Cost-effectiveness of sponge cytology test vs endoscopy for esophageal squamous cell carcinoma screening in high-risk areas

Abstract

BACKGROUND

Recent advances in esophageal sponge cytology tests highlight their potential in esophageal squamous cell carcinoma (ESCC) screening, but their cost-effectiveness remains unclear.

AIM

To investigate the cost-effectiveness of the sponge cytology test for ESCC screening in high-risk areas.

METHODS

We built a Markov model simulating 100000 participants aged 45 years in ESCC high-risk areas to project the cost-effectiveness of several screening strategies: No screening, endoscopic screening, and sponge cytology screening (with subsequent endoscopy for positive results) with different intervals. Outcomes included ESCC cases and deaths, cost, quality-adjusted life year (QALY), and incremental cost-effective ratio (ICER).

RESULTS

All screening strategies reduced ESCC cases and prevented deaths compared with no screening (373-2962 vs 3134 and 257-2305 vs 2409 per 100000 participants, respectively). Shorter screening intervals were associated with higher QALYs gained per person (one-time to per year: 532-5972 for endoscopy and 747-7162 for sponge cytology test). An annual sponge cytology test strategy was considered the most cost-effective with the greatest QALYs gained and an ICER of 6630 USD per QALY among all strategies. Endoscopy screening strategies were dominated by the annual sponge cytology strategy.

CONCLUSION

Our findings suggest that the sponge cytology screening strategy is cost-effective in ESCC high-risk areas and may inform policy decision-making.

Key Words: Cost-effectiveness; Early detection of cancer; Diagnostic test; Esophageal squamous cell carcinoma; Markov model; Sponge cytology test; Endoscopic screening

Core Tip: This study is the first economic evaluation of a minimally invasive sponge cytology screening strategy in areas at high risk for esophageal squamous cell carcinoma (ESCC). We found that both endoscopic and sponge cytology screening programs were highly cost-effective compared with no screening. Among all evaluated strategies, annual sponge cytology screening provided the greatest health benefits at the most favorable cost-effectiveness, indicating that it may represent an optimal population-based screening approach for ESCC in high-risk regions.

INTRODUCTION

Esophageal cancer ranks 11^th and 7^th in terms of incidence and mortality rates among all cancer types worldwide, and of those, approximately 85% of esophageal cancer cases had squamous cell carcinoma[1,2]. Esophageal squamous cell carcinoma (ESCC) mostly occurs in high-risk areas, such as Eastern and South Central Asia and South Africa[2]. As it has an asymptomatic onset, most cases are detected at late stages. The survival rate remains low in most countries[3] (with 3-year survival rates lower than 30%), thus early detection and curable treatment are crucial to prognostic improvement[4].

One-time upper gastrointestinal endoscopy is the current standard method for ESCC screening. Its effectiveness in reducing esophageal cancer incidence and mortality in high-risk regions has been widely confirmed[5-7]. However, large-scale endoscopic screening faces significant challenges, including unequal resource distribution and unavoidable discomfort. Notably, the novel esophageal sponge cytology test has emerged as a promising pre-endoscopic screening tool due to its minimally invasive nature and operational simplicity. The procedure involves swallowing a compressed sponge encapsulated within a gelatin capsule and attached to a thin string; when the capsule dissolves, the sponge expands in the stomach and is gently withdrawn through the esophagus, enabling the collection of exfoliated esophageal epithelial cells for cytological and biomarker analysis[8]. This test can facilitate risk stratification by identifying high-risk individuals for prioritized endoscopic referral and has already been implemented in real-world screening programs in China[8,9].

A series of studies have been conducted to evaluate the economic benefits of the sponge cytology test in Barrett’s esophagus and gastroesophageal reflux disease screening and surveillance in the United Kingdom and United States[10-13]. However, economic evaluation research on the sponge cytology test in ESCC screening is notably absent. Given that variations in disease subtypes and healthcare systems can significantly influence cost-effectiveness outcomes, further research is needed to assess the value of sponge cytology in the context of ESCC. In light of the globally uneven geographic distribution of ESCC incidence[2,14,15], we conducted this cost-effectiveness study of the sponge cytology test in target areas to fill this gap in evidence. We further compared the cost-effectiveness of sponge cytology testing with that of endoscopic screening to provide a comprehensive understanding of their economic and clinical impacts.

