Rates and Factors Associated With Hepatocellular Carcinoma Diagnosis, Stage, Treatment, and Survival

Abstract

Background and Aims

Up-to-date information is needed on hepatocellular carcinoma (HCC) diagnosis, stage, treatment, and survival.

Methods

Of > 2000 patients with a new diagnosis of HCC in 2023 in the US Veterans Health Administration, a random subsample of 194 confirmed HCC cases were selected for a structured medical record review by expert hepatologists.

Results

Among 194 confirmed HCC cases in 2023, mean age was 73 years, and only 56.7% had cirrhosis diagnosed before HCC, while 12.9% had cirrhosis diagnosed after HCC and 22.2% did not have cirrhosis. Stage at diagnosis was T1 in 17.5%, T2 in 42.3%, and beyond T2 in 40.2%. Early-stage diagnosis (T1 or T2) was more common in the following groups: cirrhosis diagnosed before HCC (70.9%), HCC diagnosed by screening (86.3%), high performance status (73.0%), receipt of Veterans Affairs (VA) primary care (63.3%), or VA liver care (72.6%). Among 147 of 194 patients (75.8%) who received HCC-directed treatments, the most common, first-line treatment was Y-90 radioembolization (28.6%), followed by ablation (21.1%), transarterial chemoembolization (20.4%), systemic therapy (17.0%), surgical resection (7.5%), and external beam radiation (5.4%). Mortality (29.9% at 1 year, 44.8% at 2 years) was lower in those with early-stage diagnosis, diagnosis via screening, Child-Turcotte-Pugh class A, Model for End-Stage Liver Disease ≤ 10, absence of cirrhosis, cured hepatitis C virus, receipt of curative treatments, VA primary or liver care, and good performance status.

Conclusion

These results highlight the importance of HCC screening and engagement in liver care for early HCC diagnosis, curative treatment, and improved survival while demonstrating the feasibility of a national quality improvement program for addressing persistent gaps in the HCC screening, diagnosis, and treatment.

Graphical abstract

Introduction

The landscape of hepatocellular carcinoma (HCC) treatments has been changing very rapidly, with an expansion in treatment modalities, indications for treatment, and available expertise.^1,2 Multiple treatments, including potentially curative treatments, have to be carefully considered to determine the best option for each patient, ideally in the setting of a multidisciplinary liver tumor board (MLTB).³ However, treatment eligibility, treatment options, and survival largely depend on the stage at diagnosis. When HCC is diagnosed at very early (T1) or early (T2) stage, curative treatment options are available and survival can be excellent. Conversely, when HCC is diagnosed at late stage, treatments are rarely curative and survival is poor, especially in the setting of vascular invasion, infiltrative HCC, lymph node, or distal metastasis. While professional liver society guidelines recommend HCC screening for patients with cirrhosis,^1,2 advanced stage at diagnosis is common, likely due to under-recognition of cirrhosis, low HCC screening rates, and suboptimal performance characteristics of current HCC screening tests.4, 5, 6

Few studies performed systematic assessments of HCC diagnosis and management in large, nationwide healthcare systems. We previously performed a comprehensive assessment of HCC cases diagnosed in calendar year 2021 in the Veterans Health Administration (VHA),⁷ the largest comprehensive healthcare system in the United States, which also has the highest burden of HCC cases.⁵ We reported rates and factors associated with HCC diagnosis stage, treatment, and survival to develop plans to address gaps in care, as part of the VHA’s National Gastroenterology and Hepatology Program (NGHP), which includes an HCC Field Advisory Board (FAB). We also aimed to establish an ongoing, biannual HCC data surveillance program to track trends in HCC diagnosis and management and evaluate the results of quality improvement projects aiming to address gaps. We hereby report the characteristics of a random sample of Veterans Affairs (VA) enrollees diagnosed with HCC in calendar year 2023 (CY2023), compare those to the cases diagnosed in 2021, and highlight the rates and most important factors associated with early stage at diagnosis, receipt of HCC-directed treatments, and outcomes.

Methods

This project was approved, authorized, and executed as a quality improvement project by the VA NGHP. The VA Pittsburgh Healthcare system approved this project as a Quality Improvement project. Hence, no institutional review board approvals were required. Below we present the process for developing our iterative data monitoring system for quality improvement, following system redesign principles.⁸

Study Population and Data Extraction: Patients With a New HCC Diagnosis in Calendar Year 2023

