Cost-Effectiveness of Blood-Based Fibrosis Screening in High-Risk Metabolic Liver Diseases With Emerging Therapies

Abstract

Background and Aims

Metabolic dysfunction–associated steatotic liver disease (MASLD) remains a global public health threat. With emerging effective pharmacologic therapies, early identification of significant fibrosis in primary care is critical. We evaluated the cost-effectiveness of blood-based noninvasive tests incorporating real-world diagnostic performance in the contemporary MASLD treatment era.

Methods

We developed a decision-analytic model for MASLD natural history and treatment to compare 6 noninvasive test strategies using Fibrosis-4 (FIB-4) and/or enhanced liver fibrosis (ELF) tests at varying cutoffs. Patients identified with significant fibrosis were referred for hepatology staging (biopsy and/or imaging-based), and eligible individuals were considered for Resmetirom ($47,400/y; mean duration: 1.6–1.8 year). Model parameters were informed by a real-world cohort (n = 400) of high-risk primary care patients with suspected MASLD who underwent both tests. Outcomes included quality-adjusted life-years (QALYs), lifetime costs, incremental cost-effectiveness ratios, and adverse liver outcomes averted.

Results

The modeled population had a mean age of 64 years, mean body mass index of 32 kg/m², with 82% having type-2 diabetes and 15% significant fibrosis. Sequential screening with ELF (cutoff 9.80) following indeterminate FIB-4 (1.3–2.67) was the most cost-effective, yielding 8.610 QALYs and $96,990 in lifetime costs at a $100,000/QALY threshold. ELF-alone screening at a 9.00 cutoff maximized QALYs and individuals treated but increased unnecessary referrals. Resmetirom cost was most influential on results: if cost fell below $11,570/y (base case, $47,400/y), ELF-alone screening became the preferred strategy. Findings remained robust across sensitivity analyses, including in low-risk populations.

Conclusion

By integrating real-world diagnostic performance with new MASLD therapies, this translational modeling study identifies a scalable, cost-effective fibrosis screening pathway using FIB-4 and ELF. These findings support implementation of blood-based fibrosis screening in general primary care populations.

Introduction

Global prevalence of metabolic dysfunction–associated steatotic liver disease (MASLD) is projected to reach 56% by 2040.¹ In the United States, MASLD is a leading cause of cirrhosis, liver cancer, and liver transplantation,² incurring $1.6 billion annually in direct medical costs.^3,4 MASLD encompasses a spectrum of diseases, including metabolic dysfunction–associated steatohepatitis (MASH) and cirrhosis. Type 2 diabetes (T2D) and obesity are tightly linked to MASLD disease progression⁵ with a hazard ratio of 1.4–2.6 compared to those without these metabolic risk factors.^6,7 Three-quarters of individuals with obesity and 69% of individuals with T2D have MASLD.⁸

Intervention at an early stage of fibrosis is imperative to prevent future complications, namely cirrhosis, decompensation, and liver cancer.⁹ Noninvasive tests (NITs), including blood-based markers (eg, Fibrosis-4 [FIB-4] and enhanced liver fibrosis [ELF]¹⁰), and imaging-based tests (eg, vibration-controlled transient elastography [TE], magnetic resonance elastography [MRE]), comprise some of the diagnostic tools for fibrosis assessment.^11,12 Globally, there is growing recognition that readily available laboratory data upstream of primary care, such as elevated alanine aminotransferase (ALT > 30 U/L), can serve as automated triggers to warrant further primary care-based fibrosis risk assessment.13, 14, 15 Meanwhile, noninvasive imaging-based fibrosis assessment downstream of primary care, particularly TE, has gained increasing real-world application in specialty care settings, reflecting expanding capacity and demand for fibrosis staging.16, 17, 18 At the moment, ELF is the only authorized commercial test in the United States for advanced fibrosis due to MASH, demonstrating high prognostic value.¹⁹ Major liver societies recommend a stepwise approach using FIB-4 followed by ELF or TE for risk stratification in primary care.^20,21

The therapeutic landscape of MASLD has shifted rapidly. In 2024, Resmetirom became the first FDA-approved pharmacological treatment for MASH.²² Several other promising agents are in advanced stages of development, including Semagultide (Wegovy), which just received FDA approval.^23,24 With emerging effective treatments, early identification of at-risk individuals is increasingly critical. Due to a strong association with T2D and obesity, MASLD is often first encountered in primary care or endocrine clinics.²⁵ Although guidelines advocate NIT-based risk stratification in these settings, utilization remains low due to limited provider awareness, unclear care pathways after diagnosis, concerns of over referral, and uncertainty about cost-effectiveness.^20,25,26

Prior cost-effectiveness analyses have supported sequential FIB-4/ELF screening for reducing unnecessary referrals27, 28, 29; however, long-term results vary due to differing assumptions about the diagnosis impact on treatment.30, 31, 32 Only 1 study, to our knowledge, incorporated Resmetirom as the downstream management after identification of significant fibrosis³²; however, real-world diagnostic performance data were not directly linked to therapeutic decision-making.

