Right-First-Time Manufacturing KPIs for Cell Therapy Sites

Why Cell Therapy Breaks the Conventional RFT Model

Conventional pharma RFT works because the batch is the natural unit of measurement. A tablet batch contains hundreds of thousands of units, runs through standardized equipment, and produces statistical noise that smooths individual variability into a manageable signal. RFT calculated as the percentage of batches released without deviation or rework is meaningful because the denominator is large and the failure modes are predictable.

Cell therapy inverts this. An autologous CAR-T batch is one patient. The cells are living biological material with intrinsic donor-to-donor variability. The manufacturing process is open at multiple steps, manipulation-heavy, and dependent on individual operator skill in ways that tablet manufacturing eliminated decades ago. A single-batch failure is a patient who does not receive their treatment, which carries clinical and ethical weight that does not exist in small-molecule manufacturing.

The result is that conventional RFT calculations produce one of two unhelpful outcomes when applied directly. Either the RFT denominator is too small to be statistically meaningful (a site producing 200 patient batches per year has too few data points for traditional control charts), or the definition of “deviation” is stretched so broadly that almost every batch has at least one minor finding, driving RFT toward zero in a way that does not reflect actual performance.

The PDA’s work on cell and gene therapy manufacturing quality has consistently emphasized that the quality framework needs to be adapted, not just translated. Industry consortia including BioPhorum’s Cell & Gene Therapy workstream have published reference frameworks that articulate how conventional concepts including RFT need to be reformulated for the modality.

The Cell Therapy RFT KPI Set

A defensible RFT program for a cell therapy site rests on a set of complementary metrics that together capture what conventional RFT was designed to capture, calibrated to single-patient batch realities.

These KPIs are designed to be tracked together rather than individually. A site with a 95% Manufacturing Success Rate but a 70% First-Pass Release Rate is telling a different story than a site at 92% success and 95% first-pass. The pairing reveals whether process variability is concentrated in the bioprocessing phase or in the testing and release phase, which drives very different remediation priorities.

Single-Patient Batch Metrics

Single-patient batches require a metric that conventional pharma does not have: a metric that captures patient-level outcome rather than batch-level performance. A failed autologous batch is not just a manufacturing event; it is a patient who has to undergo apheresis again, or who progresses to alternative treatment, or who in the worst case loses the clinical window for the intended therapy. RFT programs at cell therapy sites should include a Patient-Level Manufacturing Success metric that tracks the percentage of enrolled patients who ultimately receive their manufactured product, including rerun batches.

This metric is meaningfully different from Manufacturing Success Rate. A site with a 95% batch success rate might still have only 92% patient-level success if rerun batches occasionally fail or if apheresis material is insufficient to support a rerun. The patient-level metric captures the lived experience of the patient population the site serves and is the metric that clinical and commercial stakeholders should be reading.

Single-patient batches also require careful thinking about the unit of analysis for control charts. Conventional Statistical Process Control assumes that successive batches are independent samples from a stable process. In cell therapy, the patient is the source of starting material, and patient-to-patient variability is a primary driver of process variation. Control charts should be calibrated to recognize this and avoid flagging biological variability as process drift.

The right framing is to treat patient-attributable variability separately from process-attributable variability. Apheresis quality metrics (cell viability, CD3+ count, contamination flags) should be tracked as inputs to the manufacturing process, and downstream KPIs should be stratified or normalized against them where the population is large enough. Sites with sufficient volume can build empirical correction factors that adjust for known starting-material effects.

Deviation Tracking at Cell Therapy Velocity

Conventional pharma deviation programs run at a cadence that suits batches with 30 to 90 day disposition cycles. Cell therapy operates at 7 to 14 day cycles for autologous products, which means deviation investigations need to close within days, not weeks, to keep the production pipeline flowing. The conventional CAPA cycle, with its multi-week investigation and effectiveness check timeline, is fundamentally incompatible with cell therapy operating velocity.

Sites running successful cell therapy QMS programs have developed a tiered deviation classification that triages investigations by their potential to affect product quality. Minor deviations follow a streamlined documentation and disposition flow that completes within 48 hours; major and critical deviations follow the full investigation cycle but are designed to complete within the patient-relevant timeline.

