Article Text

Original research
Cost-effectiveness of endovascular thrombectomy in patients with acute stroke and M2 occlusion
1. Mihir Khunte1,
2. Xiao Wu1,
3. Sam Payabvash1,
4. Chengcheng Zhu2,
5. Charles Matouk3,
6. Joseph Schindler4,
7. Pina Sanelli5,
8. Dheeraj Gandhi6,
9. Ajay Malhotra1
1. 1 Department of Radiology and Biomedical Imaging, Yale University School of Medicine, New Haven, Connecticut, USA
2. 2 University of Washington School of Medicine, Seattle, Washington, USA
3. 3 Department of Neurosurgery, Yale University, New Haven, Connecticut, USA
4. 4 Department of Neurology, Yale University, New Haven, Connecticut, USA
5. 5 Hofstra Northwell School of Medicine at Hofstra University, Hempstead, New York, USA
6. 6 Department of Interventional Neuroradiology, University of Maryland, Baltimore, Maryland, USA
1. Correspondence to Dr Ajay Malhotra, Department of Radiology and Biomedical Imaging, Yale University School of Medicine, New Haven, CT 06520-8055, USA; ajay.malhotra{at}yale.edu

## Abstract

Background The cost-effectiveness of endovascular thrombectomy (EVT) in patients with acute ischemic stroke due to M2 branch occlusion remains uncertain.

Objective To evaluate the cost-effectiveness of EVT compared with medical management in patients with acute stroke presenting with M2 occlusion using a decision-analytic model.

Methods A decision-analytic study was performed with Markov modeling to estimate the lifetime quality-adjusted life years and associated costs of EVT-treated patients compared with no-EVT/medical management. The study was performed over a lifetime horizon with a societal perspective in the Unites States setting. Base case, one-way, two-way, and probabilistic sensitivity analyses were performed.

Results EVT was the long-term cost-effective strategy in 93.37% of the iterations in the probabilistic sensitivity analysis, and resulted in difference in health benefit of 1.66 QALYs in the 65-year-old age groups, equivalent to 606 days in perfect health. Varying the outcomes after both strategies shows that EVT was more cost-effective when the probability of good outcome after EVT was only 4–6% higher relative to medical management in clinically likely scenarios. EVT remained cost-effective even when its cost exceeded US$200 000 (threshold was US$209 111). EVT was even more cost-effective for 55-year-olds than for 65-year-old patients.

Conclusion Our study suggests that EVT is cost-effective for treatment of acute M2 branch occlusions. Faster and improved reperfusion techniques would increase the relative cost-effectiveness of EVT even further in these patients.

• stroke
• thrombectomy

## Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

## Introduction

Endovascular thrombectomy (EVT) has revolutionized care for patients with acute ischemic stroke with emergent large vessel occlusion.1 The current guidelines from the American Heart Association/American Stroke Association recommend EVT for patients with causative occlusion of the internal carotid artery (ICA) or proximal middle cerebral artery (MCA) M1 segment (class I recommendation).2 The benefits of EVT for more distal segments are deemed uncertain, but may be reasonable in carefully selected patients (class IIb recommendation).2 Acute M2 occlusions present with relatively smaller infarcts and hypoperfused volumes, and have smaller infarct volume regardless of recanalization.3 However, M2 segment occlusions can present with severe symptoms and lead to large infarct volumes, with resultant significant morbidity and mortality.4 The pivotal randomized trials in the HERMES Collaboration either excluded or undersampled patients with M2 segment occlusions. However, good outcome (modified Rankin Scale (mRS) score 0–2) was reported in 58.2% of patients after EVT (n=67) compared with 39.7% in the control arm (n=64) in the included patients.5 This is similar to results from a recent meta-analysis by Saber et al of 12 studies with 1080 patients with M2 occlusions, where functional independence (mRS score 0–2) was reported in 59% (95% CI 54% to 64%) and mortality in 16% (95% CI 11% to 23%).6 Recanalization rate (81%) similar to M1 thrombectomy was noted, with successful M2 recanalization associated with greater rates of favorable outcome compared with poor recanalization (OR=4.22).6 The recanalization rate (81%) was higher than noted in the HERMES trials (59.2%); however, the meta-analysis by Saber et al also reported higher rates of symptomatic intracranial hemorrhage (10% vs 0%) and death (16% vs 11.9%) than HERMES data. Multiple cohort studies have used comparable outcomes after EVT for M2 occlusion relative to M1 occlusions to advocate increased use of EVT for patients with a stroke with acute M2 occlusions.4–8

