Esophageal adenocarcinoma is a cancer with increasing

Transcription

1 GASTROENTEROLOGY 2013;144:62 73 CLINICAL ALIMENTARY TRACT Health Benefits and Cost Effectiveness of Endoscopic and Nonendoscopic Cytosponge Screening for Barrett s Esophagus TATIANA BENAGLIA, 1 LINDA D. SHARPLES, 1 REBECCA C. FITZGERALD, 2 and GEORGIOS LYRATZOPOULOS 3 1 Medical Research Council, Biostatistics Unit, Cambridge, United Kingdom; 2 Medical Research Council Cancer Cell Unit, Hutchinson-MRC Research Centre, Cambridge, United Kingdom; 3 Cambridge Centre for Health Services Research, Institute of Public Health, Cambridge, United Kingdom BACKGROUND & AIMS: We developed a model to compare the health benefits and cost effectiveness of screening for Barrett s esophagus by either Cytosponge TM or by conventional endoscopy vs no screening, and to estimate their abilities to reduce mortality from esophageal adenocarcinoma. METHODS: We used microsimulation modeling of a hypothetical cohort of 50-year-old men in the United Kingdom with histories of gastroesophageal reflux disease symptoms, assuming the prevalence of Barrett s esophagus to be 8%. Participants were invited to undergo screening by endoscopy or Cytosponge (invitation acceptance rates of 23% and 45%, respectively), and outcomes were compared with those from men who underwent no screening. We estimated the number of incident esophageal adenocarcinoma cases prevented and the incremental cost-effectiveness ratio of quality-adjusted life years (QALYs) of the different strategies. Patients found to have high-grade dysplasia or intramucosal cancer received endotherapy. Model inputs included data on disease progression, test accuracy, post-treatment status, and surveillance protocols. Costs and benefits were discounted at 3.5% per year. Supplementary and sensitivity analyses comprised esophagectomy management of high-grade dysplasia or intramucosal cancer, screening by ultrathin nasal endoscopy, and different assumptions of uptake of screening invitations for either strategy. RE- SULTS: We estimated that compared with no screening, Cytosponge screening followed by treatment of patients with dysplasia or intramucosal cancer costs an additional $240 (95% credible interval, $196 $320) per screening participant and results in a mean gain of (95% credible interval, to 0.029) QALYs and an incremental cost-effectiveness ratio of $15.7 thousand (K) per QALY. The respective values for endoscopy were $299 ($261 $367), ( ) QALYs, and $22.2K. Screening by the Cytosponge followed by treatment of patients with dysplasia or intramucosal cancer would reduce the number of cases of incident symptomatic esophageal adenocarcinoma by 19%, compared with 17% for screening by endoscopy, although this greater benefit for Cytosponge depends on more patients accepting screening by Cytosponge compared with screening by endoscopy. CONCLUSIONS: In a microsimulation model, screening 50-year-old men with symptoms of gastroesophageal reflux disease by Cytosponge is cost effective and would reduce mortality from esophageal adenocarcinoma compared with no screening. Keywords: Cancer Prevention; Early Detection; Esophagus; Neoplasia. Esophageal adenocarcinoma is a cancer with increasing incidence and poor prognosis. 1,2 The potential for primary prevention of esophageal adenocarcinoma appears to be limited: the only established potentially modifiable lifestyle risk factors are smoking and obesity; however, associated relative risks are small. 3,4 Moreover, there are currently no effective population-based interventions to control increasing obesity trends. By 2015, it is estimated that, globally, more than 1.5 billion people will be overweight. 5 Esophageal adenocarcinoma arises from Barrett s esophagus; therefore, screening of patient groups at higher risk of developing symptomatic esophageal adenocarcinoma is appealing. For those patients an ideal screening strategy would use a minimally invasive test with the ability to provide tissue samples and appropriate diagnostic accuracy. Conventional endoscopy is expensive and depends on the availability of specialist clinical expertise and supporting services, whereas videocapsule endoscopy relies on visualization alone (providing no tissue sampling) and has poor diagnostic accuracy. 6 However, a novel test combining a nonendoscopic cytologic sampling device (Cytosponge; PA Consulting Group, Melbourn, United Kingdom) with an immunohistochemical biomarker offers promise. 7,8 Briefly, the Cytosponge is an ingestible gelatin capsule containing a compressed mesh attached to a string. The capsule is swallowed and moves by peristalsis into the gastric cardia; after 5 minutes the capsule dissolves, releasing the sponge. The device then is retrieved, collecting a cytologic specimen from the esophageal mu- Abbreviations used in this paper: BEST, Barrett s Esophagus Screening Trial; GERD, gastroesophageal reflux disease; HGD, high-grade dysplasia; IC, intramucosal cancer; ICER, incremental cost-effectiveness ratio; LGD, low-grade dysplasia; NDBE, nondysplastic Barrett s esophagus; QALY, quality-adjusted life-year by the AGA Institute /$