To bridge this translational gap, we integrated real-world diagnostic accuracy of blood-based NITs with contemporary Resmetirom treatment to evaluate the cost-effectiveness of fibrosis screening strategies. Given limited TE availability in primary care and the absence of consensus on optimal screening approaches, we sought to identify the most cost-effective NITs using FIB-4 and/or ELF. Our modeled cohort was informed by an empiric population with high prevalence of T2D and obesity, representing a high-risk group commonly seen in clinical practice. Furthermore, we focused on Resmetirom impact as it is currently the only MASH-targeting therapy with published instructions for use. This analysis thus provides evidence that is broadly generalizable to high-risk primary care populations and offer practical guidance for implementing cost-effective, blood-based fibrosis screening in the evolving MASLD therapeutic landscape.

Methods

Analytic Overview

We developed an individual-level state-transition model to evaluate the long-term clinical and economic impact of NITs for identifying significant fibrosis in a high-risk, primary care population informed by a cohort with high prevalence of T2D and/or obesity. We modeled the following 6 annual NITs, using FIB-4 and ELF, individually or combined, with guideline-based cutoffs (Figure 1)^20,21,33:

• 1)FIB-4 (cutoff 1.3, “FIB-4 1.3”)
• 2)ELF (cutoff 9.00, “ELF 9.00”)
• 3)ELF (cutoff 9.80, “ELF 9.80”)
• 4)Indeterminate FIB-4 (1.3–2.67) followed by ELF (cutoff 7.70, “FIB-4 1.3–2.67/ELF 7.70”)
• 5)“FIB-4 1.3–2.67/ELF 9.00”
• 6)“FIB-4 1.3–2.67/ELF 9.80”

Figure 1.

Modeled NIT-based screening strategies among a real-world population with suspected MASLD, from top to bottom: “FIB-4 1.3”; “ELF 9.00”; “ELF 9.80”; “FIB-4 1.3–2.67/ELF 7.70”; “FIB-4 1.3–2.67/ELF 9.00”; “FIB-4 1.3–2.67/ELF 9.80”.

In strategies 1–3 above, only patients with a single test result above the cutoff will be referred to hepatology (Figure 1). In sequential testing, strategies 4–6, patients with FIB-4 score < 1.3 are classified as having nonsignificant fibrosis and managed by primary care, patients with FIB-4 score > 2.67 are classified as significant fibrosis and referred to hepatology, while patients with FIB-4 score between 1.3 and 2.67 receive an additional ELF test that determines hepatology referral (Figure 1). We did not include no screening as NIT-based screening among high-risk populations were shown to be cost-effective compared to no screening and was recommended by multiple guidelines.^20,21,31,32 Regardless of the strategy, those who are referred to hepatology will be correctly staged using liver biopsy, ELF, or imaging-based tools (TE, MRE, or liver ultrasound). Patients identified with stage 2 or 3 MASH will be offered Resmetriom with imperfect uptake and persistence, and patients identified with nonsignificant fibrosis will be sent back to primary care and retested in 1–2 years. Those who are not referred to hepatology are offered the same NIT strategy annually. Screening uptake is imperfect. For each NIT, we projected outcomes for a lifetime horizon: number of decompensated cirrhosis (DCC), hepatocellular carcinoma (HCC), liver transplant (LT) cases, undiscounted life expectancy (LE), discounted quality-adjusted life-years (QALYs), discounted costs, and incremental cost-effectiveness ratios (ICERs) in 2025 United States dollars per QALY gained. We applied an annual rate of 3% when discounting QALYs and costs. We used a health-care payer perspective,³⁴ including the costs of NITs, hepatology workup (visit and staging), and downstream care costs for clinical management of various MASLD disease stages. The most cost-effective NIT(s) is defined as having the greatest ICER below a willingness-to-pay threshold of $100,000/QALY. We provide a more detailed description of the cost-effectiveness analysis framework in Supplementary Appendix, Method. Our model was built using TreeAge Pro 2021 (Williamstown, MA, www.treeage.com). Institutional review board approval was not necessary due to no real-patient records being involved in the modeling. The reporting of this study followed the Consolidated Health Economic Evaluation Reporting Standards,³⁵ with a compliance checklist provided in Supplementary Appendix.

