Review Papers Claim by Claim

Paper: “Recent advances in gene therapy for neurodevelopmental disorders with epilepsy” (10.1111/jnc.15168)

• These counts reflect how many therapy/vector/delivery strategies are explicitly shown as rows in the paper’s Table 2 excerpts you provided—not the total number of strategies in the entire field. Table 2 scope
• Row counting is a reconstruction from the provided full-text/table excerpt; it is not a field-wide meta-analysis.

• Category A: genes affecting transcriptional regulation → often suited to gene supplementation/replacement strategies, but expression “dosage windows” and global effects can be limiting (e.g., MECP2 overexpression toxicity). Transcriptional regulation therapies and dosing window
• Category B: genes changing intrinsic excitability (notably ion channels) → sometimes require precise cell-type-biased modulation; for large genes, AAV payload limits can push toward endogenous regulation (CRISPRa/i), ASOs, or alternative editing paradigms. Ion channel therapy constraints
• Category C: genes impacting synaptic transmission → may be less amenable to classical replacement when splice/endogenous processing is disrupted, and might motivate CRISPRa/i or RNA/epigenome editing approaches. Synaptic protein gene therapy limitations

• The paper emphasizes that CNS delivery is a key hurdle and contrasts local stereotactic delivery (IC, IH, ICV) with systemic routes intended to cross the BBB (e.g., IV, IT), noting that peripheral delivery can be inefficient at current vector performance levels. Delivery route and efficiency hurdles
• It also discusses immunogenicity concerns and pre-existing anti-AAV antibodies as impediments to clinical translation. AAV immunity as a barrier

• The paper suggests gene therapy may need to be administered during development (possibly even embryonically) because mature circuits and downstream phenotypes may become harder/impossible to rescue later; however, it also states this is not confirmed and likely needs empirical testing per disorder/treatment. Early therapeutic window uncertainty
• It provides disease-specific examples of differential rescue windows in Angelman models (e.g., Ube3a reinstatement effects differing by developmental stage). Angelman window dependence

• Off-target editing and long-term expression risk: CRISPR/Cas delivery strategies are discussed with concern that persistent expression may increase off-target activity and immunogenic responses to bacterial Cas proteins; anti-CRISPR “off-switches” are proposed but not yet mature enough for all clinical contexts. Cas expression risk and anti-CRISPR concept
• Delivery-vector species/strain dependence: Engineered AAV capsids (e.g., AAV-PHP variants) can show strong CNS tropism in specific mouse strains but may lose performance outside that context. AAV-PHP strain dependence limitation
• Base editing constraints and PAM/bystander effects: The review notes that base editing is constrained by PAM recognition and that widening PAM compatibility in “non-G PAM” strategies could increase available target sites while also potentially raising off-target risks. Base editing PAM constraint and tradeoffs

• Construct validity vs face validity: The paper argues that effective translation requires both genotypic similarity (construct validity) and phenotypic similarity (face validity), and it highlights behavioral/seizure phenotyping as necessary for face validity. Model validity criteria
• Species differences: failure modes are expected, not “surprises”: It points out that animal models often fail to present the full human phenotype spectrum and can miss species-specific developmental/circuit dynamics. Species differences in phenotypes
• In vitro maturation limitations: For hiPSC-derived neurons, it flags that short differentiation/maturation windows can give counterintuitive results, and emphasizes functional maturation/coculture/glial inclusion as critical for meaningful cellular phenotypes. hiPSC maturation dependence
• Organoid batch variability & network incompleteness: For 3D organoid models, it notes stochastic differentiation and batch variability (for undirected organoids), and describes that directed forebrain spheroids may omit inhibitory interneurons unless fused with subpallial models. Organoid variability and microcircuit incompleteness

• The paper’s central structure—mechanism-based gene grouping and mapping to gene-therapy strategy classes—is coherent and repeatedly justified with examples across disorders. Mechanism→therapy alignment
• Delivery and timing constraints (plus mosaicism) are emphasized as first-order translation determinants, with explicit admission that “critical windows” remain unconfirmed in general and likely vary per intervention/disorder. Translation constraints with uncertainty

• Narrative-review bias risk: as a narrative synthesis, it can overweight well-studied disorders/models and underweight negative or inconsistent translational results; the paper does mention multiple failure modes (e.g., AAV strain dependence, hiPSC maturation limits, editing risks), but it does not provide formal systematic-review methodology or quantitative meta-analysis. Narrative synthesis nature
• Human translation evidence gap (as of the review’s publication): it frequently grounds mechanistic plausibility in preclinical findings while acknowledging that delivery, timing, off-target, and long-term safety remain core unknowns. Translational validation uncertainty
• Model-to-mechanism mapping incompleteness: for transcriptional regulators and indirect effects, models may have variable phenotypic similarity and may not reproduce the full human phenotype progression. Indirect effects and model variability