A New Pulmonary Embolism, an Unresected Stomach Tumor, and a Guideline That Changed
Her gastric cancer is still in place, and the field's own advice on anticoagulating a GI-cancer patient has shifted since the last time some of this team trained. The disagreement isn't which guideline is newer — it's whether her own tumor still fits inside the population that guideline's reassuring data was built from.
Carol M., 61, has spent most of the past three years as her husband's primary caregiver as his Parkinson's disease has advanced, and the pleuritic chest pain that brought her to the emergency department this morning is the first symptom in a long time she hasn't managed to talk herself out of, because there was simply no one else to hand his morning medications to while she waited it out at home. A neighbor is sitting with him now, a temporary arrangement Carol has already mentioned twice she doesn't want to impose on for more than a day. She was diagnosed with gastric adenocarcinoma five months ago, found unresectable at staging laparoscopy for peritoneal spread, and is currently on palliative first-line chemotherapy that has kept her disease stable on her two most recent scans.
A CT angiogram today confirmed a segmental pulmonary embolism, and the team is now choosing how to anticoagulate a patient whose own primary tumor is still sitting, untouched, in her stomach. The historical caution here is real: CLOT (Lee et al., 2003) established low molecular weight heparin as the trusted standard for cancer-associated thrombosis, and two later DOAC trials — SELECT-D with rivaroxaban and Hokusai-VTE Cancer with edoxaban — both found increased major bleeding concentrated specifically in patients with luminal GI cancers like hers. But CARAVAGGIO (Agnelli et al., 2020) tested apixaban against dalteparin directly and found no such excess, including within its own GI-cancer subgroup, which is why both ASCO's 2023 update and ITAC's 2022 guidance now list apixaban as appropriate even with an intact gastric primary. What none of those population-level numbers capture on their own is that she has had guaiac-positive stools for several weeks and reported one episode of streaky hematemesis to her oncologist at her last visit — neither worked up yet, and neither necessarily typical of the reassured population a randomized trial's average outcome describes. She mentioned both findings almost in passing at that visit, more concerned at the time with whether her chemotherapy schedule could be rearranged around her husband's upcoming neurology appointment than with a slow trickle of blood she'd mostly attributed to hemorrhoids.
On the ward, choosing an anticoagulant for a stomach that's still bleeding a little
Start dalteparin, not a DOAC. CLOT established this as the standard for cancer-associated thrombosis, and two later DOAC trials — SELECT-D with rivaroxaban, Hokusai-VTE Cancer with edoxaban — both showed real excess bleeding specifically in patients with luminal GI cancers. She has an unresected gastric tumor. That's the population those trials flagged.
Those are real signals, but they're specific to rivaroxaban and edoxaban, not DOACs as a class. CARAVAGGIO tested apixaban directly against dalteparin and found no excess major bleeding overall, and specifically not in its own GI-cancer subgroup. That's why ASCO's 2023 update and ITAC's 2022 guidance both now list apixaban as appropriate even with an intact luminal GI cancer — it's a different drug with a different result, not the same caution restated with newer letters after it.
Treating "GI cancer means LMWH" as a blanket rule is exactly the kind of extrapolation the field has moved past since CARAVAGGIO published, and she'd gain a real quality-of-life benefit from an oral drug given how much she's already managing at home.
I'd agree with you if her stomach were simply "a GI cancer" in the abstract. But she has guaiac-positive stools going back weeks and one reported episode of streaky hematemesis that nobody has scoped yet. CARAVAGGIO's reassuring subgroup result describes an average outcome across patients with intact luminal GI tumors — it doesn't tell us she individually resembles that average once she already has an active, if low-grade, bleeding source of her own.
I'd rather start dalteparin now, get the EGD moved up this week, and revisit apixaban once we actually know what that tumor is doing today rather than trusting a population result to describe someone who may not fit inside it.
Agreed: start dalteparin for the pulmonary embolism, expedite EGD this week to characterize the reported hematemesis and guaiac-positive stools, and revisit apixaban once the bleeding source is actually known rather than assumed either way.
Not agreed: the hematologist still felt the oncologist's caution effectively re-imports the old blanket "GI cancer means LMWH" rule CARAVAGGIO was specifically meant to correct, just re-justified through a different, patient-specific door, and worried this reasoning could just as easily be applied to any GI-cancer patient with mild anemia. The oncologist didn't accept that framing — an unworked-up bleeding symptom is a specific finding, not a category — and wasn't willing to start an oral factor Xa inhibitor before knowing what was actually causing it. The disagreement was left unresolved, with the EGD result set as the point at which it gets revisited.