Comparative Effectiveness of Endovascular Thrombectomy in Elderly Stroke Patients

Introduction

Strokes in patients aged ≥ 80 years are common and account for about one-third of all strokes in developed societies, with 17% occurring in patients over age 85 years.^1,2 The proportion of elderly individuals (older than 80 years) is expected to at least double in the United States by the year 2050 and more than double in the European Union by 2080.^3,4 The number of incident strokes is also expected to more than double, with the majority of the increase among the elderly.⁵ Stroke remains a leading cause of serious long-term disability, with a greater disability reported in women.^6,7 Advanced age is particularly associated with relatively poor poststroke functional outcomes.⁸ A good outcome (modified Rankin Scale [mRS] score of 0–1) is only reported in 13.2% of elderly patients without intravenous tPA (tissue-type plasminogen activator; intravenous thrombolysis [IVT]) or endovascular thrombectomy (EVT).⁹ Several recent trials have demonstrated efficacy of EVT to improve outcomes in acute stroke patients with large vessel occlusion (LVO).¹⁰ However, the effectiveness of EVT in patients >80 years has not been established. Fifty percent (3 of the 6) major trials evaluating EVT for acute stroke excluded patients over 80 years of age.^11–13 EVT trials that included octogenarians showed better outcomes when treated with EVT compared to IVT alone, with more favorable odds ratios than among younger patients.^14,15 However, the number of patients in these trials above 80 years was small, raising concern for potential selection bias and inclusion of patients with greater chances of recovery.¹⁶ In addition, these studies only included premorbid functionally independent patients (baseline mRS score of ≤2). Other studies have raised concerns about the utility of EVT in the elderly because of a lower likelihood of functional independence at 90 days compared with younger patients.¹⁷ The 2018 American Heart Association/American Stroke Association guidelines do not recommend an upper age limit for EVT and recommend consideration of comorbidities into decision making.¹⁸

A recent meta-analysis assessing EVT in octogenarians concluded that age should not be a discriminator when offering EVT for patients with acute stroke.¹⁶ The pooled literature reported successful recanalization in 78% of patients (95% CI, 72%–85%) with good functional outcome in 27% at 3 months.¹⁶ Although procedure-related complications (11%, 95% CI, 4%–17%) and mortality at 3 months (34%, 95% CI, 23%–44%) are high in this population, this has to be considered in the background of generally poor outcomes in this subpopulation after IVT and standard care alone.¹⁶ The meta-analysis reported a significant difference for good functional outcomes between studies that only included premorbid independent patients, versus those that did not have this prerequisite.¹⁶

Previous cost-effectiveness analyses of EVT have not addressed patient age as a factor or found limited effectiveness in patients older than age 80.^19,20 A recent study based on results from HERMES trials (Highly Effective Reperfusion Evaluated in Multiple Endovascular Stroke) concluded that there is no reason to with-hold EVT because of patient age, neither from a medical nor from an economic or social perspective.²¹ However, this was based on a small sample, with selection bias, and the analysis did not assess the impact of varying the morbidity and mortality after IVT and EVT, which would be expected to impact the conclusions. Therefore, the purpose of our study was to conduct a decision-analytical study to assess the impact of these factors on the effectiveness of EVT and health outcomes in octogenarians.

Methods

The authors declare that all supporting data are available within the article and its in the online-only Data Supplement. Ethics approval from the local institutional review board was not required because no actual patients are involved in the study and input parameters were derived from published literature.

A decision analysis model was constructed using TreeAge Pro Suite 2018 (Cambridge, MA) over the lifetime span of patients. 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 elderly patients (>80 years) with a confirmed LVO stroke managed by tPA or tPA followed by EVT, and each patient was assigned a clinical care pathway from treatment options to final outcome by sampling the distributions of each input parameter in the model. We then compute the respective expected outcomes in terms of quality-adjusted life years (QALYs), which is a comprehensive utility metric accounting for both the expectancy and quality of life for patients in a specified health state.

Model Structure

The starting point of this model is a patient of age 80 presenting with acute ischemic stroke secondary to LVO and within the time frame eligible for treatment with IVT or IVT immediately followed by EVT. After either treatment strategy, the patients are categorized as 3 main health states based on the mRS: good outcome (mRS score of 0–2), poor outcome (mRS score of 3–5), or death (mRS score of 6). The utility for each outcome was extracted from a study by Chaisinanunkul et al,²² and the utility for each outcome is averaged through the corresponding mRS of the included mRS scores. We assigned differential annual mortality rates from stroke causes, as the model was of a lifetime horizon. The differential mortality rates were computed from the 2014 US Life Tables (most recent at time of analysis—the 2015 Life Tables would not be available until November 13, 2018). Patients with moderate to severe disability would have a 17% excess mortality.²³ The model was run until the entire cohort of the patients died from stroke or nonstroke related causes.