However, expanding thrombectomy to patients with acute M2 occlusions should be considered relative to outcomes after no-EVT/medical management in terms of expected health benefits and associated costs from a long-term, societal perspective. Higher rates of recanalization after intravenous thrombolysis (IVT) are reported for M2 occlusions relative to ICA or proximal M1 segment occlusions.9 10 For M2 occlusions with small ischemic cores (<30 mL) and good collaterals that achieve reperfusion, EVT may have lower odds of favorable outcomes than IVT.11 However, IVT has lower rates of complete reperfusion relative to EVT.10 Addition of IVT has been shown also to result in greater hospital encounter charges and final hospital bills.12 Although the safety and efficacy of EVT for acute M2 occlusions has been documented by many studies, the cost-effectiveness has not been established. Previous cost-effectiveness analyses have focused only on ICA/M1 occlusions. Moreover, conducting a randomized trial for M2 occlusions that includes a non-intervention arm has been deemed infeasible owing to perceived lack of clinical equipoise and because most institutions now prefer to perform EVT.6 The purpose of this study was to evaluate the cost-effectiveness of EVT compared with medical management in patients with acute stroke presenting with M2 occlusion using a decision-analytic model.

## Methods

A decision-analysis model with Markov modeling was constructed using TreeAge Pro Suite 2019 (Cambridge, Massachusetts, USA) over the lifetime span of patients from a societal perspective. By using computational simulation, decision analytic modeling can be considered as a complement to performing a large-cohort randomized control trial. With probabilistic sampling, the model simulates parallel cohorts of patients with M2 segment occlusions, treated with no-EVT/standard medical care or endovascular thrombectomy. We computed the respective costs associated with the strategies and expected outcomes as quality-adjusted life years (QALYs), a comprehensive utility metric accounting for both the expectancy and quality of life for patients in a specified health state.

Recommendations by the Second Panel on Cost-Effectiveness in Health and Medicine were adhered to in this study.13 The CHEERS (Consolidated Health Economic Evaluation Reporting Standards) statement is included.14

### Model structure

The base case scenario in this model was a patient aged 65 years presenting with an acute M2 segment occlusion. The two management strategies considered were EVT versus no-EVT/standard medical management for M2 occlusion. Patients in both strategy groups were treated with IVT if eligible. This was done to simulate real-life clinical scenarios, and maintain consistency with the literature sources used for input parameters described below. After each treatment strategy, the patients were categorized into one of seven health states based on the mRS score 0 to 6. We assigned differential annual mortality rates from non-stroke causes, as the model was of a lifetime horizon. The differential mortality rates were extracted from the 2017 United States Life Tables (most recent at time of analysis).15 The patient’s risk of mortality from non-stroke causes was adjusted for both functional status and the number of years since the stroke.16 17 The model was run until the entire cohort of patients died from stroke or non-stroke related causes. Patients who survived the stroke and treatment would have an additional annual risk of recurrent stroke, which would be highest in the first year post-stroke and decease over time (details in online supplemental). Patients with recurrent stroke would transition to the same or worse outcome state.

### Supplemental material

The complete tree diagram is presented in figure 1.

Figure 1

Structure of the decision tree. mRS, modified Rankin Scale.

### Clinical parameters

All clinical parameters were derived from the best available evidence in the literature with preference given to recently published randomized trials and large-cohort studies. The meta-analysis of data from HERMES Collaboration reported at 90 days, the proportion of good outcome (mRS score 0–2) in EVT as 58.2% (39/67) and mortality as 11.9% (8/67).5 This is similar to outcomes published in recent studies (a summary of recent meta-analyses on outcomes after EVT is presented in online supplemental table 1). Good outcomes after no-EVT/medical therapy alone was 39.7% (25/63) with mortality of 9.5% (6/63).5 The corrected values for the posterior distribution after a Bayesian analysis from a uniform prior probability distribution are provided in table 1. The risk of recurrent stroke is detailed in the online supplemental.

