International Stroke Conference
February 6–8, 2019

Robert W. Regenhardt, MD, PhD

Moderators: Carmelo Graffagnino, Edward Jauch

The first talk, “Challenges in Patient Access,” by Carmelo Graffagnino discussed the system problems we face in the thrombectomy era. There are many system models that can be used to organize a region for the care of acute stroke patients. The models that worked well in the past allowed rapid access to tPA, but as thrombectomy expertise is a scarcer resource, there are new challenges. Graffagnino described two cases: one that showed how the system can work perfectly and another that highlighted limitations as a patient experienced severe delay in transferring to the thrombectomy capable center. Some of these limitations are difficult to predict and even more difficult to fix from the perspective of physicians, such as lack of helicopter and ambulance availability.

He highlighted that the system in the USA is highly fragmented, and the pieces are difficult to coordinate. The 2018 AHA guidelines recommend that regional systems be developed but offer no specific details about how to organize such systems. Graffagnino stated that we must first think about who we are targeting for treatment; our system changes should extend all the way to the patients themselves. He described his approach to identify and implement interventions for high-risk patients. Through community engagement, they are working with barber shops and churches to increase education about the signs of stroke. They have further implemented a “pay it forward” model in which an individual is encouraged to teach 3 people about stroke and encourage them to each teach another 3 people and so forth.

He further discussed the role of developing a “manual of operations”. Likening this to an NFL playbook, Graffagnino felt there should be a general playbook with broad guidelines applicable to all regional systems. However, each system is different; the Patriots’ playbook may work better for them than the Rams’. Both playbooks need to be developed and are essential. His plan is to make a “living manual of operations” that is freely available to everyone; it categorizes evidence and is modified in real time as new data becomes available. As we establish referral networks, we need a “central air traffic control” that can collect data to improve each regional system. Process improvement may be implemented through real time feedback in the form of progress reports to comprehensive stroke centers (CSCs), primary stroke centers (PSCs), and EMS agencies.

The second talk, by Dorothea Altschul, was titled “The Closest Local Center is a Must: Drip and Ship, or Bust.” She described in detail the drip and ship model, in which patients are taken to the nearest local PSC to be evaluated for candidacy for tPA as fast as possible. After tPA is administered, patients who are eligible for thrombectomy are then transferred with the tPA drip running to the closest CSC or thrombectomy capable center. Altschul’s arguments were that a portion of stroke patients will have hemorrhage (15%), only a minority of tPA candidates also have an LVO (10-15%), and not all LVOs are thrombectomy eligible. By bypassing PSCs and transferring all stroke patients to a CSC, there will be unnecessary delay for tPA candidates who may not be thrombectomy candidates, and there will be unnecessary futile transfers. Bypass will only benefit the patients with ischemic stroke from LVOs who are candidates for thrombectomy; all other stroke patients may be harmed. Current AHA policy recommends EMS should not bypass a PSC in favor of a CSC if such diversion would add >15-20 min of transport time. There is no mention about which subgroups of patients should be brought directly to CSC.

Indeed, the goal for EMS is to “bring the right stroke patient at the right time to the right stroke center.” This is not an easy task. In NYC, there are an average of 4,000 9-1-1 calls to EMS per day; only 50 (1.25%) are strokes. Furthermore, of these 50 strokes, only about 5 are LVOs, and only 1.5 will be eligible for thrombectomy. It is unreasonable to expect EMS to develop expertise in this decision. There are many tools to help them, but none are perfect as accurate LVO diagnosis depends on imaging in field. A RACE score >4 has a sensitivity of 85% and sensitivity of 69% for LVO. A LAMS of 4 or 5 has a 62% probability of LVO. Furthermore, many focal neurologic deficits are not ischemic stroke; a FAST ED score ≥2 is associated with 26% stroke mimics.

