Review Papers Claim by Claim

Visual Paper Analysis — Riaz et al., "Tumor and Microenvironment Evolution during Immunotherapy with Nivolumab" (Cell, 2017)

This review prioritizes (1) reproducible, data-linked claims from the manuscript, (2) methodological strengths/limitations, (3) critical alternative explanations and blindspots, and (4) concrete experiments/analyses to falsify or extend the core claims. Visual first, explanation second.

Critical appraisal — strengths

• Prospective design with paired pre/on-therapy biopsies (same site) and multimodal assays (WES, RNA-seq, TCR-seq, IHC) in a human clinical cohort — high translational relevance Prospective paired-sample design
• Integrated clonality (PyClone/FACETS) and neoantigen prediction combined with TCR clonal dynamics — a comprehensive immunogenomic pipeline (code shared) Integrated methods and code availability

Critical appraisal — limitations, biases, blindspots

• Controlled-access raw sequencing data: raw reads are not open; some analyses rely on processed variant calls and neoantigen predictions supplied by authors — this limits independent re-analysis and sensitivity analyses Data access statement
• Neoantigen prediction is inherently noisy: NetMHC-style binding rank <2% was used but many predicted binders are non-immunogenic in vivo; therefore selective depletion of computational neoantigens is supportive but not definitive evidence of epitope-specific T cell killing Limitations of neoantigen prediction
• Sample sizes for some modalities (paired TCR-seq n=34; paired WES n=41; deeper on-therapy sequencing in subsets) reduce power for per-patient neoantigen→TCR mapping and subgroup analyses (Ipi-N vs Ipi-P) — many reported p-values are vulnerable to underpowering and multiple-testing effects Assay-specific sample sizes
• Potential confounders: tumor purity decreased on-therapy and anatomical sampling heterogeneity could mimic apparent loss of mutations — authors performed power/simulation checks but residual confounding is possible Purity and sampling caveat
• Industry funding and author COIs are extensive (BMS/Pharma support); while typical for trials, interpretive caution is warranted and independent replication is needed Funding and COI disclosure

Where the evidence is strongest (confidence notes)

• Robust: cohort-level association that responders show decreased detectable mutation/neoantigen counts on-therapy (genomic contraction) — supported by paired deep sequencing and power analyses (moderate-to-strong evidence) Genomic contraction evidence
• Moderate: proportional expansion of TCR clonotypes correlating with number of neoantigens that disappear on-therapy in responders (correlative but biologically plausible) TCR expansion correlates with neoantigen loss

Key alternative explanations and necessary falsifications

1. If on-therapy mutation loss is due to lower tumor cellularity / sampling heterogeneity rather than immune elimination, then deeper multi-region sequencing and single-cell DNA/RNA would show the "lost" mutations persist in other tumor regions or single tumor cells post-therapy. The authors performed power calculations but independent multi-region and single-cell data would disconfirm immunoediting claims Purity vs editing falsification
2. If predicted neoantigens are false positives, then mass-spectrometry of MHC-I bound peptides or functional T cell assays (ELISPOT / tetramer / TCR re-expression) should fail to show recognition of predicted neoepitopes; absence of functional recognition would weaken the immunoediting interpretation — such orthogonal validation is essential and was not supplied for most cases in this cohort Need for neoantigen validation

Concrete experimental follow-ups (high-priority, falsifying/confirmatory)

1. Mass-spectrometry immunopeptidomics on paired pre/on biopsies to directly observe which predicted neoantigens are presented pre-therapy and whether their presentation disappears on-therapy; correlate MS-validated neoantigen losses with expansion of cognate TCR clonotypes (requires matching TCRseq/T cell functional assays). (Feasible: targeted immunopeptidomics on cryopreserved tissue or larger fresh samples.)
2. Single-cell DNA + TCRαβ sequencing (paired scDNA + scTCR) from pre-and on-therapy samples to directly observe tumor-cell clones being eliminated and concurrent expansion of T cells with matched specificity — this would decisively test immunoediting at cellular resolution.
3. Spatially resolved multiplexed imaging (e.g., CODEX/MIBI/Visium + TCR in situ) of pre/on biopsies to show colocalization of expanding clonotypes with regions where predicted neoantigen-bearing tumor clones disappear, and to rule out replacement-of-sample effects.
4. Functional assays for top predicted neoepitopes: synthesize candidate peptides shown to be lost on-therapy, test pre-therapy TIL or peripheral blood T cell reactivity (ELISPOT, tetramer, TCR-transduced T cells) to validate antigen specificity of expanded clones.