The tiered classification has to be defensible to inspectors. A site that uses streamlined disposition for true major deviations will not survive inspection. The right framing is that the classification is rigorous, the streamlined path is reserved for genuinely minor events, and the documented criteria for classification are clear enough that an inspector can audit them.

The ICH Q10 and Q9 frameworks support this kind of risk-based deviation management when implemented with discipline. The Q9 quality risk management principles, in particular, give sites a defensible basis for tiered deviation handling that is calibrated to actual risk rather than to operational convenience.

The 24-Hour Release Window Constraint

Autologous cell therapy products typically have a release window measured in hours rather than days. The cells have to be infused into the patient within a clinically defined window after manufacturing completes, which means release testing and disposition decisions have to compress into a 24 to 48 hour window. This constraint reshapes what RFT can practically measure and what counts as a “minor” finding.

The first practical implication is that release testing has to be designed for speed. Compendial tests with multi-day incubation periods (such as conventional sterility testing) require either rapid alternative methods or risk-based disposition pathways that release product on a documented basis while the conventional test runs in parallel. The compendia and regulators have been increasingly accommodating of rapid microbial methods for cell therapy specifically because of this constraint.

The second implication is that the QA review cycle for batch records has to be designed for compression. Conventional batch record review at 4 to 8 hours per batch with sequential reviewer signoffs does not fit a 24-hour release window. Sites running successful cell therapy programs have flattened the review hierarchy, used parallel reviews where possible, and pre-validated batch record templates that minimize the variable content requiring review.

The third implication is that RFT classification has to account for findings discovered after the release window has closed. A finding that emerges during retrospective review of a batch that has already been infused into the patient is a different category of event than a finding discovered before release. Both need to be tracked, but they are different metrics with different operational implications.

Governance and Reporting Cadence

RFT programs at cell therapy sites need a reporting cadence that matches the operational tempo. Monthly Quality Council reviews are appropriate for the conventional pharma cadence; cell therapy programs benefit from weekly trending reviews supplemented by monthly executive summaries. The weekly cadence is short enough to detect emerging patterns before they generate a cluster of failures and long enough to allow meaningful trend analysis.

The weekly review should be structured around the KPI set, with each KPI presented with its current value, trend over the past quarter, and any flagged anomalies. The discipline of presenting the same metric set every week, in the same format, allows the team to develop pattern recognition that is impossible when the format changes. The monthly executive summary should aggregate the weekly trends into a narrative that the broader leadership team can read in 15 minutes.

The reporting infrastructure has to support this cadence. A site running cell therapy RFT on spreadsheets and manual aggregation will not sustain the cadence; the manual workload exceeds the available QA bandwidth. The investment in a real-time dashboard that pulls from the MES, LIMS, and QMS is, in practice, a prerequisite for running the program at the cadence that the modality requires. Industry experience documented in BioPharm International and similar publications has consistently emphasized that dashboard infrastructure is not optional for cell therapy QMS at scale.

Building the Program from Scratch

For sites building an RFT program from scratch, the right sequence is to install the KPI definitions and reporting cadence first, then build the dashboard infrastructure, then layer on the deviation tracking discipline and patient-level metrics, and finally integrate the program into the formal Quality Management Review cycle. Attempting to install everything simultaneously almost always produces a program that is partially complete in every dimension and fully complete in none.

The first 90 days should focus on getting the seven core KPIs defined, the data sources identified, and an interim reporting cadence in place even if the data flow is still partly manual. The second 90 days should focus on automating the data flow and adding the patient-level metric. The third 90 days should focus on integrating the program into the formal governance cycle and adding the more advanced metrics including cycle time variance and patient-attributable variability stratification.

The investment is meaningful but bounded. A site running 200 to 400 autologous batches per year should expect to invest 4 to 6 months of dedicated work from a quality engineer plus 2 to 3 months of integration work from IT to land the program. The payoff is a quality system that gives leadership real visibility into manufacturing performance and gives the site a defensible posture during inspection.

One additional point worth emphasizing: the KPI set should be revisited every 12 to 18 months. Cell therapy manufacturing is still maturing as a modality, and the metrics that captured what mattered in 2024 may not be the right metrics in 2027. The right governance posture is to treat the KPI set as a living artifact, with periodic review by the Quality Council and explicit decisions to add, retire, or recalibrate metrics as the program matures.

References & Sources