as a consultant on advisory boards to Intercept Pharma and Novo-Nordisk. A.G.S. has served as a consultant or on advisory boards for Bayer, FujiFilm Medical Sciences, Exact Sciences, Helio, Roche, Abbott, Glycotest, DELFI, and GRAIL. M.K. has received research grants (to her institution) from Gilead Sciences and Intercept Pharmaceuticals and has served as a consultant for Gilead Sciences and GSK Pharmaceuticals.

Study Highlights.

WHAT IS KNOWN

• ✓ Hepatocellular carcinoma (HCC) is a leading cause of cancer-related morbidity and mortality.

• ✓ HCC surveillance in patients with cirrhosis leads to early tumor stage, improved treatment, and better patient outcomes.

• ✓ Safety-net populations experience inherent barriers and challenges in timely access to care, which may also affect timely HCC surveillance.

WHAT IS NEW HERE

• ✓ Among 14,556 patients with cirrhosis across 4 safety-net health systems, 29.1% and 34.0% of patients received surveillance within 6 months and 12 months for specific HCC surveillance indication.

• ✓ However, when evaluating the receipt of imaging agnostic to indications, 70.9% and 78.1% received abdominal imaging.

• ✓ The presence of indigent care coverage, concurrent drug use, or concurrent mental health diagnoses further exacerbated gaps in HCC surveillance among this vulnerable population.

ABBREVIATIONS:

CI

confidence interval

CPT

Current Procedural Terminology

CT

computed tomography

HCC

hepatocellular carcinoma

HCV

hepatitis C virus

ICD-9-CM

International Classification of Diseases, Ninth Revision, Clinical Modification

ICD-10-CM

International Classification of Diseases, Tenth Revision, Clinical Modification

MASH

metabolic dysfunction–associated steatohepatitis

MASLD

metabolic dysfunction–associated steatotic liver disease

OR

odds ratio

US

ultrasound