Choosing the Opening Move in a New Diagnosis of Relapsing MS
Fourteen lesions, two relapses, an EDSS of 2.0 — and no treatment yet, which makes today the opening move rather than a correction. Start with the strongest drug, or start moderate and escalate when she breaks through? The registries favor the first. The trials built to settle it have not reported.
K.M. was highlighting a torts casebook four months ago when a gray smear opened across the center of her left visual field. Optic neuritis, confirmed on exam and MRI, resolved over several weeks with a short steroid course, and was filed at the time as an isolated event. Eight weeks later a band of numbness climbed from her waist to her ribs over two days and stayed a week. That second attack, anatomically separate from the first, met criteria for relapsing-remitting multiple sclerosis on its own. She is 27, in her second year of law school, and sitting for the bar in eighteen months.
The MRI from that second presentation is carrying more weight in this conversation than either relapse. Fourteen T2-hyperintense lesions is a substantial burden for a first real look at the disease, but the count is the least informative thing about it. Two sit infratentorially — one in the pons, one at the cervicomedullary junction — with a third in the cervical cord on a separate sequence. Those locations, not the total, are what the natural-history cohorts tie to a shorter interval before disability becomes permanent. Three of the fourteen enhance with gadolinium, meaning inflammation happening now rather than scar being counted twice.
So she arrives with three poor-prognosis features stacked independently: twin relapses four months apart, infratentorial and cord involvement, and an EDSS already at 2.0 before a single dose of anything. He and colleagues' 2020 cohort in Lancet Neurology compared early high-efficacy treatment against an escalation strategy and found less disability accumulated over ten years in the early-treated group, with the gap widest in patients whose baseline looked like hers.
What that study cannot do is randomize. Clinicians in it chose who got the potent drug, and the patients who looked worse are precisely the ones who got it — confounding that runs in the same direction as the finding rather than against it. The two trials designed to remove that problem, DELIVER-MS and TREAT-MS, have not reported their primary results. Not reported and come out neutral; simply not reported. Her decision is being made in the gap between an association that describes her well and a randomized answer that does not yet exist.
At diagnosis, before the first prescription is written
Start ocrelizumab now, not an injectable with a plan to escalate later. He and colleagues' 2020 Lancet Neurology cohort study compared patients started on early high-efficacy therapy against a matched escalation-strategy group and found meaningfully less disability accumulation over ten years in the early-treated cohort — and K.M.'s own MRI already carries two of the features that cohort's worse-outcome subgroup shared: infratentorial lesion involvement and active gadolinium enhancement at baseline.
If she'd presented with a single lesion and no enhancement, I would not be making this argument nearly as forcefully — this isn't a case for treating every new diagnosis maximally, it's specific to what her own scan already shows.
I'd start an interferon or glatiramer acetate and watch closely, and I want to be precise about why that isn't the same as being cautious for its own sake. DELIVER-MS and TREAT-MS — Ontaneda and colleagues' and Mowry and Newsome's trials respectively — are the two randomized studies actually built to test early-high-efficacy against escalation head to head, and I want to be careful about how I put this: neither has reported its primary result yet. Not reported and come out neutral. Simply not reported. DELIVER-MS is measuring 36-month brain volume loss, TREAT-MS time to sustained disability progression, and TREAT-MS only closed enrolment in 2025.
The registry comparison you're citing is a real, useful signal, but it's still observational — clinicians in that cohort were already choosing which patients got high-efficacy therapy, and sicker-looking patients at baseline are exactly who gets frontloaded in real practice. That's confounding by indication working in the same direction as the finding, not against it.
I don't think this needs to be settled as a general policy question at all, and I don't think it has to be to make today's decision. Two relapses four months apart, infratentorial and cord lesions on the same scan, and an EDSS of 2.0 before any treatment has even started — each of those is independently tied to faster disability progression in the natural-history literature, not just "a lot of lesions," which on its own is actually a weaker predictor than these specific findings are.
The trial-evidence gap you're both arguing about is real and neither of you is wrong about what it does and doesn't show yet. But her own numbers already sit inside the group both of those pending trials are specifically trying to identify — we don't need DELIVER-MS to finish enrolling to act on the prognostic markers that already exist for the patient actually in front of us.
Agreed: ocrelizumab, first infusion scheduled within the week, with baseline labs and the pre-treatment vaccination already in hand. Follow-up MRI planned at six months rather than the usual twelve, specifically to confirm the enhancing lesions have resolved and no new activity has appeared — an earlier check than routine practice, agreed to precisely because the decision rested on a baseline scan read as unusually active.
Not agreed, and named rather than smoothed over: whether this same reasoning should generalize to the next patient with a smaller lesion burden and no enhancement. The General Neurologist's position was accepted as the right general caution, not overruled — the group's actual basis for today's choice was the Neuroimmunologist's read of this patient's specific markers, not a house policy of high-efficacy-first for every new diagnosis.