Study reference

Study characteristics

Patient characteristics

Index test

(test of interest)

Reference test

Follow-up

Outcome measures and effect size

Comments

Kim, 2016

Type of study[1]: Prospective study

Setting and country: Single academic tertiary care center, Korea

Funding and conflicts of interest: Supported by Bayer Healthcare, which also provided the study drug (gadoxetic acid). Bayer was permitted to review the manuscript and suggest changes but had no role in study design, data collection, analysis, decision to publish or preparation of the manuscript

Inclusion criteria:

- Age of 20 years or older

- Presence of cirrhosis with estimated annual HCC risk of more than 5% (Risk of Hepatocellular carcinoma (HCC) was estimated by using a model, if risk index was greater than 2.33 that was estimated to correspond to an annual risk of developing HCC of more than 5%)

- Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1

- Absence of previous history of current suspicion of HCC

Exclusion criteria:

- Child-Pugh class C liver function

- Estimated glomerular filtration rate < 30ml/min/1.73m²

N=407

Prevalence: 10%

Age in years, median (IQR): 56 (52-62)

Sex: 56.5% Male / 43.5% Female

Other important characteristics:

Cause of cirrhosis, N (%):

Hepatitis B virus: 288 (70.8%)

Hepatitis C virus: 37 (9.1%)

Alcohol: 52 (12.8%)

Other: 30 (7.4%)

Child-Pugh score class A, N (%):

320 (78.6%)

Child-Pugh score class B, N (%):

87 (21.4%)

Alpha-fetoprotein in mg/dl, median (IQR): 3 (2-5)

Index test: Liver Magnetic Resonance Imaging (MRI) with 1.5-T scanner (Magnetom Avanto; Siemens). Gadoxid acid (Primovist; Bayer) was administered at a dose of 0.025 mmol/kg. Axial T1-weighted images of the arterial, portal, delayed and hepatobiliary phases were obtained at 4-mm slice thickness. Three rounds of MRI screening every six months.

Cut-off point(s):

Positive screening criterion category 5 on a 5-point scale for MRI indicating the likelihood of HCC.

Comparator test: Ultrasonographic (US) examinations. Three rounds of US screening every six months.

Cut-off point(s):

Positive screening criterion category 4 on a 4-point standardized scale for US indicating the likelihood of HCC.

Reference test[2]: Dynamic Computed Tomography (CT) scan and biopsy US guided or if a lesion was detected only by MRI, real-time US-CT fusion image guided.

Cut-off point(s): Confirmation of HCC was based on results of histologic examination and/or typical CT images with nodule > 1 cm with arterial hypervascularity and portal/delayed-phase washout as recommended by practice guidelines.

Time between the index test en reference test:

When MRI or US examination detected a nodule scored as category 5 or 4, a recall process with dynamic 4-phase CT scan was performed within three months.

At 6-months after last screening round, all study patients were followed up with dynamic CT.

For how many participants were no complete outcome data available:

N=49

• N=38: Logistic problems
• N=10: Death
• N=1: Other

Outcome measures and effect size (include 95%CI and p-value if available):

Diagnostic accuracy MRI:

Detection rate for any HCC (sensitivity): 60.5%

Detection rate for very-early and early-stage HCC (sensitivity): 59.5%

Detection rate very-early stage HCC (sensitivity): 54.5%

Specificity: 99.3%

False-negative rate: 39.5%

False-positive rate: 0.7%

Positive predictive value: 78.8%

Diagnostic accuracy US:

Detection rate for any HCC (sensitivity): 27.9%

Detection rate of very early and early stage HCC (sensitivity): 26.2%

Detection rate very early stage HCC (sensitivity): 27.3%

False negative rate: 72.1%

Specificity: 94.4%

False positive rate: 5.6%

Positive predictive value: 16.9%

Incidence rate detected HCC:

N=43/407 (10.6%)

Authors conclusion: Results of this study support our hypothesis that MRI with liver-specific contrast is more sensitive than US to detect early stage HCC in high-risk patients with cir- rhosis. For very early stage HCC (single lesion <2 cm), MRI screening yielded a detection rate of 84.8%, significantly higher than the 27.3% detected by US.

Only 27.9% of the cancers were detected by US, which is far lower than reported in other meta-analyses. Possible explanation is that MRI was able to detect tumors far earlier in development than US.

Model to estimate risk of HCC: Risk index=1.41 (if age > 50 years) + 1.65 (prothrombin activity <75%) + 0.92 (platelet count is <100x10³/mm³) + 0.74 (if anti hepatitis C virus antibody or hepatitis B virus surface antigen test is positive)

Sponsor (Bayer) partially funded the study and review the manuscripts and suggest changes.