Altschul also underscored the importance of early tPA. Indeed, 18% of LVOs will get early reperfusion with tPA alone, and if a patient presents to a PSC first, they are more likely to get tPA. She also cited several studies suggesting that prior tPA may increase the rate of recanalization after thrombectomy and decrease the number of passes. Ultimately, the SWIFT-D trial will help to answer this question. She concluded that the “drip and ship model is here to stay.” There will always be inpatient LVOs, mis-triaged LVOs, overcrowding at CSCs, and long transport times and weather that will necessitate stroke patients first presenting at a PSC. She also warned that bypassing PSCs may limit their growth as developing hospitals. Transfers are increasing; 43% of LVOS are transferred first from a PSC. Another study showed that 54% of transfers do not ultimately undergo thrombectomy at the CSC. Instead of bypassing the PSC, she argued that we should focus on the inter-hospital transfer process to improve it.

“Skip the Local Center and Head Straight to the Mothership” was the third talk, presented by Raul Nogueira. He started off by stating that he doesn’t agree with this in all cases, but there are certain situations in which it may make sense. There are “shades of gray” in how we plan systems, and each decision will need to be made not only at the individual regional system level but also at the patient level. Nogueira argued that patients without high stroke severity may be better served at a PSC regardless of time. Those with high stroke severity not in an early time window should likely be taken to a CSC mothership, as they wouldn’t be candidates for tPA anyway, may be candidates for thrombectomy, and may need other services such as hemicraniectomy and ICU care.

The real question is what to do with patients that have high stroke severity and present in an early time window. Should they go to a PSC to receive faster tPA or bypass the PSC and go straight to the CSC mothership for faster thrombectomy? Nogueira argues that the latter makes more sense in most situations. Even with regards to the tPA decision, Nogueira showed data to support that “closer is not always better.” There is evidence that tPA is delivered faster at CSCs with high volume experience, by 28 minutes in one report. There is also evidence that if you consider only patients with LVOs treated with thrombectomy, they have worse outcomes when they do not first present to a thrombectomy capable center. One study of the STRATIS registry showed patients taken directly to CSCs had onset to reperfusion times that were 110 minutes faster and improved 90-day mRS. Furthermore, a recent analysis from GWTG data of 37,260 thrombectomy patients at 639 hospitals from January 2012 to December 2017 showed that about 43% were inter-hospital transfers, underscoring the magnitude of this question. For many thrombectomy candidates the “drip and ship” model may cause harm.

Nogueira discussed his thoughts about the future of LVO in-field diagnosis and treatment, including devices to detect large strokes and hemorrhages using radiofrequency waves to measure electrical impedance and using TCD, robotics, and AI. He also discussed the smart phone app “FAST-ED” designed to be used in-field with less specialized EMTs. his can aid in the diagnosis of LVO and determine “likely a candidate for tPA only,” “likely a candidate for EVT only,” “likely a candidate for both,” or “likely a candidate for neither.” The app then utilizes this information along with GPS to help EMS route the patient to the nearest appropriate hospital, weighting the likelihood of LVO against the additional time needed to get to a CSC for those who are candidates for both. Meanwhile, the receiving clinicians and interventionalists are notified before arrival and can track the ambulance using the app on their end.

He further discussed ongoing work to enhance collaterals and sustain penumbra, including the possibility of induced hypertension with the SETIN trial and sphenopalatine ganglion stimulation with the ImpACT 24B trial. Finally, he discussed remote intervention with the use of telerobotics; the first coronary “telestent” was recently completed with the interventionalist 20 miles from the patient. He concluded that especially for patients within an early time window and high stroke severity, bypass seems reasonable in a “broad range of situations.”

The fourth talk, by Johanna Fifi, was titled “Bring the Intervention to the Patient: The Trip and Treat Model.” She opened with rather than “getting the right patient to the right place at the right time,” maybe we should “get the right team to the right patient at the right place at the right time.” Many variables are important for infarct progression during transfer, such as time and collaterals. She discussed clinical and imaging factors associated with infarct progression; 20% of transferred patients suffered ASPECTS decay in one study in Boston. In their data, they observed 36% in NYC. Most of the time delay occurs from PSC CT scan to departure. They are also working to minimize futile transfers in NYC. As other presenters mentioned, it has been shown that LVO patients treated with thrombectomy do worse if they first present to a PSC.