Patients who survived the stroke and treatment would have an additional annual risk of recurrent stroke and death from other causes. Patients remaining alive after recurrent stroke events were reallocated to greater disability patients in the mRS score of 0–2 group would progress to the mRS score of 3–5 group, and the mRS score of 3–5 group patients would progress to death.²⁴ For patients with baseline disability, it is unlikely that they will have improved outcome compared with their baseline function and possible outcomes incorporated in the model are persistent disability (mRS score of 3–5) or death.

The complete tree diagram is presented in Figure 1 and Figure I in the online-only Data Supplement.

• Download figure
• Download PowerPoint

Clinical Parameters

All clinical parameters were derived from recently published large-cohort studies or meta-analyses. A recent meta-analysis specific to elderly patients (>80 years) reported percentage of good functional outcome after EVT to be 27% and mortality rate of 34%.¹⁶ We used the results from the study by Purroy et al,²⁵ where good outcome (mRS score of 0–2) after IVT in different hospital settings was 38.7%, and the mortality was 23.2%. However, this was not confined to patients with LVO. There is a lack of specific literature on outcomes after IVT in elderly patients with LVO. A previous meta-analysis by Emberson et al⁹ reported good outcome after IVT of 17.6%, and the mortality was 17.9% in patients >80 years but defined good outcome as mRS score of 0–1. For patients with baseline disability, we assume that the mortality risks after each intervention are the same as in the general population in the base case calculation.

The risk of recurrent stroke is reported to be higher shortly after the initial stroke and decreases over time.²⁶ In the study by Pennlert et al,²⁷ the recurrent stroke risk is 6% in the first year, 16% in the first 5 years, and 25% in the first 10 years. In the model, we assigned differential risk of stroke recurrence in the first, second, to fifth year and sixth to tenth year.

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

Statistical Analysis

Base case calculation was performed using the most probable value of each parameter. Probabilistic sensitivity analysis was performed to test the robustness of the model by allowing simultaneous alteration of multiple model input parameters. We performed separate subgroup analysis for elderly patients who are functionally independent at baseline (mRS score of 0–2) versus those with baseline disability (mRS score of 3–5) as starting health states. Various 1-way sensitivity analyses were undertaken to assess the robustness of the conclusion against key variables identified.

Results

The base calculation showed that IVT only is the better strategy as compared to EVT, with a 3.76 QALYs for patients receiving IVT only and 2.93 QALYs for patients undergoing EVT. The difference in outcome is 0.83 QALYs, which is equivalent to 303 days of life in perfect health, 340 days of life in mRS score of 1–2, or 918 days of life in mRS score of 3–5. For patients with baseline disability, IVT only yields a utility of 1.92 QALYs and EVT yields a utility of 1.65 QALYs. The difference is 0.27 QALYs which is equivalent to 99 days of life in perfect health, 111 days of life in mRS score of 1–2 or 299 days of life in mRS score of 3–5 (Figure II in the online-only Data Supplement).

The results of probabilistic sensitivity analysis are presented in Table 2. As shown by the results, the worst outcome (minimum) in the IVT strategy is 3.43 QALY which is higher than the median utility of patients undergoing EVT (2.93 QALY). Patients undergoing EVT also have a wider range of possible outcomes, with the best and worst outcome differing by 1.54 QALYs. For patients with baseline disability, the conclusion is similar.

Sensitivity Analysis

Patients in Good Health at Baseline

The functional outcomes (mRS scores) after IVT and EVT strategies are the main drivers of the model results. Thus, we varied each of them over a wide range individually in the sensitivity analyses to further understand the model conclusions.

For outcomes after EVT, we varied the percentage of a good outcome (mRS score of 0–2) from 0% to 50%, and the model results show that when the likelihood of achieving a good outcome is >44.4%, EVT becomes more favorable than IVT (Figure 2A). When the risk of mortality after EVT is varied from 0% to 50%, IVT remains the better strategy when the mortality risk of EVT is above 0.96% after EVT (Figure 2B). A 2-way sensitivity analysis was performed to vary the proportion of good outcome and mortality risk of EVT. The result shows that as the good outcome after EVT increases, the threshold of EVT mortality above which IVT will be favored also increases. When the proportion of good outcome is 30%, the threshold of mortality is 8.5%. When the proportion of good outcome is 40%, the threshold becomes 25% (Figure 3A; Figure IIIA in the online-only Data Supplement). However, IVT becomes the better strategy when the likelihood of good outcome exceeds 21.3% after IVT and remains the better strategy when the risk of mortality from IVT is <56.3% (Figure 2C and 2D). A 2-way sensitivity analysis performed by varying the proportion of good outcome and mortality risk of IVT shows that as the proportion of good outcome after IVT increases, the threshold above which EVT will be favored also increases (Figure 3B; Figure IIIB). When the proportion of good outcome after IVT is >35%, IVT remains the optimal strategy even when its mortality is up to 50%.