Park, 2020

Type of study: Retrospective analysis of prospectively collected data^a

Setting and country: Single, academic, tertiary center, Korea

Funding and conflicts of interest: Original study was funded by Bayer. For this retrospective analysis, authors received no financial support and declare no conflicts of interest

Inclusion criteria:

- Age of 20 years or older

- Presence of cirrhosis with estimated annual HCC risk of more than 5% (Risk of Hepatocellular carcinoma (HCC) was estimated by using a model, if risk index was greater than 2.33 that was estimated to correspond to an annual risk of developing HCC of more than 5%)

- Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1

- Absence of previous history of current suspicion of HCC

Exclusion criteria:

- Child-Pugh class C liver function

- Estimated glomerular filtration rate < 30ml/min/1.73m²

- Withdrawn from study or died without subsequent follow-up examinations

N=382

Prevalence: 5%

Age in years, median (IQR): 56.4 (29-77)

Sex: 56.8% Male /43.2% Female

Other important characteristics:

Cause of cirrhosis, N (%):

Hepatitis B virus: 276 (72.3%)

Alcohol-induced: 46 (12.0%)

Hepatitis C virus: 34 (8.9%)

Other: 26 (6.8%)

Alpha-fetoprotein in ng/ml, median (IQR): 7.0 (2.5-5.6)

Index test: Liver Magnetic Resonance Imaging (MRI) with 1.5-T scanner (Magnetom Avanto; Siemens). The stimulated non-enhance MRI set consisted of axial DWI and T2WI. Three rounds of MRI screening every six months.

Cut-off point(s):

Lesion ³ 1 cm. with either diffusion restriction or mild to moderate T2 hyperintensity.

Comparator test: Ultrasonographic (US) examinations by 4-board certified abdominal radiologist using a convex probe (SC6-1, Supersonic Image SA; Aixplorer, France). Three rounds of US screening every six months.

Cut-off point(s):
Focal lesions ³ 1 cm on US that met 1 or more of the following criteria, were considered positive: i) discrete focal mass distinguishable from the adjacent parenchyma, ii) peripheral low echoic halo, iii) mosaic pattern, and iv) definite tumor thrombi visible on US.

Reference test: Dynamic Computed Tomography (CT) scan, biopsy and/or subsequent surveillance round(s) after six months.

Cut-off point(s): Confirmation of HCC was based on results of histologic examination and/or typical CT images with nodule > 1 cm with arterial hypervascularity and portal/delayed-phase washout as recommended by practice guidelines.

Time between the index test and reference test:

When MRI or US examination detected a nodule scored as category 5 or 4, a recall process with dynamic 4-phase CT scan was performed within three months.

At 6-months after last screening round, all study patients were followed up with dynamic CT.

Average duration of follow-up after last surveillance: 32.9 months (range 1-60 months)

For how many participants were no complete outcome data available:

N=43

• N=34: Withdrawn from study
• N=9: Death

Outcome measures and effect size (include 95%CI and p-value if available):

Diagnostic accuracy non-enhanced MRI:

Per-lesion sensitivity: 77.1% (N=37/48) (95%CI 63.2-86.8)

Per-lesion positive predictive value: 56.9% (N=37/65) (95%CI 44.9-69.0)

Per-exam sensitivity: 79.1% (N=34/43) (95%CI 64.4-88.7)

Per-exam positive predictive value: 61.8% (N=34/55) (95%CI 49.0-74.7)

Per-exam specificity: 97.9 (N=993/1014) (95%CI 96.8-98.7)

Diagnostic accuracy US:

Per-lesion sensitivity: 25.0% (N=12/48) (95%CI 14.8-39.1)

Per-lesion positive predictive value: 16.7% (N=12/72) (95%CI 8.1-25.3)

Per-exam sensitivity: 27.9% (N=12/43) (16.6-43.0)

Per-exam positive predictive value: 17.7% (N=12/68) (95%CI 8.6-26.7)

Per-exam specificity: 94.5% (N=958/1014) (95%CI 92.9-95.7)

Authors conclusion: This study demonstrated that non-enhanced MRI showed significantly better performance than US as a surveillance tool for HCC in a prospectively gathered cohort at a high risk of HCC.