The primary objective of this study was to evaluate the cost-effectiveness of sponge cytology screening in ESCC high-risk areas. We examined a range of screening strategies with varying modalities and intervals to identify the most optimal approach under the modeled conditions and to provide further evidence for ESCC prevention and control efforts.

MATERIALS AND METHODS

Overview

This economic evaluation study was conducted based on a Markov model using TreeAge Pro version 2024 R1.1 (TreeAge Software; Williamstown, MA, United States), and the analyses were conducted based on Excel version 2021(Microsoft Corporation, Redmond, WA, United States). This study was reported based on the Consolidated Health Economic Evaluation Reporting Standards reporting guideline[16]. Ethical approval was not required as no human participants were involved in this study.

Natural diseases progression

A Markov model was used in our analysis to simulate disease progression over time, providing a structured framework to evaluate the long-term outcomes and cost-effectiveness of different screening strategies. A hypothetical cohort of 100000 individuals aged 45 years residing in ESCC high-risk areas was projected. The natural history of ESCC is depicted in Figure 1. Initial prevalence and annual transition probabilities between health states are summarized in Table 1.

Figure 1 Markov model structure illustrating the natural history and treatment-related transitions of esophageal squamous cell carcinoma. Red boxes represent disease health states, including healthy, mild dysplasia, moderate dysplasia, severe dysplasia/carcinoma in situ, early esophageal cancer, and advanced esophageal cancer, while the blue box denotes the treatment state. Solid arrows indicate allowed disease progression between adjacent health states, whereas self-loop arrows represent remaining in the same state within a cycle. Dashed arrows indicate transitions following treatment, including return to the Healthy state after successful treatment of dysplasia/carcinoma in situ and post-treatment transitions from cancer states. Once esophageal squamous cell carcinoma has developed, backward transitions to pre-cancerous states are not permitted. Death is modeled as an absorbing state applicable to all disease stages, as indicated in the figure.

Table 1 Model input.

