What the vitamin C pre-leukemia trial found

"Vitamin C may benefit people with pre-cancer blood disorders" ran across health news in September. The trial behind it is real, randomized and open access. It also missed the one goal it was designed to test. The survival number in the headlines came from an analysis added later. Here's how to hold both results at once.

The trial is called EVITA, and it was published in the journal Cancer on 21 September 2026 by a team from Copenhagen University Hospital, the Van Andel Institute and the University of Southern California. It's a careful study, and the press release was more careful than most. But the story most people carried away ("vitamin C helps people live longer") rests on the weakest kind of result a trial can produce. That's worth understanding, because the same pattern shows up in supplement research all the time.

Who was in it, and what they took

EVITA enrolled 109 adults between November 2017 and September 2022: 101 at three Danish hospitals and 8 at USC in Los Angeles. All had clonal cytopenia of undetermined significance (CCUS) or a lower-risk myeloid malignancy such as low-risk myelodysplastic syndrome (MDS) or early chronic myelomonocytic leukemia. In plain terms: their blood counts were low, their blood-forming cells carried leukemia-associated mutations, and the usual plan is to watch and wait.

They were randomized to 1,000 mg of oral vitamin C a day (two 500 mg capsules) or a matching placebo for 12 months, double-blind, then followed long-term. Anyone already taking vitamin C had to stop on entry. The rationale is real biology: vitamin C is a cofactor for TET enzymes, which help remove DNA methylation, and TET2 is one of the most commonly mutated genes in these conditions. In mice, oral vitamin C slowed leukemia progression.

The primary endpoint: no difference

A trial gets one pre-specified main question, and EVITA's was whether vitamin C slowed the growth of the mutated clones over the treatment year. It didn't. The median clonal growth rate was 0.046 per year on placebo and 0.030 on vitamin C, a between-group difference of −0.016 (95% CI −0.096 to 0.064; p = .70). In the paper's words, that "indicated no difference between treatment groups."

The authors go a step further in their discussion, which I respect. They now think clone growth over a single year is too slow and too variable to measure treatment effects in these patients, and write that "clonal growth rate is not a suitable surrogate end point for treatment efficacy in this patient population." That's an honest admission that the main yardstick was the wrong one. It also means the trial's designed test came back empty.

What the vitamin C did do

It fixed low levels. At baseline, 25 of the 109 patients were deficient (below 23 µmol/L) and another 37 were inadequate (23 to 49 µmol/L), so about 57% were below the adequate range. In the vitamin C group, median plasma levels rose from 45.85 to 81.90 µmol/L by 12 months, essentially saturation. Placebo stayed flat (43.75 to 48.73).

It shifted some inflammation markers. Of 19 cytokines and chemokines measured, six changed differently between groups after correction for multiple testing. IL-6, IL-10, CXCL10, M-CSF and G-CSF rose more on placebo. That's a biological-activity signal, not a clinical one.

Fewer serious adverse events. 18 of 55 patients (33%) on vitamin C had a serious adverse event, versus 30 of 53 (57%) on placebo. Pneumonia alone hit 8 placebo patients and 1 on vitamin C.

No clear effect on progression. Nine placebo patients and five on vitamin C progressed to a higher-risk disease, mostly acute myeloid leukemia. The hazard ratio was 0.51, but the confidence interval ran from 0.17 to 1.51 (p = .22). Not significant.

The survival number, and why it's a lead, not a finding

Over a median 33.6 months of follow-up, 24 of 54 placebo patients died versus 11 of 55 on vitamin C. The hazard ratio was 0.35 (95% CI 0.17 to 0.71; p = .0025). Median survival was 42.2 months on placebo and not yet reached on vitamin C. That is a big gap, and it's why the story spread.

Now the fine print, all of it from the paper itself. Overall survival "was analyzed post hoc as an exploratory end point; it was planned before the end of the study but not included in the protocol." The authors also write that the study "was neither powered nor designed to assess survival outcomes." Four things make me hold this number loosely:

1. It wasn't the question. When a trial misses its main endpoint and then finds a dramatic effect in an outcome added later, the odds that the second result is partly chance go up. It may be real. It's exactly the kind of result that needs a second trial built to test it.

2. The groups weren't perfectly matched. The placebo group was a bit older (median 75.0 vs 72.3 years) and more male (78% vs 65%), with more patients short on several blood cell types. The vitamin C group, on the other hand, had more mutations and worse prognostic scores. The authors report the survival signal held up in models adjusting for these imbalances, which helps. With 35 deaths in total, though, a handful of patients moving either way changes the picture.

