The Extracranial-Intracranial Bypass — Two Decades of Faith a 1,377-Patient Trial Demolished in 1985

Summary

On October 30, 1967, in Zurich, the neurosurgeon M. Gazi Yaşargil sutured a scalp artery to a cortical branch of the middle cerebral artery under the operating microscope, rerouting blood around a blocked vessel to feed a starving brain; the operation was elegant, technically dazzling, and — for the prevention of stroke in patients with carotid and middle-cerebral disease — almost entirely unproven, and that gap between surgical beauty and clinical benefit is the entire case. For nearly two decades the extracranial-intracranial (EC-IC) arterial bypass spread on the strength of its own plausibility and on case series reporting open grafts, until a single randomized trial showed it prevented nothing it claimed to prevent.

The operation was never a fraud and never a mass killer in the lobotomy sense. It killed and disabled quietly, at the margins: a procedure with a roughly 12 percent thirty-day rate of stroke or death imposed up front on patients who, the trial would show, were no better protected afterward. The surrogate that sustained it was graft patency — the bypass stayed open in about 96 percent of cases, a number surgeons and angiograms could see and celebrate. A patent vessel looked like a prevented stroke. It was not the same thing, and conflating the two is the mechanism that kept the operation alive.

The reckoning arrived not from a regulator or a court but from an eight-year, NIH-funded randomized controlled trial led by the Canadian neurologist Henry J. M. Barnett of London, Ontario. Published in the New England Journal of Medicine on November 7, 1985, the International EC/IC Bypass Study randomized 1,377 patients at 71 centers in 14 countries and found that surgery added to best medical care did not reduce fatal or nonfatal stroke; two subgroups — patients with severe middle-cerebral stenosis and those with persisting symptoms after carotid occlusion — actually fared worse with the operation. Within a few years the procedure collapsed from a flourishing subspecialty to a narrow, rarely-indicated salvage technique. It was not banned. It was disconfirmed, and it became one of medicine's foundational lessons in why a trial must precede an operation, not follow it.

Timeline

Oct 30, 1967

Yaşargil performs the first STA-MCA bypass

In Zurich, M. Gazi Yaşargil anastomoses the superficial temporal artery to a middle cerebral artery branch in a patient with MCA occlusion, having trained in microvascular technique with R. M. P. Donaghy at the University of Vermont. The era of cerebral revascularization begins.

Late 1960s

Donaghy and Yaşargil codify the microsurgical method

The operating microscope, bipolar coagulation, and fine microsutures make a once-impossible anastomosis routine. The technical achievement is real and seductive.

1970s

Rapid worldwide adoption

EC-IC bypass spreads through neurosurgical centers as a logical defense against stroke in patients with carotid or intracranial occlusion. Adoption rests on physiological reasoning and uncontrolled case series, not comparative outcomes.

1977

The International EC/IC Bypass Study opens enrollment

Funded by the U.S. National Institute of Neurological and Communicative Disorders and Stroke and led by Henry Barnett, the trial sets out to test, with randomization, whether the operation prevents stroke.

1977–1982

1,377 patients randomized at 71 centers in 14 countries

Roughly 714 are assigned to best medical therapy (aspirin, blood-pressure control); about 663 to bypass plus medical therapy. Mean follow-up exceeds four and a half years.

During the trial

perioperative cost recorded

Graft patency reaches about 96 percent, but the thirty-day stroke-and-death rate in the surgical arm runs near 12 percent — harm paid in advance against a benefit not yet demonstrated.

Nov 7, 1985

The verdict is published

NEJM reports the trial found no reduction in fatal or nonfatal ischemic stroke; surgery is not effective, and two high-risk subgroups do worse. The hypothesis is rejected.

1985–1987

Procedure volume falls sharply

Neurosurgical use of EC-IC bypass for atherosclerotic stroke prevention drops steeply within months of publication as the trial's authority takes hold.

Mar 1987

Entry-bias critique published

In NEJM, Thoralf Sundt and others argue the trial may have excluded the patients most likely to benefit by allowing eligible cases to be operated on outside randomization.

