Tralokinumab for Moderate-to-Severe Atopic Dermatitis: Results from Two 52-Week, Randomized, Double-Blind, Multicentre, Placebo-Controlled Phase III Trials (ECZTRA 1 and ECZTRA 2)
British Journal of Dermatology, 2020 · paper · Clinical trials
Two 52-week phase III placebo-controlled trials of tralokinumab, an interleukin-13 antibody, as monotherapy in adults with moderate-to-severe atopic dermatitis. They assessed efficacy and safety at week 16 and maintenance of response over the year.
Checked: PDF from europepmc.org, fetched 4 Oct 2026. Page numbers are those of that PDF.
4 findings
Reference or label SCORAD differences are given a percent sign although they are differences in adjusted mean score change: "(difference −10·4%, 95% CI −14·4 to −6·5; P < 0·001" and "(difference −14·0%, 95% CI −18·0 to −10·1; P < 0·001"; Table 2 prints −10·4 and −14·0 without %.
- p. 5 Key secondary outcomes at week 16, SCORAD sentence (ECZTRA 1)
difference −10·4%, 95%
- p. 5 Key secondary outcomes at week 16, SCORAD sentence (ECZTRA 2)
(difference −14·0%, 95% CI −18·0 to −10·1; P < 0·001;
- p. 5 Table 2, SCORAD row
−10·4 (−14·4
Judge's note: Read on the page 5 image and the C005 crop. The sentence reports adjusted mean changes in SCORAD score (−25·2 vs −14·7; −28·1 vs −14·0), and Table 2 labels the row 'Adjusted mean change (SE) from baseline in SCORAD score' with the differences unit-free. The difference of two score changes is in SCORAD points, so the '%' is a wrong unit. The values themselves agree with Table 2.
Second check: another model agreed
Report a problem with this finding
Reference or label DLQI differences are given a percent sign although they are differences in adjusted mean score change: "(difference −2·1%, 95% CI −3·4 to −0·8; P = 0·002" and "(difference −3·9%, 95% CI −5·2 to −2·6; P < 0·001"; Table 2 prints −2·1 and −3·9 without %.
- p. 5 Key secondary outcomes at week 16, DLQI sentence (ECZTRA 1)
(difference −2·1%, 95% CI −3·4 to −0·8; P = 0·002;
- p. 6 Continuation of DLQI sentence below Figure 1 (ECZTRA 2)
(difference −3·9%, 95%
Judge's note: Read on the page 5 image and the page 6 crop. Table 2 labels the row 'Adjusted mean change (SE) from baseline in DLQI score' and the differences are exact subtractions of the two changes, so the unit is DLQI points, not percent. Same kind of defect as C003 but a different endpoint and different sentences, so kept separate.
Second check: another model agreed
Report a problem with this finding
Reference or label Table 5 row labels "SCORAD score of 75 at week 16, n/N (%)" and "SCORAD score of 50 at week 16, n/N (%)" describe responder endpoints that are evidently ≥ 75% and ≥ 50% improvement in SCORAD (SCORAD 75 / SCORAD 50), not absolute scores.
- p. 9 Table 5, row label
SCORAD score of 75 at week 16,
- p. 9 Table 5, row label
SCORAD score of 50 at week 16,
Judge's note: Read on the page 9 image. The rows are binary responder outcomes with non-responder imputation (footnote a) and favour tralokinumab (8·8% vs 3·0%; 26·0% vs 11·7%). Baseline median SCORAD is about 70 (Table 1), so an absolute score of 75 or 50 cannot be a response criterion met more often on active treatment. The label is misworded by analogy with EASI 75/50. These endpoints are not defined in the Methods.
Second check: another model agreed
Report a problem with this finding
Wording Discussion: "fewer skin infections requiring systemic treatment and a lower frequency of eczema herpeticum were seen with tralokinumab" — Table 6 shows the AESI rate lower only in ECZTRA 2 (11·0% vs 3·5%); in ECZTRA 1 it is 4 (2·0) placebo vs 13 (2·2) tralokinumab.
- p. 10 Discussion, paragraph 4 (runs on to page 11)
fewer skin infections requir-
- p. 11 Table 6, row 'AESI – skin infections requiring systemic treatment', ECZTRA 1 columns
4 (2·0)
- p. 11 Table 6, same row, ECZTRA 1 tralokinumab column
13 (2·2)
- p. 9 Results, Safety
for tralokinumab vs. placebo in ECZTRA 2 (Table 6)
Only the authors can say what is right here.
Judge's note: Read Table 6 on the page 11 image and the Discussion on pages 10-11. The Results text (page 9) restricts the claim to ECZTRA 2: 'lower rates of skin infections requiring systemic treatment were reported as an AESI for tralokinumab vs. placebo in ECZTRA 2'. The Discussion states it without that qualification. A pooled reading would support it (26/396 vs 34/1194), but no pooled figure is presented. So the Discussion wording generalises beyond what Table 6 shows for ECZTRA 1.