Model to estimate risk of HCC: Risk index=1.41 (if age > 50 years) + 1.65 (prothrombin activity <75%) + 0.92 (platelet count is <100x10³/mm³) + 0.74 (if anti hepatitis C virus antibody or hepatitis B virus surface antigen test is positive)

Reference testing is CT, biopsy and/or subsequent surveillance round(s) after six months.

Park, 2021

Type of study: Retrospective analysis of prospectively collected data^a

Setting and country: Single, academic, tertiary center, Korea

Funding and conflicts of interest: Original study was funded by Bayer. For this retrospective analysis, dr. Amit Singal has served as a consultant for Bayer, Wako Diagnostics, Glycotest, Exact Science, Roche and TARGET pharmasolutions. Dr. Sang Hyun Choi received grants from Bayer outside the study. None of the authors declare conflicts of interest that pertain to this work

Inclusion criteria:

- Age of 20 years or older

- Presence of cirrhosis with estimated annual HCC risk of more than 5% (Risk of Hepatocellular carcinoma (HCC) was estimated by using a model, if risk index was greater than 2.33 that was estimated to correspond to an annual risk of developing HCC of more than 5%)

- Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1

- Absence of previous history of current suspicion of HCC

Exclusion criteria:

- History of suspicion of any type of malignancy, significant comorbidities with a predicted survival period < 3 years

- Estimated Glomerular Filtration Rate (GFR) < 30 ml/min/1.73m²

- Child Pugh class C

- Contra indications for MRI

N=382

Prevalence: 5%

Age in years, median (IQR): 56 (29-77)

Sex: 56.8% Male /43.2% Female

Other important characteristics:

Cause of cirrhosis, N (%):

Hepatitis B virus: 276 (72.3%)

Alcohol-induced: 46 (12.0%)

Hepatitis C virus: 34 (8.9%)

Other: 26 (6.8%)

Alpha-fetoprotein in ng/ml, median (IQR): 7.0 (2.5-5.6)

Index test: The CAA approach consisted of first round using Magnetic Resonance Imaging (MRI) with 1.5-T scanner (Magnetom Avanto; Siemens). and included T1-weighted imaging, DWI, T2WI and contrast-enhanced T1-weighted imaging and the second and third round using Abbreviated MRI (AMRI) which consisted of DWI, T2WI and HBP imaging

Cut-off point(s):

LI-RADS, CMRI-3 which included LR-4, LR-5, LR-TIV and LR-M

Index test 2:

The AAA approach consisted of three rounds using AMRI

Cut-off point(s):

LI-RADS, AMRI-3 which included observation(s) ³ 10 mm that were not definitely benign, changes in imaging characteristics or threshold growth of previously noted observation(s) and new thrombus in vein.

Comparator test: Ultrasound approach consisting of three rounds of ultrasound.

Cut-off point(s):

 US LI-RADS with US-3 including observations  ³ 10 mm that were not definitely benign, new thrombus in vein and threshold growth

Reference test: Dynamic CT-scan, biopsy and/or

Cut-off point(s): Observations showing features of definite HCC on dynamic CT (nodules > 10 mm with arterial hyperenhancement and washout).

Time between the index test and reference test:

Observations found on MRI or US fulfilling the positive screening criteria underwent a recall process with dynamic CT scan performed within 3 months.

Observations that were not diagnostic on CT underwent pathologic confirmations. After the third round all patients underwent dynamic CT at six months.

Identity of false positive observations was determined by thoroughly review all available follow-up imaging and pathology data.

Median follow-up after last round of surveillance: 40.6 months (IQR 17.5-51.3)

For how many participants were no complete outcome data available:

N=20

• N=13: Withdrawn from study
• N=5: Death
• N=2: Incomplete MRI

Outcome measures and effect size (include 95%CI and p-value if available):

Diagnostic accuracy CAA approach:

Per-examination sensitivity round 1: 96.4% (N=27/28) (95%CI 78.6-99.5)

Per-examination sensitivity round 2-3: 80% (N=12/15) (95%CI 53.0-93.4)

Per-examination sensitivity total: 90.7% (N=39/43) (95%CI 77.7-96.5)

Per-examination specificity round 1: 96.9% (N=343/354) (95%CI 94.5-98.3)

Per-examination specificity round 2-3: 97.3% (N=640/658) (95%CI 95.6-98.3)

Per-examination specificity total: 97.1% (N=983/1012) (95%CI 95.8-98.0)

Per-examination accuracy round 1: 96.9% (N=343/382) (95%CI 94.6-98.2)

Per-examination accuracy round 2-3: 96.9% (N=652/673) (95%CI 95.2-98.0)

