The International Pharmaceutical Federation section for laboratories and medicines control mounted a half-day symposium at the Vienna congress on August 31. The symposium examined four aspects of good control laboratory practice: the value of harmonising pharmacopoeial specifications, reliance on reference materials, the “qualification of analytical equipment and how to treat results which appear to fall outside compliance limits. Professor Geoffrey Phillips reports
Dr Mike Morris (Irish Medicines Board, Dublin) examined “the relevance
of ICH topic Q6A to the control laboratory”. He first summarised the history
and outcomes of the ICH (International Con-
ference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals
for Human Use) process and then explained how group Q6A sought globally uniform
pharmacopoeial specifications. These would enhance harmonisation of traffic
between countries, minimise regulatory differences between compendia, and reduce
overall costs to supplier and to patient. He said that common specifications
should (i) contain tests, with attributes, that specified analytical procedures
and defined acceptance criteria, (ii) establish criteria with which to determine
whether a drug substance or product was acceptable for its intended purpose,
and (iii) provide legally binding standards agreed between industry and regulators.
To maintain efficacy of product (for its intended purpose) and assure patient
safety, such specifications had to evolve, reflecting current capability of
product technology and analytical development, while recognising the total database
experience with that product. Dr Morris examined some key factors in that total
experience — the elucidation of impurities and setting limits thereto,
selection of dissolution conditions, the effect of particle size, sufficient
early data for realistic compromise between limits sufficiently tight to satisfy
the regulator and broad enough for realistic production, and justification of
excluding tests such as limit of extractables and microbiology.
Levels of specification
He distinguished several levels of specification. An initial release specification,
which guaranteed marketing quality, relied on precision of manufacture, critical
review of batch data and a tight assay range (eg, the EU requirement for ±5
per cent), which was refined in the light of experience. A full product shelf-life
specification took account of impurities, content of active drug and storage
guidance; wider limits were substantiated by probably ongoing stability data.
Verification by the control laboratory invoked decision on the frequency and
proportion of batch-testing, and required alternative test procedures to be
fully validated and compared with the methods approved at registration. Similarly,
the methods — usually compendial — employed by control laboratories
had to be validated and independently reproducible. Dr Morris reviewed the ICH
quality guidelines on analytical validation (Q2) — demonstration of the
now widely agreed characteristics of precision, accuracy, “specificity”
(or “selectivity”), linearity and range of calibration, limits of
“detection” and “quantification”, and “ruggedness”
(ie, limited sensitivity to small changes in conditions or equipment). These
parameters variously applied to test methods for identity, assay and impurity
content, and frequently relied on correlation with suitable traceable authentic
reference substances. Dr Morris discussed the different levels, and purpose,
of reference material available [more fully described in Dr Hildebrand’s
presentation — see below]. He confirmed the role of the control laboratory
which had to verify its analytical methods, decide on realistic batch analysis,
demonstrate compliance with the shelf-life specification, and employ independently
transferable — preferably international compendial — procedures. Mutual
acceptability (and transferability) was enhanced if the pharmacopoeias could
harmonise their optimal test methods, designate exchangeable reference materials,
and regularly participate in suitable inter-laboratory proficiency studies.
He concluded that with easier comparability of compliance, quality could be
maintained while costs would be reduced.
The theme of essential reliance on the “use of reference materials in
the control laboratory” was developed by Dr Michael Hildebrand (head of
QC at Schering, Berlin). The common impression of reference materials (RMs)
was that of substances of the highest obtainable purity with internationally
defined properties. These existed; but Dr Hildebrand said that they were at
the top of a functional hierarchy. Below the international standard there might
be a national “primary standard”, which was a specially purified authentic
substance available from the competent national authority. Below this was a
“working standard”, possibly prepared in-house, which was calibrated
against the primary standard but was much cheaper and readily available for
daily use. There might also be “impurity standards”, which did not
themselves have to be of the highest purity; these might be isolated chromatographically
by the prime manufacturer, or specially synthesised.
