Thread by @TotoMynell

1/ Endovascular stroke treatment (EVT) is one of the biggest success stories of medicine!
In clinical reality, the question is no longer “Who to treat?” but: “Who should we NOT treat???”

A case-based tweetorial about futile recanalization …
#NeuroTwitter #SharedDecisionMaking

2/ 🧠💔Futile Recanalization (FR) 🧠💔 usually refers to very poor outcome (modified Rankin Scale 5–6) despite a technically successful intervention. It is a bit the elephant in the room in the modern thrombectomy era.

3/ 86yo woman presents with acute severe right-sided hemiplegia and aphasia, NIHSS 14, pre-stroke mRS 1, BP 178/95, glucose 9.5mmol/L, ASPECTS 7, left M1-occlusion, poor collaterals, time since onset 320 minutes.

4/ What do you think is her risk of mRS 5/6 with and without EVT?

5/ Most likely D. Using the helpful MR PREDICTS tool for Endovascular Treatment in Acute Ischemic Stroke, her chances for FR are around 50% with EVT or even higher depending on whether you chose RCT data or observational data as the basis.
mdmtest.shinyapps.io/RRRR_2/

6/ What do you think is the percentage of EVT patients suffering 🧠💔FR 🧠💔overall ?

7/ Whereas in the randomized controlled trials with strict in/exclusion criteria, only about 1/5 will suffer FR, in real-world data the % is higher at around 1/3 because of more liberal indications.
pubmed.ncbi.nlm.nih.gov/26898852/
bmj.com/content/360/bm…

8/ In basilar artery occlusion, it is even higher at around 50% given the devastating nature of this disease.
ahajournals.org/doi/10.1161/ST…
jnis.bmj.com/content/11/12/…

9/ In several countries, access to MT and qualified personnel is limited, but strain and workload are rising. Reliable FR prediction could accommodate the rising demand and increasing healthcare costs.

10/ We and others identified several predictors, of which the most important ones seem to be clinical (stroke severity, older age, active cancer, prestroke disability), laboratory (higher glucose, higher CRP, higher creatinine), imaging (WMH), and time from onset-to-treatment.

11/ The relative importance of those features helps stroke physicians to rapidly clarify the most relevant items emergency setting
n.neurology.org/content/99/10/…

12/ The problem with those algorithms: you wanna be REALLY sure, you are not withholding a potentially life- and disability saving treatment. Therefore, although the AUC (0.86) and specificity (0.93) are nice, they are USELESS in most contexts! Why?

13/ You should ask yourself: at what threshold of probability for FR would I refrain from treatment? In our high-ressource setting @StrokeBern 🇨🇭🫕, this level is likely to be >95% and there, the model has 0⃣ added benefit.

14/ In low income countries where access to MT is limited by shortage of MT devices, staff, or other hurdles, you might find yourself in the unfortunate setting, where you would refrain from treatment at much lower thresholds, let’s say 50%, the model might be more helpful...

15/ With the positive large-core trials it has become clear, that no single biomarker is helpful to identify patients that should not have EVT. However, RCTs might wanna address, whether the cost-effectiveness is preserved in patients with multiple bad prognostic factors.

16/ IMPORTANT limitations
- mRS 5 at 90 days is certainly suboptimal, delayed recovery might happen! #NeverLoseHope
- true "clinical utility” needs weighing for treatment effect, costs, and other factors
- Validation, refinement, & inclusion of advanced imaging features necessary

17/ Did we treat this patient? Of course, we did, but the presence of several poor prognostic features allowed us to incorporate them in shared decision making and help to set realistic expectations. Unfortunately this patient died on the stroke unit despite EVT.

18/ Take home 1🏡
- FR is frequent and NOT a failure of your team, it can even be an opportunity
- FR poses enormous logistic, economic, & ethical challenges
- The most important predictors are stroke severity, age, prestroke status, time to treatment and extend of brain injury

18/ Take home 2🏡
- FR models SHOULD NOT be used to exclude patients from EVT: IN DUBIO PRO EVT
- FR models can be helpful for shared decision making and for setting realistic expectations

19/ Other thoughts? Learn more about outcome prediction in stroke and FR ahajournals.org/doi/10.1161/ST…
n.neurology.org/content/99/10/…

@CheesemakerMD @TDobrocky @johanna_ospel @DavidSeiffge @MicieliA_MD @adnan_mujanovic @caseyalbin @rkchoi @Diana_A_Sousa @StrokeBern @gboulouis

@Dr_No_Charite @Eikeip @FishingNeurons @Fie0815 @mbgoeldlin @uta_hanning @Jan_FriSch @manonkappelhof @kiliantreurniet @NguyenThanhMD @johanna_ospel @marco_pasi85 @a_charidimou @mckinley_scan @Ellis_van_Etten @SuseWegener @ScanNeuroradBE