Per-examination accuracy total: 96.9% (N=1022/1055) (95%CI 95.5-97.8)

Diagnostic accuracy AAA approach:

Per-examination sensitivity round 1: 89.3% (N=25/28) (95%CI 71.6-96.5)

Per-examination sensitivity round 2-3: 80.0% (N=12/15) (95%CI 53.0-93.4)

Per-examination sensitivity total: 86.0% (N=37/43) (95%CI 72.2-93.6)

Per-examination specificity round 1: 92.4% (N=327/354) (95%CI 89.1-94.7)

Per-examination specificity round 2-3: 97.3% (N=640/658) (95%CI 95.6-98.3)

Per-examination specificity total: 95.6% (N=967/1012) (95%CI 94.0-96.7)

Per-examination accuracy round 1: 92.1% (N=352/382) (95%CI 89.0-94.5)

Per-examination accuracy round 2-3: 96.9% (N=652/673) (95%CI 95.2-98.0)

Per-examination accuracy total: 95.2% (N=1004/1055) (95%CI 93.6-96.4)

Diagnostic accuracy US:

Per-examination sensitivity round 1: 39.3% (N=11/28) (95%CI 23.3-58.0)

Per-examination sensitivity round 2-3: 6.7% (N=1/15) (95%CI 0.9-35.2)

Per-examination sensitivity total: 27.9% (N=12/43) (95%CI 16.6-43.0)

Per-examination specificity round 1: 93.8% (N=332/354) (95%CI 90.7-95.9)

Per-examination specificity round 2-3: 97.7% (N=643/658) (95%CI 96.3-98.6)

Per-examination specificity total: 96.3% (N=975/1012) (95%CI 95.0-97.4)

Per-examination accuracy round 1: 89.8% (N=343/382) (95%CI 86.3-92.5)

Per-examination accuracy round 2-3: 95.7% (N=644/673) (95%CI 93.9-97.0)

Per-examination accuracy total: 93.5=6% (N=987/1055) (95%CI 91.8-94.9)

Authors conclusion: The performance of ultrasound, the current primary surveillance modality for HCC, is suboptimal in detecting early-stage HCC. We found that in a prospective cohort of high-risk patients, AMRI-based approaches had significantly higher sensitivity for detecting early- stage HCC than an ultrasound-only approach.

Model to estimate risk of HCC: Risk index=1.41 (if age > 50 years) + 1.65 (prothrombin activity <75%) + 0.92 (platelet count is <100x10³/mm³) + 0.74 (if anti hepatitis C virus antibody or hepatitis B virus surface antigen test is positive)

If a patient had any lesion(s) fulfilling the positive criteria of the simulated surveil- lance test without confirmatory CT or histologic results and proceeded to the subsequent surveillance round, we used the CMRI of the subsequent round as a reference test for the previous observations. In this way, all simulated surveillance tests of each round had an available reference standard test (histology, dynamic CT, and/or subsequent surveillance CMRI within 6 months)

Sutherland, 2017

Type of study: Prospective cohort study

Setting and country: University-affiliated tertiary hospital, Australia

Funding and conflicts of interest: The authors received no funding for this research and no authors have conflicts of interest to declare

Inclusion criteria:

- Patients with age ³ 18 years

- Referred by the gastroenterology department with chronic liver disease for hepatocellular carcinoma screening liver ultrasound

Exclusion criteria:

- Presence of a known mass as indicated on the ultrasound request form

- Non-English speaking

- Contraindications to MRI such as pacemaker or contraindicated metallic implant

N=192

Prevalence: 3% (N=6)

Age in years, median (IQR): 58 (22-80)

Sex: 72% Male / 28% Female

Other important characteristics:

Cause of chronic liver disease:

Hepatitis B virus: N=108 (56%)

Hepatitis C virus: N=56 (29%)

Alcohol: N=21 (11%)

Hepatic steatosis: N=8 (4%)

Other: N=8 (4%)

Index test: MRI scan sequence comprised respiratory-gated DWI with the following parameters: TR 2500; TE 80; slice thickness 8 mm; distance factor 30%; FOV read 400 mm with effective voxel size of 2.6 x 2.1 x 8 mm; and b values of 100, 400, 800 acquired with 8 averages

Cut-off point(s): MRI lesions were considered suspicious if they had elevated signal on high b value DWI and were iso or hypointense to background liver on the ADC map

Comparator test: Ultrasound

Cut-off point(s): Lesions were suspicious if they were solid and not clearly focal fat infiltration or focal fat sparing.