MASLD Natural History

Our model captured the MASLD natural disease spectrum across 7 stages with increasing severity: MASLD/MASH fibrosis stage 0-1 (F0-1), MASH fibrosis stage 2-3 (F2-3), compensated cirrhosis (CC), decompensated cirrhosis (DCC), hepatocellular carcinoma (HCC), post-LT, and death (Supplementary Figure). We used a 1-year cycle length due to slow progression of MASLD. Within each cycle, patients in states F0-1, F2-3, or CC can progress or regress by 1 stage or remain stable. DCC is considered irreversible. Patients in F2-3, CC, or DCC can develop HCC, with increasing risk at later stages. Those in DCC or HCC may receive LT, transitioning to the post-LT state. All patients are subject to background mortalities based on life tables.³⁶ Patients in F2-3, CC, DCC, and HCC are subject to both liver- and non–liver-related (ie, cardiovascular and nonhepatic neoplasm–related) mortalities. Those in F0-1 are only subject to non–liver-related mortalities. Mortality in post-LT state is modeled as time-varying all-cause mortality based on registry data.³⁷

MASLD Treatment

Resmetirom, FDA-approved in 2024 following a phase 3 trial, improves fibrosis regression and reduces progression in noncirrhotic MASH (F2-3).³⁸ We quantify treatment effect using the concept of relative risk (RR), defined as the ratio of fibrosis progression/regression probabilities with treatment to those without treatment. An RR > 1 of fibrosis regression with treatment indicates that treatment is effective in improving fibrosis; while a RR < 1 of fibrosis progression with treatment suggests that treatment is effective in slowing down progression. Specifically, we assume that treatment increases the likelihood of fibrosis regression from state F2-3 to F0-1 and decreases the likelihood of fibrosis progression from state F2-3 to CC, DCC, and HCC. To reflect real-world suboptimal compliance with Resmetirom, we model imperfect initial uptake and subsequent discontinuation (due to side effects) for those offered the treatment option.^39,40

Model Inputs

Cohort characteristics

We simulated a 1-million cohort representative of the participants from a cross-sectional study, which prospectively recruited 400 people with T2D and/or obesity from primary care clinics in the Veteran Affairs Palo Alto Healthcare System (VAPAHCS, Palo Alto, CA), from May 2022 to June 2024 (Siemens Healthineers and VA Palo Alto, unpublished data, 5/2022-10/2023).²⁹ All study participants underwent a physical exam and lab testing using FIB-4 and ELF. Mean age of the study cohort was 64 years (61%, ≥65 years), mean body mass index 32 (standard deviation = 6), and 82% had T2D. FIB-4 distributions were <1.3 (62%), 1.3–2.67 (34%), and ≥2.67 (4%); ELF distributions were 7.70–9.00 (16%), 9.00–9.80 (34%), and ≥9.80 (50%).

Test characteristics

MRE was obtained from a subset (n = 101) of the participants. We derived test accuracy (ie, sensitivity and specificity) for 6 investigated NIT strategies using MRE results as the gold standard for adjudicating true disease state: F2-3 (MRE ≥ 3.14 kPa) and F0-1 (MRE < 3.14 kPa), where a cutoff value of 3.14 kPa was recommended by recent guidelines.²⁰ The resulting diagnostic accuracy is shown in Table 1.

Table 1.

Test Characteristics for 6 NIT Strategies Using MRE as the Adjudicator for True Significant Fibrosis for a Real-World Population With Suspected MASLD, Both the Full Population and the Subgroup Aged ≥65 Years

Annual NIT-based screening strategies	Sensitivitya, %		Specificitya, %
	Full population	Subgroup aged ≥65 y	Full population	Subgroup aged ≥65 y
FIB-4 1.3–2.67/ELF 9.80	73	30	43	81
ELF 9.80	73	80	38	29
FIB-4 1.3–2.67/ELF 9.00	93	30	24	83
FIB-4 1.3–2.67/ELF 7.70	93	30	20	86
FIB-4 1.3	93	30	20	81
ELF 9.00	100	100	12	7

^aSensitivity and specificity were derived using MRE, as the adjudicator for true disease state based on a subset of 101 patients. A cutoff of 3.14 kPa was used for state F2-3, or significant fibrosis.

Natural history transitions

We derived probabilities of progression, regression, and cause-specific mortalities by averaging estimates from multiple published sources (Table 2).^9,32,37,40,41, 42, 43, 44, 45, 46 When available, we used estimates that were based on systematic review or large observational studies (including the Global NASH Registry) of people with T2D or obesity at elevated risk of MASLD as our base case values. This approach reflects the higher baseline progression risks expected in populations with a high prevalence of metabolic comorbidities.

Table 2.

Key Model Inputs for a Cost-Effectiveness Analysis of Blood-Based Fibrosis Screening in a Real-World Population With Suspected MASLD