Second check: another model agreed
Report a problem with this finding
11 minor (typos and the like; most checked by the judge only)
Wording Summary says "In phase II studies, tralokinumab combined with topical corticosteroids provided early and sustained improvements", while the key-points box says "a 12-week, phase IIb trial" and the Introduction says "In a phase II trial, tralokinumab provided early and sustained improvement" (single reference 15).
- p. 1 Summary, Background
In phase II studies, tralokinumab combined with topical corticosteroids
- p. 2 What is already known about this topic?
efficacy and safety in a 12-week, phase IIb trial
- p. 2 Introduction, paragraph 2
In a phase II trial, tralokinumab
Judge's note: Checked the page 1 crop, the page 2 image and the text of pages 10-12. The abstract uses the plural "studies"; the box, the Introduction and the Discussion ("the phase II study", "the phase II trial") all refer to one trial, and only reference 15 is cited. Phase II versus IIb is loose wording only. A minor plural/singular inconsistency; either form could be defended.
Second check: another model agreed
Report a problem with this finding
Wording Table 1, ECZTRA 1 tralokinumab (n = 603), Race 'Other or missing data': "16 (2·6)" — 16/603 is 2·65%, which rounds to 2·7%.
- p. 4 Table 1, Race 'Other or missing data', ECZTRA 1 tralokinumab column
16 (2·6)
Judge's note: Read the cell on the page 4 image: "16 (2·6)". The four race counts sum to 603, so the count fits the column. 16/603 = 2.653%, not a rounding tie, so 2·7 is expected. The neighbouring cells I recomputed (3/199, 9/201, 22/593) match as printed. The discrepancy is 0.1 percentage point in a baseline table, so it is minor. No footnote explains it; the row combines 'other' and 'missing', so the printed figure may be a sum of separately rounded parts.
Second check: another model agreed
Report a problem with this finding
Wording Text: "In ECZTRA 1 and ECZTRA 2, 185 and 227 patients were rerandomized 2 : 2 : 1"; Table 3 (maintenance analysis set) arm sizes n = 35, 68, 76 and n = 46, 91, 89 sum to 179 and 226, and the gap is not explained.
- p. 6 Maintenance outcomes at week 52, first sentence
185 and 227 patients were reran-
- p. 8 Table 3 header, ECZTRA 1 arms
n = 35
- p. 8 Table 3 header, ECZTRA 2 arms
n = 46
- p. 6 Maintenance outcomes at week 52, first sentence
185 and 227 patients were rerandomized 2 : 2 : 1
Judge's note: Read the Table 3 headers on the page 8 image and the sentence on page 6. The sums are 179 and 226 against 185 and 227. The table is explicitly labelled 'maintenance analysis set', which may legitimately differ from the number rerandomized (for example, those rerandomized but not dosed). The article does not define that set, and Figure S3 and Appendix S2 are not among the files. So this is an unexplained difference between two differently labelled populations, not a demonstrated error.
Second check: another model agreed
Report a problem with this finding
Wording "A greater percentage change in EASI was observed with tralokinumab vs. placebo at week 16, with a separation between arms (P < 0·001) from week 2 onwards (Table 5, Figure 3; and Figure S7…)" — Table 5 gives adjusted mean absolute change in EASI and Figure 3 is a photograph series of one patient.
- p. 7 Additional efficacy analyses, last sentence
(Table 5, Figure 3; and Fig-
- p. 9 Table 5, row label
baseline in EASI at week 16
- p. 10 Figure 3 caption
Example of improvement in Eczema Area and Severity Index (EASI) from baseline to week 16 in a patient receiving tralokinumab
Judge's note: Checked Table 5 on page 9 (row 'Adjusted mean change (SE) from baseline in EASI at week 16': −9·0 vs −15·5; −7·0 vs −16·9, which are absolute changes given baseline medians of about 28-30), Figure 3 on page 10 (photos of one tralokinumab-treated patient with EASI scores at weeks 0, 8 and 16), and the page 13 list (Figure S7 is 'Percentage change from baseline in EASI'). Only Figure S7 matches the quantity described. The main-article pointers are related, loosely worded 'see also' references, so this is a minor imprecision.
Second check: another model agreed
Report a problem with this finding
Wording Discussion paragraph 1 says, with no regimen qualifier, "the majority of patients who achieved response criteria for improvement in the extent and severity of AD at week 16 maintained response at week 52 without need for rescue medication, including TCS"; in Table 3 only the continued-Q2W arms consistently exceed 50% (51%, 59%, 60%, 56%), while Q4W is "14/36 (39)", "22/49 (45)", 28/57 (49), 38/74 (51) and placebo is 21-47%.