Parameter	Base case	Range1	Distribution2	Ref.
Prevalence (%)3				Zeng et al[14], 2020; Xia et al[18], 2021; Chen et al[37], 2017; and Xia et al[38], 2022
mD	2.04	1.02-4.52	Triangular (0.0102, 0.0204, 0.0452)
MD	0.22	0.11-0.59	Triangular (0.0011, 0.0022, 0.0059)
SD/CIS	0.12	0.06-0.38	Triangular (0.0006, 0.0012, 0.0038)
EEC	0.05	0.03-0.10	Triangular (0.0003, 0.0005, 0.0010)
AEC	0.02	0.01-0.04	Triangular (0.0001, 0.0002, 0.0004)
Annual state transition rate4				Xia et al[18], 2021; Chen et al[37], 2017; and Xia et al[38], 2022
Healthy to mD	0.0120	± 50%	Triangular (0.006, 0.012, 0.018)
mD to healthy	0.0500	± 50%	Triangular (0.025, 0.05, 0.075)
mD to MD	0.0500	± 50%	Triangular (0.025, 0.05, 0.075)
MD to mD	0.0800	± 50%	Triangular (0.04, 0.08, 0.12)
MD to SD/CIS	0.1200	± 50%	Triangular (0.06, 0.12, 0.18)
SD/CIS to MD	Age-specific
SD/CIS to EEC
EEC to AEC
PT-SD/CIS to SD/CIS	0.0050	± 50%	Triangular (0.0025, 0.005, 0.0075)
PT-EEC to EEC	0.0500	± 50%	Triangular (0.025, 0.05, 0.075)
PT-EEC to AEC	0.1000	± 50%	Triangular (0.05, 0.10, 0.15)
PT-AEC to AEC	0.2304	± 50%	Triangular (0.115, 0.23, 0.345)
Participation rate				Gao et al[8], 2023; Xia et al[18], 2021; Chen et al[37], 2017; and Xia et al[38], 2022
Endoscopy screening	0.3352	± 50%	Triangular (0.30, 0.34, 0.67)
Sponge cytology screening	0.56	± 50%	Triangular (0.50, 0.56, 0.80)
Endoscopy after sponge cytology test	0.75	± 50%	Triangular (0.40, 0.75, 0.90)
Probability of receiving treatment3				Xia et al[18], 2021; Chen et al[37], 2017; and Xia et al[38], 2022
SD/CI	0.7458	0.5625-0.9654	Beta (20.75, 7.07)
EEC	0.9405	0.7149-1.0000	Beta (5.13, 0.32)
AEC	0.9643	0.8393-1.0000	Beta (11.08, 0.41)
Probability of symptomatic detection				Xia et al[18], 2021; Chen et al[37], 2017; and Xia et al[38], 2022
SD/CIS	0.01	0.005-0.02	Triangular (0.005, 0.01, 0.02)
EEC	0.20	0.10-0.40	Triangular (0.10, 0.20, 0.40)
AEC	0.70	0.56-0.90	Triangular (0.56, 0.70, 0.90)
Cost (USD)
Endoscopic test	74.31	± 50%	Triangular (50.36, 74.31, 108.69)	Xia et al[18], 2021; Li et al[19], 2019; Chen et al[37], 2017; and Xia et al[38], 2022
Sponge cytology test	35	± 50%	Triangular (17.5, 35, 52.5)	Estimated
Endoscopic complications	119.61	± 50%	Gamma (9.75, 0.08)	Xia et al[18], 2021; Chen et al[37], 2017; and Xia et al[38], 2022
Initial treatment				Xia et al[18], 2021; Chen et al[37], 2017; and Xia et al[38], 2022
SD/CIS	1687.61	± 50%	Gamma (3.33, 0.002)
EEC	8135.03	± 50%	Gamma (2.33, 0.0003)
AEC	7701.55	± 50%	Gamma (0.03, 0.00004)
Annual treatment
SD/CIS	227.26	± 50%	Gamma (1.27, 0.006)
EEC	386.13	± 50%	Gamma (1.23, 0.003)
AEC	359.83	± 50%	Gamma (2.05, 0.006)
Characteristics of tests
Sensitivity of endoscopy	100%	0.88-1	-	Xia et al[18], 2021; Nagami et al[39], 2014; and Dawsey et al[40], 1998
Specificity of endoscopy	100%	0.59-1	-
Sensitivity of sponge cytology test	94.5%	0.888-0.975	Triangular (0.888, 0.945, 0.975)	Gao et al[8], 2023
Specificity of sponge cytology test	91.9%	0.912-0.925	Triangular (0.912, 0.919, 0.925)
Complication rate of endoscopic test	0.00023	0-0.002	Triangular (0, 0.00023, 0.0013)	Silvis et al[41], 1976 and Wang et al[42], 2019
Health state5				Xia et al[18], 2021 and Liu et al[20], 2018
mD	1	0.98-1	Triangular (0.98, 1.00, 1.00)
MD	1	0.98-1	Triangular (0.98, 1.00, 1.00)
SD/CIS	0.84	0.79-0.89	Beta (3.57, 0.68)
EEC	0.70	0.66-0.74	Beta (2.63, 1.13)
AEC	0.61	0.56-0.66	Beta (1.12, 0.71)
PT-SD/CIS	0.90	0.88-0.92	Beta (2.26, 0.25)
PT-EEC	0.81	0.79-0.83	Beta (1.77, 0.41)
PT-AEC	0.74	0.71-0.77	Beta (0.95, 0.33)
Discount rate	0.05	0-0.08		Liu et al[21], 2022

¹Upper and lower limits of parameters were determined based on a 95% confidence interval, a 50% range, or a range of estimates[22,43].

²Utilities were assigned a beta distribution, and costs were assigned a gamma distribution. A triangular distribution encompassed the range between the minimum.

³Baseline proportion of individuals in each disease state at model entry.

⁴Annual probabilities of moving between health states.

⁵Utilities scores: Quality-of-life weights for quality-adjusted life year. calculation; 0 = death, 1 = perfect health.

mD: Mild dysplasia; MD: Moderate dysplasia; SD/CIS: Severe dysplasia/carcinoma in situ; EEC: Early esophageal cancer; AEC: Advanced esophageal cancer; PT: Post-treatment.