2 January 2013 COST EFFECTIVENESS OF BARRETT S SCREENING 63 cosa. Analysis involves immunostaining for trefoil factor 3, a diagnostic marker of Barrett s. 7 Cytosponge testing can take place in the primary care setting, with favorable patient acceptability. 8 The test has a sensitivity of 73.3% and a specificity of 93.8% for Barrett s esophagus segments of 1 cm (Prague classification C1, M1) or longer (Kadri et al, 8 Modeling studies can synthesize evidence, and illustrate the potential health benefits and the cost effectiveness of interventions with limited or emerging evidence. They also can help inform priorities for future research; conversely, further research also may be deemed unnecessary if cost effectiveness cannot be shown under any plausible model assumptions. This is important in view of the controversies surrounding screening activities in general on the one hand and policy initiatives aimed at improving early detection of cancer on the other. Against this background we have conducted a modeling study to explore the potential health benefits and cost effectiveness of screening by Cytosponge or conventional endoscopy as a means of reducing mortality from esophageal adenocarcinoma. Methods Model Structure Overview Screening strategy options were evaluated for a population of 50-year-old men with a history of symptoms of gastroesophageal reflux disease (GERD) and an assumed Barrett s prevalence of 8%. We did not include women because they constitute a low-risk group for the development of esophageal adenocarcinoma, with an incidence that is approximately one fifth of that in men. 9,10 The main model outcomes were the number of patients who develop incident esophageal adenocarcinoma and the incremental cost per quality-adjusted life-year (QALYs) for screening vs no screening. We initially compared 3 screening strategies: no screening; endoscopy screening; and initial Cytosponge screening, followed by endoscopy as a confirmatory test for Cytosponge-positive participants. We also explored screening by ultrathin nasal endoscopy in a supplementary analysis. We did not include videocapsule endoscopy because it provides no tissue sampling and has poor diagnostic accuracy for detection of Barrett s. 6 In both endoscopy and Cytosponge screening strategies, Barrett s esophagus patients with high-grade dysplasia (HGD) or intramucosal cancer (IC) received radical treatment by endotherapy, that is, radiofrequency ablation with or without endoscopic mucosal resection, as the increasingly accepted standard of care for those patients. 11 Nevertheless, in a supplementary analysis we also considered management of HGD or IC by esophagectomy. Subsequent health status, health care use, and survival experience of this cohort were estimated using microsimulation over a time horizon of 50 years or intervening death (Figure 1). The model comprises a decision tree (for the initial screening and treatment) followed by a semi-markov model (for disease progression, and subsequent screening and management strategies). Different stages of disease (ie, no Barrett s, nondysplastic Barrett s (NDBE), low-grade dysplasia (LGD), HGD, IC, and symptomatic esophageal adenocarcinoma), treatment, post-treatment, and death are repre- Figure 1. Model structure. Decision tree followed by a semi- Markov model. Transition rates between stages (progression or regression) are determined by natural history of the Barrett s esophagus. Patients with nondysplastic Barrett s and lowgrade dysplasia will undergo surveillance. In the base case, patients with treatable symptomatic cancer are managed by esophagectomy and patients with surveillance-detected highgrade dysplasia or intramucosal cancer are managed by esophagectomy or endotherapy.

3 64 BENAGLIA ET AL GASTROENTEROLOGY Vol. 144, No. 1 sented as discrete states in the model. The model has cycles of 30 days duration and movements between different states are determined by transition probabilities. Estimates of between-state transition rates, and the cost and utility associated with each state, were based on those used by the 2010 National Institute for Health and Clinical Excellence (NICE) Guideline Development Group on Barrett s esophagus Costs and utilities both were discounted at an annual rate of 3.5%. The management of Barrett s esophagus without LGD or NDBE was by follow-up surveillance endoscopy. Microsimulation was used because surveillance regimens depend on time spent in a specific state (eg, LGD) and memory of the time each patient has spent in a state is required by the model. 17 Model input estimates, uncertainty (standard errors), and distributions of each variable are shown and referenced in Table 1. Barrett s Prevalence Estimates We assumed a population prevalence of 8% for any type of Barrett s esophagus among screening invitees (ie, 50-year-old men with a history of GERD symptoms), based on estimates recently reported by Rubenstein et al. 18 For Barrett s esophagus patients, we assumed that 86% would have NDBE, 10% would have LGD, 2% would have HGD, and 2% would have IC, using conservative estimates based on a systematic review. 13 Natural Disease Progression Individuals in the no-screening scenario were exposed to risk of developing esophageal adenocarcinoma given the assumed transition probabilities, and esophageal adenocarcinoma could be detected only symptomatically (Table 2). 10 Mortality from causes other than esophageal adenocarcinoma was calculated from UK age- and sex-specific mortality statistics. 19 We used a very short duration of the microsimulation cycle (30 days). The model assumes that patients can move only between adjacent states (eg, from NDBE to LGD, or vice versa), however, because of the very short cycle length (30 days) used, potential progression or regression to nonadjacent states within relatively short follow-up periods is possible. For example, in our model, NDBE patients have a 0.02% chance of developing HGD within 6 months, a 0.05% chance within 12 months, and a 0.9% chance within 5 years. Screening Test Accuracy and Uptake Endoscopic screening with biopsy was considered a perfect (gold standard) test for the purposes of the modeled pathway. This consideration takes into account the face validity of a negative endoscopy test, that is, although the sensitivity of endoscopy is less than 100%, clinicians typically will not re-order an endoscopic investigation when a patient had a recent endoscopy with negative findings. Further, any bias potentially resulting from this assumption is minimized because Cytosponge screening is followed by confirmatory endoscopy, therefore endoscopy is used in both screening strategies. The sensitivity and specificity of Cytosponge was assumed to be 73.3% and 93.8%, respectively, for Barrett s esophagus segments of 1 cm or longer. 