VA uses a single electronic health record (EHR) system, which enables cross-facility record access to other VA facilities, the Department of Defense, and non-VA facilities through a system called Joint Legacy Viewer to facilitate complete medical records access. Of more than 2000 patients with a new diagnosis of HCC documented in the VA’s EHR in CY2023, we identified randomly 194 patients (ie, an approximately 10% subsample) who were confirmed to have HCC after review of their medical records by expert hepatology reviewers, to undergo a detailed structured review of their EHRs as outlined below. The size of this subsample of ∼200 patients for chart review was selected a priori as being both feasible and large enough for representative estimates. HCC was defined as (1) Liver Imaging Reporting And Data System 5 lesion(s) on a multiphasic, contrast-enhanced, liver-protocol abdominal computed tomography (CT) or magnetic resonance imaging (MRI) (n = 127)⁹ or (2) liver biopsy consistent with HCC (n = 72) or (3) MLTB consensus diagnosis of HCC (n = 15), in rare cases where 2 previous criteria were not met. The date of HCC diagnosis was the earliest date at which criteria 1 to 3 above were first documented. These 194 patients originated from 24 VA facilities around the country (Altoona, PA; Atlanta, GA; Clarksburg, MD; Pittsburgh, PA; Puget Sound, WA; West Haven, CT; Togus, ME; Long Beach, CA; Loma Linda, CA; San Diego, CA; Tampa, FL; Portland, OR; Miami, FL; Orlando, FL; Washington, DC; San Francisco, CA; Ann Arbor, MI; Richmond, VA; San Antonio, TX; Brooklyn, NY; Birmingham, AL; Gainesville, FL; Martinsburg, VA; and Prescott, AZ), which included low/high complexity and small/large facilities.

One member of the HCC FAB completed an in-depth chart review of the patients from each facility, using a collaboratively developed data collection instrument, implemented in REDCap. This instrument guided clinicians to complete a structured EHR extraction to confirm the diagnosis of HCC and to determine staging and treatment characteristics, incorporating both VA and non-VA medical records. The instrument was identical to the one we used in our 2021 HCC extraction, thus enabling direct comparison of the CY2023 with the CY 2021 results, except for some augmented questions added in 2023 on how HCC was diagnosed and reasons for the lack of HCC-directed treatment. Questions regarding data elements of the chart review were discussed and resolved during scheduled monthly meetings of the HCC FAB.

Statistical Analysis

Univariable and multivariable logistic regression analysis was employed to determine the associations between selected characteristics and the likelihood of HCC being diagnosed at early stage (T1 or T2) rather than late stage (beyond T2). Characteristics that were statistically significant (P < .05) in univariable analysis were included in the multivariable analysis. Multinomial regression models were used to determine the associations between selected characteristics and the likelihood of receiving potentially curative treatments or noncurative treatment or no treatment. Multinomial regression was employed since the outcome has 3 categories. Characteristics found to be significant in univariable models were included in the multivariable model. Univariable and multivariable Cox proportional hazards models were used to assess the characteristics associated with time from diagnosis to death or end of data collection (04/01/2025).

Results

Sociodemographic Characteristics, Liver Disease Etiology, and Cirrhosis Status

The CY2023 HCC cohort (n = 194) was exclusively male, mean age 73 years, predominantly non-Hispanic White (63.5%), with substantial representation from persons of non-Hispanic Black (18.2%) and Hispanic-Latino (6.2%) race/ethnicity (Table 1). The most common liver disease etiology was cured hepatitis C virus (HCV) (42.8%) followed by metabolic dysfunction-associated steatotic liver disease (MASLD) (24.2%), alcohol-associated liver disease (13.9%), active HCV (5.2%), and MASLD and alcohol-associated liver disease (MetALD) (4.1%). Approximately 30% did not have cirrhosis or were unsure, another 12.9% had cirrhosis but it was only diagnosed after HCC, and only 56.7% had cirrhosis that was diagnosed before HCC. Most patients with cirrhosis had CTP class A (80.4%) or B (14.4%), very few had class C (3.6%), and 1.6% were unknown.

Table 1.

Comparison of Patients With HCC Diagnosed in the VA Healthcare System in 2023 Versus 2021: Patient Characteristics, Diagnosis, and Treatment

