How Sponsors Calculated Prevalence for FDA Orphan Drug Designation and Where FDA Pushed Back

How sponsors calculated prevalence for FDA orphan drug designation, and where FDA pushed back

FDA's orphan drug designation depends on one number: fewer than 200,000 affected persons in the United States ⁴¹. The designation files for approved drugs that appear in Drugs@FDA review packages show that sponsors reached that number in a small set of repeatable ways. The same records show that FDA's Office of Orphan Products Development (OOPD) usually objected to how the sponsor defined the population, not to the arithmetic. The costliest disputes came from subsets that were too narrow, patient groups left out of the count, outdated denominators and claims data likely to undercount cases, and, for rare pediatric disease requests, the wrong age metric. In two pediatric-hypertension cases, designations granted in January 2013 and October 2015 were revoked on April 28, 2016, a few months before FDA approved Qbrelis (July 2016) and Epaned oral solution (September 2016).

The standard FDA applies

• Point prevalence of diagnosed patients. Prevalence is the total number of patients with the disease at a given time (point prevalence). Incidence, by contrast, counts newly diagnosed patients over a period such as a year ⁶⁴. For a therapeutic drug, FDA counts individuals in the United States who have been diagnosed with the disease or condition ⁴¹.
• Genetic conditions. When a genetic abnormality is present from birth but clinical disease appears later, or never, who counts as "diagnosed" depends on the intended use. The product may treat the abnormality itself, prevent or slow progression, or treat the manifestations ⁴¹.
• Vaccines, diagnostics and preventive drugs. These use a different denominator: the number of persons of all ages in the United States who will receive the drug each year ⁴¹.
• Timing. The estimate must be below 200,000 when the request is submitted ⁶³. FDA generally does not re-evaluate prevalence at NDA/BLA submission. It may revisit the determination if relevant information that existed at the time of the request was not provided or was not known to FDA ⁶².
• Documentation. Requests should include authoritative references, a list of sources, and the dates of the information provided ⁶³. 21 CFR 316.20(b)(8) requires information on the prevalence of the target population ²⁰⁵.
• Orphan subsets. For a disease affecting 200,000 or more persons, a drug qualifies only for a subset in which use outside the subset would be inappropriate because of a property of the drug, such as toxicity, mechanism of action or prior clinical experience ⁴⁸³³. Defining a narrower patient group is not enough on its own. The sponsor must show why the remaining patients are not appropriate candidates ⁴⁸.
• Grounds for deficiency or denial. FDA may issue a deficiency letter if information is missing, inaccurate or incomplete ⁶². It may deny a request if prevalence is 200,000 or more and there is no valid subset, if the evidence does not support the estimate or the subset, or if material facts are untrue or omitted ⁶⁵.

FDA's guidance on rare disease drug development makes a related point, though it expressly does not apply to designation requests. It says prevalence estimates vary with case definition, geography, and advances in diagnosis and treatment. It recommends citing individual current estimates rather than calculated averages ⁵⁶.

Methods sponsors used in approved rare-disease programs

Some records disclose only the result of the estimate. Hetlioz's sponsor cited 65,000 to 95,000 U.S. patients with non-24-hour sleep-wake disorder, without a documented derivation ¹⁴⁴. Sucraid's projection of 100 to 500 patients rested on one physician's referral experience, and it appeared in a post-approval letter rather than a documented designation calculation ¹⁰³.

Methods that drew FDA questions, refusals or revocations

1. Subsets defined by market or preference rather than by the drug

• Epaned (enalapril oral solution), pediatric hypertension. The sponsor proposed counting only children prescribed enalapril, particularly those aged 10 to 14 and younger who would need a liquid ²⁰. OOPD rejected this. The condition was pediatric hypertension as a whole, and a subset needs an intrinsic drug property that precludes use outside it. User preference for a liquid over tablets does not qualify ²⁰.
• Defitelio, hepatic VOD. The sponsor limited its population to post-HSCT VOD. OOPD said a narrower intended marketing indication cannot, on its own, restrict the population. Unless a product feature prevented broader use, the count had to include all hepatic VOD ⁷⁴. The sponsor's rationale was that post-HSCT VOD had the most favorable benefit-risk and that no data existed in other settings. It agreed to recalculate across the entire hepatic VOD population ⁷⁴.
• Ellence (epirubicin), Stage II node-positive and Stage III breast cancer. In an April 16, 1999 letter, FDA found the stage-defined group was not a medically plausible subset. FDA saw no evidence that epirubicin worked only in those stages or that those patients were uniquely responsive. It counted all patients who could benefit from adjuvant or palliative chemotherapy (Stages I to IV), about 1,993,000 ⁷. OOPD denied the request on July 26, 1999 ²⁰²¹⁴. The sponsor argued that adjuvant and metastatic disease are distinct clinical settings and cited prior orphan designations in metastatic breast cancer ¹⁹⁶. FDA's oncology and drug evaluation offices then agreed. In August 1999 minutes they described adjuvant therapy as distinct and the population as medically plausible, because the drug is used once, for less than a year, with use limited by cardiac toxicity ¹⁹⁹. FDA designated epirubicin on September 14, 1999 ¹⁹⁷ and confirmed seven-year exclusivity for the adjuvant, node-positive indication ²⁰¹. The subset argument succeeded once it rested on properties of the drug (cumulative cardiotoxicity, single use) rather than on labeling scope.