Dr Hildebrand reported that ICH Group 6A distinguished RMs for purposes of assay,
identity and impurity limits. In practice, initial working standards arose early
in the R&D process, whereas the primary standard was developed and its properties
established by collaborative evaluation. When a drug substance was nearing the
end of its patent life, and if it was likely to be adopted by a pharmacopoeia,
the innovator company might often provide sufficient quantity of the potential
primary standard for adoption. The compendial agency would check the producer’s
characterisation of the substance (spectroscopic and other evidence for molecular
structure and stereochemical purity) and accurately assay the content.
Challenge
Dr Hildebrand said that the big challenge was authenticating RMs for products
of biotechnology and he quoted the ICH Q6B expectation that the innovator would
develop a series of appropriately characterised in-house standards. For biotechnology
identity standards, there were additional tests for their genetic and immunological
character and their bioactivity was assessed. Biotechnology impurity references
were also screened for cellular contamination and competent viruses, whereas
the potency standards required a statistically endorsed bioassay.
He then reviewed the sources of RMs and the extent of in-house, and subsequent
compendial, validation needed according to the intended purpose of the RM. He
stressed that in the laboratory there should be restricted access to fully accounted
RMs, with a “quality” person designated for their custody and care,
including a protocol for storage and periodic retesting. For their storage,
he emphasised the need for documented procedures (eg, refrigerated storage),
optimum closure and, where appropriate, restriction from light or air.
Dr Hildebrand concluded that RMs were “part of the overall quality concept,
to ensure safety of patient and user”.
What are the “4Qs” and how do they apply to “the qualification
of equipment in the control laboratory”? Dr John Loren (head of analytical
R&D, Pfizer UK) encapsulated the collective purpose of “qualification”
as the need “to satisfy regulatory requirements and meet business needs”,
by making sure that “the analytical equipment does the job you want done,
reliably and accurately — and that you can prove this”. He traced
the concept from the reference in Article 8 of the “Good manufacturing
practice” directive 91/356/EEC, which stated: “Equipment . . . which
[is] critical for the quality of the products shall be subjected to appropriate
qualification.” He cited specific guidance on “good manufacturing
practice” for active pharmaceutical ingredients, supplied by ICH group
Q7. They required that manufacturing equipment “should be of appropriate
design, adequate size and suitably located for its intended use, cleaning, sanitisation
and maintenance”, that “control, weighing, monitoring and test equipment
. . . should be calibrated according to written procedures and an established
schedule”, that for computerised systems, the “suitability of hardware
and software to perform assigned tasks be validated, and that “before starting
process validation activities, appropriate qualification of equipment and ancillary
systems should be completed”.
Dr Loren introduced the concept of “good automated practice”, which
was applicable to automated systems now frequently found in control laboratories.
This aided interpretation of regulatory guidance intended for manual systems
and (eg, from the US Environmental Protection Agency) addressed the reliability
of laboratory information management systems data, as well as requiring assurance
for the design and capacity of components. These should undergo acceptance trials
at installation, followed by on-going testing, inspection and maintenance. For
software, he prescribed a “life-cycle” invoking all four “Qs”:
design qualification (DQ) at the requirements, design and programming stage,
installation qualification (IQ) at the testing, quality checks and final installation
stages, with operational qualification (OQ) and performance qualification (PQ)
continuing during operation, maintenance and system enhancement. He said that
the PQ stage inevitably led to incipient obsolescence and retirement, which
started the cycle again, with DQ for updated software systems.
The remainder of Dr Loren’s presentation provided useful detailed examples
of the “4Qs”. In stressing that DQ ensured suitability for purpose,
he amusingly compared the Royal Albert Hall and the Millennium Dome — both
were interesting circular designs, admitting thousands, but they were very different
in appearance. The application of DQ invoked detailed specification, data acquisition
needs, environmental health and safety, GMP, operational constraints and a cost/benefit
analysis. Effective IQ verified that all systems complied with design and specification
requirements and included adequate training and instruction manuals from the
supplier. He saw IQ as a key interface between equipment supplier and customer.