Characteristic	Year of HCC diagnosis		P value
	2021 N = 199	2023 N = 194
Sociodemographics and liver disease
Age, mean ± SD	71 ± 7	73 ± 8	.010
Male, n (%)	196 (98.5%)	194 (100%)
Race and ethnicity, n (%)			.302
Hispanic or Latino	16 (8.0%)	12 (6.2%)
Non-Hispanic Black or African American	61 (30.7%)	34 (18.2%)
Non-Hispanic White	102 (51.3%)	125 (63.5%)
Other/unknown	20 (10.0%)	20 (10.3%)
Etiology of liver disease, n (%)			.147
Cured HCV	93 (46.7%)	83 (42.8%)
Active HCV	32 (16.1%)	10 (5.2%)
ALD	31 (15.6%)	27 (13.9%)
MetALD	N/A	8 (4.1%)
MASLD	31 (15.6%)	47 (24.2%)
Othera	12 (6.0%)	19 (9.8%)
Cirrhosis status, n (%)			.841
No cirrhosis	45 (22.6%)	43 (22.2%)
Cirrhosis diagnosed before HCC	120 (60.3%)	110 (56.7%)
Cirrhosis diagnosed at the same time or after HCC	34 (17.1%)	25 (12.9%)
Cirrhosis, unsure	0 (%)	16 (8.2%)
MELD-Na score, mean ± SD	11 ± 5.0	10.7 ± 4.6	.536
CTP score at diagnosis			.276
5–6 (class A) or no cirrhosis	139 (69.8%)	156 (80.4%)
7–9 (class B)	46 (23.1%)	28 (14.4%)
≥ 10 (class C)	10 (5.0%)	7 (3.6%)
Missing	4 (2.0%)	3 (1.5%)
HCC diagnosis
HCC stage at the time of diagnosisb, n (%)			.783
T1	28 (14.1%)	34 (17.5%)
T2	79 (39.7%)	82 (42.3%)
Beyond T2	90 (45.2%)	78 (40.2%)
Missing	2 (1.0%)	0 (0%)
HCC diagnosed as a result of:			-
Screening ultrasound or serum AFP	N/A	51 (26.3%)
Screening CT or MRI	N/A	30 (15.5%)
Symptomatic presentation	N/A	36 (18.6%)
Incidental finding on imaging performed for other reasons	N/A	68 (35.1%)
Other	N/A	9 (4.6%)
Multidisciplinary liver tumor board review, n (%)	155 (77.9%)	138 (71.1%)	.181
Infiltrative HCC, n (%)b	37 (18.6%)	26 (13.4%)	.203
Vascular invasion, n (%)b	36 (18.1%)	20 (10.3%)	.051
Lymph node metastasis, n (%)b	20 (10.1%)	17 (8.8%)	.865
Distal metastasis, n (%)b	19 (9.5%)	22 (11.3%)	.799
Any high-risk characteristic (infiltration, vascular invasion, lymph node, or distal metastasis)b	61 (30.7%)	49 (25.3%)	.392
HCC treatment
Was HCC-directed treatment administered, n (%)			.693
No	52 (26.1%)	47 (24.2%)
Yes	147 (73.9%)	147 (75.8%)
First HCC-directed treatment administered, n (%)	Of 147	Of 147	.551
Ablation	36 (24.5%)	24 (16.3%)
TACE	36 (24.5%)	30 (20.4%)
Ablation + TACE	6 (4.1%)	7 (4.8%)
Transarterial radioembolization (TARE) using Yttrium-90 (Y90)	25 (17.0%)	42 (28.6%)
External beam radiation therapy	9 (6.1%)	8 (5.4%)
Surgical resection	17 (11.6%)	11 (7.5%)
Systemic therapy	16 (10.9%)	25 (17.0%)
Liver transplantation	2 (1.4%)	0 (0%)c
Untreated	52 (26.1%)	47 (24.2%)
Time (d) from date of diagnosis to date of administration of first HCC-directed treatment, n (%)	Of 147	Of 147	.646
0 to < 30	20 (13.6%)	23 (15.6%)
30 to < 60	38 (25.9%)	40 (27.2%)
60 to < 90	18 (12.2%)	33 (22.4%)
90 to < 120	16 (10.9%)	22 (15%)
> = 120	18 (12.2%)	28 (19%)
Untreated	52 (26.1%)	47 (24.2%)
Unable to determine	37 (21.4%)	1 (0.7%)
Reasons for no HCC-directed treatmentd, n (%) (more than 1 possible)	Of 52	Of 47	-
Patient declined	N/A	12 (25.5%)
Too many comorbidities	N/A	12 (25.5%)
Poor functional status	N/A	13 (27.7%)
Severity of underlying liver disease	N/A	10 (21.3%)
Very advanced stage of HCC	N/A	9 (19.1%)
Other reason/unable to tell	N/A	12 (25.5%)
Was the patient referred for liver transplantation, n (%)	18 (9.0%)	15 (7.7%)
Was the patient listed for liver transplantation, n (%)	8 (4.0%)	8 (4.1%)
Did the patient undergo liver transplantation, n (%)	2 (1.0%)	3 (1.5%)
Palliative care consultation (%)	66 (33.2%)	44 (22.7%)
Healthcare utilization
Facility at which HCC was diagnosed, n (%)			.067
VA facility	179 (89.9%)	160 (82.5%)
Non-VA facility	20 (10.1%)	34 (17.5%)
Prediagnosis primary care, n (%)			.740
VA facility	172 (86.4%)	166 (85.6%)
Non-VA facility	16 (8.0%)	15 (7.7%)
No PCP documented	11 (5.5%)	5 (2.6%)
Unable to tell	0 (0%)	8 (4.1%)
Prediagnosis liver care, n (%)			.433
VA facility	131 (65.8%)	124 (63.9%)
Non-VA facility	15 (7.5%)	23 (11.9%)
No prior liver care documented	53 (26.6%)	41 (21.1%)
Unable to tell	0 (0%)	6 (3.1%)

AFP, alpha fetoprotein; ALD, alcohol-associated liver disease; CTP, Child-Turcotte-Pugh; MELD-Na, Model for End-stage Liver Disease - Sodium; N/A, not applicable; PCP, primary care provider; SD, standard deviation.

^aOther etiologies in 2021 were: hepatitis B virus (n = 2), hereditary hemochromatosis (n = 1), cryptogenic cirrhosis (n = 2), primary biliary cirrhosis (n = 1), and no known underlying liver disease (n = 6). Other etiologies in 2023 were: hepatitis B virus (n = 3), hemochromatosis (n = 2), and no known underlying liver disease (n = 14).

^bTumor stage and characteristics at the time of HCC diagnosis.

^cThis refers to liver transplantation as first treatment. Three patients eventually underwent transplantation.

^dReasons for not receiving HCC treatment were not determined in the 2021 cohort.