Parameter	Base value	Range	Source
Natural history transition probability from state F0-1, %/y
F0-1 to F2-3	6.3	1.3–9.6	32,41
Non–liver-related death	1.6	1.3–1.8	32
Natural history transition probabilities from state F2-3, %/y
F2-3 to F0-1	2.7	0.9–6	32
F2-3 to CC	9.5	5.0–19	32,41,42
F2-3 to HCC	0.3	0.1–0.5	32,41,43
Liver-related death	3.3	2.8–5.2	32
Non–liver-related death	3.3	2.9–4.6	32
Natural history transition probabilities from state CC, %/y
CC to F2-3	5.4	4.1–14	32,40
CC to DCC	5.7	3.7–7.7	32,41,42
CC to HCC	2.4	0.1–4.6	32,41,44
Liver-related death	7.3	±20%	32
Non–liver-related death	6.2	±20%	32
Natural history transition probabilities from DCC, %/y
DCC to HCC	2.4	0.7–4.0	32,41
Liver transplantation from DCC	1.8	0.7–2.9	9,32
Liver-related death	22	8.4–24	9,32,41
Non–liver-related death	4.8	0.8–13	9,32,41
Natural history transition probabilities from HCC, %/y
Liver transplantation from HCC	3.8	2.7–5.0	9,32
Liver-related death	35	26–55	9,32,41
Non–liver-related death	6.0	0.0–12	9,32
Post liver transplantation, all-cause mortalities, %/y
Year 1–2	6.8	–	37,45
Year 3–4	12.7	–	37,45
Year 5–9	19.8	–	37,45
Year 10+	35.7	–	37,45
Treatment effect, relative risk (RR) of Resmetirom vs placebo
RR of fibrosis regression	1.82	1.02–2.89	40,46
RR of fibrosis progression	0.56	0.38–0.96	40,46
Behavior characteristics
Resmetirom uptake, %	75	20–100	39
Resmetirom discontinuation, %/y	10	0–20	39
Annual screening uptake, %	50	0–100	Assumption
Direct medical cost in various health states, 2025 USD/y
F0-1	500	480–510	31,41
F2-3	1180	640–1720	41,42
CC	23,360	18,210–28,510	32,41,47
DCC	36,290	30,190–42,390	41,42
HCCa	85,840	60,900–110,790	32,41,47
Post liver-transplantation	7630	0–15,250	32,42
Resmetirom	47,400	10,000–90,000	48
One-time cost, 2025 USD
Liver transplantation	419,000	416,200–421,860	32,41
Liver-related death	46,590	–	32
ELF test	200	170–230	29,32
FIB-4 test	0	–	–
Hepatology workup

	Cost, $	Resource use, %	Source
Hepatology consultation	257	100	29
ELF test	185	25
TE	32	25
Ultrasound liver	136	200
Endoscopy	367	50
MRI/CT abdomen/liver	323	5
Liver biopsy	354	15
Health utilities
F0-1	0.88	0.84–0.91	31,32,41
F2-3	0.76	0.68–0.84	32,42
CC	0.74	0.68–0.81	32,42
DCC	0.57	0.52–0.63	32,42
HCC	0.5	0.45–0.55	42
Post liver transplantation	0.825	0.81–0.84	32

CT, computed tomography; MRI, magnetic resonance imaging; USD, United States dollar.

^aAnnual care cost in state HCC includes hospitalization and partial utilization of resection, radiotherapy, and radiofrequency ablation.

Treatment effect

The phase 3 trial suggested that 24% from the 80-mg Resmetirom arm had fibrosis improvement over 1 year (14%, placebo).⁴⁰ Based on this result, we derived an RR of 1.82 (24/14) as the treatment effect on fibrosis regression compared to no treatment. We estimated the RR of fibrosis progression to be 0.56, by using the difference in percentage of participants with a 25% or greater increase in liver stiffness between the 2 arms from the same trial.

Costs and utilities

Our study was conducted from a health-care payer perspective and included direct medical costs associated with each health state (Table 2), including inpatient and outpatient, medications, and procedures. Hepatology workup costs comprised of specialist consultation and fibrosis staging, which reflected standard practice using a combination of liver biopsy, ELF, and imaging-based modalities (TE, MRE, or liver ultrasound). Resource use frequencies for each staging component were estimated from the same real-world cohort and are reported in Table 2. Resmetirom treatment costs were obtained from Centers for Medicare & Medicaid Services ($47,400/y at the time of analysis) and applied exclusively to individuals with F2-3 fibrosis. All costs were sourced from published literature using a combination of micro-costing and gross-costing methods, adjusted using the Consumer Price Index, and reported in 2025 United States dollars. We sourced health utilities scores from published literature using preference-based utility scores.

Sensitivity Analysis

We assessed the impact of parameter uncertainty on the comparative cost-effectiveness of strategies.⁴⁹ We used probabilistic sensitivity analysis to gauge decision uncertainty, where all model parameters were varied across evidence-based ranges according to probability distributions. In one-way sensitivity analysis, we focused on 2 strategies identified as most likely to be cost-effective in probabilistic sensitivity analysis and examined the impact of varying individual model parameters on the ICER between the 2 strategies. Key inputs varied were as follows: MASLD disease natural history–related inputs (disease progression/regression from significant/advanced fibrosis), initial prevalence of significant fibrosis, as well as treatment-related inputs, which included the annual cost of Resmetirom, treatment efficacy as measured by RR of fibrosis progression/regression with Resmetirom, and uptake and discontinuation rate of Resmetirom which affected treatment duration. For each input, we used the highest and lowest values found in the literature as sensitivity analyses ranges.

