$1,115,605 for One Extra Hemoglobin Response. Pain Missed the Bar.

$1,115,605.

Divide 100 by the 38.0-point adjusted gap in hemoglobin response, multiply by thirteen cartons at the $32,610 wholesale acquisition cost on NDC 71334-0235-00, and the product is $1,115,605.26, which rounds to that dollar per extra responder over a full year of dosing. Lancet published RISE UP on October 24, 2026, with Biree Andemariam at UConn Health leading a 52-week comparison of mitapivat 100 mg twice a day against placebo: 207 people randomized, 138 on the drug and 69 on placebo, at 69 sites in 15 countries (ClinicalTrials.gov NCT05031780). A response, meaning the week-24-to-52 average sat at least 1.0 g/dL above baseline, occurred in 56 of 138 patients on mitapivat and 2 of 69 on placebo, an adjusted difference of 38.0 percentage points (95% CI 29 to 47, two-sided p<0.0001). Annualized pain crises, the co-primary, were 2.6 versus 3.1, a rate ratio printed as 0.9 and a 14% reduction whose interval runs from a 4.1% increase to a 29.2% decrease (p=0.1213). Pain missed.

What the trial was built to detect

That crisis test carried its own slice of type I error, 0.03 against 0.02 for the blood count, and later endpoints opened only after a primary had cleared. Methods planned 198 patients for 90% power to move the annualized rate from 3.0 on placebo to 1.95 on mitapivat, a drop of 1.05, or 35%, even after assuming 35% dropout and only 0.55 years of average follow-up. What arrived was a drop of 0.5 crises a year, which is 0.5 divided by 1.05, or 47.6% of the absolute effect in that plan, and 14 divided by 35, or 40% of the relative effect, while 120 of 138 patients on mitapivat (87.0%) and 56 of 69 on placebo (81.2%) stayed through all 52 weeks, so a short clock does not explain the miss. Agios designed the trial, analyzed it with the authors, paid for medical writing, and employs several authors who hold company stock, on a population whose median age was 25, whose hemoglobin started at 8.6 g/dL versus 8.5, whose prior-year crises were 3.4 in both arms, and of whom about three quarters were already taking hydroxyurea.

Thirteen cartons, and a strength that does not match

Check the carton. Aqvesme's label sets 100 mg twice a day for anemia in adults with alpha- or beta-thalassemia, puts 56 tablets in carton NDC 71334-235-00, and says to stop if hemolytic anemia has not improved, while sickle cell disease is not the indication. FDA's approval note and a prescriber start form both repeat one 100 mg tablet twice daily, quantity 56, and the 2026 prescribing information prints the same 56-count carton. Agios dates that thalassemia approval to December 23, 2025, and keeps Pyrukynd as the name for pyruvate kinase deficiency.

Colorado's January 8, 2026 disclosure lists NDC 71334-0235-00 at a wholesale acquisition cost of $32,610 and, on that same row, prints the strength as "Tablet 5 MG," which fights the label's 100 mg carton. Vermont's prescriber list, sourced to Medi-Span on January 8, 2026, and defined there as wholesale acquisition cost plus 20%, shows that NDC as Aqvesme 100 mg at a package average wholesale price of $39,132 and a unit price of $698.78571. Divide 39,132 by 1.2 and the result is 32,610. Divide 39,132 by 698.78571 and the result is 56.00 tablets. Fifty-two weeks at two tablets a day is 728 tablets, 728 divided by 56 is 13 cartons, and 13 times $32,610 is $423,930 of list price for the year the endpoints cover, if every dose is swallowed and nobody pays less than list. Pyrukynd's Colorado sheet from the same morning prices the 5 mg, 20 mg, and 50 mg bottles at $25,760 each, so assembling 200 mg a day from 50 mg tablets prices a different bottle. Do not swap them.

One hundred divided by 38.0, times $423,930, equals 42,393,000 divided by 38, which is $1,115,605.26, or $1,115,605 to the nearest dollar. Raw counts sit nearby: 56/138 minus 2/69 is 37.7 percentage points, and $423,930 divided by that share is $1,125,045 exactly, $9,440 more than the rounded adjusted figure, and both are catalogue prices rather than a rebate.

Two units, or twenty-four

Two units every month is 24 units a year, and the discussion calls the 1.6 g/dL mean rise among responders "comparable to a regular monthly transfusion of approximately two units of red blood cells." ASH's red-cell pocket guide puts one unit at about 1 g/dL in an average-size adult, which a sickle cell transfusion review repeats for adults (Tanhehco, Annals of Blood), so 1.6 g/dL is 1.6 units in a single sitting and 0.4 units short of two, which means "approximately" can describe a level and still fail once the claim is a monthly calendar. Ad-hoc means were 0.70 units on mitapivat and 1.59 on placebo, 97 units against 110, a recorded saving of 0.89 units per patient, so 0.89 divided by 24 is 3.7% and 24 divided by 0.89 is 26.97, on totals that were not filled in after dropout and that the hierarchy footnote refuses to call significant. Across everyone randomized, least-squares mean hemoglobin rose 0.77 g/dL on mitapivat and 0.03 on placebo, a difference of 0.7 g/dL that sits under the 1.0 line used to define a response. Check the units.