Notable changes from the CY2021 cohort included a substantial increase in MASLD from 15.6% to 24.2%, with an additional 4.1% attributed to MetALD, a definition that did not exist in 2021, with a concomitant decline in cured HCV and especially active HCV.

HCC Stage at Diagnosis and Factors Associated With Early-Stage Diagnosis

The majority of HCC cases in the CY2023 cohort were diagnosed at stage T1 (17.5%) or T2 (42.3%); however, 40.2% were diagnosed beyond T2 and 25.3% had at least one high-risk characteristic (infiltration, vascular invasion, lymph node, or distal metastasis). This was a slight improvement from the CY2021 cohort in which 45.2% were diagnosed beyond T2 and 30.7% had at least one high-risk characteristic.

Only a minority of cases were diagnosed as a result of screening by ultrasound and AFP (26.3%) or screening by CT or MRI (15.5%). The most common pathway for HCC diagnosis was as an incidental finding in imaging performed for other reasons (35.1%), while symptomatic presentation accounted for 18.6% (Table 1).

The proportion of patients diagnosed with early-stage HCC (ie, T1 or T2, overall 59.8%) was higher in patients with a known cirrhosis diagnosis prior to HCC (70.9%) than in those with cirrhosis diagnosed after HCC (40.0%) or no cirrhosis (48.8%) (Table 2). Also, the proportion with early-stage HCC was higher among those diagnosed by ultrasound or AFP (86.3%) or screening CT or MRI (83.3%) than those presenting symptomatically (19.4%) or incidentally (51.1%). The proportion with early-stage HCC was higher in those diagnosed in VA vs non-VA facility (63.1% vs 44.1%), in those with a VA primary care provider (PCP) (63.3%), and in those with VA liver care (72.6%) prior to HCC diagnosis.

Table 2.

Factors Associated With HCC Diagnosis at Very Early (T1) or Early (T2) Stage Among 194 Patients Diagnosed With HCC in 2023 in the VA Healthcare System

Characteristic	HCC cases, N	Diagnosis at early stage (T1 or T2)	Odds ratio for early-stage diagnosis (95% CI)
All patients	194	116 (59.8%)	N/A
Sex, male	194	116 (59.8%)	N/A
Age, y
< 65	25	14 (56.0%)	1
65–75	100	58 (58.0%)	1.09 (0.44–2.62)
> 75	69	44 (63.8%)	1.38 (0.54–3.51)
Race and ethnicity, n (%)
Non-Hispanic White	125	80 (64.0%)	1
Hispanic or Latino	15	8 (53.3%)	0.64 (0.22–1.94)
Non-Hispanic Black or African American	34	19 (55.9%)	0.71 (0.33–1.55)
Other/unknown	20	9 (45.0%)	0.46 (0.17–1.19)
Etiology of liver disease
Cured HCV	83	55 (66.3%)	1
Active HCV	10	6 (60.0%)	0.76 (0.20–3.19)
ALD	27	15 (55.6%)	0.64 (0.26–1.56)
MASLD + MetALD	55	33 (60.0%)	0.76 (0.38–1.55)
Other	19	7 (36.8%)	0.30 (0.10-0.82)
Cirrhosis status
No documented cirrhosis	43	21 (48.8%)	1
Cirrhosis diagnosed at the same time or after HCC	25	10 (40.0%)	0.80 (0.25–1.88)
Cirrhosis diagnosed pre-HCC	110	78 (70.9%)	2.55 (1.24-5.32)
Cirrhosis, unsure	16	7 (43.8%)	0.82 (0.25–2.58)
MELD-Na
< = 10	116	70 (60.3%)	1
> 10	73	43 (58.9%)	0.94 (0.52–1.72)
CTP score (among cirrhosis)
5–6 (CTP A cirrhosis)	156	98 (62.8%)	1
> = 7 (CTP B or C cirrhosis)	35	16 (45.7%)	0.49 (0.24–1.04)
Unknown	3	2 (66.7%)	N/A
HCC diagnosed as a result of:
Incidental finding on imaging performed for other reasons	68	35 (51.5%)	1
Screening study (ultrasound or AFP) in a patient with known cirrhosis	51	44 (86.3%)	5.93 (2.45-16.07)
Screening CT or MRI	30	25 (83.3%)	4.71 (1.72-15.27)
Symptomatic presentation	36	7 (19.4%)	0.23 (0.08–0.57)
Other/unable to determine	9	5 (55.6%)	1.18 (0.29–5.12)
Functional/performance status Eastern Cooperative Oncology Group (ECOG)
0 (fully active, high-performance status)	63	46 (73.0%)	1
1–4 (low performance status)	86	49 (57.0%)	0.49 (0.24-0.98)
Unable to determine	45	21 (46.7%)	0.32 (0.14-0.72)
Site of HCC diagnosis
Non-VA facility	34	15 (44.1%)	1
VA facility	160	101 (63.1%)	2.17 (1.03-4.65)
Prediagnosis primary care, n (%)
Non-VA PCP or no PCP	20	8 (40.0%)	1
VA PCP	166	105 (63.3%)	2.58 (1.01-6.92)
Prediagnosis liver care
Non-VA liver care or no liver care	64	24 (37.5%)	1
VA liver care	124	90 (72.6%)	4.41 (2.34-8.49)

Bold values represent statistically significant results.