2. Leaving out patients the sponsor itself planned to treat

• Genotropin (somatropin), growth failure in children born small for gestational age. The sponsor's estimate excluded children treated during puberty, even though it described continued therapy through puberty as part of the expected treatment outcome ⁷⁸. On June 29, 2000, OOPD found that including pubertal children would push the population over the threshold. It held further review in abeyance, required a response within 90 days, and asked for supporting references with any resubmission ⁷⁸.

3. Outdated denominators and claims data likely to undercount (resulting in revocation)

The pediatric hypertension designations for Qbrelis (lisinopril oral solution) and Epaned are the clearest cautionary cases.

• An earlier lisinopril oral solution request (designation #12-3655; the holder's name is redacted in the review), submitted in February 2012 and covering primary hypertension with complications plus secondary hypertension, reported a prevalence of 196,453 and was granted in January 2015 ⁷⁶. Silvergate's own requests, dated July 2012 (enalapril) and May 2015 (lisinopril), led to designation #12-3767, granted January 30, 2013 ¹⁷, and designation #15-4819 (ages 0 to 16), granted October 14, 2015 ³.
• FDA consulted its Division of Pediatric and Maternal Health (DPMH) on the lisinopril estimate. DPMH identified four problems:
  • The sponsor used the 2010 Census pediatric population instead of 2014 data ⁶⁸.
  • No U.S. publication directly estimated how many children with primary hypertension need drug therapy after lifestyle measures fail ⁶⁷.
  • The sponsor relied on a cross-sectional, ICD-9 claims-based study that likely underdiagnosed hypertension and used a definition less rigorous than NHBPEP criteria ⁷⁰.
  • That study was limited to privately insured children at a single national insurer, so it may not represent U.S. children demographically or socioeconomically ⁷⁰.
• DPMH's independent estimate was 1,265,495 children with primary hypertension and 590,304 to 927,309 with secondary hypertension ⁷⁰. DPMH showed that even if only 15.8% of its 1,265,495 primary-hypertension estimate (which the review describes as obese U.S. children) needed drugs, the total would be 199,949, essentially at the threshold. Its clinical view was that the true share is far higher ⁶⁸.
• In a separate reconstruction, OOPD started from 70.8 million U.S. children aged 0 to 16, applied the 3.6% hypertension prevalence from Hansen et al., and then the 15.8% of those children who had a true hypertension diagnosis in the medical record, giving 402,710 diagnosed children ⁸³. Secondary hypertension alone came to 203,107 to 257,452 (or 208,201 using another Flynn estimate), which exceeds 200,000 before primary hypertension is counted ⁶.
• OOPD proposed revocation of both designations on April 8, 2016. Silvergate chose not to respond, and FDA revoked both designations on April 28, 2016 under 21 CFR 316.29(a)(3). Benefits were terminated because the designations had been improperly granted ³¹⁷. OOPD stressed that prevalence is measured as of the date of the request, using documentation available by then ⁷³.
• DPMH also recommended reviewing other pediatric hypertension designations granted after February 2014 and declining future requests unless there is conclusive evidence of prevalence below 200,000 ⁷⁰.

4. Unsupported assertions

• ChiRhoStim. The sponsor claimed that historical U.S. secretin use was below a certain annual number but supplied no source data ⁶⁹. The request was granted anyway, because even its maximum procedure-based estimate was under the threshold ⁶⁹. An unsupported figure did not sink this request, but it added nothing to it.