As with the other Qs, for OQ, Dr Loren supplemented the formal ICH Q7 definition
with a punchier statement from the British industry’s Pharmaceutical Analytical
Sciences Group, namely, “confirmation that the equipment functions as specified
and operates correctly”. This definition had to be seen in the context
of the local working environment, integration with other systems, performing
with real (ie, known) samples and conforming to the user’s standard operating
procedures for training, routine use, calibration and maintenance.
The fourth “Q”, PQ, was envisaged as documenting acceptable ongoing
performance, with records of compliant calibration, maintenance and standards.
Dr Loren concluded that qualification of equipment both ensured regulatory compliance
and made good business sense. He offered a thought for his audience: should
there be a fifth “Q”, value qualification (VQ)? His VQ would calculate
return on investment at the design stage, check return during use, consider
net present value, and most of all focus on the best investment at a time of
unlimited opportunity in equipment procurement.
Dr Alastair Davidson (head of the Medicines Testing Laboratory, Edinburgh, Scotland) provided a background to consideration of out-of-specification results (OSRs) in the control laboratory. These he defined as any result falling outside the specification criteria of manufacturer, agency or compendium. He explained that, prior to the Barr judgment (see below) in 1993, there had been no recognised consistent practice for retesting. Pass/fail decisions were subjective and management pressure to release batches might persuade their control laboratories to retest until compliant results were obtained.
Barr judgment
Dr Davidson said that the New Jersey Court ruling, which in 1993 adjudged Barr
Laboratories to have breached GMP guidelines by routinely retesting, resampling
and reprocessing, had effectively persuaded all pharmaceutical industry laboratories
to standardise their protocols for dealing with OSRs. He said the court did
not accept the FDA submission that an OSR necessarily meant a batch failure.
Instead, such a result might derive from error in the control laboratory, or
in a manufacturing process, or by an operator. A laboratory error, which ought
to be relatively rare, might result from inadequate operator training, careless
work or poorly maintained or calibrated equipment. A process-related error could
arise from variations in reactor conditions or changes in product composition,
for which their test results should be accepted without retesting, although
the process might need to be revalidated. Finally, the court had ruled that
all OSRs must be reported and their investigation fully documented.
Dr Davidson then concentrated on the position of the test laboratory. Retesting
was appropriate when replacing a recognised laboratory error, or supplementing
an OSR that could not be conclusively explained. In the latter case, the extent
of permitted replication should be statistically predefined in the test protocol.
However, he drew attention to the residual uncertainty in just what extent of
replication would be acceptable. The Barr ruling had not accepted a total of
five passes from six tests; how many more replicates were needed? He suggested
a case-by-case approach, as part of the analytical method validation study.
In practice, five to 12 replicates might be preordained, according to the method
variability.
Dr Davidson went on to deal at length with the statistical considerations and
their implication for the control laboratory — such as the consequences
of averaging, which the Barr judge had accepted as “valid and rational”:
one could base pass/fail decisions on the standard error of the test mean. Dr
Davidson believed that there was not general support for the FDA view that outlier
tests (while appropriate in biological assays) were not generally acceptable
in more precise chemical analysis, or in tests of product uniformity. There
was more general agreement on resampling, which could only be justified where
there was firm evidence that the original analytical sample was unrepresentative
of the product batch. He emphasised that, for supply to the US market, the industry
had to comply with the FDA guidelines on treatment of OSRs. He gave examples
illustrating that almost half of GMP warning letters from FDA inspection fell
in this area.
For non-US markets, he said that in practice most industry laboratories would
implement some, if not all of the FDA requirements for replication and for documenting
and evaluating concurrent investigation. He recommended that official laboratories
should also follow the FDA guidance; otherwise, in his opinion, official pass
criteria might be less demanding than those of (most) manufacturers’ laboratories.
He foresaw GMP inspection in Europe increasingly employing FDA guidelines, to
improve world harmonisation of testing, and including extension to other test
criteria, such as system suitability tests and equipment calibration.