For 12 months, 109 adults with clonal cytopenia or lower-risk myeloid cancers took either 1,000 milligrams of vitamin C a day or a placebo. The supplement raised blood vitamin C levels into the saturated range and changed several inflammatory signals, but it did not produce the reduction in clonal growth that investigators had chosen as the trial’s primary measure of biological effect. An exploratory analysis also found longer overall survival in the vitamin C group, although that analysis was not a prespecified primary or secondary endpoint.
The EVITA trial was designed around an early stage of myeloid disease in which abnormal blood-forming cells can carry cancer-associated mutations without necessarily having progressed to advanced leukemia.
One group included people with clonal cytopenia of undetermined significance, or CCUS. In this condition, persistent unexplained low blood counts occur alongside mutations associated with myeloid malignancies, but the bone marrow does not yet meet the diagnostic criteria for a myeloid cancer. The trial also included patients with lower-risk myelodysplastic syndromes and certain lower-risk myelodysplastic/myeloproliferative neoplasms.
The investigators were interested in vitamin C because it serves as a cofactor for TET enzymes, which participate in DNA demethylation. TET2 is among the epigenetic regulators frequently mutated in myeloid disease. Earlier laboratory and animal work cited by the authors had suggested that vitamin C might influence the behavior of abnormal blood-forming stem cells, particularly when given before advanced disease developed.
That raised a more specific question than whether vitamin C could treat established advanced cancer: could supplementation alter the biology of an early myeloid clone?
EVITA was a double-blind, randomized, placebo-controlled phase 2 trial conducted at three Danish hospitals and one U.S. center. Between November 2017 and September 2022, 109 adults were enrolled. Fifty-five were assigned to receive oral vitamin C and 54 to receive placebo.
Participants took two 500-milligram capsules of vitamin C, or matching placebo, once daily for up to 12 months. They then entered long-term follow-up. During that follow-up period, vitamin C supplementation and other treatments were not restricted.
Participants were monitored with blood samples every three months. Bone marrow examinations were performed before treatment and at 12 months, with additional bone marrow sampling for research at three months.
The main test was whether abnormal clones grew more slowly
The primary endpoint was clonal growth rate.
The investigators measured mutations in blood-forming cells using targeted next-generation sequencing covering 151 genes associated with clonal hematopoiesis, myelodysplastic syndromes and acute myeloid leukemia. After predefined filtering, analyses were restricted to pathogenic variants in 38 genes. Sequencing had a median coverage of about 1,600-fold.
For each mutation, the researchers calculated the yearly growth rate from its variant allele frequency, or VAF, which represents the proportion of sequencing reads carrying the variant. The primary analysis then used the median growth rate across a patient’s detectable mutations.
The analysis included 98 patients who had both a follow-up sample and at least one detectable mutation at baseline. CD34-positive hematopoietic stem and progenitor cells were used when available, accounting for 74 of the 109 participants. Other analyses used peripheral blood or bone marrow granulocytes, peripheral blood mononuclear cells, or CD34-negative bone marrow cells when CD34-positive cells were unavailable.
The difference between the treatment groups was small.
The median clonal growth rate was 0.046 per year in the placebo group and 0.030 per year in the vitamin C group. The estimated difference was −0.016, with a 95% confidence interval from −0.096 to 0.064 and a p value of .70.
The corresponding median proportional increases in VAF were 4.6% per year in the placebo group and 3.03% per year in the vitamin C group.
The trial therefore did not demonstrate a difference in the primary endpoint.
The same general result held when the five most frequently mutated genes in the cohort were considered individually: TET2, SRSF2, DNMT3A, SF3B1 and ZRSR2. After correction for multiple comparisons, none showed a statistically significant treatment-group difference.
The authors note that the trial was designed as an exploratory phase 2 study and did not have a formal sample-size calculation because there were insufficient prior data to estimate the effect size for clonal growth. They therefore acknowledge that the study could have lacked power to detect modest differences.
They also caution against treating one year of clonal growth as a surrogate for clinical benefit. The authors point to evidence that clone dynamics in CCUS and lower-risk myeloid malignancies can be relatively stable over a year, vary according to the mutated gene and have a complex relationship with clinical outcomes.