Paper-level scores (critical, evidence-anchored)

• paper_novelty: 9
• paper_novelty_explanation: First sizeable prospective, paired multi-omic study in melanoma showing therapy-associated genomic contraction and proportional TCR expansion consistent with immunoediting; extends earlier small case reports into a cohort-level observation (novel integration of paired WES + TCRseq in human ICB).
• paper_quality: 8
• paper_quality_explanation: High-quality technologies, rigorous statistical checks, and code sharing; limitations are: subset sample sizes for some assays, controlled-access raw data, and reliance on in silico neoantigen predictions without systematic orthogonal validation for the majority of candidates; funding and author COIs well disclosed but abundant.
• paper_generality: 6
• paper_generality_explanation: Findings are specific to melanoma patients treated with nivolumab and affected by prior ipilimumab exposure; conceptual generality (immunoediting under ICB) is broad, but quantitative biomarkers may not generalize across tumor types without validation.
• paper_usefulness: 8
• paper_usefulness_explanation: Provides actionable biomarkers (clonal mutation load, early genomic contraction metrics, on-therapy transcriptomic shifts) and points to combination targets (additional checkpoints/upregulated genes) informing trial design.
• paper_reproducibility: 6
• paper_reproducibility_explanation: Analytical code and processed data were shared; raw sequencing is controlled-access which raises friction for external replication; methods are clearly described enabling replication if data access is granted.
• explanatory_depth: 7
• explanatory_depth_explanation: Integrates evolutionary clonal analysis, neoantigen predictions and TCR dynamics to propose an immunoediting model; mechanistic depth limited by lack of direct antigen validation and single-cell lineage tracing.

Key insight (concise)

Early genomic contraction (loss of clonal neoantigens) plus focused intratumoral TCR clonal expansion is the strongest available human evidence that PD-1 blockade can produce measurable immunoediting within weeks — but proving antigen-specific killing requires orthogonal MHC-peptidomics and single-cell lineage tracing.

How to improve this review / what I would add next

Request or re-analyze raw controlled-access WES/TCR/BAM files (with approved access) to: (1) re-run neoantigen pipelines including mass-spectrometry-informed filters, (2) perform deeper multi-region comparisons, and (3) apply recent frequency-dependent selection models to the observed neoantigen-CCF distributions to distinguish NFDS vs neutral dynamics (see recent work linking frequency-dependent selection and ICB outcomes) NFDS neoantigen modeling relevance

Selected primary citation (this paper)

Selected secondary citations (context & methodological cautions)

• Neoantigen validation caveats
• Neoantigen selection models

Novel hypotheses (testable)

1. In Ipi-naive patients, early on-therapy genomic contraction selects against clonal neoantigens preferentially presented on HLA-A alleles with high peptide-binding promiscuity, producing durable benefit; falsify by combining immunopeptidomics and HLA-allele-specific T cell assays across responders/non-responders.
2. Patients whose tumors show NFDS-like neoantigen frequency patterns (many low-frequency antigenic variants) are less likely to exhibit genomic contraction and durable responses to single-agent PD-1 blockade; falsify by applying frequency-dependent selection metrics to pre-therapy WES and correlating with on-therapy contraction and outcomes (testable with existing cohorts).

Novel experiments (concise)

1. Mass-spectrometry immunopeptidomics of matched pre/on biopsies to directly measure MHC-I bound neoantigens and test whether predicted neoantigens that 'disappear' are no longer presented on-tumor cells post-therapy (high specificity test for immunoediting).
2. Single-cell DNA + TCRαβ multi-omic assay on pre/on biopsies to map tumor subclonal extinction to expansion of TCR clones at single-cell resolution, resolving tumor-purity confounding and enabling per-neoantigen/cognate-TCR mapping.