8 Uptake of Cytosponge screening was assumed to be 45% and that of endoscopy was assumed to be 23% (ie, approximately half of the uptake value assumed for Cytosponge). This assumption was based on recent evidence indicating substantially greater uptake of nonendoscopic compared with endoscopic screening methods for colorectal cancer, 21 and evidence suggestive of low uptake of endoscopy screening among patients with GERD symptoms. 21 We assumed that 80% of Cytosponge-positive patients will accept the offer of subsequent confirmatory endoscopy, informed by similar levels of uptake of colonoscopy invitation (79% to 84%) after positive fecal occult blood testing for colorectal cancer. 22,23 In a supplementary analysis for ultrathin nasal endoscopy we assumed a sensitivity of 91% and a specificity of 100% based on a recent cross-over randomized controlled trial 24 ; and an uptake of 23% (as for conventional endoscopy) based on recent evidence suggesting insubstantial differences in patient preferences for either conventional or nasal endoscopy. 24 Management Strategies Among patients with symptomatic esophageal adenocarcinoma it was assumed that 25% would be treatable by esophagectomy. 25,26 The postoperative mortality rate for esophagectomy in symptomatic patients was assumed to be 4.5% based on recent UK national audit data. 25 Survivors entered a postsurgery state until death from either nonlocal recurrence or other causes. For patients untreatable by esophagectomy an annual probability of death of 0.78 was assumed. 13 Patients with HGD or IC were treated by radical endotherapy (treatment by esophagectomy also was explored in a supplementary analysis). For those patients endotherapy was assumed to be effective in eradicating 93% of all dysplasia, 4% of all patients initially treated by endotherapy required esophagectomy (Table 1). 27 We assumed an average of 3.5 endotherapy treatment sessions, and subsequent endoscopic surveillance follow-up evaluation (at 3, 6, and 12 months in the first year, and annually thereafter) with re-treatment as required. 28,29 Patients with NDBE or LGD underwent surveillance every 3 years or 6 months, respectively, and those progressing to surveillance-detected HGD or IC were treated by endotherapy. We assumed a 1:10,000 risk of mortality associated with endotherapy. 30 In a supplementary analysis encompassing esophagectomy (as opposed to endotherapy) management, for patients with screeningor surveillance-detected HGD or IC undergoing esophagectomy, we assumed a postoperative mortality rate of 2.7%. 29 Costs and Utilities Costs associated with Cytosponge screening information were provided by one of the authors (R.C.F.) using direct knowledge of manufacturing ($15/unit), test administration (including staff time costs of $11/test), and laboratory costs ($61/ test), based on the Barrett s Esophagus Screening Trial (BEST) study. 8 All other costs in the main analysis relate to published National Health Service (NHS) reference costs in from a UK NHS perspective. 12,31 The costs were converted to US$ using the exchange rate adjusted for power purchasing parity of 1 Great Britain pound equals 1.52 US dollars. 32 Throughout the article, endoscopy denotes endoscopy and biopsy, and relevant endoscopy costs also include costs of biopsy processing and pathology reporting. Utilities used were based on the National Institute for Health and Clinical Excellence guideline. 12 In a supplementary analysis we also explored various cost ratios between ultrathin nasal endoscopy and conventional endoscopy, assuming that the cost of the former represents 100%, 75%, and 50% of conventional endoscopy. We used this heuristic approach because we could not obtain authoritative cost estimates for that procedure, and because there is large heterogeneity in the structure of these costs (eg, depending on the exact type of nasal endoscopes used, whether the procedure is performed in either primary care or hospital environments, and whether it is performed by either a trained generalist or a gastroenterologist).

4 January 2013 COST EFFECTIVENESS OF BARRETT S SCREENING 65 Table 1. Model Inputs Description Base Standard error or range Distribution Source Barrett s prevalence estimates Prevalence in 50-year-old men with 8% 1.6% 20% a DSA Rubenstein et al 18 a history of GERD symptoms Subtype distribution NDBE 86% Based on Garside et al 13 LGD 10% HGD 2% IC 2% Transition rates (annualized) NBE to NDBE b Beta Inadomi et al, 14 Rubenstein et al 15 NDBE to NBE b Inadomi et al, 14 Shaheen et al 16 NDBE to LGD b Beta Garside et al 13 LGD to NDBE b LGD to HGD b HGD to LGD b HGD to IC b IC to symptomatic cancer b Accuracy of screening tests Endoscopy sensitivity and 100% Assumed gold standard specificity Cytosponge sensitivity 73.3% 60% 90% a DSA Kadri et al 8 Cytosponge specificity 93.8% 60% 100% a Endoscopy uptake rate 23% 20% 60% DSA Informed by Kadri et al, 8 Inadomi et al, 20 and Weller et al 23 Cytosponge uptake rate 45% 20% 60% DSA Endoscopy uptake after positive Cytosponge testing 80% Based on UK experience (uptake of colonoscopy after positive FOBT screening) 22,23 30 days Cycle length Treatment Proportion of symptomatic cancer patients suitable for surgery Annual mortality in untreatable NICE guideline 12 cancer Mortality from other causes Age-dependent ONS life interim tables Mortality from esophagectomy a DSA Mortality from esophagectomy for a Inadomi et al, 14 Shaheen et al 16 HGD/IC detected via surveillance Efficacy of endotherapy Shaheen et al 27 HGD to no Barrett s 0.89 HGD to NDBE 0.04 HGD to LGD Proportion of patients who need to be treated by esophagectomy after endotherapy 0.04 Assumption based on data by Shaheen et al 27 Mortality from endotherapy Informed by Inadomi et al, 14 c Percentage of symptomatic EAC NICE guideline 12 patients requiring chemotherapy Percentage of screen-detected 0.00 HGD/IC patients requiring chemotherapy after esophagectomy Rate of re-operations after esophagectomy for 10.2% 95% CI, 8.9% 11.6% NHS National Oesophago-Gastric Cancer Audit symptomatic cancer Discount rate (per annum) Surgery for cancer d $11, ( ) Gamma NICE guideline 12 Surgery for HGD/IC $11, ( ) $4, ( ) Gamma Average length of stay in hospital for surgery, days Average length of stay considered by the cost code for surgery, days 8.69