She proposed that one novel model to minimize futile transfers and minimize the need to transfer patients altogether is to instead transfer the interventionalist, which can be done more efficiently. In this way, a PSC becomes thrombectomy capable upon the interventionalist’s arrival. In their system in NYC, they utilize a shared Mobile Interventional Stroke Team (MIST) that travels to PSCs in their network. Fifi makes the point that pre-hospital triage tools (LAMS, RACE, etc.) have a low sensitivity and may not be the best answer. NYC has plans to start bypassing PSCs using a modified NYC S-LAMS. However, the exclusion criteria include: trauma cause, wheelchair/bedbound, loss of consciousness, seizure cause, LKW>5 hours, transport time to thrombectomy capable >30 minutes. She mentioned a Wall Street Journal article describing thrombectomy: “A breakthrough stroke treatment can save lives – If it’s available.”

With the NYC exclusion criteria, there will likely be thrombectomy eligible patients that end up at PSCs. What makes the “trip and treat” model unique is that MIST is based on both a fixed and traveling team. The fixed team includes a stroke neurologist, nurse, anesthesiologist, and IR room. The traveling endovascular team includes an attending interventionalist, a fellow or mid-level assistant, and a neuro-technologist. They utilize several modes of transport in NYC, from Uber to subway. Another group is flying the interventionalist by helicopter. MIST allows for staffing with one condensed team. While the traveling team is en route, the local team, with the help of the ED, can stabilize the patient pre-procedure with parallel processing. The “trip and treat” model led to faster treatment times; first door to recanalization was improved by 68 minutes compared to the “drip and ship” model. Furthermore, having one team across multiple hospitals allows concentration of skill (as it is the same interventional team), reduced cost, does not overwhelm one mothership, and keeps patients local near their families.

The last talk was titled “Lessons Learned while Studying Multiple Delivery Models” by Marc Ribo. Ribo summarized the points made by the previous speakers, adding his impressions from his work in Catalonia. An analysis of the Catalan registry showed that of all acute ischemic strokes, 34% were eligible for tPA, 18% were eligible for thrombectomy, and only 8% were eligible for both. He emphasized that most ischemic strokes are not LVOs. As in other parts of the world, tPA is much more readily available compared to thrombectomy in this region. His group’s work has nicely mapped the region’s towns of initial alerts, PSCs, and CSCs. Their RACECAT study compares “drip and ship” to “direct mothership” models. So far, they have enrolled 860 patients, and the first interim analysis is scheduled soon.

Using their ESACAL RACE prescreening tool they can enrich for an LVO population. Using this tool, their at-hospital diagnosis is quite good with 40.5% LVOs. For those patients that are transported with a drip and ship model, emphasis needs to be placed on “door in-door out” times because this has consistently been shown to be a problem epoch in multiple studies. Ribo suggested that perhaps the initial ambulance could stay parked at the PSC until the LVO status is determined so that the same ambulance could then deliver the patient to the CSC. Systems should also make attempts to then transfer patients directly to the interventional suite at the CSC and bypass the ED and repeat imaging.

Ribo also discussed the recent data that higher volume endovascular centers with >50 thrombectomies per year have improved outcomes. He questioned if there is a number that is too much. At some point he feels a CSC could become over-worked. If a center is performing too many thrombectomies, it is probably transferring patients from too far a distance, in his opinion. He also favors the system in which perhaps “trip and treat” models could be used to transfer interventionalists to avoid diluting expertise. If “trip and treat” is not possible, new thrombectomy capable centers should be built based on the needs of a regional system and not too close to existing CSCs.

In the discussion, many great questions were raised, highlighting how unique regions share similar challenges. At the end, all speakers appeared to agree that one size does not fit all when it comes to acute stroke care systems.