Maralixibat: EXPAND and the limits of cross-disease extrapolation

Biology and disease heterogeneity

Maralixibat inhibits the ileal bile-acid transporter, interrupting enterohepatic recirculation. The rationale is reduction of the circulating bile-acid burden associated with cholestasis and pruritus. It does not mechanically correct biliary obstruction. Accordingly, a shared symptom of itch does not make biliary atresia, Alagille syndrome and every genetic cholestatic disorder the same therapeutic population. [1] [2] [6]

Analytical interpretation: available bile flow, transporter biology and residual liver function can determine how much substrate reaches the intestine. These factors make cross-disease extrapolation particularly uncertain. A successful EXPAND result would establish symptomatic benefit in its studied mixture; it would not demonstrate reversal of the underlying anatomical or genetic cause in every subgroup.

Dose and local pharmacology

EXPAND uses 285 micrograms/kg twice daily during a 20-week controlled period. This should not be substituted with the higher final dose tested in MARCH-PFIC. Drug salt conventions also matter: the corporate presentation distinguishes maralixibat from its chloride-equivalent dose. Unit consistency is essential before comparing studies or assuming equal delivered exposure. [1] [2] [4]

Its intestinal mechanism means gastrointestinal tolerability is directly relevant to adherence. Dose interruptions caused by diarrhea can alter both bile-acid effects and symptom measurement. A reduction in serum bile acids may support mechanism, but it is not a substitute for the prespecified pruritus endpoint and should not be converted into a survival claim without corresponding clinical evidence.

MARCH-PFIC randomized evidence

MARCH-PFIC randomized 93 children:47 active and 46 placebo. The primary analysis was a selected BSEP cohort of 14 versus 17, not all 93. Its adjusted itch difference favored treatment by 1.1 points. This distinction prevents a common reporting error in which the overall enrollment is presented as the primary endpoint denominator. [2] [6]

The all-PFIC and full cohorts answer broader secondary questions. INDIGO's earlier open-label experience also suggested genotype-dependent response, with bile-acid responses concentrated in nontruncating BSEP disease. These findings motivate careful biological stratification; they do not prove equal benefit in the additional conditions that EXPAND now studies.

EXPAND study and statistical reading

The trial includes patients aged six months or older with additional cholestatic conditions, including biliary atresia. Enrollment completion was announced in March 2026; the earlier design planned approximately 45 participants. That design target is not an exact completed randomized count, which should be taken from the eventual participant-flow table. [1] [3] [4]

The primary endpoint is change in itch severity through week 20, with serum bile acids and other disease markers as supporting measures. Caregiver-reported symptoms in very young children require consistent diary completion and rater understanding. Differences in sleep, concomitant antipruritic treatment and disease severity can influence observed scores without representing a direct drug effect.

The mixed-disease design should report subtype composition and prespecified stratification. A pooled mean can be driven by the largest or most responsive subgroup. Subgroup estimates will be imprecise in a small rare-disease study, so they should inform consistency rather than be selectively promoted as separate definitive successes after an overall miss.

Missing diaries after hospitalization, transplant or treatment discontinuation need explicit handling. These intercurrent events can correlate with both disease burden and adverse effects. A model that assumes missingness is benign should be tested with sensitivity analyses, especially when a few patients can materially alter the overall effect estimate.

Liver and nutritional safety

The known safety framework includes diarrhea, abdominal symptoms, liver-test changes and fat-soluble-vitamin deficiency. Some overlap with the underlying disease, requiring baseline values and longitudinal trajectories rather than isolated abnormal counts. Improved itch does not negate deteriorating bilirubin, impaired growth or nutritional deficiency in a child. [1] [2] [5]

The prior randomized diarrhea difference is clinically relevant even when events are mild or moderate. Display discontinuations and actual exposure, not only severity labels. EXPAND's lower dose and different etiologies mean its incidence cannot be predicted by copying the PFIC percentage; the older result is a safety precedent, not the new trial's outcome.

