Platinum-Resistant Ovarian Cancer: Bevacizumab Against a Healed Anastomosis
AURELIA showed a real progression-free survival benefit from adding bevacizumab to chemotherapy in platinum-resistant ovarian cancer — and specifically excluded the kind of bowel-adjacent disease this patient now has. The question isn't whether the drug works. It's whether she is the patient the exclusion was written for.
Carol T., 66, moved into a small duplex with her younger sister eighteen months ago, not long after her husband died, and the two of them have settled into a routine of driving each other to appointments that neither much enjoys but both have come to depend on. She was diagnosed with Stage IIIC high-grade serous ovarian carcinoma two years ago, underwent optimal debulking that included resection of a segment of involved small bowel with primary anastomosis, and completed six cycles of carboplatin and paclitaxel with an initial response. Four months after finishing that platinum-based regimen — inside the six-month window that defines platinum resistance rather than platinum sensitivity — her CA-125 began climbing, and a restaging CT confirmed recurrence.
The recurrence itself is small-volume, but its location is what is actually driving today's disagreement: a peritoneal nodule sits directly adjacent to the old anastomotic site, close enough that the surgeon reading the same scan flags it as a bowel-wall-involvement concern rather than a straightforward peritoneal deposit. AURELIA, the trial that established adding bevacizumab to chemotherapy in platinum-resistant disease, found a real doubling of median progression-free survival — roughly 6.7 months against 3.4 — and a meaningfully higher response rate, though no overall survival benefit. That same trial excluded patients on two separate grounds, and the gap between them is what the team is actually arguing about: any history of bowel obstruction, fistula, or perforation, and — independently, with no time limit attached to it — evidence of rectosigmoid involvement or bowel involvement on computed tomography. Both existed because bevacizumab's antiangiogenic mechanism measurably raises gastrointestinal perforation risk; AURELIA's screening held perforation to 2.2 percent, where Cannistra's earlier series, enrolling on looser criteria, reported it in roughly one patient in nine. Carol clears the first ground cleanly, with no obstruction and an anastomosis that healed without incident two years ago. She does not obviously clear the second, because the finding at issue is precisely a CT read of possible bowel involvement, and the trial attached no healing interval that would let a two-year-old anastomosis age out of the criterion. Carol has said she would rather start something imperfect this week than wait through another round of imaging she has grown to dread. Nobody reads that as consent to a perforation, but it does set the currency the delay has to be argued in: weeks of treatment foregone, not scans acquired.
Reading one scan two different ways
AURELIA nearly doubled median PFS — 6.7 months against 3.4 — and Carol clears the exclusion ground I'd weight most, which is obstruction history. No obstruction, no perforation, nothing beyond a nodule sitting close to a segment that healed without complication two years ago. I don't want to withhold a drug with this much benefit based on a proximity read rather than a documented contraindication.
I operated on that segment. What I'm seeing on this scan isn't just proximity — it's a nodule that appears to be tracking along the same tissue plane as the anastomosis, which is exactly the anatomic picture that precedes a bevacizumab-associated perforation in my experience, obstruction history or not.
And I'd contest "she clears the criteria" even as stated — AURELIA excluded bowel involvement on CT as its own separate ground, with no healing interval attached to it, which is exactly the finding I'm describing. Reading her as eligible doesn't mean the trial fails to cover her; it means deciding my read of this scan is wrong.
You're both reading the same static image and reaching different conclusions about a risk that isn't actually static — it's a function of how this nodule grows. Start single-agent chemotherapy now, whichever backbone the oncologist prefers, and re-image at the first response assessment. If the nodule shrinks or holds steady without encroaching further on the anastomotic plane, add bevacizumab then, with real anatomic information instead of a judgment call made off one scan.
Agreed: single-agent weekly paclitaxel started this cycle, with restaging imaging at the first response assessment specifically oriented to the anastomotic-plane question, not just overall response.
Not agreed: whether bevacizumab would ultimately have been safe to start today. The surgical oncologist's anatomic concern was never resolved against the medical oncologist's trial-criteria reading — both positions were left standing, with the sequencing plan adopted as a way to gather more information rather than as a verdict on which read was correct.