5 66 BENAGLIA ET AL GASTROENTEROLOGY Vol. 144, No. 1 Table 1. Continued Description Base Standard error or range Distribution Source Excess inpatient stay day cost $ ( ) $88.89 ( 58.48) Gamma Postsurgery follow-up 2 $ ( ) $ ( ) outpatient visits/y Endotherapy (RFA EMR) $ ( 1,135.00) $ ( ) Endoscopy biopsy (screening/ $ ( 517) $ ( ) surveillance) Cytosponge screening $152 ( 100) $38 $304 a ( ) DSA Local estimates available to authors Utilities No Barrett s Beta NICE guideline 12 Nondysplastic Barrett s LGD HGD/IC Symptomatic cancer Untreatable cancer Surgery for HGD/IC Surgery for perforation Surgery for cancer Post surgery state Endotherapy (RFA EMR) DSA, deterministic sensitivity analysis; EMR, endoscopic mucosal resection; FOBT, fecal occult blood testing; NBE, no Barrett s esophagus; NICE, National Institute for Health and Clinical Excellence (UK); RFA, radiofrequency ablation. a Ranges are the range of inputs used for the DSA. b Ranges are estimates from systematic review unless otherwise stated. c The estimate used here is conservative. d Mean costs of esophagectomy for cancer used here also encompass chemotherapy costs for a proportion of patients on top of the base cost for surgery. 12 Probabilistic Sensitivity Analysis Probabilistic sensitivity analysis for the base-case scenario was performed for uncertainty surrounding costs, utilities, and transition rates. Because of the computational time required for the patient-level simulation, it is prohibitive to perform robust sensitivity analysis on the transition rates. To overcome that issue, we applied a Gaussian process technique 33 to obtain model outcomes for 5000 sets of transition rates drawn from distributions, which then were combined with 5000 values of costs and utilities. The R package BACCO, a bundle for Bayesian Analysis of Computer Code Output, 34 was used to implement this. Costs and utilities were assumed to follow and distributions, respectively. 12 For each sample the average difference in costs between the screening strategies and no screening was plotted against the mean difference in QALYs on the costeffectiveness plane. Two further plots were produced: the costeffectiveness acceptability curve, which plots the probability that each strategy is cost effective compared with no screening, against the amount of money that a health provider is willing to pay per additional QALY; and the cost-effectiveness acceptability frontier, which plots the probability that each strategy is the most cost effective (optimal choice) against the health care provider s willingness to pay per additional QALY. Deterministic Sensitivity Analysis The following model inputs were varied to assess their impact on cost effectiveness: age at initial screening, Barrett s esophagus prevalence, Cytosponge diagnostic accuracy and cost, number of endotherapy sessions required for HGD eradication, duration of postendotherapy surveillance, and postoperative mortality rate (esophagectomy management supplementary analysis). In addition, we have explored different scenarios of equal or differential uptake of endoscopy and Cytosponge screening, using a maximum value of 60% (informed by the uptake of colorectal cancer screening methods in the United Kingdom 23 ) and a minimum value of 20% (informed by the combined uptake of offer of both endoscopy and Cytosponge testing in the BEST feasibility study 8 ). Table 2. Transition Matrix of Barrett s Esophagus Subtype Within 1 and 5 Years of Follow-Up Evaluation, Starting in Subtypes Indicated by the Rows Initial subtype No Barrett s Nondysplastic Barrett s LGD HGD IC Transition probability within 1 year No Barrett s Nondysplastic Barrett s LGD HGD Transition probability within 5 years No Barrett s Nondysplastic Barrett s LGD HGD

6 January 2013 COST EFFECTIVENESS OF BARRETT S SCREENING 67 Table 3. Cost Effectiveness Analysis of Screening Methods No screening (symptomatic management only) Cytosponge screening followed by endotherapy Endoscopy screening followed by endotherapy Symptomatic EAC cases per 100,000 patients (219%) 788 (217%) screened during subsequent lifetime Mean survival a Mean cost a $132 $373 $431 Mean QALY a ICER compared with no screening N/A $15,724 $22,167 ICER endoscopy vs Cytosponge N/A N/A Dominated b EAC, esophageal adenocarcinoma. a These figures represent the mean survival or cost or QALY associated with the postscreening health care use patterns and survival experience of the average patient in the screening cohort. All costs and utilities are discounted at 3.5% per annum (see Methods section for more detail). b The term dominated denotes that the Cytosponge screening strategy costs less and is more effective than endoscopy screening. Computation and Validation The model was programmed in the freely available statistical software R. Because the proportion of persons invited for screening who have Barrett s esophagus is relatively low we used stratified sampling, with robust estimates achieved by simulating 100,000 patients in each subtype. Model validity was assessed by comparing the model-predicted survival without offer of screening with survival estimates obtained from population statistics (Supplementary Figure 1) and by an extensive range of internal validity checks. We also compared the incidence of esophageal adenocarcinoma among NDBE and LGD patients estimated by our model with those reported by recently published large cohort studies or meta-analyses. Results Base-Case Results Compared with no screening/no