Describe reference test: Arterial phase hyperenhancement followed by washout on either CT or MRI or by histology (biopsy or resection)

Cut-off point(s): AASLD guidelines

Time between the index test and reference test:

Any suspicious lesion was documented with respect to size, features and hepatic segment. Prior imaging was reviewed to determine the aetiology of the lesion and assess stability. If lesion was new it sparked further investigation as per the AASLR guidelines, being repeat imaging in three months by the modality that identified it for lesions under 10 mm and cross-sectional contrast-enhanced multiphase imaging with MRI or CT for new lesions over 10 mm

For how many participants were no complete outcome data available:

N=31 (16%)

Compliance rates for:

• Two examination rounds: N=91 (47%)
• Three examination rounds: N=45 (23%)
• Four examination rounds: N=23 (12%)
• Five examination rounds: N=6 (3%)

No reasons for incomplete outcome data described

Outcome measures and effect size (include 95%CI and p-value if available):

Diagnostic accuracy MRI:

Sensitivity: 83% (N=5/6)

Specificity: 98%

Positive predictive value: 63%

Negative predictive value: 99%

Diagnostic accuracy US:

Sensitivity: 100% (N=6/6)

Specificity: 90%

Positive predictive value: 23%

Negative predictive value: 100%

^a Data from study of Kim (2016)

Study reference

Study characteristics

Population²

Interventions

Outcome measures and effect size ⁴

Incremental Costs/Effects

Comments

Kim, 2019

Study design

(Trial/Model): Cohort-based Markov model

Country: Korea

Setting: Surveillance

Perspective: Healthcare

Time horizon: 20 years

Price year (currency): 2018

Discounting:

Costs: 5% per year

Outcomes: 5% per year

Source of funding: Supported by grants from Korean National Health Clinical Research project of the Ministry of Health and Welfare, the Korean Health Technology R&D Project, Ministry of Health and Welfare, the National Research Foundation of Korea, the Korean Gastroenterology Fund for Future Development, the National Evidence-Based Healthcare Collaborating Agency and the Technology Innovation Program funded by the Ministry of Trade, Industry, and Energy of the Republic of Korea.

Inclusion criteria:

- Patients with compensated cirrhosis (Child-Pugh A)

Exclusion criteria:

Not specified

N total at baseline:

Interventions: 10,000

Important prognostic factors²:

Age: 50  years

Surveillance cycle length: 6 months

Number of health states included in the model: 11

HCC incidence: 3%

HCC stage at time of detection – very early stage:

MRI: 72.1%

US: 18.5%

HCC stage at time of detection – very early + early stage:

MRI: 97.7%
US: 61.8%

Sensitivity MRI: 85.7%

Sensitivity US (95%CI): 47.0% (33-61)

Describe interventions and comparators (treatments/procedures/tests):

MRI and ultrasound were performed concurrently

Outcome measures and effect size Annual HCC incidence 3% (base-case):

MRI: 11.101 Life Years

US: 10.717 Life Years

Outcome measure and Costs Annual HCC incidence 3% (base-case):

MRI: 62,287 $

US: 56,725 $

Cost-effectiveness outcome measures

Annual HCC incidence 3% (base-case):

MRI: 6.783 QALYs

US: 6.562 QALYs

ICER:

Annual HCC Incidence 1%: 101,586 $/QALY

Annual HCC Incidence 2%: 44,026 $/QALY

Annual HCC Incidence 3%: 25,202 $/QALY

Annual HCC Incidence 3.5%: 20,000 $/QALY

Annual HCC Incidence 4%: 16,039 $/QALY

Annual HCC Incidence 5%: 10,721 $/QALY

Authors conclusion: The present study demonstrated that MRI surveillance for HCC can be cost-effective in high-risk patients with compensated cirrhosis.

Since MRI and ultrasound were performed concurrently and one HCC had been diagnosed with MRI, US detection was no longer an opportunity. Therefore detection rate data from US were extracted from another resource.

Higher incidence of HCC, more patients developed HCC and further surveillance was not required, which had a considerable effect on the costs of surveillance.

Lima, 2019

Study design

(Trial/Model): decisional Markov model

Country: Canada

Setting: Surveillance

Perspective: Healthcare

Time horizon: Lifetime

Price year (currency): 2017

Discounting:

Costs: 1.5% per year

Outcomes: 1.5% per year

Source of funding: Supported by a New Researcher Startup Grant from the Centre de recherche du Centre hospitalier de l’Université de Montréal.