What the extension would establish

A persuasive result would connect a controlled itch improvement to tolerable sustained dosing across the enrolled disease mix. It would support a defined symptom indication rather than universal disease modification. Reporting the population boundaries and the negative or uncertain subgroup findings is essential to an honest extension of an already established mechanism. [1] [3]

Caregiver-reported itch deserves special scrutiny when gastrointestinal effects may reveal treatment assignment. That possibility does not invalidate the endpoint, but consistency with diary completion, sleep and biochemical measures strengthens interpretation. Children who undergo transplantation or stop treatment cannot simply vanish from the efficacy denominator. In a small heterogeneous trial, a few such outcomes can materially change the apparent average. The report should show those trajectories alongside the planned statistical estimate rather than rely only on a favorable completer analysis.

October 12: actual EXPAND results and analysis populations

The October 12 disclosure replaces the anticipated Q4 readout with observed results. The original mechanism, prior trials and pre-readout interpretation above remain intact. This randomized, double-blind, placebo-controlled study enrolled 52 participants, including 46 children and six adults. The primary analysis used the pediatric population, not all 52 patients. Pediatric analysis sets were 22 receiving maralixibat and 24 receiving placebo; the overall treated safety sets were 25 and 27. These denominators must not be interchanged when interpreting efficacy and harms. The intervention was 285 micrograms/kg twice daily during the 20-week blinded comparison. [7] [8]

The retrieved registry still describes estimated enrollment of 90 and dosing of 300 micrograms/kg, whereas the actual sponsor results use 52 participants and 285 micrograms/kg. These are retained as distinct source versions, not blended into a synthetic trial. The older report discusses salt-equivalent dosing, but this update does not independently reconcile every registry unit or amendment. The registry identifies the primary measure as change in the observer itch scale averaged over weeks 13–20. Neither its planned enrollment nor its longer open-label treatment description supplies an observed result or safety denominator.

Twenty-week safety: gastrointestinal burden remains visible

The sponsor reports no new safety signal, but that statement does not mean equal tolerability: diarrhea occurred in 16/25 treated patients versus 3/27 controls. Overall serious-event proportions were similar in this small trial; the study cannot rule out uncommon or delayed harms. The listed serious events in treated patients were pneumonia, pyrexia, large-intestine infection, acute kidney injury and viral infection with bronchiolitis, described as unrelated to treatment. The publication gives zero TEAEs leading to death in both groups. This is a specifically disclosed outcome, not an assumption that every unreported safety category was zero. [7]

What the topline establishes—and the remaining evidence

A lower ItchRO(Obs) score indicates less itch on the 0–4 scale. The 0.98-point placebo-adjusted change supports symptom improvement in the studied pediatric population. It is not a reduction in transplant risk or proof of correction of biliary obstruction. Days meeting the improvement definition are repeated observations within people, not independent responder counts. The bile-acid contrast was estimated after log transformation and back-transformation; subtracting −3.5% from −70.1% does not reproduce the reported −69.0% effect. [7] [8]

The company states that the primary and all five prespecified secondary endpoints met statistical significance. The supplied table uses MMRM for itch LS means, SEs and p values. The complete SAP, alpha allocation, missing-data assumptions, withdrawal disposition and confidence intervals were not independently extracted, so no additional multiplicity-adjusted or sensitivity claim is made. Native-liver biliary-atresia results are a prespecified subgroup within the same study, not an independent replication. Mirum plans an sNDA in the first half of 2027; this is neither a submitted application nor a new approval.

Sources / References

1. Mirum · EXPAND enrollment completion · March16,2026
2. Miethke et al. · MARCH-PFIC randomized trial · 2024
3. EXPAND · NCT06553768
4. Mirum · January2026 clinical presentation
5. MARCH-PFIC registered safety results
6. INDIGO long-term Phase2 study · 2022
7. Mirum · EXPAND efficacy and safety tables · October 12, 2026
8. ClinicalTrials.gov · NCT06553768 protocol snapshot retrieved October 19, 2026