Table 1

List of all parameters. all probabilities are calculated on an annual basis

### Costs and outcomes

All cost parameters were assigned a gamma distribution with 10% SD to allow for variation. The cost of thrombectomy was derived from Kunz et al who extracted the cost from the National Inpatient Sample.18 The cost of acute and long-term costs associated with mRS scores of 0, 1, 2, 3, 4, 5, and 6 were derived from a study by Simpson et al and adjusted to January 2020 values.18 19 The annual follow-up costs based on functional states were also from Simpson et al, as it provided follow-up costs within 1 year after a stroke. We extrapolated the follow-up cost to the subsequent follow-up years by keeping the costs constant due to lack of specific data in the literature on this topic.

Annual earning loss was approximated by the median earnings in the USA in 2018 and adjusted to 2020 values.20 Different employment rates were assigned for different age groups, starting from 65 years old and ending in 80 years old, derived from the Bureau of Labor Statistics.21 The probability of returning to work in each health state was from a study by Tanaka et al.22 The indirect cost of each heath state was calculated by the earnings loss multiplied by both employment rate and (1 – probability of returning to work). Earning loss was also assigned to patients with premature mortality.

The utility for each outcome was extracted from a study by Chaisinanunkul et al which provided the corresponding QALYs for each mRS score. The utility for each outcome is averaged through the corresponding mRS score.23

An annual discount rate of 3% was assigned for both costs and utilities.

All parameters used in the model are listed in table 1.

### Statistical analysis

The two primary measures used in this study were incremental cost-effectiveness ratio (ICER) and net monetary benefit (NMB). The former is defined as

