A single rare-disease indication usually looks like one market and behaves like several. The diagnosis code is shared, but the therapy each patient can receive is not, because genotype, organ involvement, or antibody status splits the population into groups that cannot be added together. Sizing the indication as one number is the most common error in rare-disease forecasting, and among the most expensive, because it counts patients no single therapy can reach.
The split happens at the point of prescribing
In a common disease you can treat the indication as one addressable pool and refine later. In rare disease the split is upstream, written into the label and the biology, so it fixes who a given asset can treat before any commercial assumption is applied. A drug approved for one subgroup does not address the others, however large the shared diagnosis looks.
Three ways a rare indication splits
Genotype is the first. In Fabry disease the oral therapy migalastat is indicated only for patients whose specific genetic variant is amenable to it, an estimated 35 to 50 percent of the population, so a validated assay partitions the market before any sizing begins and leaves an enzyme-replacement-only remainder.
Organ and phenotype is the second. Hereditary ATTR amyloidosis splits by variant into different organs and different specialties: the Val122Ile variant is predominantly cardiac and sits in cardiology, while early-onset Val30Met is polyneuropathy-dominant and sits in neurology. The same label maps to two referral pathways, two sets of specialists, and two positioning problems.
Antibody status is the third. In generalized myasthenia gravis, roughly 80 to 85 percent of patients are AChR-antibody positive and about 5 percent MuSK-positive, and the newer complement and FcRn therapies are written against specific subsets, so serostatus rather than the diagnosis sets each mechanism's ceiling.
Adding them together is the expensive error
A blended indication number overstates every asset's opportunity, because it credits each therapy with patients it cannot clinically or legally treat. It also misdirects the commercial build, sending a team to the wrong KOLs, the wrong centres, and the wrong channel, because a cardiac amyloid market and a neuropathic one do not share a customer. And it corrupts the value case, since the HTA argument for an amenable-mutation oral is not the argument for an infused therapy in the non-amenable remainder. One number, applied to several markets, is wrong in every one.
Segment first, then size
The discipline reverses the usual order. You segment the indication at the point of prescribing, into the genotype, phenotype, and biomarker groups a therapy can actually reach, and you size each as its own funnel with its own diagnosis pathway, KOLs, channel, and value case. Only then do the numbers add up to something defensible, because each one describes a market a real patient can be treated in.
A shared diagnosis is not a shared market
The single-number instinct imported from common disease is a liability in rare disease, where the biology splits the market before the commercial team ever sees it. The forecasts that survive contact with the label are the ones that size the subgroups before the indication. This is one of the shifts behind the rare-disease sizing discipline: segment before you size, or the number counts patients no therapy can reach.
AXLRx sizes each treatable subgroup as its own market, by genotype, phenotype, and biomarker. See the Fabry and myasthenia gravis briefs, or commission a model scoped to your asset.