Inclusion criteria:

- Patients with compensated cirrhosis (Child-Pugh A)

- Age 50 years

Exclusion criteria:

Not specified

N total at baseline:

Interventions: Not specified

Important prognostic factors²:

Age: 50 years

Surveillance cycle length: 6 months

Number of health states included in the model: Not specified

HCC incidence: 3%

Sensitivity MRI (range): 86% (82-93)

Sensitivity US (range): 78% (60-89)

Specificity MRI (range): 86% (79-91)

Specificity US (range): 89% (80-94)

Describe interventions and comparators (treatments/procedures/tests):

Seven surveillance and diagnostic strategies were investigated with combinations of four imagine techniques (US, CT, complete MRI, abbreviated MRI).

A) Surveillance US; diagnosis CT

F) Surveillance MRI; if inadequate CT; diagnosis MRI

Strategies were evaluated for:

- Optimal scenario: 100% of the patients with compensated cirrhosis (Child-Pugh A) and an assumed compliance rate of 100%.

- Conservative scenario: 29% of the patients with compensated cirrhosis (Child-Pugh A) and an assumed compliance rate of 52%.

Cost-effectiveness outcome measures:

Optimal scenario (100% Child-Pugh A; 100% surveillance compliance)

US: 7.269 QALY

MRI: 7.424 QALY

Conservative scenario (29% Child-Pugh A; 52% surveillance compliance)

US: 4.300 QALY

MRI: 4.354 QALY

Incremental Effects:

Optimal scenario: 0.011 QALY

Conservative scenario: 0.022 QALY

Incremental Costs:

Optimal scenario: 7293 Canadian dollars

Conservative scenario: 873 Canadian dollars

ICER:

Optimal scenario: 663,000 Canadian dollars/QALY

Conservative scenario: 39,681 Canadian dollars/QALY

Authors conclusion: In conclusion, in a scenario that assumes optimal patient compliance and takes into account inconclusive imaging examinations, CT for HCC surveillance and diagnosis and complete MRI for inadequate CT was most cost-effective

Nahon, 2022

Study design

(Trial/Model): Markov model with data from one RCT and three prospective cohort studies: CHC2000 trial, ANRS Co12 CirVir cohort, CIRRAL cohor and ANRS CO22 Hepather cohort

Country: France

Setting: Surveillance

Perspective: Healthcare

Time horizon: 20 years

Price year (currency): 2020

Discounting:

Costs: 2.5% per year

Outcomes: 2.5% per year

Source of funding: The cohorts were funded by the French ministry of health, French Ligue de Recherche contre le cancer, National Agency for Research on HIV and Hepatitis, the French national institute of Cancer and the French association for Research in Cancer

Inclusion criteria:

- Patients with biopsy proven compensated cirrhosis (Child-Pugh A)

- Age 50 years

Exclusion criteria:

Not specified

N total at baseline:

Interventions: 10,000

Important prognostic factors²:

Age: 50 years

Surveillance cycle length: 3 months

Number of health states included in the model: 14

HCC incidence: 3%

Sensitivity MRI: 85.7%^a

Describe interventions and comparators (treatments/procedures/tests):

MRI versus Ultrasound

Outcome measures and effect size:

MRI: 13.8 discounted life years

US: 13.4 discounted life years

Outcome measure and Costs:

MRI: € 106,873

US: € 100,739

ICER:

Incidence rate 3% per year:

€ 15,477/Life Year

Incidence rate 2% per year:

€ 23,338/Life Year Gained

Incidence rate 1% per year:

€ 47,194/Life Year Gained

Authors conclusion: Our prospective cohort- and model-based evaluation of very early HCC detection found that MRI monitoring was cost- effective for a baseline yearly incidence of 3% and over a range of assumptions on incidence, costs and treatment choice.

Tan, 2021

Study design

(Trial/Model): Markov model

Country: Singapore

Setting: Surveillance

Perspective: Healthcare

Time horizon: 40 years

Price year (currency): Not reported

Discounting:

Costs: 3% per year

Effects: 3% per year

Source of funding: Not reported

Inclusion criteria:

- At risk patients

Exclusion criteria:

Not reported

N total at baseline:

Interventions: 482,000

Important prognostic factors²:

Average age: 40 years

Surveillance cycle length: 6 months

Number of health states included in the model: 7

HCC incidence: 1.1%

Sensitivity MRI: 90%

Sensitivity US: 55.6%

Specificity MRI: 91.5%

Specificity US: 93%

Describe interventions and comparators (treatments/procedures/tests):

