Is Endovascular Thrombectomy for the Very Elderly?

Claire J. Creutzfeldt, MD; Michael R. Levitt, MD; Thabele M. Leslie-Mazwi, MD


Stroke. 2022;53(7):2227-2229. 

Endovascular thrombectomy (EVT) is now established as the most effective treatment for a select group of patients with acute ischemic stroke and large vessel occlusion. Many subsequent research efforts have focused on expanding the eligibility criteria—from a time window of 6 hours after stroke onset to up to 24 hours,[1,2] and from a minimum ASPECTS (Alberta Stroke Program Early CT Score) of 6 to possibly as low as 2.[3]

Stroke is a disease that carries increasing risk with increasing age. The role of EVT in the elderly or very old has remained a point of debate. Although most of the landmark thrombectomy trials excluded patients who were older than 80 years, the HERMES collaboration's (Highly Effective Reperfusion Evaluated in Multiple Endovascular Stroke Trials) per-patient analysis suggests not withholding EVT on the basis of age alone.[4] The most recent guidelines from the American Stroke Association/American Heart Association[5] admit that the number of patients >90 years in the thrombectomy trials was very small, and a recent international survey of clinicians suggests that age still plays a key role when making individual thrombectomy decisions.[6]

The recent article in Stroke, "Functional Outcomes of Patients 85 Years or Older With Acute Ischemic Stroke Following EVT: A HERMES Substudy," provides further support to those of us who offer thrombectomy to eligible patients regardless of age.[7] Using data from the HERMES collaboration trials that included a total of 77 patients older than 85 years (out of 1764 patients total), the authors analyzed the effect of age on the relationship between thrombectomy and outcome. Overall, compared with patients <85 years of age, those ≥85 years suffered worse 90-day functional outcomes (as defined by modified Rankin Scale score) and higher mortality. A mere 18% of patients ≥85 years achieved an eventual modified Rankin Scale score of 0–2, compared with 40% of those <85 years. However, the beneficial effect of EVT on 90-day outcome was still seen in the older age group, with a common odds ratio of 4.20 (95% CI, 1.56–11.32). The procedure was lifesaving; elderly patients undergoing thrombectomy had a mortality of 31%, compared with 54% in the control group. Advanced age further did not decrease the likelihood of procedural success or increase the risk of the procedure (as measured by symptomatic intracranial hemorrhage).

A few recent real-world observational studies paint a less optimistic picture for thrombectomy of patients ≥85 years,[8–10] likely because the cohort of elderly patients in the thrombectomy trials was not representative of real-world patients. After all, the ASPECTS in the thrombectomy trials was higher in the elderly compared with the younger group,[7] suggesting that elderly patients were preselected based on their likelihood of succeeding (selection bias). In addition, patients with prestroke disability (modified Rankin Scale score >1) were excluded from all of the HERMES trials, although this subgroup represents 25% to 50% of elderly patients in real-world studies.[8,10,11] The focus on independence as the primary outcome is limiting, especially for people who might have been dependent before the stroke, and for whom the return to prior function should be considered a good outcome.

Why do large, randomized, controlled trials not fully represent the patients we encounter in the real world? Because it is complicated: in addition to premorbid disability that would confound the primary (functional) outcome of many thrombectomy trials, older patients are also more likely to have preexisting comorbidities and suffer poststroke complications, increasing the risk of death or poor functional outcome. They are also more likely to have advance directives documenting a preference towards less aggressive medical care than younger patients, and possibly a lower tolerance for new disability.[8]

These observations do not mean that we should not or would not offer a potentially lifesaving and function-preserving treatment to those elderly patients who were not represented in the large, randomized trials that we quote to them. But they should at least caution us in our approach to offering a resource-intense treatment to everyone and support the recommendation that timely access to high quality patient and family centered palliative care is a core component of stroke care.[12] As we engage in a shared decision-making process with a patient or (much more often) their decision-maker, we often have only one chance to get a life-changing conversation right; this conversation is often hurried, over the phone, somewhere between the emergency department and the angiography suite. Further analyses of existing trial data will not remove the uncertainties. Instead, future research efforts should:

  1. Include evaluation of preference-sensitive outcomes (eg, a patient's likelihood of returning to their previous function or quality of life or their chance of still being able to do what is most important to the individual).

  2. Introduce, at an early stage in the consent or post-procedure setting, the concept of a time-limited trial[13] after aggressive stroke treatment, that will prepare the decision-maker for additional goals-of-care discussions when a clearer picture emerges of the trajectory of stroke recovery.

  3. Study the effect of a standardized approach to the informed consent process that considers the range of likely outcomes as well as a patient's acceptable burden of treatment.

  4. Study the effect of a communication intervention for elderly patients that includes more detailed advance care planning and provides a preamble to potential future emergent care conversations.[8]

By meeting patients where they are (including prestroke disability) and attempting to preserve baseline functional status, EVT stands a better chance of servicing patients throughout the age spectrum compared with not offering it. There is more individual variation in the elderly than at any other time point across the human lifespan. A single approach based on numerical age will never effectively address this variation and will leave patients behind who could otherwise recover, albeit with lower success rates than younger patients. Improved trial design and posttrial real-world studies will eventually identify risk factors that aid in patient selection within the elderly population. Until then, as this analysis supports, we should all be resistant to the approach of withholding treatment based solely on numerical age.