Patients diagnosed with esophageal dysplasia/carcinoma in situ or severe stages received appropriate therapy, such as endoscopic submucosal dissection, endoscopic mucosal resection, and radiofrequency ablation for esophageal dysplasia/carcinoma in situ and early esophageal cancer (including intramucosal carcinoma and submucosal carcinoma), or radical surgery, chemotherapy, radiotherapy, and other options for advanced esophageal cancer[17]. Among post-treatment patients, subsequent annual endoscopic surveillance was performed. Patients with esophageal dysplasia/carcinoma in situ were considered to be in a healthy state if no recurrence occurred within ten years[18], whereas those who refused treatment followed the natural progression of the disease.

The simulation continued until participants reached 75 years of age or died. In the model, it was assumed that participants with healthy, mild, or moderate dysplasia were assumed to die only from causes other than ESCC, those with severe dysplasia/carcinoma in situ or early ESCC followed all-cause mortality, and patients with advanced ESCC were subject to ESCC-specific mortality. A cycle length of one year and a half-cycle correction were applied in our simulation.

Screening strategies and surveillance management

Model pathways are illustrated in Supplementary Figure 1. Screening strategies simulated in this study include no screening (natural history), endoscopic screening, and sponge cytology screening. Six screening intervals were evaluated: Once, every 10 years, every 5 years, every 3 years, every 2 years, and annually.

In the no screening pathway, participants followed the natural history of ESCC, and diagnoses were made only through incidental detection. The endoscopic screening pathway represented the current gold standard in the real world, and thus if a participant received a negative result, no further examination and treatment was performed. For participants with positive sponge cytology results (defined as high-grade intraepithelial lesions or more advanced stages), confirmatory endoscopy was recommended. Those with negative results (including healthy individuals, mild or moderate esophageal dysplasia stages for endoscopy, or normal squamous esophageal cell for sponge cytology test) continued regular screening according to the assigned interval (none, every 10 years, every 5 years, every 3 years, every 2 years, or annually).

Test accuracy and participation rate

The endoscopic test was assumed to be the gold standard test, with both sensitivity and specificity set at 100%, consistent with previous economic evaluation studies[10-12]. The sensitivity and specificity of the sponge cytology test were 94.5% and 91.9%, respectively[8]. Positive results of the sponge cytology test were defined as high-grade intraepithelial lesions or more advanced states, and confirmatory endoscopy was recommended for these individuals. The participation rates for screening were set at 33.52% for endoscopy and 56% for sponge cytology, based on population-based screening studies[5,8]. The probability of referral for confirmatory endoscopy following a positive sponge cytology result was 75%, as reported in the trial by Gao et al[8]. Further details are provided in Table 1.

Costs and utilities

A healthcare sector perspective was adopted in our study. Screening costs included equipment costs, examination fees, administration expenses, staff time costs, and related resources[19]. Costs for endoscopic screening were derived from a population-based screening study[19], and costs related to the sponge cytology test comprised screening mobilization and administrative expenses (5 USD/person), examination costs (20 USD/person), and manufacturing costs (10 USD/person). The costs of corresponding treatment were obtained from a previous study by Xia et al[18]. All costs were adjusted to 2022 USD according to the health and medical service component of the Consumer Price Index and the exchange rate (1 Chinese Yuan = 0.1484 USD). Based on published literature, utility scores for each stage were estimated and converted into quality-adjusted life years (QALYs; Table 1)[18,20]. An annual discount rate of 5% was applied to both costs and QALYs in this study, in accordance with the Chinese Guidelines for Pharmacoeconomic Evaluations[21].

