Review Papers Claim by Claim

Visual synthesis of key trial results cited in the paper

Bar plot: primary musculoskeletal endpoint responses (ACR20 or ACR50 per trial primary outcome). Data points taken from the review's extracted trial table (FUTURE, SPIRIT, DISCOVER, BE OPTIMAL/COMPLETE, MAXIMISE).

Quick visual takeaways

• IL‑17 agents (secukinumab, ixekizumab, bimekizumab) and IL‑23 agents (guselkumab, risankizumab) both show large absolute treatment effects vs placebo in peripheral arthritis (ACR endpoints) and psoriasis (PASI endpoints) FUTURE-5 secukinumab resultsDISCOVER-1 guselkumab results
• Axial disease: direct RCT evidence is stronger for IL‑17 (MAXIMISE: secukinumab ASAS responses) while IL‑23 inhibitors have inconsistent or negative results in axial spondyloarthritis RCTs — a major domain where IL‑17 currently has superior trial support MAXIMISE axial PsA trialRisankizumab AS trial (negative)
• Radiographic outcomes: mixed signals — secukinumab trials showed early inhibition of vdH‑mTSS progression; IL‑23 trials (DISCOVER-2) initially negative at 24 wks but later APEX design and 24‑wk APEX signals suggest IL‑23 can also reduce progression with adequate power/selection — underscores dependence on trial design and follow-up length FUTURE-5 radiographic resultsAPEX guselkumab radiographic study

Structured critical appraisal (what the paper did well — and where it falls short)

Strengths

• Comprehensive domain‑based synthesis linking immunobiology (IL‑23 upstream Th17 maintenance vs IL‑17 downstream effector) to clinical endpoints across skin, peripheral joints, axial disease, enthesitis/dactylitis, radiographic progression and IBD comorbidity, referencing primary phase‑3 RCTs and meta-analyses IL-23/IL-17 pathway review.
• Transparent acknowledgment of the lack of direct head‑to‑head randomized trials and careful domain‑based clinical recommendations (e.g., prefer IL‑23 where IBD present; prefer IL‑17 for axial disease) supported by trial evidence cited in the review Mehta & Chandran review guidance.

Limitations, biases, and blindspots (critical)

• Reliance on indirect comparisons and post‑hoc matching‑adjusted indirect comparisons (MAICs) — these methods are hypothesis‑generating but sensitive to population differences, endpoint timing, background csDMARD/TNFi exposure, and reporting choices; the review states MAIC findings (bimekizumab vs risankizumab) but does not fully quantify residual confounding or heterogeneity across included trials MAIC comparative analysis.
• Underemphasis of conflicts of interest/sponsor influence: one author (V. Chandran) discloses multiple industry relationships; many primary RCTs are industry‑sponsored — the review mentions funding but could discuss how sponsorship might shape trial design, comparator selection, and endpoint choice (common blindspot in narrative reviews) COI disclosures in review.
• Heterogeneity in primary endpoints and timing across trials: many IL‑17 trials used ACR20 at 24 weeks while some bimekizumab trials used ACR50 at 16 weeks (BE OPTIMAL primary outcome). Cross‑trial comparisons mixing different primary endpoints or timepoints can mislead if not normalized — the review reports these results but stops short of standardized re-analysis or patient‑level pooling.
• Insufficient quantification of safety signal uncertainty: the review correctly summarizes that IL‑17 agents carry rare IBD exacerbation risk while IL‑23 agents are used in IBD, yet it could better present incidence rates and confidence intervals from pooled safety analyses and registries to clarify magnitude of risk Pooled IBD incidence (secukinumab).

Missing / useful additional analyses the paper could have included

1. A standardized network meta‑analysis (prespecified endpoints and timepoints, risk of bias weighting) rather than narrative summary of NMA results — to quantify comparative effect sizes and credible intervals across agents.
2. Patient‑level pooled radiographic analysis (or calling for data sharing) to resolve whether differences in radiographic outcomes are due to drug mechanism vs trial selection and duration; authors mention APEX but older trials had low power/time horizon.
3. Explicit safety incidence tables (e.g., serious infection, Candida, new/worsening IBD) with person‑years denominators drawn from pooled trial and registry data to show absolute risk differences.
4. Clearer discussion of biologic sequencing strategies (e.g., switching class after IL‑23 failure) and what the mechanistic data predict about cross‑resistance or retained responsiveness.

