SMA Therapy Selection: Nusinersen, Risdiplam, or Onasemnogene Abeparvovec
A five-month-old girl newly diagnosed with spinal muscular atrophy needs a disease-modifying therapy started before more motor neurons are lost. All three approved agents work — the disagreement is which one, and no trial has ever put them head-to-head.
M.R. is five months old, the second child of parents who run a small dairy farm an hour outside town, and had been an easy, alert baby until about six weeks ago, when her mother noticed she had stopped trying to roll from her back to her side — something her older brother had done weeks earlier at the same age. A well-child visit three weeks later found a weak cry, a bell-shaped chest, and tongue fasciculations, and sent her to neurology the same afternoon. Genetic testing confirmed a homozygous SMN1 deletion with two copies of SMN2, and no c.859G>C modifier — the variant that explains most of the minority of two-copy infants who track milder than their genotype predicts. She does not have the thing that would let anyone hope this runs slower than it looks.
Two numbers in her chart are doing more work than they appear to. The first is her weight: 6.1 kilograms sits near the tenth percentile for a five-month-old girl, and although her chart records her as feeding adequately with no aspiration signs, weight faltering in SMA type 1 characteristically precedes any visible bulbar failure — so "feeding adequately" describes a snapshot, not a trend, and the trend is the part that matters. The second is the six weeks. ENDEAR, which established nusinersen in symptomatic infants, found its motor-milestone benefit concentrated in the children with the shortest disease duration at enrollment; M.R.'s symptoms began six weeks ago and her respiratory function is intact on room air, which places her at the favorable edge of that trial's own population rather than at its median. She is, in other words, close to the best version of a symptomatic infant that any of these trials actually studied.
That is the argument for moving now, and it does not by itself settle which drug. All three restore functional SMN protein, none has ever been compared head-to-head, and the matching-adjusted indirect comparisons favoring risdiplam and gene therapy over nusinersen are confounded by enrollment era: FIREFISH and STR1VE recruited years after ENDEAR closed, into better supportive airway and nutritional care, with patients treated younger on average. What can be said precisely is what would foreclose an option. At five months she is well inside onasemnogene's labeled window, which turns on age — under two years — and not on weight, leaving her better than a year and a half of nominal eligibility. The constraint that could actually close today is the anti-AAV9 titer drawn this morning: above 1:50 the drug has simply never been studied, and roughly one in eight screened US infants crosses that line, with the highest rates in the youngest, where the antibody is maternally transferred.
At the family meeting, three days after diagnosis
Onasemnogene abeparvovec today, or as close to today as we can get her scheduled. And I want to be precise about why, because it isn't the reason people usually give. The label's ceiling is age — under two years — so she has well over a year in hand, and there is no labeled weight cutoff she is about to outgrow. The thing that can actually take this option away is the anti-AAV9 antibody titer we drew today and don't have back. The label requires a titer at or below 1:50; above that, gene therapy is off the table entirely, because it has never been studied there. AAV9 is a common childhood virus and antibody is usually maternally transferred, so roughly one in eight infants screened for this drug has an elevated titer, and it's commonest in the youngest ones. RESTORE, the ongoing real-world registry following children treated with this agent, now has patients several years out still holding the motor gains they made in the first year — this isn't just the pivotal trial's short follow-up anymore, it's a real durability signal.
I'd argue the same urgency for nusinersen or risdiplam if either carried an eligibility criterion that could come back disqualifying on a lab we've already sent. The point isn't that gene therapy is intrinsically the better drug — it's that only one of these three options has a pending result that can remove it from the table entirely.
You're right that the eligibility window is real, and that we shouldn't let a titer we haven't even gotten back make this decision for us by default. But I'd start risdiplam today regardless of what that titer shows, and keep gene therapy on the table as a later option rather than the other way around. She's five months old with a fragile respiratory and nutritional status — an oral, once-daily suspension her mother can give at home avoids both an intrathecal procedure under sedation and a systemic AAV infusion with its own hepatotoxicity-monitoring burden, at a point where every additional procedure carries real risk. FIREFISH's own infant cohort showed motor-milestone gains at this age comparable to what's been reported for gene therapy, and risdiplam reaches tissue nusinersen's intrathecal delivery doesn't — cardiac and GI SMN-deficiency effects are increasingly recognized in SMA1 even when the motor picture looks stable.
The RESTORE durability argument cuts both ways, too. A few years of follow-up on a one-time, irreversible treatment is reassuring, but it's also all the data that will ever exist before children are already committed to it. Risdiplam's exposure is titratable and reversible if something unexpected turns up; a transgene isn't.
I don't think either position needs the indirect comparisons doing as much work as they're being asked to do. FIREFISH and STR1VE enrolled years after ENDEAR closed, into a world with earlier diagnosis and better supportive care than nusinersen's own pivotal trial had — an apparent edge for the newer agents could just as easily be enrollment era as drug effect, and I don't think we can honestly tell from the published data which one it is. What I actually know is that nusinersen has the longest continuous track record of the three — SHINE's extension cohort is now past seven years — and for a five-month-old, that isn't a small thing to set aside for a comparison neither of you can really defend as clean.
None of that is an argument against acting quickly. I agree completely that whichever agent we choose should start now, not after further deliberation — it's specifically an argument for the agent with the most years actually watched, not the one the indirect data numerically favor.
Agreed within the visit: start risdiplam today rather than wait on the antibody titer, both because M.R.'s motor unit loss is ongoing and because nothing about starting it forecloses gene therapy afterward if the titer comes back permissive.
the family will be offered onasemnogene abeparvovec as a planned addition, not a replacement — the geneticist's argument for acting inside the eligibility window remains live, just resequenced behind today's decision rather than ahead of it.
risdiplam continues alone, and nusinersen becomes the next option to revisit if her motor trajectory plateaus — gene therapy will no longer be available to her at that point.