Vitamin C levels changed sharply with supplementation
Although vitamin C did not alter the primary clonal-growth measurement, the treatment clearly changed vitamin C exposure.
At the beginning of the study, 62 of the 109 participants, or 57%, had plasma vitamin C concentrations classified as either deficient or inadequate. Deficiency was defined as less than 23 micromoles per liter, inadequate levels as 23 to 49 micromoles per liter, and adequate levels as at least 50 micromoles per liter.
The median plasma concentration in the vitamin C group rose from 45.85 micromoles per liter at baseline to 81.90 micromoles per liter after 12 months. The change was statistically significant, with p < .0001.
The authors describe the 12-month level as plasma saturation.
In the placebo group, median vitamin C concentration changed from 43.75 to 48.73 micromoles per liter, a difference that was not statistically significant.
Measurements from peripheral blood and bone marrow also closely tracked one another. Across baseline, three-month and 12-month measurements, the Spearman correlation was at least 0.79, with p < .0001 at every time point.
The concentration measurements also indicated that participants generally followed their assigned treatment, consistent with the high compliance reported by the participants.
Six inflammatory signals moved differently
The investigators next looked for biological effects that might not be captured by the clonal-growth endpoint.
They measured 19 cytokines and chemokines involved in inflammation, immune regulation and cell movement. Six showed significantly different changes between the vitamin C and placebo groups after correction for multiple testing.
Five of them increased more in the placebo group than in the vitamin C group.
For IL-6, the placebo group had a median increase corresponding to a fold change of 1.35, while the vitamin C group had a median decrease, with a fold change of 0.96. The false-discovery-rate-corrected p value was .0002.
IL-10 increased by a median fold change of 1.20 in the placebo group and decreased to a fold change of 0.93 in the vitamin C group, with an adjusted p value of .0011.
CXCL10 rose to a fold change of 1.16 with placebo and 0.99 with vitamin C, with an adjusted p value of .0032.
Macrophage colony-stimulating factor increased to a fold change of 1.19 in the placebo group compared with 1.03 in the vitamin C group, with an adjusted p value of .0001.
Granulocyte colony-stimulating factor increased to a fold change of 1.13 in the placebo group but decreased to 0.88 in the vitamin C group, with an adjusted p value below .0001.
RANTES/CCL5 moved in the opposite direction. Its median level decreased to a fold change of 0.96 in the placebo group but increased to 1.20 in the vitamin C group, with an adjusted p value of .023.
The investigators interpret this pattern as evidence of biological activity and note that several of the cytokine changes are consistent with profiles they previously associated with lower-risk myeloid disease and healthy controls. They also discuss experimental evidence linking inflammatory signaling, particularly IL-6, with expansion of blood-forming stem cells carrying TET2 or DNMT3A mutations.
But the trial did not establish that the cytokine changes caused slower disease progression or longer survival. The authors describe the inflammatory findings as supporting a possible modulatory effect of vitamin C on pathways involved in clonal expansion and disease progression. Whether the changes reflect epigenetic alterations in particular immune-cell populations remains under investigation.
The expected global epigenetic change was not seen
Because vitamin C can support TET enzyme activity, the investigators also examined global DNA methylation and hydroxymethylation in CD34-positive hematopoietic stem and progenitor cells.
At baseline, participants carrying TET2 or IDH1/2 mutations had a lower global 5-hydroxymethylcytosine-to-5-methylcytosine ratio than other participants. The mean difference was 0.0029, with a 95% confidence interval of 0.0018 to 0.0039 and an FDR-corrected p value below .0001.
However, treatment with vitamin C did not produce an overall difference from placebo in either global 5-methylcytosine or the 5-hydroxymethylcytosine-to-5-methylcytosine ratio.
The estimated between-group difference in the change in global 5-methylcytosine was −0.0004, with a 95% confidence interval from −0.0014 to 0.0005 and p = .36.