Subgroup Analysis

Older age was suggested as a confounding factor for the diagnostic accuracy of NITs.⁵⁰ We performed a subgroup analysis of patients ≥65 years, using 2.0 as the lower cutoff for FIB-4. The diagnostic accuracy of NITs associated with the subgroup of patients ≥65 years is shown in Table 1.

Scenario Analysis

Our base-case cohort was characterized by a high prevalence of T2D or obesity. To investigate the generalizability of results to a lower-risk primary care population with suspected MASLD, we conducted a scenario analysis representing individuals with fewer metabolic comorbidities. Lower prevalence of T2D or obesity was implemented in the model by applying reduced progression and mortality risks. This approach was chosen to reflect the aggregate effect of metabolic risk factors on disease trajectory without explicitly characterizing individual’s comorbidity status, consistent with prior modeling studies.^31,32 Specifically, studies have reported that T2D or obesity increases fibrosis progression risk by a RR up to 1.8^6,25,32; we thus scaled down base-case progression rates from early to advanced stages by the inverse of this RR. Additionally, recognizing higher mortality risks in MASLD patients with T2D, we adjusted background, liver-, and non–liver-related mortality rates similarly using the inverses of published RRs of 2.14, 22.8, and 3.25, respectively.³²

Results

Diagnostic Accuracy

Across populations, 2-tier NITs with confirmatory ELF at high cutoffs had lower sensitivity and higher specificity than single-tier NITs using FIB-4 or ELF at low cutoffs (Table 1). In older patients, FIB-4-based NITs showed reduced sensitivity and increased specificity, suggesting age confounded its accuracy. The strategy “FIB-4 1.3–2.67/ELF 9.80” minimized referrals (lowest sensitivity, highest specificity), whereas “ELF 9.00” identified all treatment-eligible patients (100% sensitivity) but generated most referrals (lowest specificity).

Base Case

Clinical

Clinical outcomes improved with strategies of higher sensitivity and lower specificity. Annual screening using “FIB-4 1.3–2.67/ELF 9.80” resulted in 8.610 QALYs (Table 3), or 12.775 undiscounted LYs (Supplementary Table). This increased to a maximum of 8.635 QALYs (12.818 LYs) with “ELF 9.00”. Number treated with Resmetirom ranged from 32.2 with “FIB-4 1.3–2.67/ELF 9.80” to 35.7 with “ELF 9.00” (per 100) (Supplementary Table). Number of adverse outcomes was lowest with “ELF 9.00” (435–447 DCC, 286–297 HCC, and 43.6–45.8 LT per 10,000).

Table 3.

Base Case Results for a Cost-Effectiveness Analysis of Blood-Based Fibrosis Screening in a Real-World Population With Suspected MASLD, for the Full Population, the Subgroup Aged ≥65 Y, and the Lower-Risk Population With Fewer Metabolic Comorbidities

Annual NIT-based screening strategies	QALYs, y	Screening cost, $	Hepatology workup cost	Resmetirom cost, $	Total cost, $	ICER, $/QALY
Full population
FIB-4 1.3–2.67/ELF 9.80	8.610	650	3200	57,540	96,990	–
FIB-4 1.3	8.630	0	4290	63,670	102,960	301,270
ELF 9.00	8.635	860	4390	65,500	105,560	457,620
Dominateda
ELF 9.80	8.610	890	3570	57,530	97,570	–
FIB-4 1.3–2.67/ELF 9.00	8.630	640	3970	63,670	103,270	–
FIB-4 1.3–2.67/ELF 7.70	8.630	640	4290	63,670	103,600	–
Subgroup aged ≥65 y
FIB-4 1.3–2.67/ELF 9.80	8.538	140	980	35,420	74,300	–
ELF 9.00	8.635	860	4900	65,490	106,040	327,660
Dominated
FIB-4 1.3	8.538	0	1310	35,400	74,480	–
FIB-4 1.3–2.67/ELF 9.00	8.538	140	1170	35,410	74,480	–
FIB-4 1.3–2.67/ELF 7.70	8.538	140	1310	35,400	74,610	–
ELF 9.80	8.616	880	3860	59,880	99,940	–
Lower-risk population with fewer metabolic comorbidities
FIB-4 1.3–2.67/ELF 9.80	11.317	810	3910	70,210	114,890	–
FIB-4 1.3	11.324	0	5300	75,270	119,820	747,150
ELF 9.00	11.326	1070	5420	76,660	122,260	1,162,270
Dominateda
ELF 9.80	11.316	1090	4390	70,190	115,620	–
FIB-4 1.3–2.67/ELF 9.00	11.323	790	4880	75,270	120,180	–
FIB-4 1.3–2.67/ELF 7.70	11.324	790	5300	75,270	120,610	–

ELF, enhanced liver fibrosis; FIB-4, fibrosis index; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-years.