ALD, alcohol-associated liver disease; CI, confidence interval; N/A, not applicable.

HCC-Directed Treatments

Most patients in the CY2023 cohort were referred to an MLTB (n = 138, 71.1%) and 147 (75.8%) received HCC-directed treatment, which were similar proportions to the 2021 cohort. However, the distribution of HCC-directed treatments changed substantially from 2021 to 2023. The most common first treatment in CY2023 was selective internal radiation therapy Y90 (28.6%) followed by ablation with or without transarterial chemoebolization (TACE) (21.1%), TACE (20.4%), systemic therapy (17.0%), surgical resection (7.5%), and external beam radiation therapy (5.4%); as compared to CY2021 when the most common first treatment was ablation with or without TACE (28.6%), followed by TACE (17.7%), SIRT-Y90 (17.0%), surgical resection (11.6%), systemic therapy (10.9%), and external beam radiation therapy (6.1%).

Reasons for not receiving HCC-directed treatment included almost equally patient declining, too many comorbidities, poor functional status, severity of underlying liver disease, or very advanced HCC stage (Table 1).

Only 42 patients (21.6%) received a first-line potentially curative treatment (ie, ablation, surgical resection, or liver transplantation), while an additional 3 patients eventually underwent transplantation. Receipt of potentially curative treatment was more common in the following groups: early-stage diagnosis (35.8%), diagnosed by screening (35.3%), and prediagnosis liver care (28.2%) (Table 3).

Table 3.

Factors Associated With HCC-Directed Treatment Among 194 Patients Diagnosed With HCC in 2023 in the VA Healthcare System

Characteristic	HCC cases, N	Number (%) of patients who received HCC-directed treatment
		Potentially curativea treatment	Noncurativea treatment	No treatment
All patients	194	42 (21.6%)	105 (54.1%)	47 (24.2%)
HCC stage at diagnosis
Beyond T2	78	4 (5.1%)	50 (64.1%)	24 (30.8%)
T1 or T2	106	38 (35.8%)	55 (51.9%)	23 (21.7%)
Age, y
< 65	25	8 (32.0%)	12 (48.0%)	5 (20.0%)
65–75	100	19 (19.0%)	57 (57.0%)	24 (24.0%)
> 75	69	15 (21.7%)	36 (52.2%)	18 (26.1%)
Race and ethnicity, n (%)
Non-Hispanic White	125	33 (26.4%)	58 (46.4%)	34 (27.2%)
Hispanic or Latino	15	0 (0%)	11 (73.3%)	4 (26.7%)
Non-Hispanic Black or African American	34	6 (17.6%)	21 (61.8%)	7 (20.6%)
Other, unknown or missing	20	3 (15.0%)	15 (75.0%)	2 (10.0%)
Etiology of liver disease
Cured HCV	83	19 (22.9%)	46 (55.4%)	18 (21.7%)
Active HCV	10	0 (0%)	7 (70.0%)	3 (30.0%)
ALD	27	6 (22.2%)	10 (37%)	11 (40.7%)
MASLD + MetALD	55	14 (25.5%)	32 (58.2%)	9 (16.4%)
Other	19	3 (15.8%)	10 (52.6%)	6 (31.6%)
Cirrhosis status
No cirrhosis	43	10 (23.3%)	28 (65.1%)	5 (11.6%)
Cirrhosis diagnosed at the same time or after HCC	25	1 (4.0%)	18 (72.0%)	6 (24.0%)
Cirrhosis diagnosed before HCC	110	30 (27.3%)	49 (44.5%)	31 (28.2%)
Cirrhosis, unsure	16	1 (6.3%)	10 (62.5%)	5 (31.3%)
MELD-Na
< = 10	116	32 (27.6%)	67 (57.8%)	17 (14.7%)
> 10	73	8 (11.0%)	36 (49.3%)	29 (39.7%)
CTP score (among cirrhosis)
5–6 (CTP A cirrhosis)	156	39 (25.0%)	94 (60.3%)	23 (14.7%)
> = 7 (CTP B or C cirrhosis)	35	1 (2.9%)	10 (28.6%)	24 (68.6%)
HCC diagnosed as a result of:
Incidental finding on imaging performed for other reasons	68	11 (16.2%)	43 (63.2%)	14 (20.6%)
Screening study (ultrasound or AFP) in a patient with known cirrhosis	51	18 (35.3%)	25 (49.0%)	8 (15.7%)
Screening CT or MRI	30	9 (30.0%)	10 (33.3%)	11 (36.7%)
Symptomatic presentation	36	2 (5.6%)	20 (55.6%)	14 (38.9%)
Other, unable to determine or missing	9	2 (22.2%)	7 (77.8%)	0 (0%)
Functional/performance status ECOG
0 (fully active)	63	22 (34.9%)	33 (52.4%)	8 (12.7%)
1–4	86	13 (15.1%)	46 (53.5%)	27 (31.4%)
Unable to determine	45	7 (15.6%)	26 (57.8%)	12 (26.7%)
Facility at which HCC was diagnosed
Non-VA facility	34	7 (20.6%)	24 (70.6%)	3 (8.8%)
VA facility	160	35 (21.9%)	81 (50.6%)	44 (27.5%)
Prediagnosis primary care, n (%)
Non-VA PCP or no PCP	20	2 (10.0%)	10 (50.0%)	8 (40.0%)
VA PCP	166	39 (23.5%)	92 (55.4%)	35 (21.1%)
Prediagnosis liver care
Non-VA liver care or no liver care	64	7 (10.9%)	43 (67.2%)	14 (21.9%)
VA liver care	124	35 (28.2%)	58 (46.8%)	31 (25.0%)

Noncurative treatment includes SIRT/Y-90, TACE, external beam radiation, and systemic treatments.