Model outputs and validation

Health outcomes included the number of ESCC cases and the number of ESCC deaths throughout the simulation. Cost-effectiveness outcomes included QALYs, costs, and the incremental cost-effective ratio (ICER). A QALY is a widely used outcome measure in health economic evaluations that integrates both the quantity and quality of life into a single metric. One QALY represents one year of life lived in perfect health, and utility values typically range from 0 (equivalent to death) to 1 (perfect health), with negative values possible for health states perceived as worse than death. The ICER is defined as the ratio of the incremental cost to the incremental health benefit, calculated as the difference in costs divided by the difference in QALYs gained between two strategies[22]: ICER = Incremental cost/QALYs gained. Comparisons of ICERs were assessed in two ways: (1) Each screening strategy was compared against the no screening strategy; and (2) All evaluated screening strategies, including both endoscopy and sponge cytology at different intervals, were compared to identify the optimal strategy.

A willingness-to-pay (WTP) threshold of one to three times per-capita gross domestic product (GDP) was adopted in this study, in accordance with the World Health Organization recommendations[23]. An ICER lower than one-time per-capita GDP was considered highly cost-effective, whereas an ICER between one to three times per-capita GDP was considered cost-effective. The per-capita GDP in high-risk areas of China was 9217 USD in 2022[24].

Model validation was performed using the Assessment of the Validation Status of Health-Economic Decision Models (Supplementary material)[25].

Scenario analysis

We further conducted scenario analyses across a range of possible screening ages to explore the cost-effectiveness of different screening strategies under various settings. A wide range of possible starting ages, from 40 years to 65 years, was examined in this analysis. Age-specific prevalence values and corresponding cycle lengths were varied, and the detailed parameter settings are summarized in Supplementary Table 1.

Deterministic and probabilistic sensitivity analyses

Deterministic one-way sensitivity analyses were performed for all parameters within predefined ranges to identify the most influential parameters, and the results are presented in Tornado diagrams. Probabilistic sensitivity analysis was further conducted using Monte Carlo simulation with 10000 iterations to evaluate the robustness of the model results and to identify the optimal strategy under different WTP thresholds; results are presented as a cost-effectiveness acceptability curve. Model parameters were varied based on 95% confidence interval, a 50% range, or ranges reported in the literature[11,12]. Utility values were assigned beta distributions, and costs were assigned gamma distributions, and a triangular distribution was used to represent uncertainty between minimum and maximum values when appropriate. Details of parameter uncertainties are summarized in Table 1.

RESULTS

Base-case results: Health outcomes

Health outcomes are summarized in Table 2. All screening strategies resulted in fewer ESCC cases and deaths compared with no screening. Endoscopic screening reduced ESCC incidence by 6%-76%, corresponding to 191-2379 fewer cases per 100000 participants as screening frequency increased from once to annual screening. Similarly, endoscopic screening reduced ESCC-related mortality by 4%-79%, equivalent to 104-1895 fewer deaths per 100000 participants, compared with no screening.

Table 2 Base case: Health outcomes.

Strategy	Health outcomes (per 100000 participants)
	ESCC cases	ESCC deaths
No screening	3134	2409
Sponge cytology screening
Once	2962	2266
Every 10 years	2515	1874
Every 5 years	2011	1469
Every 3 years	1510	1088
Every 2 years	1060	750
Every year	373	257
Endoscopic screening
Once	2943	2305
Every 10 years	2606	2006
Every 5 years	2277	1685
Every 3 years	1867	1362
Every 2 years	1470	1052
Every year	755	514

Health outcomes are presented per 100000 participants for each screening strategy. ESCC: Esophageal squamous cell carcinoma.

For sponge cytology screening, ESCC incidence was reduced by 5%-88% and ESCC-related deaths by 6%-89%, corresponding to 172-2760 fewer ESCC cases and 143-2152 fewer deaths per 100000 participants, respectively. Among all strategies, annual sponge cytology screening yielded the greatest health benefit, preventing 2761 ESCC cases and 2152 deaths per 100000 participants compared with no screening.

Base-case results: Cost-effectiveness outcomes

Cost-effectiveness analysis shows that all screening strategies incurred higher costs but yielded greater QALYs compared with the no-screening strategy (Table 3 and Figure 2). All screening strategies were considered highly cost-effective compared with no screening, with ICERs ranging from 2571 USD to 6318 USD.

Figure 2 Cost-effectiveness plot. Undominated, positive incremental cost-effective ratios are regarded as undominated. EDS: Endoscopic screening; SCS: Sponge cytology screening; QALY: Quality-adjusted life year.

