Bevacizumab for Recurrent Glioblastoma: More Time, or Just Easier Time
A real, replicated finding: bevacizumab extends the time before a recurrent glioblastoma progresses again, without extending how long the patient lives. For a mother racing a deadline that has nothing to do with a tumor scan, the two aren’t the same question.
Naomi K., 47, has kept a countdown taped to her refrigerator since spring: seventy-one days until her oldest walks across a stage in a cap and gown she helped pick out, three sizes too big, at freshman orientation four years ago. Fourteen months ago a seizure at her desk led to an MRI, a resection, six weeks of concurrent chemoradiation, and six completed cycles of adjuvant temozolomide, all of it tolerated well enough that she went back to running the front office of a pediatric dental practice within a month of finishing. The surveillance scan two weeks ago changed the countdown’s meaning: new, thicker enhancement along the old resection margin, with enough surrounding edema to explain the headaches and the word-finding slips she’d chalked up to being tired. Her outpatient oncologist started dexamethasone 4mg twice daily four days ago; the headaches are better, the word-finding isn’t, and her blood pressure — well-controlled for years on a low dose of lisinopril — has run in the 150s since the steroid started.
What bevacizumab actually offers her is a real but specific kind of benefit, and the literature is unusually clear about where the line sits. The phase 3 EORTC 26101 trial (Wick et al., 2017), which randomized patients at first glioblastoma recurrence to lomustine alone or lomustine plus bevacizumab, found progression-free survival more than doubled with the combination — 4.2 months versus 1.5 — while overall survival showed no difference between the arms. That gap is the entire debate: more scan-quiet time, reliably, without a proven change in how much time she has left. The same trial’s own quality-of-life substudy found no meaningful cost to that added time either, which cuts the other way — whatever those extra months look like, they don’t appear to come at the price of how the average patient feels day to day. What the trial can’t tell anyone is whether that average holds for a specific ten-week deadline that has nothing to do with a tumor marker.
At recurrence, with a date on the calendar
Seventy-one days from now, Naomi wants to be standing in a gymnasium, not lying in a hospital bed. The phase 3 trial that tested exactly this — EORTC 26101, bevacizumab plus lomustine against lomustine alone at first recurrence — found progression-free survival of 4.2 months against 1.5, essentially tripling the time before the next crisis point. That trial’s own quality-of-life data found no measurable cost to that added time either. For a patient whose actual goal has a calendar date on it rather than a survival curve, more predictable months before the next scan crisis isn’t a consolation prize next to overall survival — it may be the more relevant number.
If she’d told me her only goal was living as long as possible regardless of how those months went, I’d be having a completely different conversation with her about this drug.
You’re right that the PFS number is real and replicated — I won’t argue with the trial data. But I’d slow down before we let a secondary endpoint reframe a negative primary one. EORTC 26101’s actual primary endpoint was overall survival, and it was negative — no difference between the arms. That same trial recorded grade 3 to 5 adverse events in almost two-thirds of the combination group versus just over a third with lomustine alone: hypertension, proteinuria, wound-healing and bleeding complications, clotting risk. She is already showing an early hypertension signal on four days of dexamethasone alone. Layering a drug with a real, mechanism-based blood-pressure effect onto that isn’t a free option just because it’s approved for this indication.
Two things about that 4.2 months. The quality-of-life substudy you’re citing measured group averages across the whole trial population — it says nothing about whether a patient already trending hypertensive on steroids alone will be one of the ones who tolerates it that well. And the number itself belongs to bevacizumab plus lomustine, not to bevacizumab. The randomized trial that did run a monotherapy arm, BELOB, did not find the single agent the winner. If we give her the drug alone, we should say plainly that we are giving her a regimen neither of those trials tested.
You’re both arguing about a decision we don’t actually have to make in its most extreme form yet. The single most consistent finding across the bevacizumab literature, randomized and retrospective both, isn’t a survival number at all — it’s steroid-sparing, and EORTC 26101 measured it directly: of the half of its patients on corticosteroids at randomization, 23 percent in the bevacizumab arm came off them entirely, against 12 percent on lomustine alone. She has a real, live problem right now: her blood pressure is already moving on a dexamethasone course that’s four days old, not fourteen months old. If bevacizumab lets us bring that dose down, we’ve solved today’s problem regardless of who’s right about November. I’d start it as monotherapy, not add lomustine given her current performance status — and you’re right that this puts me off EORTC 26101’s actual regimen, which is precisely why I’m not promising her its progression-free number. I’m promising her a steroid taper. I’d set the reassessment at her first follow-up MRI in six to eight weeks — which happens to land close to the actual date on her refrigerator — rather than commit either of you to a position about survival we don’t need to settle today.
Agreed: bevacizumab monotherapy starts this week, with a formal reassessment at her first follow-up MRI in six to eight weeks — both for radiographic response and for whether the dexamethasone dose can safely come down, the more immediate problem the team actually agreed was driving today’s decision.
Not agreed, and named rather than settled: how the team should weigh a ten-week personal deadline against a trial literature that was never designed to measure it. The palliative care specialist wants that deadline written into the chart as a real clinical goal, on the theory that a treatment plan should openly serve what the patient is actually trying to reach, not just what a survival curve optimizes for. The clinical pharmacologist is uneasy naming a specific date as a treatment target at all, worried it quietly substitutes hope for evidence in a way that could make it harder to stop the drug later if it stops helping. Both sides of that disagreement are being told to Naomi directly rather than resolved for her first.