Where the evidence is strongest — and lowest — confidence

• Highest confidence: superiority of IL‑17 agents for axial inflammation (based on dedicated RCTs such as MAXIMISE) and robust psoriasis skin clearance (multiple large RCTs) MAXIMISE axial evidence.
• Moderate confidence: comparable peripheral joint efficacy across IL‑17 and IL‑23 classes when judged by ACR endpoints in individual trials (but indirect comparisons remain vulnerable to heterogeneity). Examples: DISCOVER (guselkumab) and SPIRIT/FUTURE series (ixekizumab/secukinumab) show overlapping absolute effects DISCOVER-1 peripheral arthritis evidenceFUTURE-5 peripheral arthritis evidence.
• Low confidence: claims that one class unequivocally reduces radiographic progression more than the other — evidence is mixed and trial design / follow-up length strongly influence apparent differences; definitive resolution requires head‑to‑head RCTs or pooled patient‑level radiographic analyses with standardized scoring and risk adjustment Bimekizumab review and NMA.

Practical clinician‑facing synthesis (evidence‑weighted)

1. For predominant axial PsA: favor IL‑17 agents (strongest RCT evidence; MAXIMISE) — confidence: high/medium MAXIMISE support for IL-17 axial.
2. For peripheral dominant PsA (synovitis, enthesitis, dactylitis): both IL‑17 and IL‑23 classes show clinically meaningful improvements in ACR and domain endpoints; bimekizumab trials show particularly large effect sizes for some endpoints but cross-trial comparisons limited — confidence: moderate BE OPTIMAL bimekizumab efficacy.
3. If IBD comorbidity or strong IBD risk: favor IL‑23 inhibitors (efficacy in Crohn’s/UC trials; mechanistic safety signal); avoid IL‑17 where possible in active IBD due to rare but plausible exacerbation risk — confidence: moderate/strong Risankizumab CrohnIBD incidence with secukinumab.

What evidence would change these conclusions (falsification tests)

• A large, randomized, adequately powered, head‑to‑head trial of IL‑17 vs IL‑23 inhibitors in PsA with prespecified multi‑domain endpoints (ACR, PASI, enthesitis/dactylitis scores), centralized radiographic scoring through ≥2 years, and stratification by baseline axial involvement would either confirm or refute class superiority claims (authors note this need) Author call for head-to-head RCTs.
• Patient‑level pooled radiographic datasets showing no class effect after adjustment for baseline progression risk (sex, CRP, erosions) would undercut claims of superior structural protection by IL‑17 agents.

Bottom-line judgement (evidence-weighted)

The review is a high‑quality, clinically useful narrative synthesis that accurately maps mechanisms to domain‑specific evidence and makes reasonable treatment recommendations. Its main limitation is dependence on indirect comparisons and heterogenous trials — therefore conclusions about between‑class superiority (especially for radiographic progression) should be considered provisional until head‑to‑head RCTs or patient‑level pooled analyses are performed Review conclusions and limitations.

Actionable next steps & reproducible analyses I recommend

1. Perform a standardized network meta‑analysis using trial patient‑level covariates where available, harmonizing endpoints (e.g., converting to odds ratio for ACR20 at 24 wks) and performing sensitivity analyses by prior TNFi exposure and csDMARD background.
2. Request patient‑level radiographic data from sponsors (DISCOVER, FUTURE, BE trials) to run pooled mixed models with consistent vdH‑mTSS scoring and covariate adjustment (sex, baseline damage, CRP) to evaluate structural effects robustly.
3. Generate safety incidence tables (events per 1000 patient‑years) using pooled trial + registry data to quantify absolute IBD and candidiasis risks across classes.

Selected primary sources used in this critique (examples)

• Mehta & Chandran review (source)
• FUTURE-5 secukinumab RCT
• DISCOVER-1 guselkumab RCT
• MAXIMISE axial PsA trial