Cost-effectiveness analysis of hepatocellular carcinoma screening by combinations of ultrasound and alpha-fetoprotein among Alaska Native people, 1983–2012

Figures & data

Table I. Epidemiologic data for a cost analysis from a prospectively followed cohort of AN persons with chronic hepatitis B virus infection, 1983–2012

Epidemiologic input	Data
Number of cohort participants^a	839
Males (%)	564 (67%)
Number of cohort participants with HCC (%)	21 (2.5%)
Early-stage tumour^b	10
Median (mean) years survival by tumour stage
Early stage^b	3.1 (5.0)
Late stage	0.2 (0.8)
Total number of potential screening opportunities^c	21,226
Total number of AFP measurements in cohort^d	10,931
Median AFP measurements/person	11
% of total potential screening opportunities	51%

AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma; US, ultrasound.

^aComprises males attaining age ≥40 years and females attaining age ≥50 years during January 1, 1983 to December 31, 2012. ^bSingle tumour ≤5 cm in diameter or ≤3 tumours ≤3 cm in diameter. ^cAssumes persons received screening for HCC every 6 months. ^dAssumes persons received screening for HCC on dates for which an AFP measurement was documented (excluding AFP measurements occurring <4 months after a prior measurement).

Table II. Base-case model assumptions for estimating the costs for hepatocellular carcinoma screening

Model input	Assumption
Number of screening tests performed by AFP→US^a	10,931
Number of AFP measurements	9,378
Number of US screenings	1,553
Number of ultrasounds performed by US-alone method^b	10,931
Tumours detected at an early stage by AFP→US method^c	10
Tumours detected at an early stage by US-alone method according to size^c,d	14
Direct costs per test
(Medicare reimbursement rates in 2012)
AFP	$109.94
US	$26.76
Roundtrip cost/patient to an US-equipped facility^e	$200
% cohort members requiring transportation to US facility	60%

AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma; US, ultrasound.

^aAssumes patients received screening for HCC by serum AFP measurements initially and switched to ultrasound for all subsequent screenings if AFP >10 ng/mL. ^bAssumes patients received screening for HCC by ultrasound on dates for which AFP measurements are recorded. ^cSingle tumour ≤5 cm in diameter or ≤3 tumours ≤3 cm in diameter. ^dAssumes 33% (4) of tumours identified at a late stage by the AFP→US method were identified by the US-alone method at an early stage. ^eFor patients living in rural Alaska areas without ready access to US.

Table III. Comparing the costs of 2 hypothetical screening scenarios for hepatocellular carcinoma (HCC) – Alaska, 1983–2012^a

	Costs/benefits without transportation expenses^b		Costs/benefits with transportation expenses^b
	AFP→US^c,d	US-alone^d	AFP→US^c,d	US-alone^d
Analysis without discounting
Total cost for cohort (Base Year 2012)	$528,000	$1,203,000	$868,000	$2,517,000
No. of early-tumours detected^e	10	14	10	14
Median (mean) YLG for Cohort	29.6 (42)	41.4 (58.8)	29.6 (42)	41.4 (58.8)
Cost/early-stage tumour detected	$53,000	$86,000	$87,000	$180,000
Cost/YLG at median (mean)	$18,000 ($13,000)	$29,000 ($20,000)	$29,000 ($21,000)	$61,000 ($43,000)
Incremental cost-effectiveness ratios^f
Extra cost ($)/extra early-tumour detected	$169,000		$412,000
Extra cost ($)/Extra YLG at Median (Mean)	$57,000 ($40,000)		$139,000 ($98,000)
Analysis with discounting^g
Total cost for cohort (Base year 2012)	$357,000	$814,000	$587,000	$1,702,000
No. of early-tumours detected^e	10	14	10	14
Median (mean) YLG for Cohort	27.8 (38.1)	38.9 (53.3)	27.8 (38.1)	38.9 (53.3)
Cost/early-stage tumour detection	$36,000	$58,000	$59,000	$122,000
Cost/YLG at median (mean)	$13,000 ($9,400)	$21,000 ($15,000)	$21,000 ($15,000)	$44,000 ($32,000)
Incremental cost-effectiveness ratios^f
Extra cost ($)/extra early-tumour detected	$114,000		$279,000
Extra cost ($)/extra YLG at median (mean)	$41,000 ($30,000)		$100,000 ($73,000)

AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma; US, ultrasound; YLG, years of life gained.

