Roche is filing an oral drug that matched its own best-seller in the one indication where Ocrevus has no competitor. Non-inferiority to Ocrevus in PPMS is a cannibalisation risk Roche has chosen to take, because the alternative is leaving the oral progressive market to someone else.

• Sanofi’s tolebrutinib is the cautionary comparator. FDA turned it down in non-relapsing secondary progressive MS at the end of 2025, after it had failed two relapsing trials. Roche arrives with positive relapsing and progressive data, the package Sanofi never had.
• Liver safety is the review’s central question. Fenebrutinib was placed on clinical hold after liver enzyme elevations in two FENhance patients, and FENtrepid recorded more transient elevations than Ocrevus. Expect a monitoring requirement on the label.
• FENtrepid’s 12% numerical reduction is not a superiority claim. Fatal events ran 1.4% against 0.2%, none attributed to treatment. Reviewers will still ask, and the label language on both points will shape how neurologists switch patients.

The company’s explanation is specific and testable: the Enroll-HD control set had more than 50% missing data by year four, and participants who stayed in the natural history study longer appeared to progress more slowly than those who left. Survivor bias in an external control flatters the control and shrinks the treatment effect, so the argument runs in uniQure’s favour if it holds.

• The filing rests on year three, not year four. FDA’s question is whether the four-year attenuation reads as waning durability or as a weakening comparator, and the agency has the patient-level data to decide.
• Functional capacity held while the composite weakened. A 61% slowing in daily functional capacity is the endpoint patients and payers care most about, and it is the one that stayed nominally significant.
• External controls are now the defining design question in Huntington’s. Skyhawk’s SKY-0515 posted a 1.59-point cUHDRS gap against a weighted external control in Week 38. Both programmes will be judged on how their comparators behave over time.

A missed primary, a significant subgroup and a registry survival comparison is a familiar rare disease filing, and it rarely clears FDA without a second trial. The subgroup is 83% of the study, which makes it harder to dismiss than a typical post hoc cut, but the 17% left out are precisely the patients not on standard therapy.

• NPC already has two approved drugs. Levacetylleucine and arimoclomol were both approved in 2024, so Trappsol Cyclo has to show benefit on top of an existing standard, which is partly what the background-therapy subgroup does.
• Hearing is the safety question cyclodextrins carry. Hearing-related events ran 15.6% against 13.3%, mostly mild or moderate, but one severe bilateral deafness case in a paediatric-heavy population will dominate the review.
• The survival analysis is the stronger card and the weaker evidence. An 85% mortality reduction against registry controls in infantile-onset NPC (HR 0.154, p=0.044) is large, but it rests on 41 treated patients across the programme against external comparators.

$94 million for an acute migraine Phase 3 is a statement that investors see room beyond gepants and triptans. A mechanism outside CGRP is the pitch to payers, because the CGRP class already covers most patients who respond to it.

• Competitive with marketed therapies is not a number. Phase 3 has to show two-hour pain freedom against placebo with a margin large enough to win formulary placement against generic triptans.
• Triptan-resistant patients are the commercial target. Nearly a third of the Phase 2b population was triptan-resistant, and a drug that works where triptans fail has a defined population rather than a crowded first line.
• Complementary rather than competitive with CGRP is the claim to watch. Distinct neuronal populations suggest combination use, which would expand the acute market rather than split it.

Buying a company mid-pivotal against the market leader is a bet on the trial, and Terumo has chosen to own the outcome rather than license it. A second approved embolic would turn MMA embolisation from a single-vendor procedure into a competitive one.

• 24 patients is a feasibility dataset. The resolution rates are strong but uncontrolled; RADIANT’s head-to-head design against Onyx decides whether NeoCast’s distal penetration claim translates into fewer recurrences.
• Pain-free injection is the practical differentiator. Arsenal designed NeoCast to be injected without pain under conscious sedation, which widens the procedure to patients who cannot have general anaesthesia.
• Neurovascular is consolidating around full-portfolio players. Terumo Neuro, formerly MicroVention, is adding a liquid embolic to compete in a procedure Medtronic currently anchors.

Europe is a harder test of the same evidence. HTA bodies, not only EMA, will judge a 17-patient external-control dataset, and national benefit assessments routinely discount uncontrolled comparisons. Authorisation may follow the US; reimbursement will take longer, country by country.

• EMA has a route built for this evidence shape. Exceptional circumstances exists for diseases where complete data cannot be obtained, as Nezglyal showed in cerebral adrenoleukodystrophy in Week 39.
• Treatment-centre readiness is again the bottleneck. A single AAV infusion with steroid cover and liver and platelet monitoring needs designated centres in each country, which is slower to build than one US network.
• The voucher economics do not travel. The US Priority Review Voucher has no European equivalent, so the EU launch has to stand on product revenue alone.

TBI has defeated decades of neuroprotection trials, and a Phase 3 IND is permission to try rather than evidence of efficacy. Heterogeneity, not mechanism, has sunk most TBI trials, so enrolment criteria and the treatment window will matter more than the molecule.

• Two TBI programmes advanced in one week. Pharmazz at Phase 3 and Astrocyte at Phase 2 suggest renewed investor interest in a field most large companies abandoned.
• Ask about the endpoint. The Glasgow Outcome Scale-Extended at six months is the standard and has defeated most predecessors; biomarker enrichment is the modern way around it.
• Small companies carry this field now. Neither sponsor has a large partner, which keeps the trials affordable and the readouts high-risk.

A dosing patent is lifecycle management, and a legitimate one in a rare paediatric disease where the dose scales with the child. Fifteen more years of exclusivity on the only approved Rett therapy sets the horizon every gene therapy in Rett now has to beat.

• Method patents are easier to design around than compound patents. A generic sponsor could seek a label that avoids the claimed dosing scheme, so the practical protection depends on how the claims read once issued.
• Gene therapy is the real competition. Taysha presented encore data for TSHA-102 in Rett at the Child Neurology Society meeting this week, and a one-time treatment changes the value of a chronic oral drug more than a generic would.
• Payers will notice the horizon. A 2041 date removes the expectation of near-term generic pricing for a chronic paediatric therapy.

A designation is positioning, not evidence. The voucher is the economic reason small companies chase this status, and with the programme now running to 2029 the incentive outlasts most Phase 1 timelines.

• Angelman is crowded at the antisense end. Several programmes aim to unsilence the paternal UBE3A allele with antisense oligonucleotides, so a new entrant needs a mechanism or delivery advantage to matter.
• The voucher extension changed the arithmetic. Egetis requested a voucher with Emcitate, approved this week, and every successful rare paediatric approval now mints a tradeable asset.
• Ask for the data behind the designation. Rare Pediatric Disease status requires only that the disease qualifies; it says nothing about MVX-220’s activity.

Endpoint alignment before Phase 3 is the right sequence in a disease with no regulatory precedent. With no approved DLB therapy there is no agreed endpoint to copy; cognition, fluctuations, neuropsychiatric symptoms and motor features all compete to be primary.

• The outcome matters more than the meeting. Watch whether Cognition discloses the agreed primary endpoint, because that defines the trial’s size and cost.
• Acadia’s Alzheimer’s psychosis miss is the neighbouring lesson. Neuropsychiatric endpoints in dementia carry large placebo responses, as RADIANT showed in Week 39.
• A small company running a dementia Phase 3 alone is rare. Expect partnering to be part of the Phase 3 plan.