^aA strategy is said to be a dominating strategy, if it is on the cost-efficiency frontier. That is, there exists some willingness-to-pay threshold such that this strategy would be the most cost-effective strategy. Otherwise, a strategy is said to be dominated.

Cost and cost-effectiveness

Lifetime costs ranged from $96,990 to $105,560, with highest cost for “ELF 9.00” due to treating most with Resmetirom and rendering longest LE (Table 3). Lifetime spendings on Resmetirom at $47,400/y (Total: $57,540–$65,500; mean duration: 1.6–1.8 year) accounted for most of the lifetime costs. Three strategies were cost-effective, with increasing costs: “FIB-4 1.3–2.67/ELF 9.80,” “FIB-4 1.3,” and “ELF 9.00.” The ICERs were $301,270/QALY (“FIB-4 1.3” vs “FIB-4 1.3–2.67/ELF9.80”), and $457,620/QALY (“ELF 9.00” vs “FIB-4 1.3”). Given a willingness-to-pay threshold of $100,000/QALY, the optimal strategy was “FIB-4 1.3–2.67/ELF 9.80.”

Subgroup Analysis

The clinical, cost, and cost-effectiveness outcomes for the subgroup analysis of patients ≥65 years are shown in Table 3 and Supplementary Table. LE ranged from 8.538 to 8.635 QALYs. Total lifetime costs ranged from $74,300 to $106,040. As in the base case, the strategies “FIB-4 1.3–2.67/ELF 9.80” and “ELF 9.00” remained cost-effective, while “FIB-4 1.3” did not. Annual screening using “FIB-4 1.3–2.67/ELF 9.80” remained the most cost-effective strategy at $100,000/QALY for this older subgroup.

Scenario Analysis

For a cohort with lower prevalence of T2D or obesity, projected LE increased to 11.316–11.326 QALYs (Table 3 and Supplementary Table). Total lifetime costs also increased correspondingly and ranged from $114,890 to $122,260. At $100,000/QALY, strategy dominance remained the same as base case, with “FIB-4 1.3–2.67/ELF 9.80” being the most cost-effective strategy and increased ICERs between strategies on the cost-efficiency frontier.

Probability Sensitivity Analysis

In probability sensitivity analysis, the optimal base case strategy, “FIB-4 1.3–2.67/ELF 9.80”, remained most likely (≥60% of simulations) to be cost-effective for both the full population and those aged ≥65 years at willingness-to-pay thresholds of $0–$160,000/QALY (Figures 2 and 3). With higher thresholds, “ELF 9.00” became likely cost-effective in growing plurality of simulations. The strategy “FIB-4 1.3” was likely more cost-effective with higher thresholds in the full population but not among an older cohort (Figure 3).

Figure 2.

Probability sensitivity analysis for a cost-effectiveness analysis of blood-based fibrosis screening in a real-world population with suspected MASLD: the full population. This plot displays, for the full population, the probability that each strategy was the most cost-effective (y-axis) as willingness-to-pay (x-axis) increased from $0 to $160,000/QALY. ELF, enhanced liver fibrosis; FIB-4, fibrosis-4; MASLD, metabolic dysfunction–associated steatotic liver disease; NIT, noninvasive test.

Figure 3.

Probability sensitivity analysis for a cost-effectiveness analysis of blood-based fibrosis screening in a real-world population with suspected MASLD: the subgroup aged ≥65. This plot displays, for the subgroup aged ≥65 years, the probability that each strategy was the most cost-effective (y-axis) as willingness-to-pay (x-axis) increased from $0 to $160,000/QALY. ELF, enhanced liver fibrosis; FIB-4, fibrosis-4; MASLD, metabolic dysfunction–associated steatotic liver disease; NIT, noninvasive test.

One-Way Sensitivity Analysis

The parameter most influential on the ICER between “ELF 9.00” and “FIB-4 1.3–2.67/ELF 9.80” was Resmetirom cost: varying from $10,000 to $90,000/y increased the ICER from $89,660/QALY to $616,690/QALY (Figure 4). At a $100,000/QALY willingness-to-pay threshold, the threshold cost of Resmetirom was $11,570/y (76% reduction from base case at $47,400/y), below which “ELF 9.00” would replace “FIB-4 1.3–2.67/ELF 9.80” as the most cost-effective strategy. Treatment effect was influential on the ICER but did not alter this conclusion; for example, even if Resmetirom reduced fibrosis progression risk 3-fold or increased fibrosis regression risk 3-fold, both compared to no treatment, the ICER remained above $100,000/QALY, leaving “FIB-4 1.3–2.67/ELF 9.80” as most cost-effective. Other parameters that impacted the ICER but also did not change this conclusion included the following: the effect of Resmetirom on fibrosis regression, transition probability from F2-3 to F0-1, transition from F2-3 to CC, discount rate for cost and QALYs, and treatment uptake.

Figure 4.

