WHEN THE COMPARISON BREAKS: UNANCHORED INDIRECT COMPARISON IN A RARE PAEDIATRIC CANCER
Author(s)
Asma Jouini, Biology engineer1, Malgorzata Wojtal, MSc2, Lylia Chachoua, PharmD, PhD3, Mondher Toumi, MSc, PhD, MD4.
1Clever Access, Paris, Tunisia, 2Analyst, Clever Access, Kraków, Poland, 3Clever Access, Paris, France, 4Aix-Marseille University, Marseille, France.
1Clever Access, Paris, Tunisia, 2Analyst, Clever Access, Kraków, Poland, 3Clever Access, Paris, France, 4Aix-Marseille University, Marseille, France.
OBJECTIVES: Without a common comparator, indirect comparison rests on assumptions that rarity makes hard to satisfy. This analysis examined the unanchored comparison at the heart of the first orphan oncology JCA (tovorafenib for paediatric low-grade glioma) to show where, and why, it became fragile.
METHODS: A methodological case analysis examined the matching-adjusted and population-adjusted indirect comparisons in the published JCA report and dossier, focusing on the similarity assumption, the positivity/overlap requirement, prognostic-variable and effect-modifier adjustment, effective sample size, and safety-related exposure differences.
RESULTS: Because the single-arm pivotal study and its comparator shared no common control, the comparison was unanchored and could not test consistency. Its validity hinged on adjusting for every shared prognostic variable and effect modifier — so a single unmeasured variable differing across populations could bias the estimate by an unknown amount. The positivity/overlap precondition could not be fully verified, adding uncertainty to every derived estimate; matching reduced the effective sample size to roughly six; and on safety, differences in exposure duration could not be adjusted, with the direction of bias indeterminate. Most of this fragility was structural to a rare paediatric evidence base rather than a failure of execution, so the assessment's scepticism was aligned with good methodological practice rather than in excess of it
CONCLUSIONS: Unanchored comparison can fail on a single missing variable. Where rarity makes such comparisons unavoidable, the framework should specify in advance how their structural fragility is to be weighed rather than discovered case-by-case.
METHODS: A methodological case analysis examined the matching-adjusted and population-adjusted indirect comparisons in the published JCA report and dossier, focusing on the similarity assumption, the positivity/overlap requirement, prognostic-variable and effect-modifier adjustment, effective sample size, and safety-related exposure differences.
RESULTS: Because the single-arm pivotal study and its comparator shared no common control, the comparison was unanchored and could not test consistency. Its validity hinged on adjusting for every shared prognostic variable and effect modifier — so a single unmeasured variable differing across populations could bias the estimate by an unknown amount. The positivity/overlap precondition could not be fully verified, adding uncertainty to every derived estimate; matching reduced the effective sample size to roughly six; and on safety, differences in exposure duration could not be adjusted, with the direction of bias indeterminate. Most of this fragility was structural to a rare paediatric evidence base rather than a failure of execution, so the assessment's scepticism was aligned with good methodological practice rather than in excess of it
CONCLUSIONS: Unanchored comparison can fail on a single missing variable. Where rarity makes such comparisons unavoidable, the framework should specify in advance how their structural fragility is to be weighed rather than discovered case-by-case.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
HTA198
Topic
Health Service Delivery & Process of Care, Health Technology Assessment, Methodological & Statistical Research
Disease
Pediatrics, Rare & Orphan Diseases