Late stage delays in gene therapy programmes are often blamed on manufacturing scale up, comparability failures, or unexpected regulatory questions. But in many cases, the real cause sits much further back in the programme - hidden in analytical decisions made when timelines felt flexible and expectations seemed distant. Most gene therapy programmes do not fail analytically because the science is wrong. They fail because the analytical strategy was never really a strategy at all. Instead, methods were built to answer immediate questions, not future ones. Assays were selected because they worked quickly, not because they could evolve. And analytical development progressed in parallel with the programme - rather than as an integrated driver of it. By the time those choices are exposed, they are no longer easy to change.
In early development, speed dominates. Programmes move fast, data needs are immediate, and analytical methods are often treated as tactical tools-developed to support the next study, the next batch, or the next decision point. That approach can work in simpler modalities. In gene therapy, it rarely does. Gene therapy analytics must span complex product attributes, evolving mechanisms of action, and regulatory expectations that sharpen dramatically as programmes advance. Yet many analytical strategies are assembled piece by piece, often in silos: one team focusing on vector characterisation, another on potency, another on release testing.
Early decisions, assay formats, reference standards, acceptance of variability, these are determiend without a clear view of how those methods will be used, validated, or defended years later. As a result, analytical development becomes a series of reactive
fixes rather than a coherent system. This challenge is magnified by the diversity of gene
therapy modalities. Viral vectors, lipid nanoparticles, gene editing systems, and mRNA platforms all bring distinct analytical demands. Applying a generic mindset, or reusing approaches from adjacent modalities without adaptation, creates fragile foundations that tend to crack under late stage pressure.
Early assays that don’t scale with the programme
Many programmes rely heavily on early stage assays that are fit for discovery but fundamentally misaligned with later needs. These assays may be semi quantitative, highly operator dependent, or sensitive to subtle changes in materials or process conditions. Early on, that variability is tolerated. Data trends are often more important than absolute values, and decisions are directional. Problems emerge when those same assays are expected to support comparability, validation, or regulatory filings.
At that point, teams discover that the assay cannot meet precision requirements, cannot be transferred cleanly, or does not truly reflect the critical quality attribute it was assumed to measure. The result is re-development under time pressure, often alongside active clinical timelines. In one common scenario, a potency assay developed around a convenient in vitro readout works well enough for early studies, only to prove poorly correlated with clinical relevance or manufacturing changes later. Re-thinking potency late in development is rarely straightforward and regulators are acutely sensitive to it.
Misalignment between R&D and QC realities
Another frequent failure point is the gap between R&D friendly methods and QC ready ones. Assays optimised for flexibility, exploration, or rapid iteration are not always compatible with the robustness, simplicity, and control expected in a QC environment. When analytical development does not actively consider the eventual QC context, method transfer becomes a bottleneck. Assays that rely on specialised expertise, bespoke reagents, or complex data interpretation can stall as they move closer to commercial expectations. This misalignment often surfaces late, when timelines are compressed and tolerance for change is low. What seemed like a reasonable compromise early on becomes a structural risk later.
Underestimating comparability and method evolution
Comparability is not a single event; it is a recurring analytical challenge across a programme’s lifetime. Process changes, scale up, site transfers, and raw material shifts all place increasing demands on analytical methods. Yet many strategies treat methods as static. Assays are developed to support a specific phase, without a clear plan for how they will evolve-or how historical data will be bridged as changes occur.
When methods must be upgraded or replaced, the absence of a lifecycle strategy can turn routine improvements into regulatory concerns. Questions arise around data continuity, relevance of legacy results, and the impact on established specifications. Late stage comparability failures are rarely caused by a single change. They are the cumulative result of analytical strategies that did not anticipate evolution.
Treating methods as tasks, not a system
Perhaps the most fundamental issue is the tendency to treat analytical methods as isolated deliverables rather than as parts of an interconnected system. Potency, purity, identity, and structural characterisation are often developed independently, each optimised within its own context. But regulators assess products holistically. They expect coherence across CQAs, consistency in how attributes are defined, and alignment between analytical data and the proposed mechanism of action.
When methods are developed in isolation, gaps appear. Data may be difficult to interpret collectively, or different assays may tell subtly different stories about the product. Reconciling those narratives late in development can be challenging-and sometimes
impossible without rework.
Why this problem is getting worse, not better
Several industry trends are amplifying these risks. Gene therapy modalities are diversifying rapidly, with increasingly sophisticated constructs and delivery systems. Analytical expectations are rising accordingly, particularly around potency, heterogeneity, and functional relevance. Regulators have also become more experienced. Early flexibility has given way to more structured expectations, especially as programmes move toward late stage development and commercialisation.
At the same time, development timelines are under constant pressure. Accelerated pathways and competitive landscapes push teams to move fast, often at the expense of long term analytical planning. The result is a widening gap between early analytical choices and late stage demands.
The shift in thinking that’s required
Avoiding these failures does not start with better assays. It starts with a different mindset. Analytical strategy needs to be treated as a core component of product development, not a supporting function. That means thinking early about how methods will need to evolve, how CQAs will be integrated, and how data generated today will be used and scrutinised - years from now.
Methods should be designed with change in mind, not locked into a single phase or purpose. Alignment across teams, functions, and development stages is essential. This is not about slowing programmes down. It is about building analytical foundations that can support speed without sacrificing credibility.
Where deeper complexity lies
Several areas consistently demand deeper consideration than they initially receive. Potency assessment remains one of the most challenging, particularly when mechanisms of action are complex or indirect. Structural characterisation, including particle heterogeneity and integrity, continues to stretch available tools and interpretations. Lifecycle strategy-how methods, specifications, and data sets evolve together-often determines whether late stage changes are manageable or disruptive.
Each of these topics warrants focused attention in its own right.
A closing thought
Gene therapy success depends on far more than biological innovation. Analytical strategy plays a quiet but decisive role in determining whether a programme advances smoothly or struggles under late stage scrutiny. The decisions made early, often under pressure and with incomplete information set constraints that are difficult to escape later.
As the field matures, programmes that recognise analytics as a strategic discipline, rather than a reactive necessity, will be better positioned to succeed. The industry is already moving in that direction. The question is how many programmes will get there in time.