One-way sensitivity analysis of the ICER between the strategies “ELF 9.00” and “FIB-4 1.3–2.67/ELF 9.80.” This tornado diagram displays the impact of varying individual parameters on the change in ICER between annual screening using ELF at a 9.00 cutoff (“ELF 9.00”) and annual screening using ELF at a 9.80 cutoff to follow indeterminate FIB-4 (“FIB-4 1.3–2.67/ELF 9.80”). For each parameter, the base case value is listed first, followed by the range in the parentheses. The base case values led to the thin vertical lines through the diagram, where the ICER was $336,050/QALY. The left value in the parentheses led to the lower ICER from the sensitivity analysis and the right value led to the higher ICER. A longer bar reflects a greater change in ICER as the parameter was varied. The black thick line at $100,000/QALY marks the willingness-to-pay threshold. The bars that extend to the right of this line mark where “ELF 9.00” was the most cost-effective compared to “FIB-4 1.3–2.67/ELF 9.80,” while the bars that extend to the left mark where “FIB-4 1.3–2.67/ELF 9.80” was more cost-effective than “ELF 9.00.” ELF, enhanced liver fibrosis; F2-3, fibrosis stage 2-3; FIB-4, fibrosis-4; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-years.

Discussion

We evaluated the clinical and cost-effectiveness of 6 blood-based NIT strategies to risk stratify an empiric population with suspected MASLD, explicitly linking real-world diagnostic test performance to contemporary treatment of F2-3 fibrosis with Resmetirom. We found that a 2-tier annual NIT strategy using ELF (cutoff 9.80) following indeterminate FIB-4 (1.3–2.67) (“FIB-4 1.3–2.67/ELF 9.8”) was the most cost-effective at a willingness-to-pay threshold of $100,000/QALY. The strategy “FIB-4 1.3–2.67/ELF 9.80” remained optimal across wide range of sensitivity analyses, including scenarios in which treatment costs were reduced by up to 76% and when screening was applied to populations with fewer metabolic comorbidities or older age distributions, where diagnostic accuracy may be confounded by age. Collectively, these findings highlight how integrating practical, scalable NIT-based pathways with emerging therapies can inform guideline implementation and improve early detection in general primary care populations.

Other NIT strategies offered modest clinical benefit by enabling earlier detection and treatment but incurred substantially higher staging and drug costs, limiting their cost-effectiveness. In probability sensitivity analysis, single-tier NITs—FIB-4 at a 1.3 cutoff (“FIB-4 1.3”) and ELF at a 9.00 cutoff (“ELF 9.00”)—gained some dominance at higher willingness-to-pay but remained less favorable than “FIB-4 1.3–2.67/ELF 9.80.” In older patients, however, declining sensitivity of FIB-4-based strategies had reduced clinical value, leaving only “ELF 9.00” as likely cost-effective besides “FIB4 1.3–2.67/ELF 9.80.” Notably, “ELF 9.00” maintained 100% sensitivity across age groups, ensuring all eligible individuals were treated early, and yielded greater LE gains than “FIB-4 1.3–2.67/ELF 9.80” in those aged ≥65 years and those with lower prevalence of T2D or obesity.

One-way sensitivity analysis revealed that Resmetirom cost was most influential and the only input that altered strategy dominance. Specifically, if Resmetirom cost fell below $11,570/y (76% reduction from base case, $47,400/y), “ELF 9.00” would become more cost-effective than “FIB-4 1.3–2.67/ELF 9.80.” Compared to “FIB-4 1.3–2.67/ELF 9.80,” higher sensitivity of “ELF 9.00” improved long-term clinical outcomes through treating 3–4 more patients per 100. However, its lower specificity led to more unnecessary referrals and higher cost primarily driven by Resmetirom. This suggests that reduced drug cost is needed to support adoption of single-tier NITs by enabling cost-effective treatment. Several US-based studies,^46,51,52 including one by the Institute for Clinical and Economic Review and another by the drug manufacturer, found treating all with significant fibrosis cost-effective at $100,000/QALY with Resmetirom priced between $15,410 and 26,280/y—well below the current $47,400/y as our threshold of $11,570/y for “ELF 9.00” to outperform “FIB-4/ELF 9.80.” While our comparison was not between treatment and no treatment but between different NIT strategies, the improvement in clinical outcomes was driven by greater treatment uptake, with costs dominated by drug price. Thus, the ICERs across NITs paralleled treatment vs no treatment. Our finding reinforces that moderate Resmetirom pricing both increases treatment value and expands cost-effective screening options. Ultimately, drug price is the main determinant of long-term screening cost-effectiveness.