intervention, endoscopy screening followed by treatment of HGD and IC patients, and endoscopic surveillance of Barrett s patients with NDBE or LGD, has an additional mean cost of $299 per screening participant and results in a gain of QALYs and an incremental cost-effectiveness ratio (ICER) of $22.2 thousand (K) (Table 3). Compared with no screening, Cytosponge screening followed by treatment of HGD and IC patients, and endoscopic surveillance of Barrett s patients with NDBE or LGD, costs an additional $240 per screening participant and results in a gain of QALYs and an ICER of $15.7K. Cytosponge screening was as effective as endoscopic screening, therefore the decision is reduced to a cost comparison and the Cytosponge is cheaper per patient screened than endoscopy. Assuming screening invitation uptake of 45% for Cytosponge and 23% for endoscopy, that is, considerably higher uptake of Cytosponge than endoscopy (see Methods section), initial screening, followed by treatment of HGD and IC patients, and endoscopic surveillance of Barrett s patients with NDBE or LGD, reduces the number of 50-year-old men with a history of GERD symptoms who would develop symptomatic esophageal adenocarcinoma during their subsequent lifetime by 17% (endoscopy screening) or 19% (Cytosponge screening). Supplementary Analysis In a supplementary analysis considering esophagectomy management (as opposed to endotherapy) for patients with HGD or IC, the cost per screening participant and cost-effectiveness ratios were greater than those observed for endotherapy management, with ICER values of $26.0K and $36.5K per QALY for Cytosponge and endotherapy screening, respectively (Supplementary Table 1). In a further supplementary analysis, if the cost of ultrathin nasal endoscopy was set at 50% of the cost of conventional endoscopy, ultrathin nasal endoscopy screening with endotherapy management compared with no screening is associated with an ICER of $19.1K per QALY compared with no screening (Supplementary Table 2). Probabilistic Sensitivity Analysis We used probabilistic sensitivity analysis to estimate 95% credible intervals around the base-case estimates for the incremental mean cost per screening participant and the number of QALYs gained for either screening strategy compared with no screening. These were $299 per screening participant (95% credible interval, $261 $367) and a QALY gain of (95% credible interval, ) for endoscopy screening, whereas the respective figures for Cytosponge were $240 (95% credible interval, $196 $320) and (95% credible interval, to 0.029). The joint distributions of the differences in costs and the differences in QALYs are plotted in cost-effectiveness planes (Figure 2), and costeffectiveness acceptability curves and frontiers (Figure 3), illustrated for both the endoscopy and the Cytosponge screening strategy, and also encompassing subsequent management with either endotherapy or esophagectomy. Incorporating the uncertainty in costs, utilities, and transition rates, all 4 strategies do not present a big difference in terms of incremental QALY, therefore the comparison should be made mainly in terms of costs. In that sense, Cytosponge screening followed by endotherapy management is the most cost-effective option (Figure 2A). For a willingness-to-pay threshold of $45K per QALY (as used in the United Kingdom 35 ) and endotherapy management, the probability of cost effectiveness is 94% for both the

7 68 BENAGLIA ET AL GASTROENTEROLOGY Vol. 144, No. 1 Deterministic Sensitivity Analysis Cytosponge screening with endotherapy management was the most cost-effective strategy in all deterministic sensitivity analysis scenarios considered (Figure 4 and Supplementary Table 3). Utility and cost effectiveness increase with decreasing age at initial screening invitation, and when targeting population subgroups with higher Barrett s prevalence. Varying all other inputs used in deterministic sensitivity analysis did not have a large impact on cost effectiveness. Different assumptions about uptake of the invitation for screening by endoscopy or Cytosponge do not affect ICER estimates of either strategy Figure 2. Cost-effectiveness plane comparing each screening method against no screening/no intervention for different management strategies: (A) endotherapy, (B) esophagectomy. The black dashed line represents the threshold of $45K per QALY. Cytosponge and endoscopy. On the other hand, when considering a care pathway involving esophagectomy management, the probability of cost effectiveness is 87% for Cytosponge and 62% for endoscopy (Figure 2B). The cost-effectiveness acceptability curve (Figure 3A) shows that if a threshold of $100K were to be chosen, all 4 possible combinations of screening strategy by subsequent management option would have a cost-effective probability of 80% or greater. The cost-effectiveness acceptability frontier (Figure 3B) indicates that, for a health care provider willing to pay up to $16.5K per QALY, no screening is the optimal choice, but above that threshold Cytosponge screening with endotherapy management is the optimal choice. Figure 3. (A) Cost-effectiveness acceptability curve. (B) Cost-effectiveness acceptability frontier. For a given threshold of incremental costeffectiveness ratio (willingness-to-pay), the cost-effectiveness acceptability curve (top) gives the probability that each strategy is cost effective, and the cost-effectiveness acceptability frontier (bottom) indicates the probability that the optimal strategy is cost effective.

8 January 2013 COST EFFECTIVENESS OF BARRETT S SCREENING 69 Figure 4. (A-D) Deterministic sensitivity analysis tornado diagrams for each of the 4 screening strategies modeled. Within each of the 4 panels, horizontal bands represent variation in ICERs for 1-way deterministic sensitivity analysis scenarios. The vertical line on each diagram represents the ICER for the main model (see also Supplementary Table 3).