For the 5-hydroxymethylcytosine-to-5-methylcytosine ratio, the estimated difference was −0.0001, with a 95% confidence interval from −0.0006 to 0.0004 and p = .77.
One subgroup did show a signal: patients with CCUS who received vitamin C had a greater reduction in global 5-methylcytosine than those receiving placebo. The authors did not report a similarly clear difference across the other examined subgroups.
Disease progression was numerically lower but not statistically different
The investigators also tracked whether participants progressed to higher-risk myeloid malignancies.
During a median follow-up of 33.6 months in the intention-to-treat population, 14 progression events were recorded. Nine occurred in the placebo group and five in the vitamin C group.
Seven placebo-group patients progressed to acute myeloid leukemia, one to CMML-2 and one to myelodysplastic syndrome with excess blasts. In the vitamin C group, three progressed to acute myeloid leukemia and two to myelodysplastic syndrome with excess blasts.
A competing-risks analysis, which treated death as a competing event, estimated a hazard ratio of 0.51 for progression in the vitamin C group compared with placebo. Its 95% confidence interval ranged from 0.17 to 1.51, and p = .22.
Thus, the trial did not demonstrate a statistically significant difference in progression.
The researchers also found no difference between groups in hematological improvement or hematological progression.
Fewer adverse events were recorded in the vitamin C group
During treatment, the vitamin C group also had fewer recorded adverse events and serious adverse events.
The safety analysis included 53 placebo-treated participants and 55 who received vitamin C. There were 82 adverse events among 32 placebo patients, compared with 54 events among 24 vitamin C patients.
Serious adverse events occurred in 30 placebo patients, accounting for 66 events, compared with 18 vitamin C patients, accounting for 39 events.
That corresponded to serious adverse events in 57% of placebo-treated patients and 33% of vitamin C-treated patients.
After accounting for differences in treatment duration, the exploratory exposure-adjusted incidence rate for adverse events was 14.2 per 100 patient-months in the placebo group and 9.1 per 100 patient-months in the vitamin C group. For serious adverse events, the corresponding rates were 11.4 and 6.6 per 100 patient-months.
Some individual events also differed between groups. Pneumonia was reported in eight placebo patients, or 15%, compared with one vitamin C patient, or 2%. Anemia occurred in eight placebo patients and four vitamin C patients. Acute aseptic arthritis and internal bleeding were each reported in three placebo patients and none in the vitamin C group.
Some gastrointestinal events were more frequent with vitamin C. Diarrhea occurred in three vitamin C patients compared with one placebo patient, esophagitis occurred in three vitamin C patients and none receiving placebo, and gastrointestinal infections occurred in two vitamin C patients and none in the placebo group.
One participant receiving vitamin C developed kidney stones requiring intervention. That patient had experienced kidney stones more than a year earlier.
Two deaths occurred during study treatment in the placebo group, both attributed to disease-related causes. No deaths occurred in the vitamin C group during the treatment period.
The study was not designed to establish that vitamin C caused the difference in adverse-event rates, and the safety findings were not the primary focus of the trial.
Survival produced the most striking exploratory signal
The most pronounced difference emerged after the investigators examined overall survival as an exploratory outcome.
Overall survival had been planned for analysis before the end of the study, but it was not included in the original trial protocol as an endpoint. The analysis was therefore post hoc and exploratory.
By the end of follow-up, 35 deaths had occurred: 24 among placebo-treated patients and 11 among those assigned to vitamin C.
Median overall survival was 42.2 months in the placebo group. It had not been reached in the vitamin C group at the time of analysis.
The unadjusted Cox model produced a hazard ratio of 0.35, with a 95% confidence interval from 0.17 to 0.71 and p = .0025.
The investigators also examined whether differences in baseline characteristics could account for the result. The treatment groups were not perfectly balanced at baseline. The placebo group had a slightly higher median age, a greater proportion of men, more patients with multilineage cytopenia and lower blood counts. The vitamin C group, meanwhile, had a higher mutational burden and poorer disease prognostic scores on some measures.
After adjustment for age, sex, baseline hemoglobin and CCUS diagnosis, assignment to vitamin C remained associated with lower all-cause mortality. The adjusted hazard ratio was 0.42, with a 95% confidence interval from 0.20 to 0.88 and p = .021.