ALD, alcohol-associated liver disease.

^aPotentially curative treatments include radiofrequency ablation (with or without concomitant TACE), surgical resection, and liver transplantation.

Median time from diagnosis to first HCC-directed treatment was 68 days (interquartile range: 39–107). Overall, 15.6% (23/147) received treatment within 30 days and another 27.2% (40/147) from 30 to < 60 days of diagnosis. There was no substantial variability in time to treatment by different characteristics (Table A1).

Survival

Over a mean follow-up of 481 days (minimum = 15 days, maximum = 819 days), 84 of 194 patients died (43.3%). Cumulative all-cause mortality was 15.5% at 6 months, 29.9% at 1 year, and 44.8% at 2 years (Table 4).

Table 4.

Factors Associated With All-Cause Mortality From the Time of HCC Diagnosis Among 194 Patients Diagnosed With HCC in 2023 in the VA Healthcare System With Follow-Up Extending to April 2025

Characteristic	Total patients, n	Hazard ratio for death (95% CI)	Cumulative incidence of mortality 6 mo 1-y 2-y
All HCC patients	194	N/A	15.5%	29.9%	44.8%
HCC treatment
None	47	1	48.9%	66.0%	74.6%
Curative treatment	42	0.14 (0.07–0.30)	4.8%	9.5%	22.4%
Noncurative treatment	105	0.27 (0.17–0.43)	4.8%	21.9%	40.3%
HCC stage at diagnosis
Beyond T2	78	1	26.9%	46.2%	66.5%
T1 or T2	116	0.36 (0.23–0.55)	7.8%	18.1%	31.4%
Age, y
< 65	25	1	20.0%	28.0%	46.6%
65–75	100	0.92 (0.47–1.78)	13.0%	29.0%	41.8%
> 75	69	0.99 (0.50–1.96)	15.9%	30.4%	45.4%
Race and ethnicity, n (%)
Non-Hispanic White	125	1	17.6%	29.6%	39.0%
Hispanic or Latino	15	1.87 (0.92–3.80)	26.7%	46.7%	53.3%
Non-Hispanic Black or African American	34	1.16 (0.66–2.03)	5.9%	23.5%	50.4%
Other, unknown or missing	20	1.20 (0.61–2.38)	5.0%	25.0%	55.4%
Etiology of liver disease
Cured HCV	83	1	12.0%	22.9%	38.4%
Active HCV	10	2.77 (1.22–6.33)	20.0%	60.0%	70.0%
ALD	27	1.71 (0.91–3.22)	25.9%	40.7%	51.9%
MASLD + MetALD	55	1.01 (0.58–1.75)	10.9%	21.8%	40.3%
Other	19	2.02 (1.01–4.02)	21.1%	42.1%	60.5%
Cirrhosis status
No cirrhosis	43	1	9.3%	25.6%	35.5%
Cirrhosis diagnosed at the same time or after HCC	25	2.20 (1.07–4.51)	16.0%	36.0%	60.4%
Cirrhosis diagnosed before HCC	110	1.27 (0.70–2.32)	15.5%	29.1%	40.9%
Cirrhosis, unsure	16	2.06 (0.89–4.76)	25.0%	31.3%	56.3%
MELD-Na
< = 10	116	1	8.6%	20.7%	34.0%
> 10	73	2.22 (1.43–3.43)	24.7%	41.1%	59.8%
CTP score (among cirrhosis)
5–6 (CTP A cirrhosis)	156	1	7.1%	20.5%	35.7%
> = 7 (CTP B or C cirrhosis)	35	5.25 (3.32–8.29)	51.4%	71.4%	86.7%
HCC diagnosed as a result of:
Incidental finding on imaging performed for other reasons	68	1	8.8%	22.1%	39.2%
Screening study (ultrasound or AFP) in a patient with known cirrhosis	51	0.72 (0.38–1.35)	3.9%	13.7%	30.8%
Screening CT or MRI	30	1.00 (0.49–2.02)	10.0%	26.7%	37.5%
Symptomatic presentation	36	3.91 (2.29–6.67)	50.0%	66.7%	83.6%
Other, unable to determine or missing	9	0.98 (0.30–3.24)	0.0%	22.2%	33.3%
Functional/performance status ECOG
0 (fully active)	63	1	3.2%	11.1%	21.9%
1–4	86	3.46 (1.87–6.40)	20.9%	38.4%	55.0%
Unable to determine	45	3.49 (1.77–6.86)	20.0%	35.6%	58.1%
Facility at which HCC was diagnosed
Non-VA facility	34	1	14.7%	29.4%	48.6%
VA facility	160	0.89 (0.52–1.54)	15.6%	30.0%	43.9%
Prediagnosis primary care, n (%)
Non-VA PCP or no PCP	20	1	30.0%	55.0%	80.0%
VA PCP	166	0.34 (0.19–0.60)	13.3%	25.3%	38.8%
Prediagnosis liver care
Non-VA liver care or no liver care	64	1	18.8%	39.1%	57.3%
VA liver care	124	0.60 (0.38–0.93)	12.9%	25.0%	38.1%

Bold values represent statistically significant results.

