From CRISPR Screens to Programmable Medicine: A Five-Year Outlook

The path from screen to medicine is usually described as a pipeline: discover a target, choose a modality, deliver it. In practice the order is reversed by feasibility. Modalities are chosen because they can be delivered, and targets are chosen because a modality exists.

That inversion explains most of what is likely to happen in the next five years, and most of what will disappoint.

Screens are no longer the bottleneck. Genome-wide perturbation with a readable output is routine, and single-cell readouts have made the output richer than viability.

Three consequences are already visible:

• • Target discovery is cheap; target validation is not. A screen nominates hundreds of genes; the cost has moved to deciding which ones matter in a relevant system.
• • Modality choice has become the first decision, not the last. Base editing, prime editing, interference, epigenetic silencing and activation are selected by what the tissue can accept.
• • The regulatory surface has shifted. Programs are judged on potency assays and durability data rather than on editing efficiency — a change that academic work is still catching up with.

1. In vivo delivery will remain the binding constraint through the period, and the winners will be tissues with solved delivery.Liver, eye, CNS via local administration, and ex vivo-manipulated cells will dominate approvals. Systemic delivery to solid tissue at therapeutic dose remains the hard problem.Refuted if: a systemic, tissue-targeted delivery platform demonstrates therapeutic-level editing in a solid organ in a registrational study.

2. Programmable regulation will outnumber programmable sequence change in early clinical programmes.Silencing a gene, activating a silenced allele, or tuning a programme avoids the irreversibility problem and the genotoxicity conversation. The tools are less mature; the clinical logic is easier.Refuted if: base and prime editing capture the majority of new IND filings in the period, which would imply that precise sequence correction is more clinically tractable than expected.

3. Potency assays will become a visible bottleneck and a publication topic.As products advance, the question "what does this batch do, quantitatively" becomes unavoidable, and the field will need mechanism-linked assays with reference standards. There is very little literature on this now.Refuted if: regulators converge on generic editing-efficiency metrics as potency surrogates, removing the need for mechanism-specific assays.

4. The most-cited papers of the period will be about durability, not about editing.Once several programmes are in patients, the decisive question becomes whether the effect persists, and in what units. Papers that establish durability standards across cell types will accumulate citations quickly.Refuted if: delivery breakthroughs dominate the citation landscape, which is plausible if a genuinely systemic platform arrives.

Three constraints are structural, and a five-year horizon is too short to dissolve them:

The third is the one most often underestimated by technologists. A better editor does not by itself produce a better potency assay; the assay is a separate scientific problem, and it is currently under-invested.

For a lab, three practical implications follow:

• • Build models that can be used by a programme, not only by a paper. Durability in population doublings, allele-resolved data, and a quantitative potency readout are the three additions that make academic work translatable.
• • Choose modality before choosing a system. A model built for knockout cannot answer a variant question, and rebuilding costs a quarter.
• • Treat delivery as a research problem, not a procurement problem. The groups that solve delivery for their cell type produce work others cannot repeat.

For a programme, one implication dominates: the target is the cheap part. Budget for the model and the assay, because that is where timelines actually go.

Five-year forecasts in this field have a poor record. CRISPR therapeutics were predicted to be routine by 2020; they are instead concentrated in a handful of tissues with excellent delivery and a few well-understood diseases.

The reason forecasts fail is consistent: they extrapolate the technology and ignore the pharmacology. Editing gets better on a predictable curve; delivery and durability do not.

So the useful version of this outlook is narrower than it sounds: expect better tools, expect the same bottlenecks, and expect the programmes that succeed to be the ones that chose a tissue where the bottleneck was already solved.

The direction of travel is toward programmable regulation as a clinical modality — not because it is more powerful than sequence editing, but because it is more forgiving of the constraints that actually bind.

If you are building the models that such programmes need — durability, allele resolution, potency readouts — that is the work we do.

Contact EDITGENE to request a project consultation or quote.

This is a forecast about bottlenecks and standards, not a review of clinical results. Programmes in this space are early-stage and modality descriptions are mechanism-based.

Predictions are offered with explicit falsifiers; they are not statements about the safety or efficacy of any candidate therapy.

• • Fellmann C, et al. Cornerstones of CRISPR-Cas in drug discovery and therapy. Nature Reviews Drug Discovery. 2017;16:89–100.
• • Anzalone AV, et al. Search-and-replace genome editing without double-strand breaks or donor DNA. Nature. 2019;576:149–157.
• • Nuñez JK, et al. Genome-wide programmable transcriptional memory by CRISPR-based epigenome editing. Cell. 2021;184:2503–2519.
• • Wang D, et al. Adeno-associated virus vector as a platform for gene therapy delivery. Nature Reviews Drug Discovery. 2019;18:358–378.