The ICER was assessed using the recommended willingness-to-pay (WTP) threshold in the USA of US$100 000/QALY.24 ICER was used in the base case calculation and probabilistic sensitivity analysis. The other indicator NMB was defined as , and it was used to compare strategies in the sensitivity analyses. Base case calculation was carried out using the mean value for each parameter. Probabilistic sensitivity analysis simulation was performed with 10 000 iterations, modeling 10 000 patients by sampling the input parameter values from their respective distributions. In addition, key variables, including outcomes and mortality after each intervention, cost of thrombectomy, annual cost of disability, and utility values assigned to people, were varied over a wide range in deterministic sensitivity analyses. Specifically, we attempted to perform a subgroup analysis for younger patients by changing the number of Markov cycles while keeping the clinical outcomes constant due to the lack of age-specific outcomes in the available literature. ### Validation Face and internal validation were performed to review problem formulation, data source, model structure, and results. A calibration validation was not applicable due to the lack of randomized control trial data on the utility of vascular imaging in minor patients with a stroke. ## Results ### Base case analysis In the short-term model (90 days), EVT had a higher cost (US$51 015) and higher effectiveness (0.16 QALY), whereas no-EVT/medical therapy had a lower cost (US$36 542) and lower effectiveness (0.14 QALY). The ICER was US$678 006/QALY, which is above the US$100 000/QALY WTP threshold, indicating that EVT was not cost-effective in the short term. The long-term model (lifetime) showed EVT to be the dominant strategy, with lower cost (US$551 159) and higher effectiveness (9.56 QALYs). No-EVT/medical therapy alone yielded effectiveness of 7.90 QALYs at a cost of US$577 181. The difference in effectiveness was 1.66 QALYs, equivalent to 606 days in perfect health or 681 days in mRS score 0–2 health state. In the subgroup analysis for 55-year-old patients presenting with acute M2 segment occlusions, EVT yielded 13.88 QALYs at a cost of US$955 332 and no-EVT/medical management yielded 11.55 QALYs at a cost of US$1 019 075. The subgroup analysis for 75-year-old patients showed that EVT yielded 5.72 QALYs at a cost of US$298 399 and medical management yielded 4.67 QALYs at a cost of US$302 281. ### Probabilistic sensitivity analysis Monte Carlo simulation with sampling of 10 000 showed EVT to be the more cost-effective strategy in 93.37% of iterations (figure 2A). EVT remained the optimal strategy regardless of the WTP threshold, with a higher threshold leading to a higher proportion of iterations where EVT was more favorable. The acceptability curve varying WTP from US$0 to US$200 000/QALY is presented in figure 2B. Figure 2 (A) Scatterplot of probabilistic sensitivity analysis. The reference strategy was medical management and all dots lying below the dotted willingness-to-pay threshold indicate iterations in which endovascular thrombectomy was more cost-effective. (B) Acceptability curve. CE, cost-effectiveness; QALY, quality-adjusted life year ### Sensitivity analysis We performed a sensitivity analysis for each of the variables and identified factors that had the largest impact on the incremental net monetary benefit. The tornado diagram is presented in online supplemental figure 1. The incremental net monetary benefit was most sensitive to the probability of a good outcome after EVT and no-EVT/medical therapy alone. The incremental net monetary benefit was also sensitive to the direct long-term medical costs and probabilities of mortality after each treatment. ### Supplemental material The good outcome of each strategy was varied independently. The result showed that when the probability of good outcome after EVT in patients with an M2 occlusion exceeded 44.9% (base case value 58.2%), it was the more cost-effective strategy (figure 3A). Varying the probability of good outcomes after no-EVT/medical management in patients with an M2 occlusion (figure 3B) indicated that the strategy became cost-effective after the probability of good outcome exceeded 56.5% (base case value 39.7%). Two-way sensitivity analysis varying probabilities of good outcomes after both treatments from 35% to 50% simultaneously (figure 3C) showed that EVT would be more cost-effective if it had better outcomes in 4–6% more patients than medical management in absolute numbers. Figure 3 (A) One-way sensitivity analyses varying good outcome after endovascular thrombectomy (EVT). A higher net monetary benefit is more desirable. (B) One-way sensitivity analyses varying good outcome after medical management. A higher net monetary benefit is more desirable. (C) Two-way sensitivity analyses varying good outcomes after EVT and medical management. The color in each region represents the area in which the corresponding strategy is more cost-effective. (D) Two-way sensitivity analyses varying differences in good outcomes and mortality between EVT and medical management. The color in each region represents the area in which the corresponding strategy is more cost-effective. Keeping the outcomes after medical management constant, we varied the difference between good outcomes of EVT in reference to medical management from −5% to 25% and the difference between mortality outcomes of EVT in reference to IVT from −5% to 10% in absolute numbers. The two-way sensitivity analysis (figure 3D) indicated that when the mortalities are equal in both strategies, the proportion of good outcomes after EVT needs to be at least 5.5% higher than that of medical management in absolute numbers for EVT to be cost-effective. The analysis also showed that EVT is the more cost-effective strategy throughout the difference in mortality range as long as the good outcome after EVT is at least 6.5% higher than that of IVT. We also performed a four-way sensitivity analysis of the absolute proportions of good outcomes and mortality risks after each strategy. The results are presented on the online interactive platform: (https://cea-radiology-app.shinyapps.io/m2_occlusions_cea/). We used effectiveness as the metric in two-way sensitivity analysis varying mortality risks due to the high cost of care associated with surviving patients, especially those with moderate to severe disability. We varied the mortality risks in each strategy from 0% to 40% in a two-way sensitivity analysis (online supplemental figure 2), which indicated that EVT yields better outcomes as long as the probability of mortality was at most 35% higher than that with medical management. ### Supplemental material When the cost of thrombectomy was varied (online supplemental figure 3), medical management became the more cost-effective strategy with a procedural cost of thrombectomy upwards of US$209 111, a value much higher than the baseline value of US$16 783. ### Supplemental material When varying the direct long-term costs associated with the mRS scores (online supplemental figure 4), the results showed that as the costs increased, the gap in cost-effectiveness between the two strategies widened in favor of EVT. ### Supplemental material One-way sensitivity analyses varying the utility values for mRS scores 0 through 4 from 0.91 to 1.0 QALYs, 0.76–0.91 QALYs, 0.65–0.76 QALYs, 0.33–0.65 QALYs, and 0–0.33 QALYs, respectively, showed no impact on the model results. ## Discussion Our study results suggest that EVT is a cost-effective strategy in the long-term (lifetime) compared with medical management in patients with a stroke with acute M2 occlusion. Despite higher initial costs, EVT is cost-effective in the long term, yielding lower costs and better health outcomes than medical management. EVT results in a difference in health benefit of 1.66 QALYs, which corresponds to 606 days in perfect health or 681 days in mRS score 0–2 health state. A recent meta-analysis reported the rate of good outcomes (mRS score 0–2) after EVT for acute M2 occlusions ranging between 40.7% and 80.5%.25 A summary of multiple recent meta-analyses is provided in online supplemental table 1. Our sensitivity analysis shows that EVT is cost-effective when the probability of good outcomes after EVT in patients with an M2 occlusion exceeds 44.9% (base case value in meta-analysis was 58.2%). Varying the outcomes after both strategies shows that EVT is more cost-effective when the probability of good outcomes after EVT is 4–6% higher than for medical management in clinically likely scenarios (possible range of good outcomes after no-EVT/medical management being 35–50%). However, the reported difference in outcomes in the literature is greater than 18%.26 Although thrombectomy is associated with higher initial costs, our analysis suggests that EVT is cost-effective in the long-term unless thrombectomy procedural costs exceed US$209 000. The cost-effectiveness of EVT is even greater in younger patients, with 55-year-old patients having an ICER of −US$27 320/QALY compared with −US$15 647/QALY for 65-year-old patients.

