Five Days of Steroids In, and Her Legs Are No Stronger
A full five-day course of high-dose IV steroids has moved nothing — her EDSS is exactly where it started. Plasma exchange has one randomized trial behind it in this precise situation, and that trial enrolled twenty-two patients. Trying steroids again has none. Neither option is well evidenced; only one has a clock attached.
Six weeks into training for a regional race, N.B. went from riding forty miles on a Saturday to needing support on both sides to cross a room. She is 29, a competitive amateur cyclist, and the race is now the least of it.
The relapse began eleven days ago, progressing over four days to bilateral lower-extremity weakness. She completed a full five-day course of IV methylprednisolone, 1g daily, finishing six days ago. Her strength has not moved. Her EDSS is 6.0 — bilateral support to walk twenty meters — unchanged from the day steroids were started, against a baseline of 1.5. Her MS has otherwise been well controlled on interferon beta-1a since her diagnosis at 25, which is part of why the failure to respond is being taken as seriously as it is — this is not a patient whose disease has been announcing itself.
That combination, a severe attack with no meaningful response to a completed corticosteroid course, is the exact clinical picture Weinshenker and colleagues' randomized, sham-controlled crossover trial was built around in 1999. Moderate or marked improvement followed 42 percent of active exchange courses against 6 percent of sham. It is the only randomized evidence in this scenario, and it rests on twenty-two patients in total.
What the trial does not address is timing, and what fills that gap is observational: Keegan and colleagues found time from attack onset among the predictors of who responds, with the response rate falling as initiation moves further out. She is at day eleven. A second steroid course, if it does what the first did, would put exchange somewhere past day sixteen — not past a cutoff, because there isn't one, but further down a slope that has already begun. Nothing about waiting is free, and nothing about proceeding is well proven.
Day 11, reviewing the lack of steroid response
I'd proceed directly to plasma exchange, not repeat steroids. Weinshenker and colleagues' 1999 trial in Annals of Neurology is the one piece of randomized, sham-controlled evidence we have in exactly her situation — acute severe attacks of CNS demyelinating disease that had already failed high-dose corticosteroids. Moderate or marked improvement followed 42 percent of active exchange courses against 6 percent of sham. She meets that entry criterion precisely: completed five days of IV methylprednisolone, day eleven from onset, EDSS unchanged at 6.0.
If she were three or four days out from onset and hadn't yet completed a full steroid course, I'd want that course finished first — it's specifically that she's already failed a completed course, eleven days in, that makes this the moment for PLEX rather than a repeat.
You're right that it's the only randomized trial in this exact scenario, and I'm not going to pretend otherwise — but I want to say what that trial actually is. Twenty-two patients. Not twenty-two per arm; twenty-two in total, across every demyelinating diagnosis, crossover design. That 42-versus-6 percent comes down to a difference of about seven patients. I take it seriously as the best evidence available. I don't take it as a settled effect size, and I notice we're about to put a central line in a twenty-nine-year-old on the strength of it.
And the comparison you're drawing isn't quite the one on the table. Weinshenker's control arm was sham exchange, not a second steroid course. So the trial tells us PLEX beats doing nothing in steroid-refractory attacks. It doesn't tell us PLEX beats trying more steroid — that comparison has never been run, which means the evidence asymmetry between our two positions is smaller than "randomized trial versus no trial" makes it sound.
That's a fair correction and I'll accept it — sham is not a repeat course, and I overstated the comparison. Where I still land differently is on what the delay costs. Keegan and colleagues looked at who actually responds to plasma exchange and found time from attack onset among the predictors, with response falling off the further out treatment starts. If we give her another five days of methylprednisolone and it does what the first five days did, we're initiating exchange somewhere past day sixteen, in the part of the window where the observational data say the response rate is already dropping.
You framed the line as a risk we're accepting on thin evidence. I'd frame the second steroid course the same way: it also has no trial behind it, and its cost isn't zero either — it's five more days of an unrecovered EDSS of 6.0 in a patient whose deficit may stop being reversible. Neither of us is choosing a risk-free option. We're choosing which uncertainty to spend her time on.
Then I'd rather lose this argument on the clock than on the evidence, because I think the clock is the honest reason. I'll agree to exchange starting today. What I won't agree to is treating the line as an afterthought — she needs the citrate and hypotension risks named to her directly, not folded into a consent form, and I want a plan for what we do at exchange five if nothing has moved.
Because that's the part neither of us has answered. Your trial tells us what happens when exchange works. It doesn't tell us when to stop if it doesn't, and I'd rather we decide that now than at the bedside on day six with a family in the room.
Agreed: plasma exchange initiated today, five to seven exchanges over the next ten to fourteen days, with a central line placed this afternoon and the citrate, hypotension and access risks discussed with her directly rather than through a form. What moved the Hospitalist was not the trial — he had already shown, correctly, that its comparison arm was sham rather than a second steroid course — but the timing argument: five more days of methylprednisolone would push exchange past day sixteen, into the part of the window where response rates fall.
Not agreed, and named rather than smoothed over: the stopping rule. Neither physician could point to evidence on when to abandon exchange if she hasn't moved by the fifth session, and both were uncomfortable for different reasons — the Neurologist unwilling to stop early in a patient who might still respond late, the Hospitalist unwilling to keep running an invasive procedure on a deficit that isn't budging. They agreed to set a reassessment point at exchange five and to make that decision with her rather than for her. The Hospitalist also recorded his view that the underlying trial is a twenty-two-patient crossover study and is being asked to carry more weight than its size supports.