A New Deficit, an MRI Six Hours Away, and Two Steroid Doses
A man with a rapidly progressing cord-compression exam and no imaging yet needs a steroid dose decided before anyone can see what they're treating. The disagreement isn't whether to start dexamethasone — it's whether an unstaged, worsening deficit justifies the higher historical dose or whether that dose's own real costs settled this years ago.
Walter K., 71, has entered his prize tea roses in the county fair every September for the past thirty years, and this year's blooms — he'll tell anyone who asks — are the best he's grown, which is exactly why the numbness that started in both feet three days ago made him wait far longer than he should have before mentioning it to his wife. He has metastatic castration-resistant prostate cancer with extensive vertebral and pelvic bone metastases, diagnosed four years ago, currently on his second-line systemic therapy after progressing through abiraterone. His most recent bone scan, six weeks ago, already showed increased uptake at L2 and L3 that his oncologist had flagged for close follow-up rather than immediate intervention, on the reasoning that his pain at the time was mild and his neurologic exam was entirely normal — a reasonable call at the time that this week's rapid progression has since overtaken.
Over three days, aching mid-back pain he'd blamed on a long afternoon kneeling among the roses progressed into bilateral leg weakness he could no longer talk himself out of, and this afternoon he had his first episode of urinary retention, requiring a straight catheterization in the emergency department. His exam shows 3/5 strength in both hip flexors and a sensory level at roughly T10 — findings consistent with cord compression severe enough to threaten permanent loss of function within hours, not days, which is the reason to treat now rather than wait, not a guess dressed up as one. The whole-spine MRI that would show exactly where and how severely is not available tonight; the only overnight slot at this hospital is committed to an emergent stroke code, and his own study won't run until the early-morning list opens in roughly six hours. He has had type 2 diabetes on basal-bolus insulin for eleven years, A1c last checked at 8.1% — a real, live consideration for whichever corticosteroid dose the team settles on, and one neither dosing study speaks to: Vecht's 1989 comparison and Heimdal's 1992 toxicity series both sorted patients by steroid dose, not by glycemic reserve, so the harm figures they produced describe a population his pancreas is not obviously part of. His wife, sitting beside the gurney, keeps returning to the same question: how did three days of numbness turn into this by tonight, and could anything have caught it sooner. Nobody in the room has an answer that changes what happens next, but the honest one is that his own bone scan six weeks ago already contained the warning nobody yet had reason to act on.
In the emergency department, before the MRI slot opens
Start dexamethasone at the higher end tonight. Sorensen et al., 1994, is the trial that actually established corticosteroids do something here — high-dose therapy against no steroid at all, with a real ambulatory benefit at the higher dose. He's worsening in front of us with no imaging to tell us how bad it already is. That's exactly the population that trial was built for.
Sorensen compared a dose to nothing. Vecht et al., 1989, compared dose to dose — 10mg against 100mg bolus — and found no significant difference in pain, ambulation, or bladder function. That's the efficacy half. The harm half is Heimdal et al., 1992, which is the study people usually misremember as part of Vecht: they loaded 96mg, saw serious complications in 14.3% of patients — a fatal ulcer bleed, a rectal bleed, two perforations — against none at standard dose, and abandoned the high-dose regimen outright. That's the comparison that actually answers tonight's question, and it's why guidelines moved to moderate dosing years ago.
The efficacy argument from Sorensen doesn't survive contact with a trial that directly compared the two doses to each other and found the higher one bought nothing.
And this isn't an abstract harm for him specifically. He's on basal-bolus insulin with an A1c of 8.1% already. A high steroid load is a real, predictable push toward severe hyperglycemia in a patient with that little glycemic reserve tonight, not a hypothetical side effect on a package insert.
You're both right that the steroid dose matters, and I don't think it's the thing that will actually determine his outcome tonight. Patchell et al.'s 2005 randomized trial found direct decompressive surgery plus radiotherapy produced better ambulatory outcomes than radiotherapy alone — and that advantage is time-sensitive, strongest when surgery happens before complete paralysis, not after. A six-hour wait for imaging isn't a scheduling inconvenience here, it's the variable most likely to cost him the window Patchell's data says matters most.
Start the moderate dose — I don't think tonight is the night to add steroid-related harm on top of an already unstable glucose picture — but let's page neurosurgery now and push hard for an expedited study rather than accepting the 6:00 slot as fixed.
Agreed: dexamethasone 10mg IV bolus followed by 16mg/day divided, insulin adjusted to frequent fingersticks, GI prophylaxis started, and neurosurgery and radiology paged immediately to push for an expedited study rather than waiting on the scheduled 06:00 slot.
Not agreed: the medical oncologist still wanted a single one-time higher bolus — a bridge dose, not a sustained high-dose regimen — given how much his exam had changed in just the last six hours before imaging even exists to confirm the extent. The clinical pharmacologist saw no trial evidence that a single higher bolus adds anything Vecht's comparison didn't already address, nor any reason to think Heimdal's toxicity signal spares a one-time dose, and didn't consider the diabetes risk trivial even as a one-time dose. Neither position was overruled; the moderate-dose plan proceeded, with the bridge-dose question left open pending what the morning's imaging actually shows.