1987

The critique is rebutted with data

Barnett's investigators (Haynes and colleagues) document that the number of eligible-but-nonrandomized operated patients was small and could not explain the null result; the trial's conclusion stands.

1990s–2000s

Narrow indications survive

The operation persists only for selected hemodynamic failure, moyamoya, and complex aneurysms requiring flow replacement, not for routine atherosclerotic prophylaxis.

2011

A second trial confirms the verdict

The Carotid Occlusion Surgery Study (COSS), which selected patients by measured hemodynamic compromise, is also stopped early for futility — closing the door the 1987 critics had tried to leave open.

The Beautiful Anastomosis and the Logic That Sold Itself

The EC-IC bypass was adopted because it made sense, and because it was hard. The reasoning was clean: if a brain artery is blocked and the tissue downstream is starved, rerouting fresh blood around the blockage should prevent the stroke that the blockage threatens. The operation that delivered this — connecting a scalp artery to a cortical vessel under magnification — was among the most demanding feats in surgery, and that difficulty became, perversely, part of the evidence. A procedure this refined, performed by this elite a cohort, producing angiograms this clean, felt like it had to work. What was missing was the only thing that mattered: a comparison group. The case series that justified the operation reported on patients who received it, not on matched patients who did not, and they reported the wrong endpoint. They counted open grafts. Whether those open grafts translated into fewer strokes, fewer deaths, or better lives than aspirin alone was never tested before the operation became standard in centers across the world. The physiology was plausible, the surgery was masterful, and the clinical question was simply assumed answered.

The Trial That Counted Strokes Instead of Grafts

What made the International EC/IC Bypass Study decisive was that it refused to score the operation on its own terms. Where surgeons measured patency, the trial measured the thing patients actually care about: stroke and death. Over eight years and 71 centers in 14 countries, 1,377 patients were randomized — a scale and rigor almost unheard of for a surgical procedure at the time. The grafts performed beautifully, staying open in roughly 96 percent of operated patients, exactly as the case series had promised. And it made no difference. Patients who received the bypass suffered fatal and nonfatal strokes at rates no better than those on medical therapy alone, and they paid a perioperative penalty — a thirty-day stroke-or-death rate near 12 percent — that medical patients did not. Two subgroups thought most likely to benefit, those with severe middle-cerebral stenosis and those still symptomatic after carotid occlusion, did measurably worse with surgery. The trial had isolated and destroyed the central confusion of the field: a patent anastomosis is a surgical success and a clinical irrelevance unless it lowers the stroke curve, and it did not.

The Critique, the Rebuttal, and the Door That Closed Twice

The 1985 result was not accepted quietly. In 1987, Thoralf Sundt and other surgeons argued in the same journal that the trial had been undermined by its own conduct: that some eligible patients, including high-risk cases that might have shown benefit, had been operated on outside the randomization and thus excluded from the analysis, biasing the comparison. It was a serious, specific objection, and the trialists answered it with data rather than rhetoric. Barnett's group, in work led by R. Brian Haynes, audited the nonrandomized operated patients and showed they were too few to overturn the result; the null finding held. The deeper test came a generation later. Critics had insisted the trial failed because it did not select patients by measured hemodynamic compromise — the very group thought most likely to benefit. The Carotid Occlusion Surgery Study set out in the 2000s to do exactly that, using PET imaging to enroll only patients with proven misery perfusion. In 2011 it too was halted early: surgery did not help, and the perioperative stroke cost was again real. The door the critics had propped open in 1987 was shut, with evidence, in 2011. The operation that had been beautiful, plausible, and patent was, for stroke prevention, simply ineffective.

Contributing Factors

01

Surrogate endpoint mistaken for benefit

The field scored the operation on graft patency — a vessel staying open, visible on angiography in about 96 percent of cases — rather than on strokes prevented or lives improved. A patent bypass is a surgical achievement, not a clinical outcome. Optimizing the measurable proxy while assuming the true endpoint follows is the recurring mechanism by which plausible interventions outrun their evidence.