Acute M2 occlusions may be seen in 4% of all patients discharged with acute ischemic stroke, and may be the second most common occlusion site (after M1).4 Rai et al using the 2016 US population reported the rate of M2 occlusions was 7/100 000 people/year, yielding an annual incidence of 21 176 M2 strokes nationally.4 Although admission National Institutes of Health Stroke Scale (NIHSS) scores are lower in M2 segment occlusions, discharge NIHSS score may not be significantly lower.27 Our analysis shows that extending EVT to patients with acute stroke with M2 occlusions is expected to result in more than US\$551 million in annual cost savings in comparison with no-EVT/medical management.

The reported natural history of untreated M2 occlusions is variable. Lima et al reported good outcomes in 54.2%, but with mortality rate of 20.8% in untreated patients with acute M2 occlusions, with stroke severity and collateral flow being significant determinants of outcome.28 Rai et al reported good outcomes after medical management in 43% and mortality in 27%, with significantly higher mortality for NIHSS score ≥9 (42%) than for NIHSS score <9 (2.7%).4 The heterogeneity in reported results after EVT in M2 occlusions is probably due to patient selection. There is probably over-representation of patients with larger, more disabling M2 occlusions in studies reporting EVT outcomes.6 We have provided a mechanism for four-way sensitivity analysis of the absolute proportion of good outcomes and mortality risks after each strategy in an online interactive platform to enable assessment of the results after outcomes in different settings (https://cea-radiology-app.shinyapps.io/m2_occlusions_cea/).

MCA branches show wide anatomic variation, which may influence the arterial territory at risk in cases of occlusion. However, Saber et al found that the definition of M2 segment did not alter efficacy or safety outcome rates after EVT.6 Dominant M2 (supplying more than 50% of MCA territory) occlusions lead to worse clinical outcomes than non-dominant M2 occlusions.29 Treatment effect favoring EVT was maximal in patients with proximal and dominant M2 segment occlusions in HERMES.5 However, it has been shown that reperfusion is associated with better outcomes in both groups. Patients with either a dominant or a non-dominant M2 occlusion should not routinely be excluded from EVT.30 Occlusion of the superior division may be an independent predictor of poor outcomes as it supplies the central and precentral areas.31 32 However, no relevant differences from a technical perspective have been reported in superior versus inferior division occlusions on EVT.31 A recent meta-analysis also found no statistically significant difference in time from last known normal to groin puncture or time to recanalization for M2 compared with M1 occlusions.25

