Lecanemab in Early Alzheimer's Disease
2023 · paper · Clinical trials
An 18-month phase 3 trial of intravenous lecanemab versus placebo in 1,795 people with early Alzheimer's disease. Lecanemab reduced brain amyloid and moderately slowed decline on cognition and function measures, but caused infusion reactions and brain imaging abnormalities.
Checked: PDF from digitalcommons.wustl.edu, fetched 4 Oct 2026. Page numbers are those of that PDF.
0 findings
None that passed a second check.
9 minor (typos and the like; most checked by the judge only)
Wording Discussion: "this trial exceeded the prospectively defined target, with an estimated treatment difference of 0.373 points" places the sample-size target (0.373, Methods p4) where the observed result would go; the observed difference was −0.45 (p8).
- p. 12 Discussion, second paragraph
exceeded the prospectively defined target, with an estimated treatment difference of 0.373 points
- p. 4 Statistical Analysis
the estimated treatment difference between lecanemab and placebo in all the participants was 0.373 points
- p. 8 End-Point Results
(difference, −0.45;
- p. 8 End-Point Results
(difference, −0.45; 95% confidence interval [CI], −0.67 to −0.23;
Judge's note: On p4, 0.373 is the assumed treatment difference used for the sample-size calculation. The p12 sentence reads as if 0.373 were the trial's estimate. The intended meaning (a target of 0.373) is recoverable, but the wording is misleading.
Second check: another model agreed
Report a problem with this finding
Wording Table 1 footnote ‡: "A score of 0.5 is considered to be the threshold for Alzheimer's disease and was required for trial enrollment", yet the table lists 165 (19.2%) and 169 (19.3%) participants with a global CDR of 1.
- p. 6 Table 1 footnote ‡
A score of 0.5 is considered to be the threshold for Alzheimer’s disease and was required for trial enrollment.
- p. 5 Table 1, Global CDR score
165 (19.2)
Judge's note: I checked the footnote on p6 and the Table 1 rows on p5. 'Threshold' can be read as a minimum score, and on that reading the footnote is consistent. Read literally, though, it says a score of exactly 0.5 was required, which conflicts with the CDR-1 rows. Wording such as 'at least 0.5' would remove the ambiguity.
Second check: another model agreed
Report a problem with this finding
Wording The text says "729 (81.2%) in the lecanemab group ... completed the trial and had data available on the primary end point", but Figure 2A shows 714 lecanemab participants at 18 months (the placebo figure, 757, matches the text).
- p. 8 Results, Participants
729 (81.2%) in the lecanemab group and 757 (84.4%) in the placebo group completed the trial and had data available on the primary end point
- p. 7 Figure 2A, No. of Participants at 18 mo
714
Only the authors can say what is right here.
Judge's note: Methods p6: the mITT population needs a baseline and at least one post-dose CDR-SB measurement, so any completer with an 18-month CDR-SB value would be in mITT and counted in Fig. 2A. The 15-participant gap in the lecanemab group is not explained; visit windowing could account for it. Only the authors can say which statement is right.
Second check: another model agreed
Report a problem with this finding
Wording "For each of these assessments, separation of the trial groups was apparent by visual inspection of graphs at 3 months", but Figure 2E (ADCS-MCI-ADL) has no 3-month visit; its first post-baseline point is at 6 months.
- p. 9 Results, paragraph after ADCS-MCI-ADL results
For each of these assessments, separation of the trial groups was apparent by visual inspection of graphs at 3 months.
- p. 7 Figure 2E x-axis
E ADCS-MCI-ADL Score
Only the authors can say what is right here.
Judge's note: On p7, the Fig. 2E x-axis has ticks at 0, 6, 12 and 18 months only. Panels A–D do have 3-month points. The sentence therefore cannot apply to ADCS-MCI-ADL as written.
Second check: another model agreed
Report a problem with this finding
Wording The Discussion (p. 13) says the ARIA sensitivity analyses "showed that ARIA had no effect on the results", but the Results (p. 9) describe the same analyses only as "generally consistent with the primary analysis (Table S2)".
- p. 13 Discussion, limitations paragraph (left column)
showed that ARIA had no effect on the results
- p. 9 Results, sensitivity analyses paragraph
were generally consistent with the primary analysis (Table S2)
Only the authors can say what is right here.
Judge's note: I opened both crops. On p. 9 (right column), the Results say the CDR-SB sensitivity analyses for Covid-19 missed doses and for possible bias from functional unblinding due to ARIA "were generally consistent with the primary analysis (Table S2)". On p. 13 (left column, limitations), the Discussion says these analyses "showed that ARIA had no effect on the results". The Methods (around line 462 of the text) only say that these analyses were planned. I found nothing else in the article that backs the absolute claim. The Discussion's wording is stronger than the Results' 'generally consistent' and should be softened or justified. Table S2 is in the supplement, which is not among the files, so I could not check how close the estimates are. Because the defect is in the wording only, I rate it S3 and leave it as a query for the authors.
Second check: another model agreed
Report a problem with this finding
Wording Discussion: "In the CSF substudy and in plasma analyses ... with the exception of NfL" leaves NfL unqualified, whereas Results (p10) exclude only "CSF NfL (Fig. S5)".
- p. 12 Discussion, first paragraph on page (left column)
with the exception of NfL, which is less sensitive to neurodegeneration
- p. 10 Results, end-point paragraph (left column)
with the exception of CSF NfL (Fig. S5)
Only the authors can say what is right here.
Judge's note: The Results imply that plasma NfL improved numerically. The Discussion sentence covers the plasma analyses too, so its unqualified 'NfL' exception reads as including plasma NfL. Fig. S5 is not among the files, so only the authors can say which statement is intended.
Second check: another model agreed
Report a problem with this finding
Typo Appendix affiliations give "Washington University School of Medicine in St. Louis, St. Louis (R.B.)", but the author is Randall J. Bateman (byline 'R.J. Bateman'), so the initials should be R.J.B.
- p. 13 Appendix, affiliations
Washington University School of Medicine in St. Louis, St. Louis (R.B.)
- p. 2 Author byline
R.J. Bateman
Judge's note: In the appendix on p13, other authors' middle initials are included (C.J.S., L.D.K., C.H.D.). Bateman's middle initial J is dropped, which is inconsistent with the rest of the list.
Second check: none (judge only)
Report a problem with this finding
Typo Figure 1, placebo box: "67 (7.5) Withdrew consent" has no % sign, although every other entry in the flow diagram has one.
- p. 4 Figure 1, placebo discontinuation reasons
67 (7.5) Withdrew consent
Judge's note: I looked at the crop. Every other entry in both boxes prints the %, e.g. '69 (7.7%) Withdrew consent'. The value is correct (67/897 = 7.5%). Only the % sign is missing.
Second check: none (judge only)
Report a problem with this finding
Typo Results: "257 in the study of tau pathologic features on PET" uses 'study', while the same sentence calls the other two 'substudy' ("three longitudinal substudies").
- p. 8 Results, Participants
257 in the study of tau pathologic features on PET
Judge's note: I checked the crop of p8. The term is inconsistent, but the meaning is clear. This is cosmetic.
Second check: none (judge only)