The authors note that the shorter survival observed in the placebo group compared with some published estimates could partly reflect the fact that EVITA included both newly diagnosed and previously diagnosed patients and measured survival from randomization rather than from diagnosis. They also calculated that the estimated median age at death in the placebo group was approximately 78.5 years, which they said was close to the 79.9-year life expectancy for Danish men.
The researchers did not find evidence that baseline vitamin C concentration substantially changed the treatment association. Increasing age appeared to attenuate the association between vitamin C treatment and longer overall survival, while the other examined variables did not show evidence of modifying the association.
Higher vitamin C levels were also associated with longer survival
The investigators conducted another exploratory analysis across the entire cohort, regardless of treatment assignment.
Because vitamin C is present in both the supplement and placebo groups as a dietary nutrient, they examined whether each participant’s median plasma vitamin C concentration during the treatment period was associated with subsequent overall survival.
After adjustment for age, sex, baseline hemoglobin and CCUS diagnosis, higher plasma vitamin C concentration was associated with longer overall survival. The reported hazard ratio was 0.45, with a 95% confidence interval from 0.21 to 0.95.
The modeled relationship was nonlinear. At lower vitamin C concentrations, relatively small increases were associated with larger predicted changes in five-year survival probability, while increases at higher concentrations were associated with more modest changes.
This analysis was observational within the randomized trial rather than a randomized comparison of different blood vitamin C concentrations. It therefore does not establish that increasing vitamin C concentration caused the survival association.
The trial’s limits shape how the results can be interpreted
The investigators identify several limitations.
The study was relatively small, with 109 randomized participants, and no formal sample-size calculation was performed. Because of that, the trial may not have had sufficient statistical power to detect modest effects on its primary endpoint.
Treatment lasted for only 12 months. After treatment ended, participants were no longer restricted from taking vitamin C. The authors therefore note that undocumented post-treatment supplementation could have produced informal crossover and potentially affected long-term outcomes.
The researchers also emphasize that clonal growth over one year may not be an appropriate surrogate for treatment efficacy in this population. In EVITA, median proportional increases in VAF were 4.6% per year in the placebo group and 3.03% per year in the vitamin C group. The authors note that these rates are comparable to approximately 4% annual increases reported in recent prospective studies and argue that clone growth has a complex relationship with clinical outcomes.
The choice of clonal growth as the primary endpoint reflected the evidence available when EVITA was designed, when mutation presence and clone size were more widely considered to be closely linked with clinical outcomes.
The investigators also deliberately used oral rather than high-dose intravenous vitamin C. Their goal was to test vitamin C as a cofactor for TET2-related epigenetic activity rather than to test potential cytotoxic effects associated with very high intravenous doses. They hypothesized that 1,000 milligrams taken orally each day would bring plasma concentrations to approximately 70 to 80 micromoles per liter, a range they considered sufficient to enhance epigenetic activity based on pharmacokinetic and preclinical evidence.
That target was reached in the trial, but the expected overall change in global DNA methylation or hydroxymethylation was not observed.
The authors also point to a previous single-arm phase 2 study of 10 people with TET2-mutated CCUS who received high-dose intravenous vitamin C. That study found no hematological responses at 20 weeks and largely stable TET2-mutated VAFs at 20 and 52 weeks. Another nonrandomized study involving members of a family carrying a germline TET2 loss-of-function mutation reported reversal of hypermethylation in enhancer regions after 12 months of oral vitamin C, although that study involved only six relatives.
Within EVITA itself, the clearest randomized biological changes were the rise in plasma vitamin C and the differences in cytokine and chemokine trajectories. The primary measure of clonal growth did not differ, global methylation measures did not show an overall treatment effect, and progression rates were not statistically different.
The survival finding, while statistically significant and still present after adjustment for several baseline variables, came from an exploratory post hoc analysis. The investigators therefore describe it as a signal rather than a definitive clinical endpoint and state that the findings provide a rationale for a future phase 3 trial.
The study was published in Cancer.
