New Group Wants 'Phase-Appropriate' Thinking To Retire

By Arnaud Deladeriere, Ph.D., Cell&Gene Consulting Inc.

The term “phase-appropriate” has become a common defense for choices made under pressure — from delaying automation to running early trials with fragile analytical packages. But what does “phase-appropriate” truly mean, and is it serving the field well?

I recently convened a group of industry leaders to tackle this discussion and other critical topics. We’re calling it the Cell&Gene Foundry, and our first conversation set out to test this framing.

I assembled the Foundry members to create an open forum where industry leaders can share hard-won lessons, debate entrenched assumptions, and collectively shape better strategies for cell and gene therapy development.

Over the course of our first conversation, a consensus emerged: the term “phase-appropriate” obscures the real challenge. Decisions in cell and gene therapy are not about aligning with arbitrary clinical phases but about balancing risk. The group proposed a shift in vocabulary and mindset — from “phase-appropriate” to “risk-appropriate” — a truer reflection of the choices developers face.

Here’s Why We Should Reframe Phase-Appropriate To Risk-Appropriate

Unfortunately, we see “phase-appropriate” routinely used as shorthand for shortcuts. Companies under investor pressure race to the clinic, telling themselves that corners can be cut because later phases will allow for correction. In practice, those later corrections rarely happen. Good data in Phase 1 often creates momentum that locks companies into fragile processes, leaving them with steep, costly comparability requirements when moving toward pivotal studies or commercialization.

This is not only a matter of inefficiency — it is a matter of survival. Companies that underestimate the complexity of scaling, automation, or analytics may find themselves stranded with promising science but no path to licensure. At the same time, overbuilding early can drain limited capital and prevent programs from reaching the clinic at all.

The balance lies in recognizing the dual risks: moving too fast with inadequate foundations vs. moving too slow and running out of resources before reaching the point of regulatory or clinical validation. Every decision — whether to automate, to invest in a new assay, or to scale manufacturing — is fundamentally a risk assessment.

Early Development Decisions Can Lock In Fatal Inefficiency

The conversation highlighted the persistent gap that exists between academic innovation and commercial readiness. Academic centers generate world-class science and often initiate first-in-human trials, yet their systems and expectations are shaped by academic norms rather than commercial demands. Assays that work well enough in preclinical labs rarely meet regulatory standards. Processes designed for small-scale proof of concept may be incompatible with industrial manufacturing environments.

Many young companies spin out of academia with angel or seed investment, eager to show clinical proof of concept. Investor expectations, sharper than ever, drive them toward speed. But in doing so, they risk building on shaky foundations. Without a road map to a BLA, they may discover too late that they have locked themselves into inefficient methods or unscalable systems.

Examples discussed illustrated how companies sometimes accelerate through creative routes — such as leveraging investigator-initiated trials in alternative geographies — to generate data that secures partnerships or acquisitions. These strategies can work, but they heighten the importance of clarity: developers must understand what should be done, what can be done, and what risks they are truly taking.

A recurring lesson was the underestimation of timelines by younger companies. Many organizations assume they are closer to GMP manufacturing readiness than they actually are.

Automation And Manufacturing Strategy Dominate The Debate

Few topics sparked more debate than automation. The prevailing industry narrative suggests that automation can wait until after early trials, once proof of concept is achieved. In theory, companies plan to pause after Phase 1, redesign processes, and implement automation. In practice, this almost never happens. Once early data is positive, momentum carries programs forward, and the appetite for disruptive process change disappears.

The group noted examples where this deferral created massive post-approval burdens, forcing companies to conduct extensive comparability studies when major changes to equipment and reagents became unavoidable. In contrast, those that embraced automation early — even at preclinical stages — often found it to be more cost-effective in the long run, saving both money and time.

Still, realities vary. Access is not uniform. Academic centers may acquire bioreactors through capital grants but find themselves unable to sustain consumable costs, reverting to manual methods despite having the right equipment. Smaller companies face the stark calculation of whether up-front automation investment is feasible with limited funds.

The Foundry converged on a pragmatic view: automation is not optional for eventual commercialization. The only question is when it will be adopted and how much cost, disruption, and risk will be absorbed by delaying it. Early integration, where feasible, is nearly always the lower-risk path.

Analytics And Quality Systems Underpin The Issues

If automation is the visible bottleneck, analytics is the invisible one. The Foundry discussion returned repeatedly to the centrality of data integrity and analytical robustness.

Studies and regulatory data were cited showing that about 50% of CMC-related issues stem from analytics — not from manufacturing mechanics. Recent FDA complete response letters reinforced this reality, blocking programs from approval despite promising clinical data because analytical packages were inadequate.

The group observed that many developers begin with self-developed assays that suffice in academic or preclinical contexts but cannot withstand regulatory scrutiny. Translation gaps emerge: assays that track endpoints in animal models fail to map onto clinically relevant human outcomes. For early-stage companies, the challenge is to implement fit-for-purpose quality systems that will govern data integrity. Full validation is unnecessary in Phase 1, but reproducibility, reliability, and basic controls are not optional. Data integrity must be ensured from the start, because every development decision ultimately rests on that data.

Participants stressed the value of moving to electronic quality systems early, especially for organizations collaborating across sites. Paper-based records, still common in startups, slow down communication, complicate product release, and generate inefficiencies that ripple across partnerships.

Analytics, the group concluded, should not be seen as a late-phase hurdle but as a core enabler of the entire development path.

CDMO And Partnership Models Can Add Or Reduce Risk

The discussion then turned to CDMOs, whose role in accelerating — or delaying — programs cannot be overstated.

Experience shows that clients approach CDMOs with vastly different expectations: some want pure execution of a predefined process, some seek guided execution with expert input, and some arrive with only a concept, asking for a process to be built from scratch. CDMOs must adapt to each, but the most successful partnerships are those that align early on both risk appetite and long-term vision.

Examples were shared of programs reaching IND submission in as little as eight to 10 months, made possible by parallelizing workstreams under a single QMS and taking risk-based approaches to regulatory engagement. These are exceptions, but they demonstrate what is possible with strategic planning, integrated capabilities, and decisive execution.

The Foundry also highlighted the persistent industry assumption that companies must own their own manufacturing. While internal manufacturing may make sense for certain strategic reasons, most therapeutic areas rely heavily on outsourcing — and for good reason. CDMOs can offer capital efficiency, regulatory experience, and scalability that young companies cannot replicate quickly.

Ultimately, collaboration — between developers, CDMOs, academics, and regulators — was identified as a critical success factor. No single organization can overcome the systemic challenges of cell and gene therapy development alone.

Key Takeaways

Through the discussion, several collective lessons emerged that could help the industry move forward:

About The Author:

Arnaud Deladeriere, Ph.D., is principal consultant at Cell&Gene Consulting Inc. Previously, he was head of MSAT and Manufacturing at Triumvira Immunologics, and before that, manufacturing manager at C3i. He received his Ph.D. in biochemistry from the University of Cambridge.