Happy birthday, Earth. (It's every day, actually.)
Earth Day (April 22)
It's Sunday morning. Here's what happened in medicine this week.
A precision viral therapy cleared aggressive brain tumors in mice with a single dose, the new AHA/ASA stroke guidelines reshape thrombolysis and add a first pediatric protocol, and a fresh wave of AI sepsis deployment data finally shows mortality moving in the right direction.
Let's get into it.
🧬 A single-shot viral therapy cleared brain tumors in 83% of mice
💊 New stroke guidelines reshape door-to-needle care, including pediatrics
📊 AI sepsis tools cross from promise to actual mortality reduction
📝 The migraine that wasn't
🔬 ON THE HORIZON
A single dose of a new viral therapy wiped out aggressive brain tumors in 83% of treated mice.
Glioblastoma has been the same story for two decades. Surgery, radiation, temozolomide, recurrence, death. Median survival sits around 15 months.
A team at Trogenix published a different kind of attack in Nature this week. They built synthetic super-enhancers that switch on only inside glioblastoma stem cells, delivered by AAV with a one-two payload of HSV-TK and IL-12.
The result in an aggressive mouse model was complete tumor elimination in 83% of animals after a single infusion, with no toxicity at 11 months and no recurrence. The IL-12 component appeared to lock in long-term immune protection.
Why it matters: Most viral therapies for cancer either lack tissue specificity or fizzle on durability. This one used cell-state-specific gene regulation to do both, and the same enhancer logic could apply to other transcription-factor-driven tumors.
Bottom line: Mice are not patients. But the first human glioblastoma trial is expected to begin dosing this quarter.
Source: Nature, April 9, 2026.
📋 CLINICAL UPDATE
The new stroke guidelines just elevated tenecteplase, extended treatment windows, and added the first ever pediatric protocol.
The 2026 AHA/ASA acute ischemic stroke guideline is the first comprehensive overhaul in years. The headline shift: tenecteplase 0.25 mg/kg now sits alongside alteplase 0.9 mg/kg as a class 1 recommendation in the 4.5-hour window.
Tenecteplase is a single bolus. Alteplase requires a one-hour infusion. That means faster door-to-needle and one less pump during transfer.
The window also stretched. Selected patients with unknown-onset or 4.5 to 9 hour strokes are eligible for thrombolysis using DWI or perfusion-mismatch criteria.
Two other meaningful changes:
Tight glucose targets of 80 to 130 mg/dL are out, due to hypoglycemia risk without outcome benefit
The first dedicated pediatric stroke section appears in any AHA/ASA stroke document
The clinical angle: If your hospital still defaults to alteplase out of habit, the case for switching just got stronger. Update the stroke protocol, the order sets, and the pump room.
Source: Stroke (AHA/ASA Guideline), 2026N PRACTICE
📊 AI IN PRACTICE
A new generation of AI sepsis tools is finally moving the mortality needle in real hospitals.
Sepsis prediction algorithms have been around for years with mixed real-world results. The deployment data from this year tells a different story.
An npj Digital Medicine study compared 97,559 stays in wards using AI-driven sepsis surveillance to 25,851 stays in matched controls. AI-flagged cases on the intervention wards saw drops in in-hospital and 90-day mortality. The control wards did not.
Earlier deployment data from UC San Diego's COMPOSER model showed similar gains:
17% relative reduction in in-hospital sepsis mortality
5% absolute increase in sepsis bundle compliance
4% reduction in 72-hour SOFA score change after onset
The pattern is consistent. Tools that integrate with workflow, surface alerts to the right person, and feed back into a quality system actually save lives. Tools that just fire alarms create alert fatigue.
Bottom line: The sepsis AI question is no longer "does it work." It is "does your hospital implement it well." The implementation gap is now bigger than the algorithm gap.
Source: npj Digital Medicine, 2026.
🩺 FROM THE FIELD
The Migraine That Wasn't
Note: This is a composite narrative for template purposes, not a reported story.
The patient was 34 and had a migraine. That's what the chief complaint said when Dr. R picked up the chart at 2 AM.
Standard story: photophobia, nausea, eight out of ten. She had a history of migraines going back to college, and Toradol was already in the order.
He walked in. She was sitting up, tired but oriented. The headache had been there since around noon.
She mentioned, almost as an aside, that for about ten minutes around 1 PM she could not get her words out. It came back. She didn't think much of it.
Dr. R sat down and asked her to describe it again, slowly. She had been mid-sentence with her sister when the words just stopped. Then they came back.
He pulled up the discharge order set he had been about to use and started a different one. NIHSS was zero, CT was clean, MRI showed a tiny diffusion restriction in the left frontal lobe.
A 34-year-old woman with a migraine had had a TIA at lunchtime. The workup found a patent foramen ovale. She left two days later on an antiplatelet, with cardiology and neurology follow-up scheduled.
He thought about it for weeks afterward. Not because he caught it, but because of how easily he might not have. The chart said migraine, the Toradol was queued, and he almost didn't sit down.
📝 BEFORE YOU GO
One question this week: the new guideline puts tenecteplase on equal footing with alteplase as class 1. Has your hospital updated its protocol, or are you still defaulting to a one-hour infusion out of habit?
Hit reply and tell us.
If you liked this, share The Consult with a colleague → theconsult.info
Until next time,
The Consult
