Intravitreal Aflibercept (VEGF Trap-Eye) in Wet Age-Related Macular Degeneration
2012 · paper · Clinical trials
Two randomized trials with 2419 patients compared aflibercept injections, monthly or every two months, with monthly ranibizumab for wet macular degeneration. All aflibercept regimens matched ranibizumab in maintaining vision at 52 weeks, with similar safety.
1 finding
Wording Abstract: "In a prespecified integrated analysis of the 2 studies, all aflibercept regimens were within 0.5 letters of the reference ranibizumab for mean change in BCVA", but Figure 3C gives 9.3 (2q4) vs 8.7 (Rq4), and the LOCF inset gives IAI 2q4 − Rq4 = 0.60.
- p. 1 Abstract, Results
all aflibercept regimens were within 0.5 letters of the reference ranibizumab for mean change in BCVA
- p. 8 Figure 3C, end-of-line values and inset (LOCF)
9.3 2q4
- p. 1 Abstract, Results
all aflibercept regimens were within 0.5 letters
- p. 8 Figure 3C inset, LOCF
IAI 2q4 - Rq4 0.60
Judge's note: I checked the abstract on p1 and Figure 3C on p8. The week-52 line labels read 9.3 2q4, 8.7 Rq4, 8.4 2q8 and 8.3 0.5q4. The inset (integrated analysis, least-squares mean with 95% CI) gives LOCF differences of 0.60, −0.43 and −0.32. LOCF is the paper's declared imputation (Methods p4). The 2q4 difference of 0.60 is outside 'within 0.5'. Only the observed-case (0.25) and completers (0.43) analyses fall inside it, and the abstract names neither. The 2013 erratum (pp13–15) replaces only a Discussion paragraph. That paragraph says 'within 1 letter', which fits the data. The erratum does not change the abstract.
Second check: another model agreed
Report a problem with this finding
10 minor (typos and the like; most checked by the judge only)
Wording Abstract Participants: "Patients (n = 2419)" is the combined safety analysis set (1215 + 1204), not the number randomized (1217 + 1240 = 2457), and the abstract does not say which set it is.
- p. 1 Abstract, Participants
Patients (n = 2419)
- p. 5 Results, Patient Disposition
In VIEW 1, 1217 patients were randomized
- p. 5 Results, Patient Disposition
In VIEW 2, 1240 patients were randomized
- p. 1 Abstract, Participants
Patients (n ⫽ 2419)
Judge's note: I checked Figure 1A (p2) and Figure 1B (p3). Randomized counts are 1217 and 1240. Safety sets are 304+304+303+304 = 1215 for VIEW 1 and 309+297+307+291 = 1204 for VIEW 2, which sum to 2419. Full analysis sets sum to 2412. So 2419 is a real population (the treated or safety set). It is not wrong, but the abstract does not label it, and the Results report randomized counts. This is a labelling ambiguity, not a wrong number.
Second check: another model agreed
Report a problem with this finding
Wording Figure 1B (VIEW 2): the IAI box "Total: 98 (10.7%)" uses the safety-set denominator (913), while the ranibizumab box "27 (8.9%)" uses the randomized count (303). The figure states neither denominator.
- p. 3 Figure 1B, 'Discontinuation all IAI groups' box
98 (10.7%)
- p. 3 Figure 1B, 'Discontinuation Ranibizumab' box
27 (8.9%)
Judge's note: I checked Figure 1B on p3. 98/937 randomized gives 10.5% and 98/913 safety-set gives 10.7%. The IAI reason rows also fit 913: 39 gives 4.3% (4.2% on 937) and 6 gives 0.7%. The ranibizumab rows fit 303: 27 gives 8.9% and 11 gives 3.6%. On 291, 11 would be 3.8%. The two boxes in one figure therefore use different, unstated denominators. The data themselves are not in error.
Second check: another model agreed
Report a problem with this finding
Wording Results: "Regardless of whether the analysis was by LOCF, by multiple imputations, by assessing completers, or by using actual observed data ... within 0.3 letters ... (Fig 3C, inset)". The cited inset and caption show only 3 analyses (LOCF, observed case, completers), with no multiple-imputation result.
- p. 6 Results, Mean Changes in BCVA
LOCF, by multiple imputations, by assessing completers, or by using actual observed data
- p. 8 Figure 3 caption/inset
using 3 different analyses: by last observation carried forward (LOCF), using observed case data, and by assessing completers
- p. 6 Results, Mean Changes in BCVA paragraph
by LOCF, by multiple imputations, by assessing completers, or by using actual observed data
Only the authors can say what is right here.