Table 3 Base case: Cost-effectiveness outcomes.

Strategy	Cost per 100000 participants (USD)	QALYs per 100000 participants	Incremental cost per 100000 participants (USD)		QALYs gained per 100000 participants		ICERs (USD per QALY gained)
			Against no screening strategy	Among all strategies	Against no screening strategy	Among all strategies	Against no screening strategy	Among all strategies
No screening	4532190	1473925	-	-	-	-	-	-
Sponge cytology screening
Once	6452020	1474672	1919830	1919830	747	747	2571	2571
Every 10 years	8691075	1475723	4158885	4158885	1798	1798	2313	2313
Every 5 years	11823542	1476865	7291352	3132467	2940	1142	2480	2743
Every 3 years	15682730	1478055	11150540	3859188	4129	1190	2700	3244
Every 2 years	20115833	1479177	15583643	4433104	5252	1122	2967	3950
Every year	32781384	1481087	28249194	12665550	7162	1910	3944	6630
Endoscopic screening
Once	7019587	1474458	2487397	567568	532	-214	4671	Dominated
Every 10 years	9716978	1475250	5184788	1025903	1324	-473	3915	Dominated
Every 5 years	13676309	1476120	9144119	185767	2194	-745	4167	Dominated
Every 3 years	18773932	1477073	1424742	3091202	3147	-982	4525	Dominated
Every 2 years	24851313	1478033	20319123	4735480	4108	-1144	4946	Dominated
Every year	42263921	1479898	37731731	9482537	5972	-1189	6318	Dominated

Dominated, negative incremental cost-effective ratios are regarded as dominated. Costs and quality-adjusted life years are reported per 100000 participants for each screening strategy. QALY: Quality-adjusted life year; ICER: Incremental cost-effective ratio.

Under endoscopic screening, one additional QALY was gained at a cost of 3915-6318 USD, whereas under sponge cytology screening, the cost per QALY gained ranged from 2313-3944 USD. In both endoscopic and sponge cytology screening strategies, shorter screening intervals were associated with greater QALYs gained per 100000 participants, increasing from 532 to 5972 QALYs for endoscopy and from 747 to 7162 QALYs for the sponge cytology test as screening frequency increased from once to annual screening.

Further comparisons among all strategies indicated that annual sponge cytology screening was the optimal strategy, yielding the greatest QALYs gained (1910 per 100000 participants) with an ICER of 6630 USD. The endoscopy screening strategies were dominated by the annual sponge cytology strategy, as they produced fewer QALYs at higher or comparable costs (QALYs: 1474 458-1479898 vs 1481087 per 100000 participants).

Comparisons between screening modalities within the same screening interval demonstrated that sponge cytology screening dominated endoscopic screening across all evaluated screening intervals, achieving higher QALYs at lower costs (Supplementary Table 2). Evaluation of screening intervals suggested that annual screening was the best option for both endoscopic and sponge cytology screening strategies, with ICERs of 6630 USD and 9340 USD, respectively, among all assessed strategies (Supplementary Table 3).

Scenario analysis

In scenario analyses, all screening strategies remained cost-effective compared with the no-screening strategy (Supplementary Figure 2A; ICERs ranged from 1397 USD to 6560 USD). Annual sponge cytology screening strategies consistently dominated among all assessed strategies, with ICERs ranging from 5400 USD to 6036 USD.

In addition, the results suggested that screening programs initiated at younger ages yielded greater QALYs in this setting (Supplementary Figure 2B), with total QALYs increasing from 718222 to 1591612 per 100000 participants as the starting age increased from 40 years to 65 years.

Deterministic and probabilistic sensitivity analyses

In deterministic one-way sensitivity, all screening strategies remained cost-effective in comparison with the no-screening strategy. The most influential parameters are summarized and presented in Tornado diagrams in Figure 3. Ranking results were sensitive to screening participation rates. When the participation rate in sponge cytology screening fell below 0.39, the annual endoscopic screening strategy yielded the highest QALYs (14.7990 per person) with an ICER of 22996 USD (Figure 4A and Supplementary Table 4). Although annual sponge cytology screening costs were approximately half as much as annual endoscopy screening (196 USD vs 423 USD per person), it was dominated by a sponge cytology participation rate of 0.28, as it yielded the second highest QALYs.