ALD, alcohol-associated liver disease; CI, confidence interval; N/A, not applicable.

All-cause mortality was lower in patients with early-stage (T1 or T2) rather than late-stage (beyond T2) diagnosis (31.0% vs 61.5%), patients who received potentially curative or noncurative treatments than no treatment (25% vs 41.3% vs 74.5%), patients who had cured HCV or MASLD/MetALD than active HCV (37.3% vs 38.2% vs 70.0%), patients with no cirrhosis than those with cirrhosis diagnosed after HCC diagnosis (32.6% vs 64.0%), patients with MELD-Na ≤ 10 than > 10 (32.8% vs 58.9%), patients with no cirrhosis or CTP class A than those with CTP B/C (34.0% vs 85.7%), patients diagnosed by screening ultrasound/AFP or screening CT/MRI rather than symptomatic presentation (29.4% vs 36.7% vs 80.6%), patients with performance status 0 than 1 to 4 (20.5% vs 54.7%), patients with VA PCP than no VA PCP (37.3% vs 75.0%), and those with VA liver care than no VA liver care (37.1% vs 54.7%) (Figure). Age was not associated with mortality.

Figure.

Cumulative incidence of death from the time of diagnosis of HCC, according to key baseline characteristics.

Discussion

Our systematic assessment of the diagnosis and management of HCC cases diagnosed in the national VA healthcare system in 2023 reveals some persistent gaps in early-stage diagnosis (59.8%), receipt of any HCC-directed treatment (75.8%), receipt of potentially curative treatment (21.6%), timely receipt of HCC-directed treatment within 30 (15.6%) or 60 days (42.8%) of diagnosis among those treated, and high all-cause 1-year (29.9%) and 2-year (44.8%) mortality. Compared to VA HCC cases diagnosed in 2021, HCC cases diagnosed in 2023 were more likely to have MASLD/MetALD (28.3% vs 15.6%), to be diagnosed at stage T1/T2 (59.8% vs 53.8%), and to receive SIRT-Y90 (28.6% vs 17.0%) or systemic therapy (17.0% vs 10.9%) as first-line treatment. In this quality improvement effort executed by the VA NGHP, we identified many factors and processes associated with HCC stage of diagnosis, treatment, and survival and demonstrated the feasibility and utility of conducting periodic evaluations in a large integrated healthcare system to better understand potential areas for system-level intervention and improve care.

Although an early-stage diagnosis (T1 or T2) of 59.8% may seem low, it is higher than what has been reported in other healthcare systems, regions, or populations.^10,11 Furthermore, this proportion must be interpreted in the context of the proportion of HCC cases who had cirrhosis. Only 56.7% of HCC cases had known and documented cirrhosis prior to HCC diagnosis and therefore only this subset would have the potential to undergo HCC surveillance aiming for early detection. Indeed, 70.9% of HCC cases in patients with previously diagnosed cirrhosis were diagnosed at early stage. This approaches the limit of current HCC screening strategies and would likely only substantially increase if improved screening strategies are introduced.

It is remarkable that 22.2% of HCC cases did not have cirrhosis and in an additional 8.2% there was uncertainty about whether cirrhosis was present based on our expert hepatologists’ chart review, consistent with increasing recognition of HCC associated with noncirrhotic metabolic dysfunction associated steatohepatitis and HCV.12, 13, 14, 15 Therefore, unfortunately, in almost one-third of HCC cases, there is little that can be done to improve the probability of early detection since screening is not recommended, until better individualized risk prediction tools are developed to identify high-risk patients who would qualify for screening among the noncirrhotic population.

Finally, cirrhosis was present but not yet diagnosed at the time of HCC detection in 12.9% of HCC cases (down from 17% in our 2021 survey and much lower than the proportions of undiagnosed cirrhosis reported in HCC patients in older, non-VA studies16, 17, 18), of whom only 40% were diagnosed at early stage. If cirrhosis is not diagnosed and documented, screening for HCC cannot occur. Strategies to improve diagnosis and documentation of cirrhosis are being planned by the national VA Gastroenterology/Hepatology Program and need to be implemented, especially in an era where MASLD is becoming the predominant underlying condition, posing additional algorithmic problems because of a hugely expanded at-risk population.

Our results reinforce the importance of screening, which is underused,¹¹ for early detection: the proportion with early-stage HCC was higher among those diagnosed by ultrasound or AFP (86.3%) or screening CT or MRI (83.3%) than those presenting symptomatically (19.4%) or incidentally (51.1%). A more novel finding was that a substantial proportion of HCCs were diagnosed by screening CT or MRI scanning (15.5%), not much lower than the proportion diagnosed by screening ultrasound or serum AFP (26.3%). Although screening by CT or MRI is not recommended in general, guidelines suggest considering CT/MRI screening when ultrasound is considered inadequate, for example, due to poor visualization.