^aIn a cohort of 839 hepatitis B virus infected AN men aged ≥40 years and women aged ≥50 years. ^bTotal costs rounded to the nearest thousand. ^cAssumes patients received screening for HCC by serum AFP measurements initially and switched to ultrasound for all subsequent screenings if AFP >10 ng/mL. ^dAssumes patients received screening for HCC on dates for which AFP measurements were recorded. ^eEarly-tumour if single tumour ≤5 cm in diameter or ≤3 tumours ≤3 cm in diameter; model assumes 33% (4) of tumours identified at a late stage by the AFP→US method were identified by the US-alone method at an early stage. ^fRatio=(costs US-alone – costs AFP→US)/(outcome US-alone – outcome AFP→US), where outcomes are the number of early-tumours detected or number YLG by early detection. Treatment costs after the detection of tumour (early or late) are not included in these estimates. ^gDiscounted direct costs of screening and YLG at 3%/year (reference year 2012) over a 30-year time horizon.

Fig. 1.  Sensitivity analysis: Impact of varying the percentage of hepatocellular carcinoma tumours that were identified by AFP→US at a late stage and potentially identified by US-alone at an early stage in hepatitis B virus infected AN persons.*^†‡§¶ Abbreviations: AFP, alpha-fetoprotein; US, ultrasound; ↓, indicates base-case assumption *Screening assumed to start for men at age ≥40 years and for women at age ≥50 years. ^†AFP→US assumes patients received screening for HCC by serum AFP measurements initially and switched to ultrasound if AFP >10 ng/mL; screening method resembles the Alaska Native Health System hepatocellular carcinoma screening program. ^‡The number of additional tumours that might have been detected at an early stage (i.e. single tumour ≤5 cm in diameter or ≤3 tumours ≤3 cm in diameter) by an US-alone method is unknown; therefore, sensitivity analysis determined the cost/early-stage tumour detected by assuming US-alone method identified 0–100% of the tumours identified by AFP→US at a late stage. ^§Direct costs of screening discounted at 3%/year (reference year 2012) over a 30-year time horizon. ^¶Assumes 60% of patients lived in a village without ready US access and required transportation to US facility.

Appendix Table I. Demographic and clinical characteristics of the cohort participants with history of hepatitis B virus (HBV) infection and those who developed hepatocellular carcinoma (HCC) – Alaska, 1983–2012.^a

	Participants with HBV^b	Participants with HCC
Characteristics	N=839	N=21
Ethnic Group
Eskimo	656 (78%)	21 (100%)
Aleut	87 (10%)	0
Indian	96 (11%)	0
Urban residence	211 (25%)	5 (24%)
Patients alive^c	553 (66%)	3 (14%)
Mean (median) age of patients alive^c	56 (55) years	66 (73) years
Mean (median) person-years of follow-up	12 (11) years	10 (9) years
Mean (median) age at HCC diagnosis	–	61 (61) years
HBV Genotype
A	105 (13%)	2 (10%)
B	44 (5%)	0
C	54 (6%)	9 (43%)
D	387 (46%)	5 (24%)
F	117 (14%)	5 (24%)
H	1 (0.1%)	0
Unknown	131 (16%)	0
Cirrhosis present at HCC diagnosis^d	Unknown	8 (38%)

^aAnalysis restricted to males aged >40 years and females aged >50 years. ^bIncludes patients with HBV infection who subsequently cleared hepatitis B surface antigen. ^cAs of 12/31/2012. ^dBiopsy confirmed.

Lederle FA, Pocha C. Screening for liver cancer: the rush to judgment. Ann Intern Med. 2012; 156: 387–9. [PubMed Abstract].

 Google Scholar