Figure 3 Tornado diagrams. A: One-way sensitivity analysis Tornado diagrams for sponge cytology strategies within different screening intervals; B: One-way sensitivity analysis Tornado diagrams for endoscopy screening strategies within different screening intervals. mD: Mild dysplasia; MD: Moderate dysplasia; SD/CIS: Severe dysplasia/carcinoma in situ; EEC: Early esophageal cancer; AEC: Advanced esophageal cancer; EDS: Endoscopic screening; SCS: Sponge cytology screening; ICER: Incremental cost-effective ratio; QALY: Quality-adjusted life year.

Figure 4 One-way sensitivity effectiveness chart. A: One-way sensitivity effectiveness chart for various participation rates of sponge cytology test; B: One-way sensitivity effectiveness chart for various participation rates of endoscopy; C: One-way sensitivity effectiveness chart for various participation rates of endoscopy after sponge cytology test. EDS: Endoscopic screening; SCS: Sponge cytology screening; QALY: Quality-adjusted life year.

The annual sponge cytology screening strategy remained the best option when the participation rate in the endoscopy test varied within the predefined ranges (Figure 4B). At an endoscopy participation rate of 0.50, the annual endoscopy screening strategy achieved a similar health benefit to the annual sponge cytology test but at substantially higher costs (QALY/cost per person: 14.8105/598 USD for annual endoscopy screening vs 14.8109/328 USD for annual sponge cytology screening). The impact of participation rate on confirmation following a positive sponge cytology test is illustrated in Figure 4C.

We also examined the impact of endoscopy accuracy, which was assumed to be 100% in alignment with previous studies. The results demonstrated robustness, as the annual sponge cytology test remained the optimal strategy across a range of assumed endoscopy accuracy (Supplementary Tables 5 and 6).

The cost-effectiveness acceptability curve (Supplementary Figure 3) illustrates the result of 10000 iterations from the probabilistic sensitivity analysis. The probabilistic sensitivity analysis results are consistent with base-case findings and indicate that annual sponge cytology screening was cost-effective in 72.36% of iterations and 99.62% of iterations at the WTP thresholds of once and three-times per-capita GDP, respectively. Sponge cytology screening every 2 years was cost-effective in 23.98% of iterations at the once per-capita GDP threshold, whereas sponge cytology screening every 3 years was cost-effective in 2.85% iterations. Annual endoscopic screening was cost-effective in 0.26% of iterations at the WTP of three times the per-capita GDP.

DISCUSSION

To our knowledge, this is the first economic evaluation study of the minimally invasive sponge cytology test in ESCC high-risk areas. In this study, we found that for participants in ESCC high-risk areas, all screening programs were associated with a reduction in ESCC cases and deaths. Our study demonstrated that screening a targeted population was highly cost-effective, and an annual sponge cytology screening strategy was the most cost-effective of all the strategies assessed. Endoscopic screening resulted in increased costs, but fewer QALYs were gained compared to sponge cytology screening in this study. In scenario analyses, QALYs increased with younger screening age and screening intensity, representing the effectiveness of cancer screening.

All tested strategies remained consistently cost-effective compared with no screening in the deterministic one-way sensitivity analyses. Marginal disparity in ranking results occurred when the participation rate of screening varied, and this concurred with the findings of an earlier study comparing the cost-effectiveness of the sponge cytology test and endoscopy in Barrett’s esophagus screening[12]. Of note, sponge cytology tests are preferred over endoscopic tests without sedation due to their minimally invasive and less time-consuming nature[26], as the latter is currently the most performed modality in China. Examination using sponge cytology tests requires no facilities, and checks outside the hospital are thereby potentially suitable for population-based screening. Therefore, a higher participation rate of sponge cytology screening makes sense.