A high proportion of patients (75.8%) received HCC-directed treatments, with documentation of adequate reasons for no treatment in almost all untreated patients (eg, patient declining, too many comorbidities, poor functional status, severity of underlying liver disease, or very advanced HCC stage). However, receipt of potentially curative treatments as first-line was relatively uncommon overall (21.6%), and more common in those early-stage diagnosis (35.8%), diagnosed by screening (35.3%), and prediagnosis liver care (28.2%). Delays in receipt of HCC treatment were unfortunately common with only 15.6% (23/147) received treatment within 30 days and another 27.2% (40/147) from 30 to < 60 days of diagnosis. In previous studies, we identified complex and multifactorial reasons for such delays including both patient-related and system-related factors.⁷ Efforts and interventions aiming to reduce the time from diagnosis to treatment will be considered by the national VA GI/Hepatology Program.

All-cause mortality was very high in our HCC cohort (15.5% at 6 months, 29.9% at 1 year, and 44.8% at 2 years despite substantial proportions diagnosed early and treated). Factors associated with lower all-cause mortality included potentially modifiable factors (early-stage diagnosis, diagnosis via screening, cured HCV, receipt of curative treatments, and prediagnosis VA primary or liver care) as well as nonmodifiable factors (CTP class A, MELD ≤ 10, absence of cirrhosis, and good performance status). These results point to multiple targets for implementation interventions aimed at ultimately improving survival, but also highlight the persistent problem of low survival in patients with HCC.

Our results highlight the importance of comprehensive, nonfragmented care and engagement with primary care and liver care to optimize outcomes for the entire continuum of HCC management. Patients with HCC diagnosed in VHA (vs non-VHA facility), those with prediagnosis VA PCP (vs no PCP or non-VHA PCP), and those with prediagnosis VHA liver care (vs no liver care or non-VHA liver care) were more likely to be diagnosed at early stage, to receive potentially curative treatments and to survive. VHA is engaging in ongoing, proactive efforts to identify Veterans with cirrhosis and to increase HCC screening rates using educational outreach, population health approaches to identifying at-risk patients, provision of advanced liver disease dashboards to facilitate cirrhosis care including HCC screening, and execution of a large national trial of the comparative effectiveness of HCC screening modalities.^19,20

Noteworthy changes occurred even in the short space of 2 years between our prior assessment of HCC cases diagnosed in VHA in 2021⁷ and the current assessment of 2023 cases. The proportion of HCC cases with underlying MASLD (24.2%) or MetALD (4.1%) in 2023 increased relative to the proportion of cases of what was then termed nonalcoholic fatty liver disease in 2021 (15.6%), mirroring national and international trends in increasing proportions of HCCs attributed to MASLD. There was a dramatic reduction in HCC cases with active HCV from 16.1% in 2021 to 4.2% in 2023 and a reduction in cases with HCV cured or active combined from 62.8% in 2021 to 48.0% in 2023. This is consistent with data demonstrating a reduction in HCC risk associated with HCC eradication21, 22, 23 and successful HCV elimination efforts achieved in VHA.^24,25 However, despite the decline in HCV and rise in MASLD/MetALD, HCV (cured and active combined) remained the most common underlying liver disease in VHA HCC cases in 2023 (48%). There was an encouraging trend toward higher rates of early-stage diagnosis in 2023 (T1 17.5%, T2 42.3%, and T1/T2 59.8%) compared to 2021 (T1 14.1%, T2 39.7%, and T1/T2 53.8%). Finally, in 2023, transarterial radioembolization Y90 emerged as the most common first treatment (28.6%), while in 2021 it accounted for 17.0% of first treatments, reflecting national trends and published evidence of Y90 effectiveness. Additionally, systemic therapy as first treatment increased from 10.9% to 17.0%, reflecting recent advances and updated guidelines for advanced-stage HCC treatment with immune checkpoint inhibitor and angiogenesis inhibitor therapies.^1,2 These trends over time support the ongoing biennial assessment of HCC by the VHA NGHP to provide timely updates aimed at optimizing care, with the next assessment planned for new HCC cases diagnosed in 2025.

Our results are observational and should be interpreted with caution, especially when attributing causality or speculating on the effects of potential interventions. For example, improved survival may be attributed to screening programs, but lead-time bias (measured survival time from diagnosis increases but actual time of death remains the same), length-time bias (slow-growing, less aggressive cancers more likely to be detected through screening), and selection bias (patients who opt into screening tends to be more health conscious and have better outcomes) certainly contribute to the observed associations between screen-detected cancers and lower mortality. Also, our results should be extrapolated to non-VA populations with caution, because of VA patient demographics (mostly male), unique exposures related to service, and higher burden of comorbidities than the general US population.

Conclusion

This biennial quality improvement project demonstrated the feasibility and value of leveraging a national network of clinicians to conduct timely chart reviews to track HCC care with the potential to improve early detection, intervention, and outcomes. We intend to repeat this quality improvement exercise to provide an ongoing assessment of potential gaps in HCC care and recommendations to bridge these gaps.