No surveillance versus Ultrasound surveillance versus Non-contrast Enhanced MRI (NCEMRI) surveillance

Outcome measure and Costs, mean (SE):

No surveillance: 4,675 $ (263)

US surveillance: 23,803 $ (367)

NCEMRI surveillance: 177,876 $ (1111)

Cost-effectiveness outcome measures, mean (SE):

No surveillance: 7.483 QALY (0.044)

US surveillance: 11.242 QALY (0.074)

NCEMRI surveillance: 11.426 QALY (0.074)

Incremental Effects:

US surveillance: 3.759 QALY

NCEMRI surveillance: 3.943 QALY

NCEMRI versus US: 0.184 QALY

Incremental Costs:

US surveillance: 19,128 $

NCEMRI surveillance: 173,201 $

NCEMRI versus US: 154,073 $

ICER:

US surveillance: 5,088$/QALY

NCEMRI surveillance: 43,924$/QALY

NCEMRI versus US: 837,353$/QALY

Authors conclusion: Despite NCEMRI having a superior diagnostic accuracy, it is a less cost- effective strategy than US for HCC surveillance in the general at-risk population, from an overall healthcare perspective. Future local cost-effectiveness analyses should include stratifying surveillance methods with a variety of imaging techniques (US, NCEMRI, CEMRI) based on patients’ risk profiles.

Relatively low transition probability of 1.1 percent compared with other studies, results in decreased cost-effectiveness.

^a Data from the PRIUS study

Kim, 2016

Was a consecutive or random sample of patients enrolled?

Yes, consecutive sample of patients at risk for HCC

Was a case-control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Yes, clear in- and exclusion criteria were reported

Were the index test results interpreted without knowledge of the results of the reference standard?

Yes, US and MRI interpretations were allocated to different radiologists who were blinded to the findings of the other imaging modalities.

If a threshold was used, was it pre-specified?

Yes, positive screening criterion category 5 or 4 on MRI or US (standardized 5-point scale for MRI or 4-point scale for US)

Is the reference standard likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index test?

Unclear, it is not clear of the assessors of the reference standards (CT and biopsy) were blinded for results of the index or comparator tests.

Was there an appropriate interval between index test(s) and reference standard?

Yes, 3 months when positive screening criterion for MRI or US. 6 months for all study patients after last screening round.

Did all patients receive a reference standard?

Yes, CT or CT and biopsy

Did patients receive the same reference standard?

Yes, patients received reference standard according to the AASLD criteria.

Were all patients included in the analysis?

Yes

Are there concerns that the included patients do not match the review question?

Yes, high percentage of hepatitis B patients in the sample.

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

No, MRI in the Netherlands is also gadoxetic enhanced.

Are there concerns that the target condition as defined by the reference standard does not match the review question?

No

CONCLUSION:

Could the selection of patients have introduced bias?

RISK: LOW

CONCLUSION:

Could the conduct or interpretation of the index test have introduced bias?

RISK: LOW

CONCLUSION:

Could the reference standard, its conduct, or its interpretation have introduced bias?

RISK: UNCLEAR

CONCLUSION

Could the patient flow have introduced bias?

RISK: LOW

Park, 2020

Was a consecutive or random sample of patients enrolled?

Yes, consecutive sample of patients at risk for HCC from study data of Kim (2016)

Was a case-control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Yes, clear in- and exclusion criteria were reported

Were the index test results interpreted without knowledge of the results of the reference standard?

Yes, MRIs were anonymized, shuffled in random order an analysed by two reviewers who were blinded to data, pathological results, final diagnosis and results of full contrast-enhance MRI

If a threshold was used, was it pre-specified?

Yes, lesions of ³ 1 cm. with either diffusion restriction or mild to moderate T2 hyperintensity

Is the reference standard likely to correctly classify the target condition?

Yes, there were different reference standards: Dynamic CT, biopsy and and/or subsequent surveillance round(s) after six months.

Were the reference standard results interpreted without knowledge of the results of the index test?

Yes, results of reference standard tests were reviews by one of the authors who was not involved in imaging analysis

Was there an appropriate interval between index test(s) and reference standard?

Yes

Did all patients receive a reference standard?

Yes

Did patients receive the same reference standard?

Yes, patients received reference standard according to the AASLD criteria.

Were all patients included in the analysis?

Yes

Are there concerns that the included patients do not match the review question?

Yes, high percentage of hepatitis B patients in this sample

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

No

Are there concerns that the target condition as defined by the reference standard does not match the review question?