3. Treatment stopped at 12 months; follow-up didn't. After the treatment year there were no restrictions on vitamin C. The authors flag "informal crossover" as a limitation: some placebo patients may have started taking it, and some vitamin C patients may have stopped.

4. Most deaths were infections. Bacterial infection was the leading cause of death: 14 of 24 on placebo, 5 of 11 on vitamin C. Combine that with the pneumonia and serious-event numbers and a plausible story is that vitamin C helped people with weak blood counts fend off infections, rather than slowing the leukemia itself. That's my reading, not the authors' conclusion, and it would still be a meaningful benefit. But it's a different claim from "vitamin C intercepts cancer," and the progression data don't yet support the second one.

To their credit, the researchers say all of this plainly. The paper calls the results "hypothesis‐generating" and says a phase 3 trial is warranted. In the press release, lead investigator Kirsten Grønbæk put it this way:

"It is too early to make definitive recommendations based on our results, but we are hopeful that a more expansive study will provide further insights."

The part that matters for everyone else

The finding I'd act on first isn't the survival curve. It's that more than half of these patients started the trial below the adequate vitamin C range, and the paper cites earlier reports that deficiency is common in blood cancers. Correcting a deficiency is a far more ordinary claim than treating pre-leukemia. It's also one any patient can raise with their hematologist today: ask whether your vitamin C level has been checked.

On safety, 1,000 mg a day sits within the adult Tolerable Upper Intake Level, which the NIH Office of Dietary Supplements lists in a range of 400 to 2,000 mg depending on age. Its most common complaints at high intakes are gastrointestinal: diarrhea, nausea and cramps. In EVITA, three patients on vitamin C and one on placebo stopped because of adverse events.

And a historical note the authors include: vitamin C has a long, disappointing record as a cancer treatment. Randomized trials decades ago "showed no efficacy of oral vitamin C in patients with advanced malignancies." The interesting idea in EVITA is timing: giving it early, in precursor conditions, when the biology may still be steerable. That's a sensible hypothesis. One phase 2 trial with a missed primary endpoint hasn't proven it.

How to read the next headline like this

Find the primary endpoint first. If it missed, everything after it is a lead, however striking.

Look for "post hoc" or "exploratory." Those words mean the analysis wasn't the trial's designed test.

Count the events, not just the percentages. A hazard ratio of 0.35 built on 35 deaths is fragile in a way the ratio alone hides.

Ask what the supplement might be fixing. If many participants were deficient, the benefit may come from correcting a shortfall, which tells you little about extra doses for people who aren't.

Not medical advice. This is educational analysis, not a recommendation — a study is not a prescription. Talk to a qualified clinician before acting on anything you read here. Full disclaimer →

Sources

S. U. Mikkelsen, A. Al-Mousawi, A. B. Puglisi et al. (corresponding authors K. Grønbæk and P. Jones), "Oral vitamin C supplementation in patients with clonal cytopenia of undetermined significance or lower-risk myeloid malignancies: Results from EVITA, a phase 2 randomized, placebo-controlled trial," Cancer 132(19):e70549 (published online 21 September 2026), DOI 10.1002/cncr.70549, ClinicalTrials.gov NCT03682029. (Primary source, open access, full text read via Europe PMC. Source of the enrollment, sites, dosing, baseline characteristics and vitamin C status, the primary endpoint result, the surrogate-endpoint admission, the cytokine, adverse-event, progression and survival results, causes of death, the limitations, the historical note on advanced-cancer trials, the authors' conflict-of-interest statement, and all quoted passages except the Grønbæk quote.)
Van Andel Institute, "Phase 2 Trial Suggests Vitamin C May Benefit People with Pre-Cancer Blood Disorders," press release via Newswise (21 September 2026). (Opened and read. Source of the headline framing and the quoted Grønbæk statement.)
NIH Office of Dietary Supplements, "Vitamin C: Fact Sheet for Health Professionals" (updated 31 July 2025). (Opened and read. Source of the 400 to 2,000 mg Upper Intake Level range and the gastrointestinal side effects.)

The paper's authors report consulting and grant relationships with pharmaceutical and research companies (listed in the paper's conflict-of-interest statement). Vitamin C is classified as a nutritional supplement, not a drug, by both the Danish Medicines Agency and the U.S. FDA, as the paper notes.

Disclosure, plainly: I founded and run Shroombiosis (a company I run), which formulates and sells supplements. That is a commercial stake in how supplement research gets read. Shroombiosis does not sell a vitamin C product, and I have no relationship with any author, institution or brand discussed here. Nothing here is sponsored and no link earns a commission; here's the full policy. A recommendation with no stake at all: for performance nutrition, Die Tryin Co. is a fellow combat-veteran-owned brand I'm glad to point people to. I don't own it and earn nothing from the link.