Twelve Years Without a Relapse, and a Question About Stopping
Twelve years on rituximab without a relapse or a lesion, and now two pneumonias and an IgG below the floor. The trial built for exactly this question came back ambiguous by design — it missed non-inferiority by half a percentage point, on an endpoint made mostly of lesions nobody felt.
M.C. taught school for three decades and now looks after two grandchildren under six several days a week, which she reports is harder. She is 67. Diagnosed with relapsing-remitting MS at 39, she spent sixteen years cycling through two disease-modifying therapies that never fully held her disease, and switched to rituximab — off-label, but well established — at 55. She remembers that as the point her MS stopped being something she planned her life around. Twelve years have followed without a relapse or a new lesion.
What brought the question to clinic is not her MS. It is two pneumonia hospitalizations in three years and an IgG of 540, below the lab's reference floor of 700, after more than a decade of continuous B-cell depletion. She wants to know whether twelve quiet years mean the drug is still the reason for the quiet — or whether her disease has done what MS does with age, becoming less inflammatory on its own regardless of what she takes.
DISCOMS was built for precisely this. Corboy and colleagues randomized 259 patients over 55, relapse-free five years and lesion-free three, to continue or stop. New disease events occurred in 4.7 percent continuing against 12.2 percent discontinuing — an absolute difference of 7.5 points against a prespecified non-inferiority margin of 8, which is to say the trial missed proving its case by half a percentage point. The composition of those events matters more than the gap. Of 22 events across the whole trial, 15 were one or two new MRI lesions with no relapse and no change in disability, and four patients in total had an acute relapse. Whichever way she decides, the evidence she is deciding on is a near-miss on a composite endpoint made largely of findings that only a scheduled scan would ever have seen.
Twelve years in, reviewing whether to stop
I'd discontinue, with close monitoring. In DISCOMS — Corboy and colleagues, Lancet Neurology 2023 — the primary event rate was 4.7 percent continuing versus 12.2 percent discontinuing over about 22 months, and the thing worth knowing is what those events were: of 22 events across 259 patients, 15 were one or two new MRI lesions with no relapse and no change in disability, and only four patients in the entire trial had an acute relapse, one of them in the continue arm. Inflammatory MS activity is independently known to decline with age. Twelve years of continuous B-cell depletion has a real, cumulative cost — two pneumonias and an IgG below the reference floor aren't nothing.
If she were fifty rather than sixty-seven, I wouldn't be making this argument the same way — the age-related decline in inflammatory activity DISCOMS's population specifically studied is doing real work in my reasoning here, not just her stability alone.
DISCOMS did not meet its prespecified non-inferiority margin — 8 percentage points, against an observed absolute difference of 7.5. That's a precise, important distinction: the trial didn't prove discontinuation is safe by the bar it set for itself, only that most patients who stopped did fine. And I'd add that the primary endpoint was a composite, relapse or a new or enlarging T2 lesion, not a clinical endpoint — so 'no difference in the clinical outcome' isn't quite what the trial reported either. Her disease has been silent for twelve years specifically on this drug.
I'm actually closer to your position than a flat disagreement suggests — the real event-rate difference in DISCOMS was itself small. I just don't think 'most patients did fine in an underpowered trial' is the same claim as 'proven safe,' and I'd want that distinction to actually matter for how confidently we counsel her.
I don't think either of you has actually proposed what DISCOMS's own finding argues for. You've both been arguing about whether 7.5 points clears an 8-point margin. But look at what the events were — almost all of them lesions found on a scheduled scan, not patients who felt anything. That is an argument for a monitored discontinuation with a clear, pre-agreed trigger for resuming therapy, not for an open-ended stop or an indefinite continuation. The trial's own failure mode is detectable by MRI before it costs her anything.
You're both reading the same trial carefully and neither of you is wrong about what it does and doesn't show. But 'discontinue' and 'continue unchanged' are both further from the actual data than a monitored middle path is, and I don't think that middle path has been given a fair hearing yet in this conversation.
Agreed: rituximab discontinued today, with MRI surveillance moved up to six and twelve months and an explicit, written trigger — any new relapse or any new lesion on either scan — for immediately resuming therapy rather than waiting for a scheduled follow-up to reconsider. IgG will be rechecked at six months given her recent infection history.
Not agreed: how the group would weigh a genuinely ambiguous surveillance MRI — a single small lesion of uncertain significance, rather than a clean new-activity or clean-negative result — which none of the three positions was built to resolve cleanly. Named directly as a real gap in today's plan rather than smoothed over, to be addressed if and when it actually arises rather than guessed at now.