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Because of the scant literature on the risk of recurrent stroke, we also varied the risk of recurrence in the first year, between the second and fifth year, as well as between sixth to tenth years. Because the risk of recurrent stroke impacts both IVT and EVT strategies to the same extent, the conclusion of the model remains unchanged from the base case scenario.

Patients With Disability at Baseline

There is a lack of specific literature on outcomes after IVT and EVT in elderly patients with disability (mRS score of 3–5) at baseline. Because these patients cannot improve from their baseline disability before stroke, they are likely to remain in the same health state (mRS score of 3–5) or die (mRS score of 6) from any stroke treatment. Therefore, we focused on varying the mortality rate in patients with baseline disability (mRS score of >2) to assess the impact of intervention in preserving their baseline disability (mRS score of 3–5). When the mortality risk associated with EVT is varied from 0% to 50%, EVT is the better strategy when the mortality risk after EVT is lower than 23.2%. When it is higher, IVT alone is better for this group of patients (Figure 4A). Similarly, when the mortality associated with IVT is varied from 0% to 50% in the sensitivity analysis, the model results show that IVT alone is the better strategy when mortality remains <34.0%. When the mortality risk exceeds 34.0%, EVT replaces IVT to be the optimal strategy (Figure 4B). A 2-way sensitivity analysis varying both mortality risks of IVT alone and EVT showed that the model results are highly sensitivity to the mortality risks from either IVT or EVT strategies. The strategy with a higher mortality risk will yield a worse utility in the model (Figure 3C; Figure IIIC in the online-only Data Supplement).

• Download figure
• Download PowerPoint

Discussion

Our study investigated the impact of relevant factors on the effectiveness of EVT in patients above 80 years of age. The study results do not support expanding use of EVT in all patients >80 years with LVO and acute stroke. The effectiveness of EVT is significantly determined by the morbidity and mortality after both IVT and EVT strategies, respectively.

The previous analysis by Kunz et al²¹ advocated for greater use of EVT based on the morbidity/mortality on a small sample from studies included in HERMES. A total of 99 patients were included each in EVT and control arms, and all patients included were functionally independent (mRS score of ≤2) at baseline.¹⁴ In the medical arm/IVT group, 13.9% elderly patients had a good outcome (mRS score of 0–2) and there was 45.2% mortality, compared with 29.8% good outcome and 28% mortality at 3 months after EVT.¹⁴ However, outcomes after IVT alone in the reported literature are better than those reported in HERMES. The ITA-90+ Collaborative, which had >80% patients with large artery infarcts, showed >27% good outcome (mRS score of 0–2) after IVT in nonagenarians (patients >90 years).²⁸ Based on the meta-analysis by Emberson et al,⁹ the rate of mortality after IVT was 17.9% in patients >80 years. In a recent real-world study, Alawieh et al¹⁷ did not find a significant improvement in rates of good outcome after EVT compared with medical management (20.5% versus 19.5%, respectively; P>0.05) in elderly patients.

Our study results indicate that for patients >80 years who are functionally independent at baseline, IVT may be the more optimal strategy when likelihood of good outcome after IVT is >21.3% or mortality is <56.3%. EVT becomes the better strategy only when good outcome can be achieved in 44.4% with EVT. Previous studies which only included patients with baseline mRS score of ≤2 (and thus were likely to have a better outcome after thrombectomy) showed good outcome in only 31.7% (95% CI, 18.4%–41.1%).¹⁶ Our study also shows that patients undergoing EVT also have a wider range of possible outcomes.

As many as 33% of octogenarians presenting with ischemic stroke are not independent at baseline.²⁹ Although higher 90-day mortality rate (38.5%) has been reported after EVT in patients with moderate baseline disability, good outcome has been previously reported in 23.9% of patients when the definition was expanded to include return to premorbid functional status.³⁰ Our model results indicate that for patients in whom return to the baseline status can be achieved, EVT may be the better strategy when mortality after IVT is >34% or after EVT is <23.2%.

