Case #002: Eighteen Months of Stability Data. One Degree of Drift.

01The study

The product was a film-coated tablet — a fixed-dose combination destined for a European regulatory submission. The development programme had taken four years. The stability study was the last major deliverable before the dossier could be filed.

The study followed ICH Q1A(R2). Long-term conditions: 25°C ± 2°C / 60% RH ± 5% RH. Accelerated conditions: 40°C ± 2°C / 75% RH ± 5% RH. Time points at 0, 3, 6, 9, 12, and 18 months. Three batches. Primary packaging — aluminium blister, as specified in the target market submission requirements.

The stability chamber used for the long-term study was a walk-in unit, installed two years earlier. It had been qualified at installation. The IQ had been completed, the OQ had been completed, and the documentation was in order. The chamber had been in continuous use since qualification. It had an alarm system. It had a continuous data logger. The data showed consistent conditions throughout the eighteen months of the study.

Or so it appeared.

02The submission

The regulatory dossier was filed at month twenty-two — four months after the final stability time point. The submission was thorough. The stability data showed no significant changes at any time point. Degradation products were within specification. The assay results were consistent. The physical characteristics were unchanged.

The regulatory assessor assigned to the dossier began her review with the pharmaceutical development section. She moved to the manufacturing section. She reviewed the analytical methods. When she reached the stability section, she read through the data, noted the results, and then turned to the qualification documentation for the chamber used in the long-term study.

She had one question.

03The question

The question arrived as a formal query in the regulatory assessment process. It was a single paragraph. It asked the applicant to confirm whether the operational qualification of the stability chamber used for the long-term study had been conducted under loaded conditions — that is, with a representative sample load present in the chamber during temperature mapping — or under empty conditions.

It was not a hostile question. It was a routine question. The assessor had asked it dozens of times. The answer was usually straightforward.

04The answer

The quality team pulled the OQ report. It was a clean document — well-formatted, fully signed, with temperature mapping data from nine probe positions across the chamber volume. The mapping had been conducted over 72 hours. The results showed uniformity of ±0.6°C across all positions. Well within specification.

The mapping had been conducted with the chamber empty.

There was no record of a loaded qualification. There had been no PQ with a representative product load. The chamber had been mapped, qualified, and placed into service for long-term stability studies without anyone verifying that its thermal performance under load matched its thermal performance when empty.

The team reviewed the continuous monitoring data from the eighteen months of the study. The data logger positioned at the chamber air sensor showed consistent conditions throughout. What it did not show — because no thermocouple had been positioned there — was the temperature at product level within the loaded chamber.

An engineer was sent to the chamber with a calibrated logger. He positioned probes at product level, loaded the chamber to a representative fill level, and ran a 48-hour mapping exercise.

The results came back two days later.

At product level, in the loaded configuration, the temperature in the upper rear zone of the chamber was consistently 1.1°C above the setpoint. Not 2°C. Not enough to trigger the alarm — the alarm was set at ±2°C on the air sensor, not at product level. But enough to mean that the product in that zone had been stored at 26.1°C for eighteen months instead of 25°C.

05Root cause

The investigation identified a single proximate cause and two systemic causes.

The OQ was conducted on an empty chamber. This is the proximate cause. A walk-in stability chamber is a large thermal mass. When loaded with product, the airflow pattern inside the chamber changes, the thermal equilibrium shifts, and hot and cold spots migrate. The temperature uniformity demonstrated in an empty qualification is not representative of the temperature uniformity achieved under operational loading. For a walk-in chamber used for regulatory stability studies, the OQ must include a loaded mapping — and the load must be representative of the maximum anticipated product density.

The qualification protocol did not specify loaded conditions. The protocol used for the OQ was a standard template. It specified the number of probe positions, the duration of the mapping, and the acceptance criteria. It did not specify that the mapping must be conducted under loaded conditions. The engineer who conducted the qualification followed the protocol as written. The protocol was wrong.

The monitoring system was calibrated to the wrong reference point. The continuous data logger monitored the chamber air sensor — a probe positioned near the chamber controller, in an area of high airflow and relatively uniform temperature. It was not positioned at product level. The alarm limits were set against this sensor. A chamber that was out of specification at product level appeared in-specification on the monitoring report, because the monitoring point was not where the product was.

06What should have happened

The OQ protocol should have required a loaded temperature mapping. A walk-in stability chamber used for ICH Q1A studies should be qualified under conditions that represent actual use — which means loaded, or at minimum mapped under both empty and loaded conditions to understand the effect of loading on thermal distribution.

The monitoring system should have included probes at product level, not only at the chamber air sensor. A monitoring programme that alarms when the air sensor deviates but does not monitor conditions at product level is a monitoring programme that protects the chamber, not the product.

The PQ should have been conducted before the stability study started, with a representative load, to demonstrate that the chamber maintained conditions within specification in its actual operational configuration. EU GMP Annex 15 is explicit: qualification must demonstrate fitness for purpose under conditions of use.

06What changed after

The regulatory submission was placed on hold. The affected stability samples were re-assigned to the valid zones of the chamber — the lower two-thirds where temperature had remained within specification. The protocol was amended to exclude data from the upper rear zone. A supplementary loading study was initiated to re-generate the missing data points.

The regulatory submission was delayed by fourteen months.

The company revised its chamber qualification procedure. The new protocol required a full loaded temperature mapping for all walk-in stability chambers, with probes positioned at product level across the full chamber volume. Monitoring probes were repositioned to product-level locations in addition to the air sensor. Alarm limits were set against product-level sensors, not air sensors.

The chamber that caused the problem was re-qualified. It passed — after the air distribution system was adjusted to correct the hot zone in the upper rear. The adjustment took two days. The information needed to make it had been available eighteen months earlier, if anyone had looked.