Improvements in technique and devices have led to high recanalization rates. Saber et al reported 87% recanalization with stent retriever thrombectomy (SRT) first and 80% with aspiration first, with functional independence in 57.6% with SRT first and 48.6% with aspiration-first.6 On the other hand, Phan et al reported recanalization in 80.5% for SRT compared with 86.8% with aspiration-first technique, with good outcomes in 59.9% with SRT and 74.5% with a direct aspiration first pass technique.33 Smaller diameter next-generation stent retrievers and aspiration catheters further improve reperfusion rates, and result in better clinical outcomes in these patients.34

Lack of clear guideline-based recommendations lead to significant variation in EVT decision-making in acute M2 occlusions.7 In an international, multidisciplinary survey in 2019 of 607 physicians, 65.4% respondents favored EVT in acute M2 occlusions under current local resources and 69.6% under assumed ideal conditions.7 Insufficient access to treatment facilities, monetary and health-policy related limitations are possible causes for low EVT decision rates in some regions. Our analysis shows that EVT is cost-effective from the long-term, societal perspective given the better outcomes, despite the initial higher cost. EVT was more likely to be favored in the survey for older patients, for those with longer time since symptom onset, higher EVT volumes, and by neurosurgeons.7 A sensitivity analysis varying the age of the patient showed that EVT is even more cost-effective in younger patients.

Limitations of this study include the paucity of age-specific outcomes after thrombectomy in patients with acute M2 occlusion in the literature. We performed subgroup analysis for younger patients by changing the number of Markov cycles and age-specific mortality and employment rates, while keeping the clinical outcomes constant, which showed EVT to be cost-effective. Current literature evidence shows that EVT has better outcomes in younger patients, and would therefore result in EVT being even more cost-effective in younger patients than demonstrated in our analysis. There is a lack of literature on the outcomes after recurrent stroke based on pre-existing mRS state. In our model, we assumed a patient would progress to an equivalent or worse functional state. The cost of thrombectomy used in the base case analysis was not specific to patients with acute M2 occlusion stroke. We therefore varied the cost of thrombectomy in the sensitivity analysis to assess its impact on the model results. Similarly, the indirect costs due to the loss of productivity may be different from those for patients with more proximal emergent large vessel occlusion strokes. However, we could not find specific literature on this topic relating to M2 occlusion strokes. Also, the probability of return to work is based on a Japanese study. This has been used in previous cost analysis also as US-specific data are not available.35 36 To compensate, we varied the utility assigned to good and poor outcomes, and the indirect costs in sensitivity analyses. The indirect costs are probably also underestimated as all the cost inputs recommended for inclusion in indirect costs by the Second Panel on Cost Effectiveness analyses are not available specifically for patients with a stroke.

## Conclusion

Our study suggests that EVT is a cost-effective approach in patients with acute ischemic stroke with M2 occlusion in the long term (lifetime horizon), considering the higher reperfusion rates and better clinical outcomes with EVT relative to medical management, with robust results in sensitivity analyses. Faster and improved reperfusion techniques would increase the relative benefit of EVT even further in these patients.

## Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

## Ethics statements

### Ethics approval

Ethics approval from the local institutional review board was not required since no actual patients are involved in the study and input parameters were derived from published literature.

• ## Supplementary Data

This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

## Footnotes

• Twitter @AjayMalhotraRad

• Contributors AM: Study concept and design, acquisition of data, analysis and interpretation, critical revision of the manuscript for important intellectual content, study supervision. MK, XW: Acquisition of data, analysis and interpretation, critical revision of the manuscript for important intellectual content. SP: Acquisition of data, critical revision of the manuscript for important intellectual content. CZ, CM, JS, PS, DG: Acquisition of data, critical revision of the manuscript for important intellectual content, study supervision.

• Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

• Competing interests None declared.

• Provenance and peer review Not commissioned; externally peer reviewed.

• Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.