It is vital to note that the over 80 years age-group is not homogeneous and outcomes change with advancing age. The STRATIS registry (Systematic Evaluation of Patients Treated With Neurothrombectomy Devices for Acute Ischemic Stroke; real-world setting of 984 patients treated at 55 sites) reported good outcome in 43.2% of patients ≥80 years without baseline disability.³¹ However, good outcomes decreased significantly with advancing age, with rates of 52.4%, 38.3%, and 26.5% for age groups 80 to 84, 85 to 90, and 90+, respectively. Mortality at 90 days also increased and was 24.5%, 27.9%, and 35.1% for the respective age groups. It should also be noted that 76% (748/984) of patients had a prestroke mRS score of 0, indicating high selection bias even in STRATIS registry in favor of doing EVT in healthier patients.

There are limitations to this study that need to be considered when interpreting these results. The scarcity of literature on specific outcomes after IVT in patients >80 years with an LVO has been detailed above. Zhu et al³² found LVO to be more frequent in patients with an unfavorable outcome compared with favorable outcome in older stroke patients but did not report specific outcomes after EVT with and without IVT. We performed wide sensitivity analyses to study impact of outcomes after IVT on the conclusions. There is also a lack of literature on outcomes after EVT in elderly patients with LVO and baseline disability. We varied the mortality rates after EVT to assess the effectiveness of EVT in returning these patients to their baseline disability (mRS score of 3–5). An mRS score of 3 may convey a relatively good quality of life in a patient above age 80 years, and the potential shift effect of EVT towards a lower mRS has been advocated as a favorable outcome after EVT.¹⁶ However, absence of specific data in the literature prevents mRS shift-analysis, and this may underestimate slightly the utility of EVT.

No significant difference in functional outcome between EVT and IVT has been reported for patients with LVO and mild stroke (National Institutes of Health Stroke Scale score of ≤5) at presentation in the overall population.³³ EVT may have greater benefit in patients with higher stroke severity at presentation, as also shown by the pooled HERMES data although this was not specific to the older population.¹⁴ The thrombectomy trials (which have been used to advocate more widespread use of EVT in older patients) have a moderate risk of bias across studies.³⁴ Patients included had a baseline National Institutes of Health Stroke Scale score of 17 (SD=14–20) and Alberta Stroke Program Early CT Score on baseline of 9 (SD=7–10).¹⁴ The ESCAPE (Endovascular Treatment for Small Core and Anterior Circulation Proximal Occlusion With Emphasis on Minimizing CT to Recanalization Times) and EXTEND IA (Extending the Time for Thrombolysis in Emergency Neurological Deficits — Intra-Arterial) trials (which contributed half of the total patients over 80 years) had eligibility criteria of moderate to good collateral circulation or perfusion mismatch and small infarct core. It is likely that these specific subgroups of patients who are functionally independent at baseline, have at least moderate stroke severity at presentation, have a small infarct core and good collateral flow, may benefit from thrombectomy. However, these trial results cannot be used to justify widespread use of EVT in all patients above 80 years. EVT may have better effectiveness in selected patients as those included in the HERMES study, or if patients are likely not to have a good outcome after IVT alone. For example, patients with high clot burden (thrombus length >8 mm) have been shown to not recanalize with IVT alone.³⁵

Discrepancy between early neurological course and mid-term outcome in older stroke patients has also been reported after mechanical EVT.³⁶ Despite similar improvements in postinterventional National Institutes of Health Stroke Scale scores of older and younger patients, high mortality rates and less frequent good functional outcome have been noted in older patients despite successful recanalization and substantial early neurological recovery.³⁶ Multiple factors have been identified that may contribute to poor functional outcome of older stroke patients including preexisting neuronal loss and reduced neuronal plasticity, leukoaraiosis, higher rate of comorbidities and poststroke complications (eg, infections and injuries due to falls).^37–40 The mid- to long-term poorer outcome of older stroke patients may be related to these factors and not directly linked to brain damage induced by incident stroke.³⁶ The overall effectiveness of EVT may be better if poststroke care can be improved and long-term outcomes more closely resemble the post-EVT early recovery period.

In conclusion, our study results demonstrate the factors influencing effectiveness of EVT in healthy as well as baseline disabled patients above 80 years of age. Importantly, the comparative effectiveness of EVT depends significantly on mortality and the likelihood of good outcomes with both IVT and EVT strategies. Better identification of patients not benefiting from IVT would optimize the selective use of EVT thereby improving its effectiveness.

Footnotes