No

CONCLUSION:

Could the selection of patients have introduced bias?

RISK: LOW

CONCLUSION:

Could the conduct or interpretation of the index test have introduced bias?

RISK: LOW

CONCLUSION:

Could the reference standard, its conduct, or its interpretation have introduced bias?

RISK: LOW

CONCLUSION

Could the patient flow have introduced bias?

RISK: LOW

Park, 2021

Was a consecutive or random sample of patients enrolled?

Yes, consecutive sample of patients at risk for HCC from study data of Kim (2016)

Was a case-control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Yes

Were the index test results interpreted without knowledge of the results of the reference standard?

Yes, readers independently interpreted the MRIs in a blinded-to-outcome manner but were allowed to view MRI images of previous round(s) when they interpreted the second and third round MRI

If a threshold was used, was it pre-specified?

Yes, LI-RADS classifications was used

Is the reference standard likely to correctly classify the target condition?

Unclear, CT and/or pathological confirmation and/or follow-up of medical records

Were the reference standard results interpreted without knowledge of the results of the index test?

Unclear

Was there an appropriate interval between index test(s) and reference standard?

Yes

Did all patients receive a reference standard?

Yes

Did patients receive the same reference standard?

Yes, patients received reference standard according to the AASLD criteria.

Were all patients included in the analysis?

Yes

Are there concerns that the included patients do not match the review question?

Yes, high percentage of hepatitis B patients in this sample

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

No

Are there concerns that the target condition as defined by the reference standard does not match the review question?

No

CONCLUSION:

Could the selection of patients have introduced bias?

RISK: LOW

CONCLUSION:

Could the conduct or interpretation of the index test have introduced bias?

RISK: LOW

CONCLUSION:

Could the reference standard, its conduct, or its interpretation have introduced bias?

RISK: UNCLEAR

CONCLUSION

Could the patient flow have introduced bias?

RISK: LOW

Sutherland, 2017

Was a consecutive or random sample of patients enrolled?

Yes, consecutive sample of patients referred by gastroenterology department with chronic liver disease

Was a case-control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Yes

Were the index test results interpreted without knowledge of the results of the reference standard?

Yes, MRI was reviewed by radiologist blinded to the US and all prior imaging

If a threshold was used, was it pre-specified?

Yes

Is the reference standard likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index test?

Unclear

Was there an appropriate interval between index test(s) and reference standard?

Yes, prior imaging was reviewed to determine aetiology and assess stability. If a lesion was new, further investigation as defined by the AASLR

Did all patients receive a reference standard?

Unclear, gold standard is reported but not if all patients received reference standard

Did patients receive the same reference standard?

Unclear, gold standard is arterial phase hyperenhancement follow by washout on CT or MRI or histology

Were all patients included in the analysis?

Yes

Are there concerns that the included patients do not match the review question?

Unclear, study population are patients with chronic liver diseases

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

No

Are there concerns that the target condition as defined by the reference standard does not match the review question?

No

CONCLUSION:

Could the selection of patients have introduced bias?

RISK: LOW

CONCLUSION:

Could the conduct or interpretation of the index test have introduced bias?

RISK: LOW

CONCLUSION:

Could the reference standard, its conduct, or its interpretation have introduced bias?

RISK: UNCLEAR

CONCLUSION

Could the patient flow have introduced bias?

RISK:  UNCLEAR

Study reference

(first author, publication year)

1

Y/N

2

Y/N

3

Y/N

4

Y/N

5

Y/N

6

Y/N

7

Y/N

8

Y/N

9

Y/N

10

Y/N

11

Y/N

12

Y/N

13

Y/N

14

Y/N

15

Y/N

16

Y/N

17

Y/N

18

Y/N

19

Y/N

20

Y/N

Comments

Kim, 2019

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

N

NR

MRI and US were performed concurrently, one HCC had been diagnosed with MRI, US detection was no longer an opportunity. Therefore detection rate data from US was extracted from another resource.

Lima, 2019

Y

Y

Y

N

Y

Y

Y

Y

Y

Y

N

Y

Y

Y

Y

Y

N

NR

NR

Disutility associated with false-positive diagnosis was based on cost-effectiveness study  of patients undergoing evaluation for coronary artery disease (overestimation of cost-effectiveness)

Nahon, 2022

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Y

Tan, 2021

N

Y

Y

Y

Y

Y

Y

Y

NR

N

NR

Y

Y

Y

N

Y

Y

NR

NR

Death as outcome is taken into account as all-cause death