9 70 BENAGLIA ET AL GASTROENTEROLOGY Vol. 144, No. 1 against no screening, but do affect the impact of either strategy on the proportion of symptomatic cases of esophageal adenocarcinoma that can be prevented among screening invitees. If the uptake of Cytosponge was either 20% or 60% (rather than the base case of 45%, Supplementary Table 4), then the proportion of symptomatic cases that could be prevented would be 8% or 25%, respectively (not 19%). Similarly, if the uptake of endoscopy was 20% or 60% (rather than 23%), the impact would be 14% or 43% (not 17%). The ICER of endoscopy over Cytosponge for different scenarios of uptake of either technique is always positive (ie, there is extra cost per QALY using endoscopy compared with Cytosponge), except for a scenario of uptake of 45% for Cytosponge and 20% for endoscopy, in which case endoscopy screening is both less effective and more costly. Discussion Summary of Main Findings We provide comparative quantitative evidence for the health benefits and cost effectiveness of different screening strategies that aim to reduce mortality from esophageal adenocarcinoma among 50-year-old men with a history of GERD symptoms. Endoscopy and Cytosponge screening strategies do not differ much in terms of QALY gain, so the comparisons are based mainly on cost differences, with Cytosponge being the most cost-effective alternative. Assuming screening invitation uptake of 45% for Cytosponge and 23% for endoscopy (ie, considerably higher uptake of Cytosponge than endoscopy), screening by endoscopy or by Cytosponge, respectively, could prevent 17% and 19% of esophageal adenocarcinoma cases expected to occur during the subsequent lifetime of these individuals. However, deterministic sensitivity analysis suggests that the potential reduction in the number of esophageal adenocarcinoma cases expected to occur during the subsequent lifetime of individuals invited for screening can vary substantially when assuming higher or lower uptakes for either screening strategy. In supplementary analyses, if esophagectomy (instead of endotherapy) is used for patients with HGD or IC, the cost effectiveness of Cytosponge compared with endotherapy is even greater. Further, compared with ultrathin nasal endoscopy screening, Cytosponge screening is more cost effective even if the cost of ultrathin nasal endoscopy is assumed to be as low as 50% of the cost of conventional endoscopy (ie, when using a conservative assumption for this cost ratio). Combined with the acceptability and ease of administration of Cytosponge, 8 these findings indicate that screening of symptomatic GERD patients based on this test is an appealing strategy worthy of further exploration. Comparisons With the Literature A small number of previous studies examined the cost effectiveness of potential screening strategies by transoral or transnasal endoscopy, or videocapsule endoscopy in middle-aged men in the general population. 14,29,36 This study examined the cost effectiveness of Cytosponge screening, and we focus on middle-aged men with a history of GERD symptoms as opposed to members of the general population. Previous studies chiefly were related to esophagectomy rather than endotherapy management. Therefore, the usefulness of comparisons of the findings of our study with previous literature is limited. Comparing the number of incident cases of esophageal adenocarcinoma occurring in patients with Barrett s esophagus estimated by our model with recent epidemiologic evidence about such incidence may be insightful. The model estimates the risk of developing esophageal adenocarcinoma among NDBE or LGD patients to be 0.15% (95% credible interval 0.02% 0.46%) and 0.54% (95% credible interval 0.07% 1.75%) per year of follow-up evaluation, respectively. This is similar to the findings of a large survey in a Danish population (0.10% [95% confidence intervals 0.07% 0.13%] and 0.51% [95% confidence interval 0.03% 0.86%] per year of follow-up evaluation, respectively). 37 It should be noted that the latter study did not require endoscopic verification of visible Barrett s lesions, 37 therefore estimates may reflect a relatively high proportion of patients with short-segment disease. A population-based study from Northern Ireland by Bhat et al 38 found that among patients with endoscopically visible Barrett s and intestinal metaplasia the incidence of adenocarcinoma is 0.33% (95% confidence interval 0.24% 0.45%) per year of follow-up evaluation. Lastly, a recent meta-analysis by Desai et al 39 indicated an adenocarcinoma incidence of 0.33% (95% confidence interval 0.26% 0.40%) per year of follow-up evaluation among NDBE patients. Despite substantial differences in these study populations and designs, the confidence intervals of reported incidence estimates overlap substantially (as do the credible limits of the incidence estimates produced by our model). Given the uncertainty around the estimates we believe that these comparisons indicate an acceptable degree of concordance of incidence estimates produced by our model compared with the best current estimates regarding progression rates from recently published large cohort studies or meta-analysis. We performed probabilistic sensitivity analysis in view of the uncertainty in respect to transition rates. Strengths and Limitations Our study had several strengths. We have used evidence for model inputs from recently published literature and up-to-date consensus statements. Among other options, we have examined the contemporary management of patients with HGD or IC with endotherapy followed by surveillance. The use of microsimulation in our article adds substantially to methodologies previously used to examine cost effectiveness of Barrett s screening interventions by allowing detailed patient histories (ie, time since entry to current state) to be modeled directly. We restricted our model to offer of screening to men. An offer of screening to women would be associated with

