Azacitidine–Venetoclax or Induction Chemotherapy for Acute Myeloid Leukemia
Peer-reviewed randomized phase 2 trial; NEJM 2026, 2026 · paper · Clinical trials
A randomized phase 2 trial in 172 chemotherapy-eligible adults with acute myeloid leukemia, comparing azacitidine–venetoclax with induction chemotherapy. Median event-free survival was 14.5 versus 6.2 months (hazard ratio 0.57).
2 findings
Substance Text: "The incidence of hematopoietic-cell transplantation was 60% (95% CI, 53 to 68) in the azacitidine–venetoclax group"; the I bar on the azacitidine–venetoclax bar in Figure 3C spans roughly 50 to 71, so the text interval and the plotted interval for the same quantity do not agree.
- p. 8 Results, Secondary End Points
cell transplantation was 60% (95% CI, 53 to 68)
- p. 8 Results, Secondary End Points
(95% CI, 32 to 47) in the induction chemothera-
- p. 8 Results, Secondary End Points (comparison interval)
in 62% of those (95% CI, 50 to 72)
- p. 9 Figure 3C, I bar on azacitidine–venetoclax bar
Hematopoietic-Cell Transplantation
Only the authors can say what is right here.
Judge's note: Measured the Figure 3C crop against its axis ticks (100 and 50, about 2.56 px per percentage point). The azacitidine–venetoclax bar is at about 61 with an I bar from about 50 to about 71. The induction bar is at about 40 with an I bar from about 33 to 47, which agrees with the text's 32 to 47. So the mismatch is in the azacitidine–venetoclax arm only; the gap of about 3 points at each end is about 8 px and clearly outside reading error. The text interval is 15 points wide, while other N=86 proportions in the same paragraph are about 22 points wide (62%, 50 to 72). The article does not state the interval method for this end point, and the induction interval is also narrower than those neighbours, so I cannot say which of text or figure is intended. This is a query for the authors about the 53 to 68 interval against the plotted bar.
Second check: another model agreed
Report a problem with this finding
Wording Methods: "results were reported at 0.1% sensitivity"; Results: "assessed at a specificity of 0.1%". The same 0.1% MRD flow-cytometry reporting level is called sensitivity in one place and specificity in the other.
- p. 3 Methods, Patients, last sentence
were reported at 0.1% sensitivity
- p. 8 Results, Secondary End Points, MRD paragraph
assessed at a specificity of
- p. 8 Results, Secondary End Points, MRD paragraph
assessed at a specificity of 0.1%
- p. 3 Methods, Patients, last sentence
results were reported at 0.1% sensitivity
Judge's note: Read both sentences on the page 3 and page 8 images. Both describe the single central multiparametric flow MRD assay and the same 0.1% level. No other passage defines a separate specificity. The two terms are not interchangeable, and the Methods wording (sensitivity) is the one that fits a detection threshold, so the evident fix is in the Results.
Second check: another model agreed
Report a problem with this finding
5 minor (typos and the like; most checked by the judge only)
Wording Introduction says "among older patients treated in clinical trials,6,8,9" but reference 8 is titled "Age and acute myeloid leukemia: real world data on decision to treat and outcomes from the Swedish Acute Leukemia Registry", a registry analysis by its own title, not a clinical trial.
- p. 2 Introduction, third paragraph
among older patients treated in clinical trials,6,8,9
- p. 13 Reference 8
real world data on decision to treat and out- comes from the Swedish Acute Leukemia Registry
Judge's note: Checked the page 2 crop and the reference list on page 13 (crop and text). Reference 6 (CPX-351 versus cytarabine plus daunorubicin) and reference 9 (high-dose daunorubicin) are trials by their titles; reference 8 is registry data by its title. The same 6,8,9 group is used aptly earlier on page 2 for "suboptimal long-term outcomes", so this looks like a reused citation group. It is a loose description, not a wrong pointer.
Second check: another model agreed
Report a problem with this finding
Wording The pre-transplantation MRD window is "within 30 days before hematopoietic-cell transplantation" in the Methods and Results but "within 4 weeks before HCT" in the Figure 3 legend describing the same Panel E data.
- p. 8 Figure 3 caption (Panel E description)
within 4 weeks be- fore HCT
- p. 8 Results, Secondary End Points, MRD paragraph
status within 30 days before hematopoietic-cell
- p. 3 Methods, Patients
within 30 days before hematopoietic-cell transplantation
- p. 8 Results, Secondary End Points, MRD paragraph
status within 30 days before hematopoietic-cell transplantation
Only the authors can say what is right here.
Judge's note: Read the Methods on page 3 and the Results text and Figure 3 legend on page 8. All three describe the same pre-HCT MRD assessment; the Results sentence cites Fig. 3E. Four weeks is 28 days, not 30. It may be a loose paraphrase, but the two stated windows differ and the authors would need to say which one applied.
Second check: another model agreed
Report a problem with this finding
Wording Text: one azacitidine–venetoclax patient "was removed shortly after initiation owing to an eligibility violation (NPM1 mutation)"; Figure 1 shows the only patient outside the 85 who discontinued treatment in that arm as "1 Withdrew consent for treatment", and lists no eligibility-violation category.
- p. 5 Results, Patient Population (top of page)
was removed shortly after ini- tiation owing to an eligibility violation (NPM1 mutation)
- p. 5 Figure 1, azacitidine–venetoclax arm side box
1 Withdrew consent for treatment
Only the authors can say what is right here.
Judge's note: Viewed page 5. In the induction arm the text (four patients declined participation) matches the figure side box ("4 Withdrew consent for treatment and data collection"), so the parallel azacitidine–venetoclax side box appears to be the same patient the text describes, with a different reason. The NPM1 patient could instead sit among the "12 Were withdrawn by physician", with the consent withdrawal being someone else, but the paper does not say. Only the authors can reconcile the two accounts.
Second check: another model agreed
Report a problem with this finding
Wording Figure 1 lists "50 Underwent HCT" and "28 Underwent HCT" under discontinued treatment, whereas the Figure 3 legend gives transition to HCT as "52 patients [60%]" and "34 patients [40%]"; the difference is not explained.
- p. 5 Figure 1
50 Underwent HCT
- p. 5 Figure 1, induction arm
28 Underwent HCT
- p. 8 Figure 3 legend
52 patients [60%]) and the induction chemo-
Judge's note: Viewed pages 5, 8 and 9. The Figure 1 counts are reasons for discontinuing trial treatment, so patients who stopped for another reason (for example "Changed to other consolidation therapy" or "Were withdrawn by physician") and were later transplanted could account for the extra 2 and 6. That makes the two sets plausibly different quantities. However, the Figure 3 legend calls its count "transition of patients in first response to HCT", and nothing in the article text states the distinction, so a reader sees two HCT counts per arm. Either form could be legitimate.
Second check: another model agreed
Report a problem with this finding
Typo Table 1, Asian row, Total column is set as "15(9)" without the space used in every other cell (e.g. "13 (8)", "10 (6)").
- p. 6 Table 1, Race or ethnic group, Asian row, Total column
15(9)
Judge's note: Checked on the crop and the page 6 image: the cell is visibly tighter than its neighbours, so this is not an extraction artefact. The value itself is consistent (15 of 172 rounds to 9%).
Second check: none (judge only)