Multiple studies assessed cost-effectiveness of NIT-based screening assuming nonpharmacological interventions as the downstream impact.^30,31,53 To our knowledge, our study is among the first to incorporate Resmetirom, a major strength.³² One prior study compared NIT strategies with no screening and found FIB-4 followed by TE or ELF (9.80 cutoff) most cost-effective (ICERs < $50,000/QALY) even at $47,400/y for Resmetirom.³² Currently in most settings, TE requires a referral to hepatology. Though we employed similar model structures for the MASLD natural history progression, the study assumed benefit from Resmetirom at F1, whereas we limited treatment effect to F2-3 per its approved use.⁴⁰ This more conservative approach likely explains the difference between our ICER findings. Another primary strength of our study lies in the use of real-world patient data on test performance. This allowed us to extrapolate, with confidence, the long-term value of NITs in risk stratifying a population with suspected MASLD. In a few short-term economic evaluations,27, 28, 29 including one based on a preliminary sample of our patient data, “FIB-4 1.3–2.67/ELF 9.80” was identified as the most cost-effective NIT strategy due to minimized unnecessary referrals. Our study extends this discovery by showing that “FIB-4 1.3–2.67/ELF 9.80” remained the most long-term cost-effective NIT strategy across wide sensitivity analysis ranges.

Our findings align with the evolving landscape of MASLD burden and clinical practice. Recent studies from Japan,¹⁸ US,^16,54 and Canada¹⁷ have reported increasing liver stiffness measurements at the national level using TE-based criteria, reflecting growing clinical recognition of advanced fibrosis and expanding real-world application of fibrosis assessment. In this broader context, our results are particularly timely: as demand for and capacity to perform fibrosis evaluation increase, scalable and accessible screening programs are needed upstream of specialist care. Blood-based NITs such as FIB-4 and ELF, both readily available in primary care, can serve as cost-effective triage tools to identify individuals most likely to benefit from specialist referral and effective pharmacological therapies, complementing downstream imaging-based fibrosis assessment.

Globally, reducing barriers to screening implementation and primary care integration remain pressing issues, further underscoring the relevance of our findings. In the United States, efforts to improve MASLD screening in primary care include the development and testing of multicomponent care pathways incorporating electronic health record-based triggers to accelerate specialist referrals,¹³ as well as pilot educational interventions aimed at increasing provider awareness and guideline adherence.⁵⁵ In parallel, the Nara Declaration and related analyses have emphasized elevated alanine ALT (ALT > 30 U/L) as a straightforward, automated upstream trigger for primary care-based risk stratification of liver fibrosis.^14,15 While ALT alone lacks sufficient specificity for fibrosis staging, its growing recognition as an informative initial filter highlights the need for downstream noninvasive risk stratification tools that are feasible in primary care. In this setting, our proposed sequential screening strategy using FIB-4 and ELF provides a scalable and pragmatic framework for refining fibrosis risk after initial identification, aligning naturally with this growing stream of primary care-focused initiative.

Our study has several limitations. First, we restricted screening impact to Resmetirom use and did not model other potential MASLD treatment options such as Semaglutide (Wegovy). Due to the recency of Wegovy’s FDA approval, it is unclear how it will be used as an alternative or simultaneous MASH therapy to Resmetirom. In addition, we assumed that different screening approaches would not lead to additional use of Wegovy, given that it is likely already prescribed for our modeled cohort with T2D and/or obesity. Second, there is paucity of data on the real-world adoption of and persistence with Resmetirom. We varied both uptake and discontinuation rates widely in sensitivity analyses and found neither to be influential on key conclusions. Third, we modeled treatment efficacy as constant over time based on 52-week clinical trial results. It is uncertain how Resmetirom can sustain fibrosis improvement in the long-term. Our sensitivity analyses showed that even with much higher efficacy, the preferred strategy would remain unchanged. Fourth, our modeled population was derived from a real-world veteran cohort that was predominantly older, with high prevalence of T2D and obesity, which may limit generalizability to other populations. However, scenario analysis focusing on a population with fewer metabolic comorbidities and subgroup analysis incorporating the age impact on screening performance confirmed that our key conclusions remained robust. Finally, we did not include imaging modalities such as TE in our comparisons because our focus is on screening in primary care with limited TE availability.

In this study, we did not consider the health equity impact of blood-based NIT strategies. Populations differing in socioeconomic status, for example, might benefit unequally from screening approaches due to varying disease prevalence and access to care. Future studies should investigate the distributional cost-effectiveness of these NIT strategies for different socioeconomic subpopulations.

In the new therapeutic era of MASLD, sequential testing with indeterminate FIB-4 (1.3–2.67) followed by ELF at a 9.80 cutoff represents the most cost-effective noninvasive screening strategy for primary care. While single-tier ELF screening at a 9.00 cutoff maximized clinical benefit through identifying all treatment-eligible individuals, it also increased unnecessary referrals and is economically justifiable only if drug costs are lower. Overall, these results support broader adoption of blood-based NIT strategies in both high-risk and general primary care populations. As new MASLD therapies emerge and guidelines evolve, ongoing evaluation of the comparative cost-effectiveness of screening approaches will be essential to optimize both clinical outcomes and health-care value.

Conclusion

Sequential testing with ELF at a 9.80 cutoff to follow indeterminate FIB-4 represents a cost-effective, noninvasive MASLD fibrosis screening strategy in primary care.