An AABB apheresis report still treats exchange, not a simple top-up, as the way to lower sickle hemoglobin without pushing hematocrit past about 30% (Sarode and colleagues, Transfusion), and RISE UP's report of no hyperviscosity events is not a viscosity reading. Indirect bilirubin fell by a least-squares 16.9 micromoles per liter (p<0.0001) and average hemoglobin is reported up at p<0.0001, after which fatigue's failure stops the chain, so hospitalizations at 1.6 versus 1.8 a year stay nominal (p=0.2498). PROMIS Fatigue 13a moved minus 2.7 points on mitapivat and minus 2.3 on placebo, a gap of minus 0.5 (95% CI minus 3.0 to 2.0, p=0.7112), or 0.5/4.1 = 12.2% of the within-person threshold set before unmasking. Ataga and colleagues, in the trial's summary of that 2020 review rather than a re-extraction here, found hemoglobin 0.4 to 0.9 g/dL lower before later cerebrovascular findings, albuminuria, higher estimated pulmonary pressure, or death, and the discussion admits those gaps do not prove a drug-made rise helps, pointing to voxelotor, which raised hemoglobin and was withdrawn after more crises and deaths than placebo.

What You Can Do

If you price this for a plan, score the pain primary as a miss and demand a net price, since $423,930 is catalogue list and Colorado already says discounts can move what a customer pays. If you treat sickle cell disease, remember that 56 of 138 people on drug cleared 1.0 g/dL, fatigue did not separate from placebo, and the prescribable label is adult thalassemia. If you referee the next paper, force a one-time unit increment and a monthly schedule into separate sentences. Watch an FDA filing, any crisis test locked before responders are split off, and the open-label extension the paper says is still running, and do not start an unapproved use from these tables.

Limitations

This analysis did not re-fit patient-level records, Agios published model rates rather than diary counts, and the 0.5-crisis gap is only a difference of those annualized rates. Net price is unpublished, so $32,610 remains list, and if Colorado's "5 MG" cell names a package other than the 100 mg carton, the thirteen-carton year has to be rebuilt, just as the ad-hoc unit totals have to be rebuilt if the missing weeks after dropout were large. Deaths are unused because Table 3 lists events leading to death in 2 of 138 and 1 of 69 while the discussion lists three and two, none judged related. CDC's data page puts the U.S. total near 100,000 from two 2010 studies and says the exact number is unknown, and this piece does not estimate how many would meet the trial's hemoglobin, crisis, transfusion, and age rules, or the label's adult limit. A thalassemia list price on sickle cell disease is a scenario, not an invoice.

The Strongest Case Against This Analysis

A fair objection lives in the paper's own post-hoc split and should be stated at full strength. Inside the mitapivat arm, responders versus non-responders had a crisis rate ratio of 0.7, a 26% reduction (95% CI 6.3 to 42.0), hospitalizations 34% lower, emergency visits 53% lower, and a fatigue change of minus 5.19 points, past the 4.1-point bar, against minus 2.55 in non-responders. If the benefit sits in the 56 whose count rose, averaging all 138 dilutes it, because 82 never cleared 1.0 g/dL, and a later count could decide who stays on the $423,930 year. That logic is real, and it is not randomized: the split was made on a post-treatment result, the paper says the study was not built to find baseline predictors, and a quieter year ahead could have been what produced the rise. Until a protocol locks the split in advance, or randomizes a run-in, 26% is a clue next to a pain primary that missed 0.05. Grant the clue.

The Bottom Line

The blood count moved and the pain primary did not: 56 of 138 versus 2 of 69 crossed 1.0 g/dL, an adjusted 38.0-point gap that costs $1,115,605 per extra response at the January 8, 2026 list price for a year of swallowed tablets, while crises fell 0.5 a year inside an interval that includes a small worsening, and fatigue moved 12.2% of the way to a 4.1-point bar. Among responders, 1.6 g/dL is about one and a half units given once, and the units actually recorded as saved are 0.89 per person, not 24. Carry the blood count. Leave the pain claim and the monthly-transfusion sentence until a pre-specified test, or a real price, says otherwise.

Related

• One Approval, 9,500 Diseases: The FDA Just Priced Gene Editing by the Platform, Not the Patient
• The $356,000 Question: What a Cure Has to Cost to Beat a Chronic Drug
• One HIV Shot, Two Prices: $28,218 a Year in America, $40 in Licensed Countries From 2027. Nine African Nations Planned About 1.37 Million Shots. Their Funders Secured 600,000 Across Every Country They Serve.