Judge's note: I checked the p6 text and the Figure 3 inset and caption on p8. The caption reads 'using 3 different analyses'. The inset panels are LOCF, Observed and Completers. The multiple-imputation BCVA result is not in the cited figure. The only other mention of multiple imputation (p6) concerns the primary end point. An appendix may hold the result, but the appendices are not among the files.
Second check: another model agreed
Report a problem with this finding
Wording Figure 2 caption: "the horizontal bars indicate 95% confidence interval (CI) range" for both studies, whereas the Methods test VIEW 1 at the 4.9% level and the Table 2 footnote states "*95.1% CI for VIEW 1."
- p. 7 Figure 2 caption
the horizontal bars indicate 95% confidence interval (CI) range
- p. 6 Table 2 footnote
*95.1% CI for VIEW 1.
Judge's note: I checked the p4 Methods ('4.9% for VIEW 1'), the Table 2 footnote on p6 and the Figure 2 caption on p7. The VIEW 1 bars in Figure 2 are the primary-end-point CIs. Consistent with the Methods, they would be 95.1% CIs, but the caption labels all of them 95%. This is a labelling inconsistency only.
Second check: another model agreed
Report a problem with this finding
Typo The instrument is expanded as "National Eye Institute 25-Item Visual Function Questionnaire (NEI VFQ-25)" in the Methods (p4) but as "National Eye Institute 25-Item Visual Functioning Questionnaire" in the Table 1 and Table 2 footnotes.
- p. 4 Methods, End Points
25-Item Visual Function Questionnaire (NEI
- p. 5 Table 1 footnote
National Eye Institute 25-Item Visual Functioning
Judge's note: I saw both forms on the page images (p4 Methods, p5 Table 1 footnote, p7 Table 2 footnote). This is a cosmetic naming inconsistency.
Second check: none (judge only)
Report a problem with this finding
Typo Methods: "Pharmaceutical and Medical Device Agency" for the Japanese regulator.
- p. 4 Methods, End Points and Statistical Analyses
Pharmaceutical and Medical Device Agency and other regula-
- p. 4 End Points and Statistical Analyses
Pharmaceutical and Medical Device Agency
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that the agency's official English name is 'Pharmaceuticals and Medical Devices Agency'. The page image on p4 does read 'Pharmaceutical and Medical Device Agency'. This is a cosmetic naming matter.
Second check: none (judge only)
Report a problem with this finding
Typo Table 2 (VIEW 1, 2q8, CNV area): "0.86 (0.15–1.58)" uses an en dash, while every other CI in Table 2 uses 'to'.
- p. 6 Table 2, VIEW 1, 2q8, LS mean difference for change in CNV area
0.86 (0.15–1.58)
Judge's note: The page 6 image confirms the en dash in this one cell. All other CIs on p6 and p7 use 'x to y'. The difference is cosmetic only.
Second check: none (judge only)
Report a problem with this finding
Typo Table 3 footnote expands APTC ATE as "Anti-platelet Trialists' Collaboration Arteriothrombolic Event". The term should be Arterial Thromboembolic (or Arteriothrombotic) Event.
- p. 10 Table 3 footnote
Anti-platelet Trialists’ Collaboration Arteriothrombolic Event
- p. 8 Results, Safety
specific arterial thromboembolic end points
Judge's note: The crop and the page 10 image both show "Arteriothrombolic". The Results text on page 8 says "arterial thromboembolic end points". This is a misspelling.
Second check: none (judge only)
Report a problem with this finding
Typo Financial Disclosure(s) lists 'MA, G.G., B.S., and R.S. are employees of Bayer HealthCare.' The initials 'MA' (Majid Anderesi) have no periods, while every other author's initials in the same sentence do.
- p. 12 Financial Disclosures
MA, G.G., B.S., and R.S. are employees of
- p. 12 Financial Disclosure(s)
MA, G.G., B.S., and R.S. are employees of Bayer HealthCare.
Judge's note: The page 12 crop shows 'MA, G.G., B.S., and R.S. are employees of Bayer HealthCare.' Every other initial in the disclosures has periods (G.D.Y., N.S., U.S.-E., etc.), so 'MA' is an inconsistency within the same sentence. The fix is 'M.A.'. This is cosmetic only.
Second check: none (judge only)
Report a problem with this finding
Typo The Funding statement reads 'Bayer HealthCare, Berlin Germany.' There is no comma between city and country, unlike 'Berlin, Germany' elsewhere in the paper.
- p. 12 Funding
Bayer HealthCare, Berlin Germany.
Judge's note: The page 12 crop shows 'Bayer HealthCare, Berlin Germany.' in the Funding paragraph. The abstract (line 14 of the text) and affiliation 11 (line 1317) both write 'Berlin, Germany', so the missing comma is an inconsistency. This is cosmetic punctuation.
Second check: none (judge only)