Review Papers Claim by Claim

Visual-first critique: Antigen processing and presentation in cancer immunotherapy (Lee et al., JITC 2020)

Figure: concise visual summary of the mechanistic steps the paper emphasizes as selective filters for natural antigen presentation (proteasome→TAP→ERAP→PLC→surface). Each step is necessary to mimic experimentally when validating candidate neoantigens Lee et al., review of MHC-I pathway

Figure: Lee et al. report many experimentally validated tumor antigens would score as weak binders by standard IC50 cutoffs (≈30% >500 nM), demonstrating why binding-only filters miss bona fide antigens Lee et al., analysis of validated antigens

Figure: Lee et al. show a correlation between MHC binding predictions and combined proteasome/TAP/loading total score — but the correlation does not equal immunogenicity, and stability/processing context matters Lee et al., prediction limitations

Critical synthesis — what the review gets right

• Mechanistic completeness: the review correctly integrates proteasome subtype, TAP specificity, ER aminopeptidases (ERAP1/2), tapasin/TAPBPR editing and pMHC stability as sequential filters that shape the presented peptidome and thus antigen discovery sensitivity Lee et al., pathway summary
• Empirical caution: using validated antigen lists and IEDB tools, the authors empirically show many validated antigens would fail simple IC50 cutoffs — a powerful empirical illustration of prediction limits Lee et al., validation analysis
• Method guidance: recommendation to favor assays that mimic natural processing (TMG constructs, long-peptide or mRNA electroporation into APCs) is accurate and supported by literature showing pulsing minimal epitopes can mislead antigen identification Lee et al., experimental validation suggestions

Main limitations, blindspots, and constructive critiques

1. Review (narrative) nature — no formal meta-analysis: conclusions rely on selective examples and a retrospective re-scoring of validated epitopes rather than pre-registered benchmarking datasets; this can introduce selection/publication bias (authors acknowledge this) Lee et al., methods & limitations
2. Underweighting of peptide‑MHC stability and MS-derived ligandome advances: the review notes stability matters but gives limited practical guidance on integrating stability predictors or mass-spec (mono-allelic immunopeptidomics) datasets that now substantially improve predictors (e.g., mass-spec training sets for NetMHCpan-like models) — more concrete recommendations would be useful IEDB update (resource)
3. Limited discussion of newer immunogenicity predictors: the review concentrates on classical MHC-binding and combined processing predictors (e.g., NetCTLpan/IEDB) but — by 2020 this was reasonable — more recent immunogenicity models that include TCR-contact features and interpretable ML (published after/around 2020) are not covered; readers should consult updates integrating stability, presentation and TCR-facing features for candidate ranking Tenzer et al., combined MHC pathway modeling
4. Translational gaps: the review notes but cannot close the gap between identified epitopes and effective clinical vaccines/TCR therapies; predicting which presented peptides will recruit high‑quality TCRs (functional avidity, cross-reactivity/off-target risk) remains an open, major challenge — the review could have more emphasised off‑target screening needs and TCR safety pipelines (e.g., T-scan and trogocytosis methods it cites) T-scan high-throughput discovery

Practical, evidence-based takeaways (what to do differently after reading)

• When prioritizing neoantigens, use combined pipelines: WES/RNA‑seq → include non‑canonical sources (splice, retained introns, endogenous retroviruses) → mass‑spec immunopeptidomics when possible → predictors trained on eluted ligand + stability data → functional screens (TMG/mRNA electroporation into autologous DCs or T-scan) Laumont et al., non-canonical neoantigens
• Use longer peptides / antigen‑encoding constructs for validation to mimic natural processing (the TMG and long-peptide approaches referenced in the paper produced clinically useful T cell specificities) Tran et al., KRAS TCR therapy
• Prioritize incorporation of peptide–MHC stability assays and mono-allelic immunopeptidomics data into predictors — this increases true-positive rates over binding-only scores (review notes this pathway and the literature supports it) Stability as immunogenicity correlate

Meta-evaluation metrics (paper scored objectively)

• paper_novelty: 6
• paper_quality: 8
• paper_generality: 8
• paper_usefulness: 8
• paper_reproducibility: 6
• explanatory_depth: 7

Why these scores?

The review consolidates extensive primary literature and provides an evidence-backed critique and practical recommendations — hence strong quality and usefulness. Novelty is moderate because mechanistic components were known; the combination and clinical focus yields incremental novelty. Reproducibility is limited by narrative approach (no primary dataset or pre-registered meta-analysis), and explanatory depth is solid but not theoretical-breakthrough level.

What evidence would disprove the review’s central recommendation?

If future large-scale, prospective trials showed that binding-only pipelines (IC50 cutoffs) consistently matched the sensitivity and specificity of comprehensive processing-aware functional screens for clinically efficacious neoantigens, that would contradict the paper’s claim that processing/context-aware validation is required; similarly, if mass-spectrometry–based presentation maps systematically failed to correlate with T cell recognition in many tumors, that would force re-evaluation.

Actionable next steps (for a lab or computational group)

1. Integrate mono-allelic immunopeptidomics and peptide‑MHC stability training data into your predictor ensemble; retrain and benchmark on held-out validated neoantigen lists.
2. Prioritize functional screens that present antigens via endogenous processing (TMG or mRNA electroporation into autologous DCs) rather than short-peptide pulsing for final validation.
3. Include off-target screening (T-scan / trogocytosis methods) early when developing TCR-based therapies to detect cross-reactivity.

Key citations used in this critique

• Lee et al., J Immunother Cancer (2020)
• IEDB resource update
• Tenzer et al., combined pathway modeling
• T-Scan functional epitope discovery
• Laumont et al., non-canonical neoantigens
• Peptide-MHC stability correlates with immunogenicity