Eight of 17 fibrotic livers improved on senolytic pills, and the interval still includes 1

Eight of 17.

On 1 October 2026, Koning, Meijnikman and colleagues reported a double-blind trial at Amsterdam UMC in which dasatinib plus quercetin, three days a week for three weeks and then four weeks off, repeated for three cycles, moved fibrosis by at least one Steatosis Activity Fibrosis stage without worsening of MASH in 8 of 17 people, against 1 of 14 on placebo, so 8/17 minus 1/14 equals 95/238, or 39.92 percentage points, which they print as 40, and 238 divided by 95 is a number needed to treat of 2.51, while (8/17) divided by (1/14) equals 112/17, or 6.59.

Nine events, one of them on placebo

Rebuild the interval on that 6.588235 ratio and the log is 1.8854, the standard error is the square root of (1/8 minus 1/17 plus 1/1 minus 1/14), which is 0.9974, and 1.96 times that half-width exponentiates to 0.933 and 46.53, rounding to the printed 0.93 and 46.53, so a ratio of 1 is still inside the band even though a two-sided Fisher test on the cells 8 and 9 against 1 and 13 equals 0.021, the P value they report.

One placebo success.

MASH resolution, 9 of 17 against that same single placebo case, has a relative-risk interval that just clears 1, from 1.06 to 51.64, while EPoS, a second fibrosis ruler on the same biopsies, does not: 8 of 17 against 2 of 14 is a gap of 78/238, or 32.77 points, and the paper's interval runs from minus 0.9 to 59.4 points, with P equal to 0.068. Score the four missing second biopsies as failures, which is what the primary analysis does, and a worst case that instead counts the missing placebo biopsy as a success drops the fibrosis relative risk from 6.59 to 112/34, or 3.29, again with P equal to 0.068, while complete cases of 8 of 14 against 1 of 13 give 7.43, the ratio in the supplement, which means nine histological events are carrying the headline.

Twenty-seven tablets, two pharmacy prices

Dosing days are 3 times 3 times 3, which is 27, and the methods name those tablets as Sprycel, from Bristol-Myers Squibb, at 100 mg, plus 1,000 mg of quercetin from Springfield, a nutraceutical supplier whose lot has no unit price in the paper, so the 27 grams of quercetin stay out of the dollars. Sprycel's label starts chronic-phase leukemia at 100 mg once a day and continues until progression or intolerance, which means 27 tablets are 27 days of that starting dose, not a leukemia course.

Amsterdam dispensed Sprycel, so $15,961.89 describes the trial's own pills at a US pharmacy acquisition survey rather than a Dutch hospital invoice, while $24.09013 is what that same file records for a generic 100 mg tablet, and neither figure is a copay or a price after rebate. Set the brand course against Madrigal's wholesale acquisition cost for Rezdiffra, $4,239.38 per 30 tablets of any marketed strength as of 1 October 2026, which is $50,872.56 for twelve packs. In the MAESTRO-NASH abstract, fibrosis improved by a stage without a worse activity score in 25.9 percent of 323 people on 100 mg and 14.2 percent of 321 on placebo, a gap of 11.7 points and an NNT of 8.55, so $50,872.56 divided by 0.117 lists one extra responder at $434,808.21 on a 360-day stack that is not this 27-day course, not this endpoint wording, and not this sample of 31.

Bigger trials already printed smaller gaps

ESSENCE reduced fibrosis without worsening steatohepatitis, at week 72, in 36.8 percent of 534 people on semaglutide 2.4 mg and 22.4 percent of 266 on placebo, difference 14.4 points, interval 7.5 to 21.3, an NNT of 6.94, or 13.3 at that floor. Tirzepatide 15 mg for 52 weeks, with missing biopsies filled in to follow placebo, put fibrosis improvement at 51 percent against 30 percent, difference 21 points, interval 1 to 42, so an NNT of 4.76 becomes 100 at the floor. FDA's approval note says two pathologists, in a trial that randomized 888 people, read scarring improvement at 24 to 28 percent on 100 mg against 13 to 15 percent on placebo. None of those comparison arms took a senolytic pill, which is the point of setting their gaps beside 39.92 points from 31 people. Younossi, Kalligeros and Henry put MASLD at 38 percent of adults, a different pool from the 6 to 8 million Americans FDA counted with NASH and moderate or advanced scarring, and this trial counted no deaths. Dulai's meta-analysis of 1,495 patients put the all-cause mortality rate ratio, versus no fibrosis, at 2.52 for stage 2 and 3.48 for stage 3, so one stage is not a fixed cut in the chance of dying.

