In Vivo Genome Editing Drug Products — CBER Regulatory Strategy for Direct In Vivo Patient Administration

In vivo genome editing programs face a safety characterization challenge that ex vivo programs do not: when the editing machinery is delivered directly to the patient, every tissue reached by the delivery vehicle is a potential site of off-target genome editing. An LNP that accumulates 90% in the liver and 10% in the spleen at the proposed clinical dose delivers genome editing machinery to hepatocytes and to splenic immune cells simultaneously — and CBER expects the off-target editing characterization plan to address both tissues, because the clinical consequence of off-target editing in splenic CD4+ T cells is different from, and potentially more serious than, the same off-target editing event in terminally differentiated hepatocytes.

In vivo genome editing CMC regulatory strategies fail at CBER pre-IND review not because the delivery platform is scientifically unvalidated, but because the off-target editing characterization plan is scoped to the target tissue only — leaving CBER without the tissue-specific off-target analysis for tissues reached by the delivery vehicle, the germline transmission risk assessment for gonadal biodistribution, and the organ selectivity documentation that justifies excluding non-target tissues from the off-target characterization scope.

In Vivo Delivery Biodistribution and the Off-Target Tissue Scope — Why the 1% Accumulation Threshold Defines the Minimum CBER-Expected Characterization Plan

Ionizable LNP systems formulated for systemic administration preferentially accumulate in the liver — typically 70–90% of the injected dose in hepatocytes — with secondary accumulation in the spleen, commonly 5–15% of the liver accumulation level for standard four-component ionizable LNP systems, driven by phagocytic uptake in splenic macrophages and dendritic cells. This secondary accumulation is not a rounding error in the biodistribution study; it represents delivery of active editing machinery to an entirely different tissue and cell population than the intended target. The off-target editing characterization scope for an in vivo LNP program should address every tissue where accumulation reaches or exceeds approximately 1% of the target organ accumulation at the proposed clinical dose, based on non-clinical biodistribution data generated at the clinical dose itself — not a surrogate tracer dose, since LNP tissue distribution can shift with dose as hepatocyte uptake machinery approaches saturation at supratherapeutic levels. For a liver-targeting program showing liver (90%), spleen (9%), lung (3%), and kidney (1.5%) accumulation, the minimum off-target characterization scope includes spleen, lung, and kidney, each requiring tissue-specific cell isolation — splenocytes by density gradient separation, for example — followed by genome-wide off-target detection (GUIDE-seq or BEACON-seq depending on the editing modality) in each tissue type separately, because a single off-target analysis performed only on the target tissue answers a narrower safety question than the one CBER is asking about a systemically delivered product.

Germline Transmission Risk Assessment — The 1-Copy-per-1,000-Cells Threshold and the Non-Clinical Study Design CBER Expects Before IND Clearance

A germline transmission risk assessment is required whenever non-clinical biodistribution detects vector genome copies or editing machinery in gonadal tissue at the proposed clinical dose — and “the delivery vehicle is non-integrating” is not, by itself, a sufficient basis for omitting this assessment, because non-integration does not preclude transient editing activity reaching dividing germ cells such as spermatogonial stem cells. The empirically defined regulatory threshold for negligible germline transmission risk is 1 vector genome copy per 1,000 cells (0.001 copies/cell average); below this level, the probability of heritable germline editing is considered negligible given the editing efficiency per cell that heritable modification would require. The study design CBER expects: non-clinical biodistribution in male and female animals at the proposed clinical dose administered by the intended clinical route, gonadal tissue isolation (testis and ovary analyzed separately) at the timepoint of maximum systemic exposure, qPCR quantification of vector genome copies per cell using transgene- or mRNA-specific primers with a quantification limit at or below approximately 0.0001 copies/cell (requiring at least 1,000 cells of input per reaction), and a direct comparison of the measured gonadal copy number against the 1-per-1,000-cells threshold. Programs whose IND submission asserts negligible germline risk based on the delivery vehicle’s mechanism of action alone, without gonadal biodistribution data to support that assertion, have made an argument CBER cannot independently verify.

Organ Selectivity and In Vivo Potency — The Biodistribution-to-Potency Bridge That Converts Non-Clinical Data Into the IND Safety Rationale

For programs using organ-selective LNP formulations — engineered through lipid composition adjustment to preferentially target an organ beyond the liver, such as lung endothelium — the organ selectivity ratio (target organ accumulation relative to the nearest off-target organ) is the parameter that determines how narrowly the off-target characterization scope can be justified. A selectivity ratio of 10:1 or greater between the target organ and the nearest off-target tissue supports a correspondingly narrower off-target analysis plan, measured in non-clinical biodistribution studies using a reporter payload (luciferase mRNA or a fluorescent lipid tracer) at the proposed clinical dose, with organ-specific signal normalized to organ mass. This selectivity data must, in turn, connect to the potency assay: for a liver-targeted in vivo editing program, the accepted Phase 1 potency surrogate systems are primary human hepatocytes (with on-target editing efficiency measured by amplicon sequencing, commonly benchmarked around ≥40% indel frequency for nuclease-based editing or an equivalent base conversion threshold for base editing, within 48 hours of treatment) or hepatocellular carcinoma cell lines such as HepG2 or Huh7 as an early-phase release surrogate. A potency assay measuring LNP uptake by flow cytometry — the percentage of treated hepatocytes positive for a fluorescent lipid label — does not predict editing efficiency, because uptake is necessary but not sufficient for editing: endosomal escape, mRNA translation, and nuclease or effector activity must all still occur downstream of uptake, and a program substituting uptake measurement for an editing-efficiency endpoint has measured a precondition rather than the drug product’s actual biological activity.

The XGene In Vivo Genome Editing Safety Architecture

1. Non-clinical biodistribution study design — target and off-target organ accumulation quantified at the proposed clinical dose using the actual clinical formulation, with organ selectivity ratio documentation (≥10:1 benchmark for targeted delivery claims). 2. Tissue-specific off-target editing characterization — cell type isolation for every tissue meeting the ~1% accumulation threshold, with genome-wide off-target detection performed separately in each tissue type. 3. Germline transmission risk assessment — gonadal qPCR at the proposed clinical dose, evaluated against the 1-copy-per-1,000-cells threshold, regardless of the delivery vehicle’s integration status. 4. In vivo potency assay design — primary hepatocyte or HepG2/Huh7 assay systems measuring on-target editing efficiency by amplicon sequencing, not LNP uptake alone. 5. Pre-IND CBER scope confirmation — a written request establishing CBER’s position on off-target tissue scope and germline study requirements before the IND CMC architecture is finalized.

An in vivo genome editing CBER regulatory strategy earns confidence not by demonstrating that the delivery platform reaches its intended target organ, but by demonstrating — through dose-appropriate biodistribution data, tissue-specific off-target characterization, and a germline risk assessment grounded in actual gonadal measurement — that every tissue the delivery vehicle reaches has been accounted for in the safety package.

For your in vivo genome editing program, can you identify today whether your non-clinical biodistribution study was conducted at the proposed clinical dose using the clinical LNP formulation — and whether your off-target editing characterization plan addresses every tissue where LNP or AAV accumulation exceeded 1% of the target organ accumulation at that dose?