Previous studies have explored the cost-effectiveness of endoscopic screening strategies in ESCC high-risk areas like China[18,27,28]. All screening strategies remained cost-effective, and more frequent screenings yielded higher QALYs (from once to every 2 years: 1572532 to 1582334 per 100000 participants)[18], this is in line with our study. However, clinical resources are constrictive in ESCC high-risk areas for implementing mass endoscopic screening. Questionnaire-based risk-stratification tools prior to endoscopic screening are an alternative approach to promoting screening efficiency[29-31]. However, given their suboptimal performance in the screening population (areas under the receiver operating curve: 0.681-0.80), these tools were not included in our model. Our analyses focused on the sponge cytology test combined with endoscopic confirmation, which showed favorable discriminative abilities for ESCC in earlier studies[8,9,32]. This could potentially understate the benefits of sponge cytology screening, as its less invasive nature and time efficiency might make it more appealing for more frequent screening intervals.

Instead of implementing universal screening across all areas, the target population we focused on involved participants in ESCC high-risk areas given the unbalanced geographical distribution of ESCC. We believe that a targeted approach is more efficient as it yields higher positive cases and avoids unnecessary screening. Also, an extensive range of screening intervals and starting ages were included in this study to provide evidence for optimizing screening programs as tailored screening frequencies and starting ages were crucial for the balance of resources and risks. Another strength of our study is that the data incorporated in this study, e.g., initial prevalence, transition probabilities, and utility scores, were derived from several prospective trials. This allowed us to simulate scenarios reflecting real-world conditions more closely.

Notably, a combination of artificial intelligence (AI) with esophageal cytology further improved the cost-effectiveness of sponge cytology tests in this study, as AI-based automatic classification achieves comparative performance to expert cytologists[33-35], which leads to a reduction in expenditure due to extra labor. Reduced demand for professionals further promotes the fair distribution of screening opportunities, especially for regions with sparse medical resources. Besides, we believe that with the development of sponge cytology tests, e.g., scaled-up productivity and increased test accuracy, the overall costs are likely to decrease and the tests will be more affordable for wider use. Also, future advances in sponge cytology test data management and analytics can promote its applicability in different medical systems or settings.

There is no doubt that endoscopy can also detect gastric and duodenal lesions in addition to esophageal lesions. Encouragingly, sponge cytology has also shown remarkable performance in identifying cancerous conditions of the esophagogastric junction as well as gastric metaplasia and gastric atrophy[8,36]. Thus, sponge cytology could also suggest endoscopic referral for individuals at increased risk of gastric cancer. However, it should be noted that sponge cytology lacks evidence to detect stomach lesions as endoscopy does. We therefore recommend that the sponge cytology screening strategy should primarily be used for individuals with an increased risk of esophageal cancer, especially in high-risk areas or for those with known risk factors. For individuals with a combined risk of esophageal and gastric cancer, endoscopic screening should still be considered as the standard strategy. In such populations, a combined approach of sponge cytology and endoscopy may be a potential consideration. For example, if participants receive a negative endoscopy test in year 1, they might undergo sponge cytology tests in subsequent years; and when the next cycle begins, a repeat endoscopy test and subsequent sponge cytology tests should continue. This approach may offer a more comprehensive monitoring strategy for individuals at higher risk of both cancers. Nevertheless, further tailored research is imperative to verify the role of sponge cytology test in this population.

This study has limitations. First, adherence to all strategies across different screening intervals in this study was assumed to remain constant throughout the simulation due to model constraints, which does not fully reflect real-world variability. To address this, we conducted a wide range of sensitivity analyses to mitigate uncertainties in the parameters. Also, the available data on the accuracy of sponge cytology tests were limited by the lack of long-term follow-up. To better establish the test’s clinical utility, future studies incorporating extended follow-up periods are essential to evaluate its real-world effectiveness.

CONCLUSION

In summary, this study indicates that both endoscopic and sponge cytology screening programs are highly cost-effective compared with no screening in ESCC high-risk areas. Among the strategies evaluated, annual sponge cytology screening emerged as the optimal approach. These encouraging findings provide evidence for policymakers and healthcare sectors in terms of new paradigms for early ESCC detection and improved treatment.

ACKNOWLEDGEMENTS

We thank Lei Yuan from the Department of Health Management, Naval Medical University for his constructive comments on this work.