What the same livers did not do

NAS improved in 12 of 17 against 7 of 14, interval minus 14.0 to 51.4 points, P equal to 0.42, and enzymes, FIB-4, ELF, stiffness and the controlled attenuation parameter did not separate the arms, while estimated GFR's 4.14 ml difference did not survive the authors' sensitivity checks. Platelets fell by 50.01 billion per liter, interval minus 90.81 to minus 9.22, with zero prespecified myelosuppression events, against any adverse event in 14 of 17 versus 6 of 14. Diabetes was 76.5 percent versus 35.7 percent, P equal to 0.033, which is 13 of 17 versus 5 of 14 because no other integer counts round to those percents, and the primary test did not adjust for it. Farr's 60-person bone trial of the same pair missed its resorption marker, P equal to 0.611. Hickson treated nine people for three days, Justice treated 14 without a placebo arm, and Nambiar randomized 12 for feasibility. James Kirkland and Tamar Tchkonia, authors, hold Mayo Clinic patents on senolytic drugs, disclosed in the competing-interest note, with NIH support under R37AG013925 and R33AG061456, while ZonMw grant 10140262110036 paid for the trial. ClinicalTrials.gov still lists 30 enrolled, age cap 65, and no results, while the paper randomized 31, writes ages 18 to 70, and ran from 18 July 2023 to February 2026.

What You Can Do

If you have fibrotic MASH, do not treat this paper as a prescription. Rezdiffra is the approved drug, on a one-year biopsy surrogate a 54-month study still has to confirm, and ESSENCE's fibrosis gap starts at 7.5 points, so ask a hepatologist which option fits the biopsy you have. If you are already taking dasatinib plus quercetin outside a trial, the harms counted here were 14 of 17 people with an adverse event and a platelet drop of about 50 billion per liter. Tell a physician. Do not schedule 27 days of a leukemia pill as a supplement. If you price senolytic companies, use the interval that includes 1, the EPoS miss, and the patent line. Nine events should not set the price.

Limitations

Published counts, not a patient-level file, are the whole input, and a plain normal approximation of the risk difference runs from 12.6 to 67.2 points rather than the paper's 8.2 to 64.5, so that band is a different method. NADAC on 11 November 2026 is a US acquisition survey, not an Amsterdam invoice and not a net price. Rezdiffra's $434,808.21 uses twelve list-price packs and rounded percentages. Quercetin is unpriced. Diabetes counts are inferred by rounding, not printed as 13 and 5. Pooled biopsies mean the gene-set scores are not one library per person, and no death or cirrhosis outcome was measured.

The Strongest Case Against This Analysis

Grant the optimistic reading at full strength: eight improvements against one, Fisher 0.021, a risk-difference interval that excludes zero, a resolution interval that just excludes 1, and in the treated pools a SenMayo score of minus 1.4 (P equal to 0.018) with a fibrosis gene-set score of minus 2.11 (P less than 0.001), arrived in 21 weeks in the arm that had more diabetes, and patents do not rewrite a blinded stage, so a short intermittent course can be exactly what the kidney and lung pilots were built to justify.

That reading can be true without being confirmed, because the primary relative-risk interval includes 1, EPoS misses, and enzymes, stiffness and the activity-score proportion do not carry the result, while one placebo biopsy separates 6.59 from a worst-case ratio the paper cannot keep significant. A lab that does not hold the Mayo patents should repeat the 8 and the 1, and until then 2.51 describes 31 people.

The Bottom Line

Amsterdam's 21-week trial improved a fibrosis stage in 8 of 17 people on dasatinib plus quercetin and in 1 of 14 on placebo, an exact gap of 95/238 and an NNT of 2.51, while the relative-risk interval still runs from 0.93 to 46.53. Twenty-seven Sprycel tablets cost $15,961.89 at the November NADAC, or $39,988.74 per extra response at that NNT, against a generic tablet price that cuts the same arithmetic to $1,629.51.

When a second fibrosis score inside the same paper includes zero, the first score has not finished the argument.

Related

• The First Company to Bet on Killing Zombie Cells Is Worth $0.06. Thirty Trials Still Think It's a $53 Billion Idea.
• Three Nature Papers. A 200% Lifespan Claim. Zero Published Survival Curves. Inside the First Trial to Rejuvenate Aging Immune Cells.
• A Single Shot Into Leg Muscle Made Old Mice Live 20.5% Longer. It Fixed Six Organs They Weren't Aiming At.