02

Plausible physiology accepted as proof of efficacy

"Blocked artery, reroute the blood, prevent the stroke" is a mechanistic story so clean it felt self-evidently true. But physiological plausibility predicts neither net benefit nor net harm; the brain's collateral circulation and the operation's own perioperative risk were not in the story. A compelling mechanism is a hypothesis to test, never a result to bank.

03

Technical virtuosity confused with clinical value

The bypass was one of the hardest, most elegant operations in surgery, performed by an elite cadre. That difficulty conferred prestige and discouraged the question of whether it worked at all — it felt unseemly to randomize away a masterwork. Skill at performing a procedure is independent of whether the procedure should be performed.

04

Adoption before the controlled comparison existed

For nearly two decades the operation became standard on the strength of uncontrolled case series that lacked a comparator group and measured the wrong endpoint. Once a procedure is entrenched, the trial that should have preceded it becomes harder to run and easier to resist. Diffusing an unproven surgery first and testing it later builds a constituency invested in the answer being yes.

05

An honest trial that scored the patient's outcome and was defended on data

The episode's saving grace is also its lesson: the harm was bounded because someone ran a large, randomized, multinational trial measuring stroke and death, then met the inevitable methodological challenge with audited numbers rather than authority — and a second trial later closed the remaining loophole. Disconfirmation only works when the disconfirming evidence is gathered and then held against pressure to explain it away.

Aftermath

The material consequence was an abrupt contraction: a flourishing subspecialty of cerebral revascularization for atherosclerotic stroke prevention shrank within a few years to a narrow salvage technique reserved for moyamoya disease, selected hemodynamic failure, and complex aneurysms needing flow replacement. The durable ripple was epistemic. The EC/IC Bypass Study became a textbook exemplar — taught in evidence-based medicine and clinical-trials courses — of how a surgical procedure can spread on plausibility and patency, and why nothing short of a randomized comparison measuring patient outcomes can establish that an operation prevents what it is designed to prevent. It reshaped expectations for cerebrovascular surgery: later interventions in the field, from carotid stenting to intracranial procedures, faced a higher evidentiary bar partly because of this precedent. The 1987 entry-bias debate became its own teaching case in how to interrogate and defend a trial, and the 2011 COSS futility result confirmed that even the best-selected patients gained nothing. What remains is a procedure that is still performed — correctly, narrowly, and rarely — and a phrase that endures in clinical teaching: the EC-IC bypass is the operation that was patent, elegant, and useless for the thing it was sold to do, the case in which a beautiful anastomosis taught medicine to count strokes, not grafts.

Lessons

Measure the outcome the patient cares about, not the one the operator can see. A patent graft, a clean angiogram, a technically perfect repair — these are operator-visible surrogates. If your evidence is the proxy you can photograph rather than the event you are trying to prevent, you have not yet shown benefit. Score the disease, not the device.
Treat a plausible mechanism as a question, not an answer. "It should work" is the beginning of a trial, never the end of one. The more obvious the physiological logic, the more disciplined you must be about testing it, because obviousness is exactly what suppresses the comparison that would reveal you are wrong.
Run the controlled trial before the procedure becomes standard, not after. Once an intervention diffuses on case series, you create a constituency that will resist the randomization and contest the null result. The cheapest, cleanest moment to test an operation is before its practitioners have staked their careers on it working.
When the disconfirming result is challenged, answer with audited data, not authority. The 1987 entry-bias critique was legitimate and was defeated by counting the nonrandomized patients, not by invoking the trial's prestige. A disconfirmation survives only if its authors will do the additional work to defend it against the most specific objection raised.
Let the well-designed trial retire the practice — and confirm it when the loophole is named. The objection that the "right" patients were never tested deserved its own trial, and got one. Build the follow-up study that targets the surviving hypothesis; do not let an unfalsified subgroup keep a disconfirmed operation alive indefinitely.