A Rebleeding Ulcer and a Drug Not Yet Written Into the Guideline
A single patient, two hours past endoscopic control of a high-risk bleeding ulcer. The disagreement is between the drug with the sharper mechanism and the drug with the trial actually built for this exact window.
T.O., a 58-year-old long-haul truck driver, was two days from a scheduled DOT physical when he collapsed in a truck-stop bathroom, hematemesis down the front of his shirt. EGD found a 1.8cm posterior duodenal bulb ulcer with a visible vessel — Forrest class IIa, the endoscopic category carrying the highest recognized risk of rebleeding among non-actively-spurting lesions — successfully treated with epinephrine injection and thermal coagulation. He has taken over-the-counter ibuprofen for chronic low back pain most days for the past several years, never mentioned it at a checkup because he's never had a formal one, and was otherwise unaware anything was wrong until he was on the bathroom floor.
The vessel is controlled, but the actual question in front of the team now is what happens over the next 72 hours, when rebleeding risk is highest. Acid suppresses clot-degrading pepsin activity and stabilizes the platelet plug directly — not incidentally, this is the entire pharmacologic rationale for aggressive acid suppression post-hemostasis, and it's why the intervention that matters most here isn't another endoscopic pass but how completely his stomach's acid production is shut down over the next three days.
High-dose intravenous PPI — typically esomeprazole as a bolus followed by continuous infusion — is the regimen every major GI bleeding guideline has specified for a decade, backed by trial data going back to Sung and colleagues' original meta-analyses. Vonoprazan's potassium-competitive mechanism achieves faster, deeper, and less variable acid suppression than any PPI, independent of CYP2C19 metabolizer status — a genuine pharmacologic advantage for a bleeding ulcer specifically, where rapid, complete suppression in the first hours matters most. What it does not have, unlike PPI infusion, is a randomized trial in a Western high-risk-rebleeding population that any current guideline actually cites. The trial usually offered in its place is Geeratragool and colleagues' 2024 study, which randomized 194 patients across six Thai centers after endoscopic hemostasis and found oral vonoprazan noninferior to IV pantoprazole for rebleeding at 30 days, 7.1% versus 10.4%. Its own investigators named the limits: generalizability to multiethnic populations remained open, only about 3% of their patients carried the very highest-risk lesions, and time to endoscopy ran longer than a U.S. service would expect. T.O. sits at the sharper end of all three — a Forrest IIa vessel scoped and treated the same morning — which is exactly the corner of the risk distribution the trial enrolled least of.
The window that actually decides this case
High-dose IV PPI infusion is what every guideline I follow specifies for a Forrest IIa lesion post-hemostasis, and I don't think the next 72 hours is the time to deviate from the evidence base we actually have outcome data for.
Sung and colleagues' meta-analyses established this regimen's rebleeding-reduction benefit years ago, in exactly this population — high-risk stigmata, post-endoscopic treatment. I'd rather start the drug with a trial behind it in patients who look like him.
I'm not disputing the PPI trial base — it's real and it's why PPI infusion is still my second choice, not a bad one. What I'd flag is that his CYP2C19 status is unknown, and a meaningful fraction of PPI-treated patients under-suppress acid because of it, a variability problem vonoprazan's mechanism doesn't have.
Rebleeding risk is concentrated almost entirely in the first 72 hours, which is exactly the window where acid suppression speed and completeness matter most — not a detail, the whole mechanistic argument for why we suppress acid at all here. Geeratragool and colleagues' 2024 trial — 194 patients across six Thai centers — found oral vonoprazan noninferior to IV pantoprazole for 30-day rebleeding, 7.1% against 10.4%. I'll say directly that it hasn't been replicated in a Western population, and the AGA's own practice update on potassium-competitive acid blockers still says the evidence is insufficient to use them first-line for a bleeding ulcer with high-risk stigmata.
You're right that CYP2C19 variability is a real gap in the PPI argument, and I don't have a clean answer for a patient we haven't genotyped. But biologic plausibility that hasn't cleared an outcome trial in our own population is exactly the pattern I've watched fail before — more complete suppression on paper, no measurable rebleeding difference in practice.
I'll go with standard IV PPI infusion as our regimen today, precisely because it's the one with a trial base behind it for a patient who looks like him. If vonoprazan's evidence base in Western populations catches up, that's a real reason to revisit this the next time.
High-dose esomeprazole infusion for 72 hours per standard post-hemostasis protocol, transitioning to oral PPI at discharge with H. pylori testing pending and formal NSAID-alternative counseling for his back pain.
Left genuinely open: the Clinical Pharmacologist's mechanistic case for vonoprazan wasn't rejected as wrong, only as not yet evidenced in the population that matters for this decision — both physicians agreed the next high-risk rebleeding patient should get the same PPI-infusion regimen unless that evidence base changes first.