- p. 10 Discussion, paragraph 1
the majority of patients who achieved response criteria
- p. 8 Table 3, IGA score of 0 or 1 at week 52, Q2W to Q4W columns
14/36 (39)
- p. 8 Table 3, IGA score of 0 or 1 at week 52, ECZTRA 2 Q2W to Q4W
22/49 (45)
Judge's note: Viewed the Discussion crop (page 10) and the Table 3 page image (page 8). Table 3 reads as the reviewer quotes. The sentence in Discussion paragraph 1 is not restricted to a regimen. Pooling all rerandomised responders gives under 50% for all four endpoint/trial combinations; pooling only the two tralokinumab arms gives under 50% for IGA 0/1 in ECZTRA 1 (34/75) and just over 50% for the other three. The paper states the result precisely elsewhere: the abstract says "with continued tralokinumab treatment", and Discussion paragraph 3 says "More than 50% ... with tralokinumab Q2W ... and 39-51% ... when receiving tralokinumab Q4W ... (21-47%) ... rerandomized to placebo". So this is a loose summary sentence that the paper itself qualifies later, not a wrong number: a minor wording overstatement.
Second check: another model agreed
Report a problem with this finding
Wording "the reduction in S. aureus colonization… was more than 10 times greater with tralokinumab vs. placebo: the median reduced from 969 to 22 gene copies per cm2 with tralokinumab vs. 649 to 238 gene copies per cm2 with placebo" — the basis of 'more than 10 times' is not stated.
- p. 9 Results, final paragraph before Table 5 ... S. aureus colonization
was more than 10 times greater with tralokinumab vs. placebo
Judge's note: The page 9 image prints 969, 22, 649 and 238 with no decimal points, so the note claiming lost decimals is wrong. On absolute median reductions the ratio is about 2.3 (947 vs 411). On fold reduction it is about 44-fold vs 2.7-fold, a ratio of about 16, which does satisfy 'more than 10 times'. Relative (fold) change is a usual way to express qPCR copy numbers, so the statement holds under a legitimate reading. What remains is an ambiguity of wording, not a demonstrated error.
Second check: another model agreed
Report a problem with this finding
Typo Missing full stop at the end of the SCORAD sentence: "Table 2; and Tables S6 and S7; see Supporting Information)" is followed directly by a new paragraph "The adjusted mean changes in DLQI…".
- p. 5 Key secondary outcomes at week 16, end of SCORAD paragraph
Table 2; and Tables S6 and S7; see Supporting Information)
Judge's note: The crop shows the closing parenthesis with no full stop, and the next sentence starting a new indented paragraph. Not an extraction artefact.
Second check: none (judge only)
Report a problem with this finding
Typo Figure 1 legend sets one P value with a baseline full stop, "†P = 0.002 vs. placebo", whereas the rest of the legend and the article use the raised decimal point ("P < 0·05", "P < 0·01", "P < 0·001").
- p. 6 Figure 1 caption, last line
†P = 0.002 vs. placebo
Judge's note: The crop clearly shows '0.002' with a baseline point next to '0·05', '0·01' and '0·001' with raised points. Cosmetic only; the value agrees with Table 2.
Second check: none (judge only)
Report a problem with this finding
Typo Figure 2 legend punctuation: "**P < 0·01, vs. placebo ***P < 0·001 vs. placebo." — a comma sits inside the first key and there is no separator before the second.
- p. 7 Figure 2 caption, last lines
**P < 0·01, vs. placebo ***P < 0·001 vs. placebo.
Judge's note: Read on the page 7 image. The Figure 1 legend punctuates the same items as '**P < 0·01 vs. placebo, ***P < 0·001 vs. placebo'. The comma is misplaced; the meaning is unaffected.
Second check: none (judge only)
Report a problem with this finding
Typo Table 4, pruritus NRS at week 1, ECZTRA 1 difference cell: "−0·4 (−0·6 to −0·3); P < 0·001)" has an unmatched closing parenthesis.
- p. 8 Table 4, pruritus NRS at week 1 row, ECZTRA 1 difference column
P < 0·001)
Judge's note: The page 8 image shows 'P < 0·001)' in that cell only; the confidence-interval bracket is already closed and no other cell has the extra character. Not an extraction artefact.
Second check: none (judge only)
Report a problem with this finding
Typo Eosinophilia threshold printed as "(> 1·5 × 10⁹ L)" with the per-litre indication (⁻¹ or '/') missing.
- p. 10 Safety, first lines of text under Figure 3
eosinophilia (> 1·5 × 10⁹ L)
Judge's note: The page 10 image shows '> 1·5 × 10⁹ L' with nothing after the L. A cell count is a concentration (10⁹ per litre), so the unit as printed is incomplete. The listed crop file for this cluster did not exist, so I used the full page. It is a dropped character and the meaning is recoverable.
Second check: none (judge only)