5. A disease plainly above the threshold

• Entocort EC (budesonide), Crohn's disease. OOPD denied the request because U.S. Crohn's disease prevalence was about 370,300 ¹¹. The record does not show a subset argument that could have saved it ¹¹.

Rare pediatric disease requests: a separate pediatric test

Rare pediatric disease designation (the basis for the priority review voucher) adds a second requirement: the disease must be serious or life-threatening, and its serious or life-threatening manifestations must primarily affect individuals aged 0 through 18. Under FDA's 2019 draft guidance, onset in childhood is not enough. FDA assesses which manifestations primarily affect children, weighing the timing and rate of progression, effects on growth and development, and whether the proportion of children with a manifestation exceeds the proportion of adults ²²⁰. The Cholbam, Defitelio and Vimizim reviews below (2014 to 2016) predate that guidance and weighed the age share of the affected population; the Defitelio reviewer, for example, required that more than 50% of cases be in people under 19 ²²¹. Read their percentages as decisions under that older reading, not as the current test. Several approved-drug records show sponsors getting this wrong.

• Cholbam. The sponsor relied on age at diagnosis. OOPD required the current age distribution of all diagnosed U.S. residents living with the disease. OOPD then obtained data showing that 69% were under 18 ¹⁸⁵.
• Defitelio. The sponsor relied on pediatric CIBMTR VOD rates of 9.2% (2008 to 2014) and 6.1% (2014), compared with 2.2% and 1.56% in adults, and on a 61.1% pediatric share in expanded-access Study 2006-05 ²¹². The chemotherapy-associated VOD figure of 79.7% pediatric used a cutoff of 16 years or younger, not the required 18 ²¹³. The reviewer also found problems with the underlying data. Extrapolating CIBMTR's research-level subset to all U.S. centers was inaccurate, because the share of patients selected for research varied by year ²¹⁵. The published incidence studies were mostly non-U.S. and used varying diagnostic criteria ²¹⁵. FDA concluded that hepatic VOD had not been shown to primarily affect people from birth through 18 and denied the voucher ^209211219.
• Vimizim. By contrast, registry, natural-history and Phase 3 data showing 64% to 80% of patients aged 18 or younger were accepted. So was reasoning from onset at ages 1 to 3 and shortened survival ⁷⁵.
• Exondys 51 (eteplirsen). FDA told the sponsor at a pre-NDA meeting that prevalence estimates are a critical element of any rare pediatric disease voucher request ⁸⁶.
• Post-approval reporting. Recent voucher approvals require sponsors to report, for each of the first four years after approval, estimates of the total U.S. disease population and the population aged 0 to 18, along with demand and actual distribution. Examples include Evrysdi ¹⁵⁸, Daybue ¹⁷⁰ and Xolremdi ¹⁶⁰.

Practical lessons from the record

1. Define the disease before the market. OOPD consistently counted the whole disease or condition unless a drug property limited use. A labeling strategy, a formulation preference or trial eligibility was not enough (Epaned, Defitelio, Ellence) ²⁰⁷⁴⁷. FDA's device-side guidance on HUD subsets lists trial eligibility, a planned narrow indication, unmet need, standard of care and price as insufficient on their own ⁴⁷. That guidance concerns devices, not orphan drugs.
2. Count everyone the drug is expected to treat. Leaving out planned treatment phases, such as puberty for Genotropin, brought a refusal ⁷⁸.
3. Use current denominators, and question sources that undercount. Outdated Census data and single-payer claims data were central to the pediatric hypertension revocations ⁶⁸⁷⁰³.
4. Match the metric to the condition. Point prevalence of diagnosed patients is the default ⁶⁴⁴¹. Annual counts apply to diagnostics, vaccines and preventive drugs ⁴¹. Acute conditions were handled with annual incidence in the Defitelio review ⁷⁴.
5. Show your work with dated, cited sources. FDA expects authoritative references with dates ⁶³. Unsupported figures (ChiRhoStim's historical-use claim) carried no weight ⁶⁹.
6. Leave margin under 200,000. Estimates close to the threshold, such as the 196,453 in the pediatric hypertension history, did not survive FDA's independent reanalysis ⁷⁶⁶. A designation granted on such an estimate can be revoked later if the evidence available at the time of the request shows the population was larger ⁶²³.

The Drugs@FDA review packages rarely reproduce the full designation request. For many orphan-designated products (for example, Darzalex, Uptravi and Spevigo), the record confirms designation but not how the sponsor estimated prevalence ^155111112. These cases therefore show patterns in how OOPD reasons. They are not a complete catalogue of accepted methods.