10 January 2013 COST EFFECTIVENESS OF BARRETT S SCREENING 71 much less favorable cost effectiveness because the incidence of esophageal adenocarcinoma is approximately one fifth 9,10 and our deterministic sensitivity analysis showed that a lower prevalence of Barrett s esophagus is associated with poorer cost effectiveness of screening. It is not uncommon for screening strategies to be sex-specific. For example, the UK government implements a national population-based abdominal aortic aneurysm screening program solely aimed at men. The previously reported acceptability of the Cytosponge appears to be very good, 8 although participation rates in the BEST study were low (19%) because all participants additionally were required to undergo an endoscopy as well. We have assumed that uptake of Cytosponge among patients with a history of GERD symptoms will be 45%. However, in the absence of other effective interventions to prevent esophageal adenocarcinoma deaths, such uptake still would be associated with a substantial decrease in mortality from esophageal adenocarcinoma. Different assumptions about screening uptake do not affect ICER estimates of either Cytosponge or endoscopy against no screening, but do influence estimates of the proportion of symptomatic cases of esophageal adenocarcinoma that can be prevented among screening invitees. Therefore, the uptake of screening invitation is a source of uncertainty regarding the health impact of either strategy, which we explored with deterministic sensitivity analysis. Clearly, contemporary primary evidence to reduce this uncertainty would be welcome, but we estimate and report the range of the impact of either strategy for a wide range of different uptake assumptions (Supplementary Table 4). Multifaceted interventions encompassing organizational changes and health care professional and patient education can help increase uptake of preventive interventions. 40 Policy and Research Implications In our study, Cytosponge screening was shown to be a useful test to detect Barrett s and capable of reducing mortality associated with esophageal adenocarcinoma, at an acceptable cost. These considerations have direct implications for policy making, particularly in countries (such as the United States and the United Kingdom) with a high prevalence of both GERD and Barrett s esophagus and a high population health burden of esophageal adenocarcinoma mortality. Further, in sensitivity analysis, we have explored the potential cost effectiveness of an offer of screening to patients with lower estimates of Barrett s prevalence. For a Barrett s prevalence of 1.6% among individuals invited for screening (ie, for a prevalence level similar to that observed in population-based studies 41 ), Cytosponge screening has an ICER of $39.4K/QALY (Supplementary Table 3). Although for an assumed Barrett s prevalence of 3% among screening invitees the respective ICER value would be $25.6K/QALY. These findings can help inform policies that aim to offer screening by Cytosponge to populations beyond the one modeled in our study (ie, 50-year-old men with a history of symptoms of GERD). We assumed that surveillance of patients with NDBE and LGD would be carried out only by endoscopy. However, through the emergence of genomic or molecular biomarkers for Barrett s esophagus and esophageal adenocarcinoma, 42,43 there is the potential for using the Cytosponge as part of surveillance protocols, which currently is being evaluated by the BEST2 study ( However, the data are not yet available to include this in the model. The use of risk-stratification biomarkers also could inform decisions both about the age at which individuals are invited for screening and the frequency and length of surveillance follow-up evaluation. Given that the cost of Cytosponge administration is substantially lower than that of endoscopy, screening protocols encompassing Cytosponge use for subsequent surveillance may in fact be much more cost effective than those observed in our study. In addition, Cytosponge-based surveillance is likely to be more acceptable to patients and feasible for the health care system. Identifying ways to improve surveillance protocols for Barrett s esophagus patients is a priority because the cost effectiveness of current protocols based on endoscopy is doubtful. 44 In addition, the estimated cost-effectiveness ratios may be conservative if the recommended length of postendotherapy surveillance follow-up evaluation is reduced in the future, as evidence is emerging suggestive of durability of the therapeutic effect. 27 Conclusions Based on simulation modeling incorporating current evidence and contemporary management options we provide quantitative evidence indicating that a nonendoscopic screening method, the Cytosponge, has the potential to prevent esophageal adenocarcinoma deaths at conventional thresholds of cost effectiveness. There is further promise about the use of biomarkers to personalize surveillance protocols. Such developments will support exploration of public health policies encompassing an offer of screening to population groups at varying levels of esophageal adenocarcinoma risk in the future. Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at and at dx.doi.org/ /j.gastro References 1. Eloubeidi MA, Mason AC, Desmond RA, et al. Temporal trends ( ) in survival of patients with esophageal adenocarcinoma in the United States: a glimmer of hope? Am J Gastroenterol 2003;98: Lepage C, Rachet B, Jooste V, et al. Continuing rapid increase in esophageal adenocarcinoma in England and Wales. Am J Gastroenterol 2008;103:

11 72 BENAGLIA ET AL GASTROENTEROLOGY Vol. 144, No Steevens J, Schouten LJ, Goldbohm RA, et al. Alcohol consumption, cigarette smoking and risk of subtypes of oesophageal and gastric cancer: a prospective cohort study. Gut 2010;59: Merry AH, Schouten LJ, Goldbohm RA, et al. Body mass index, height and risk of adenocarcinoma of the oesophagus and gastric cardia: a prospective cohort study. Gut 2007;56: World Health Organisation. Chronic diseases and health promotion. Part two. The urgent need for action. Chapter one. Chronic diseases: causes and health impact. Available at: who.int/chp/chronic_disease_report/part2_ch1/en/index16.html. Accessed: July Bhardwaj A, Hollenbeak CS, Pooran N, et al. A meta-analysis of the diagnostic accuracy of esophageal capsule endoscopy for Barrett s esophagus in patients with gastroesophageal reflux disease. Am J Gastroenterol 2009;104: Lao-Sirieix P, Boussioutas A, Kadri SR, et al. Non-endoscopic screening biomarkers for Barrett s oesophagus: from microarray analysis to the clinic. Gut 2009;58: Kadri S, Lao-Sirieix P, O Donovan M, et al. Acceptability and accuracy of a non-endoscopic screening test for Barrett s oesophagus in primary care: cohort study. BMJ 2010;341:c Shaheen NJ. Should women with heartburn undergo screening upper endoscopy for prevention of cancer? Am J Gastroenterol 2011;106: Gajperia C, Barbiere J, Greenberg D, et al. Recent incidence trends and sociodemographic features of oesophageal and gastric cancer types in an English region. Aliment Pharmacol Ther 2009;30: Bennett C, Vakil N, Bergman J, et al. Consensus statements for management of Barrett s dysplasia and early-stage esophageal adenocarcinoma, based on a Delphi process. Gastroenterology 2012;143: National Institute for Health and Clinical Excellence (NICE). Barrett s oesophagus ablative therapy (CG 106) Appendix 6: cost effectiveness analysis for Barrett s Oesophagus. Available at: pdf. Accessed: July Garside R, Pitt M, Somerville M, et al. Surveillance of Barrett s oesophagus: exploring the uncertainty through systematic review, expert workshop and economic modelling. Health Technol Assess 2006;10:1 142, iii iv. 14. Inadomi J, Sampliner R, Lagergren J, et al. Screening and surveillance for Barrett s esophagus in a high-risk population: a costutility analysis. Ann Intern Med 2003;138: Rubenstein JH, Inadomi JM, Brill JV, et al. Cost utility of screening for Barrett s esophagus with esophageal capsule endoscopy versus conventional upper endoscopy. Clin Gastroenterol Hepatol 2007;5: Shaheen NJ, Inadomi JM, Overholt BF, et al. What is the best management strategy for high grade dysplasia in Barrett s oesophagus? A cost effectiveness analysis. Gut 2004;53: Rutter CM, Zaslasvsky AM, Feuer EJ. Dynamic microsimulation models for health outcomes: a review. Med Decis Making 2011; 31: Rubenstein JH, Mattek N, Eisen G. Age- and sex-specific yield of Barrett s esophagus by endoscopy indication. Gastrointest Endosc 2010;71: Office of National Statistics. Interim Life Tables, United Kingdom Available at: index.html?nscl Interim Life Tables. Accessed: July Inadomi JM, Vijan S, Janz NK, et al. Adherence to colorectal cancer screening: a randomized clinical trial of competing strategies. Arch Intern Med 2012;172: Corder AP, Jones RH, Sadler GH, et al. Heartburn, oesophagitis and Barrett s oesophagus in self-medicating patients in general practice. Br J Clin Pract 1996;50: Lee TJ, Rutter MD, Blanks RG, et al. Colonoscopy quality measures: experience from the NHS Bowel Cancer Screening Programme. Gut 2012;61: Moss SM, Campbell C, Melia J, et al. Performance measures in three rounds of the English bowel cancer screening pilot. Gut 2012;61: Shariff MK, Bird-Lieberman EL, O Donovan M, et al. Randomized crossover study comparing efficacy of transnasal endoscopy with that of standard endoscopy to detect Barrett s esophagus. Gastrointest Endosc 2012;75: NHS National Oesophago-Gastric Cancer Audit rd annual report. Available at: NCASP/audits%20and%20reports/NHS%20IC%20OGC%20Audit% %20interactive.pdf. Accessed: July Lyratzopoulos G, Barbiere J, Gajperia C, et al. Trends and variation in the management of oesophagogastric cancer patients: a population-based survey. BMC Health Serv Res 2009;9: Shaheen NJ, Overholt BF, Sampliner RE, et al. Durability of radiofrequency ablation in Barrett s esophagus with dysplasia. Gastroenterology 2011;141: Lyday WD, Corbett FS, Kuperman DA, et al. Radiofrequency ablation of Barrett s esophagus: outcomes of 429 patients from a multicenter community practice registry. Endoscopy 2010;42: Gupta N, Bansal A, Wani SB, et al. Endoscopy for upper GI cancer screening in the general population: a cost-utility analysis. Gastrointest Endosc 2011;74: e Barbiere J, Lyratzopoulos G. Cost-effectiveness of endoscopic screening followed by surveillance for Barrett s esophagus: a review. Gastroenterology 2009;137: Department of Health, (2009) National Schedule of Reference Costs for NHS Trusts and PCTs combined. Available at: PublicationsPolicyAndGuidance/DH_ Accessed: July OECD. Purchasing power parities for GDP, economics: key tables from OECD, no Available at: org/economics/purchasing-power-parities-for-gdp-2009_ x table11. Accessed: July Stevenson MD, Oakley J, Chilcott JB. Gaussian process modeling in conjunction with individual patient simulation modeling: a case study describing the calculation of cost-effectiveness ratios for the treatment of established osteoporosis. Med Decis Making 2004;24: Hankin RKS. Introducing BACCO, an R bundle for Bayesian analysis of computer code output. J Stat Software 2005;14: National Institute for Health and Clinical Excellence (NICE). Guide to the methods of technology appraisal reference Available at: June2008.pdf. Accessed: July Nietert PJ, Silverstein MD, Mokhashi MS, et al. Cost-effectiveness of screening a population with chronic gastroesophageal reflux. Gastrointest Endosc 2003;57: Hvid-Jensen F, Pedersen L, Drewes AM, et al. Incidence of adenocarcinoma among patients with Barrett s esophagus. N Engl J Med 2011;365: Bhat S, Coleman HG, Yousef F, et al. Risk of malignant progression in Barrett s esophagus patients: results from a large population-based study. J Natl Cancer Inst 2011;103: Desai TK, Krishnan K, Samala N, et al. The incidence of oesophageal adenocarcinoma in non-dysplastic Barrett s oesophagus: a meta-analysis. Gut 2012;61: Power E, Miles A, von Wagner C, et al. Uptake of colorectal cancer screening: system, provider and individual factors and strategies to improve participation. Future Oncol 2009;5: Ronkainen J, Aro P, Storskrubb T, et al. Prevalence of Barrett s esophagus in the general population: an endoscopic study. Gastroenterology 2005;129: