As per FDA CGMP (21
CFR.211.110) an in-process testing is requiredfor powder blends to demonstrate
adequacy of mixing, but it does not state that the blend has to be directly
assessed for uniformity. But the original draft stratified sampling guidance
document allowed the use of in-process dosage unit data to demonstrate dosage
unit uniformity during routine manufacture.
AS directly testing of
in-process dosage units is discouraged, blend uniformity should be assessed
during process design (Stage 1 Validation) and process qualification (Stage 2
Validation). But the exception being when blend sampling presents severe risks
to the operators taking the samples (which should be discussed and accepted by
regulators).
The US Pharmacopeia
(USP) <905> uniformity of dosage unit (UDU) test is familiar across the
industry. Its procedure and acceptance criteria have been widely used for batch
release since its introduction in 2007, due to its convenient results reporting
and ease of determining conformance to acceptance criteria.
Although used for batch
release, the procedures and acceptance criteria in USP <905> do not
represent a statistical sampling plan. As of 2013, the US Food and Drug
Administration (FDA) withdrew its support for USP <905> procedures and
acceptance criteria for batch release, followed by the USP in 2014.
Multiple statistical
approaches have been proposed as replacements for the procedures and acceptance
criteria currently defined by USP <905>. While these approaches satisfy
the statistical shortcomings of USP <905>, they require significant
deviation from the current familiar implementation of that compendial test,
such as comparing test results against large, table-listed acceptance limits
(ASTM E2709/E2819), or determining acceptance by applying one- and two-sided
tolerance intervals to the test results. Both approaches require mathematical
manipulation of the test results followed by individual determination of
conformance to the acceptance limits.
To overcome obstacles
associated with industry acceptance of a statistical methodology that may be
unfamiliar, ISPE attempted to develop a method that provides adequate
statistical assurance in a manner that has the same familiar look and feel of
the acceptance criteria test associated with USP <905>. The process
presented here results in a single number that, when maintained within the
defined numerical limit, satisfies the minimum statistical assurance provided
by the acceptance criteria, a process very similar to that of the current
application of USP <905>.
The CU approach
presented here was selected and developed for two reasons:
First, 50% confidence,
95% probability to pass USP <905> has been suggested as appropriate for
release testing.
Second, the methodology
was developed to have the same practical look and feel as the current
application of USP <905>.
A stepwise summary of the proposed commercial testing procedure and application of the CU criteria:
- During manufacturing, collect at least one sample from at least 30 locations spaced equally across the batch, including the beginning and end of the run.
- Assay a total of 10 dosage units from approximately equal locations across the batch, including the beginning and end of run. These samples should be taken from the 30 samples collected during manufacturing.
- Calculate the average (x̄) and SD of the 10 results.
- Calculate AV50/95 = |100.0 – x̄ | + 2.664 × SD
- The sample complies if all individual values are within 75.0–125.0 %LC (Label claim) and AV50/95 ≤ 15.0.
- If AV50/95 > 15.0, assay 20 additional dosage units (one dosage unit from each of the remaining locations collected during manufacturing).
- Calculate the average (x̄) and SD of the 30 total results
- Calculate AV(50/95) = |100.0 – x̄ |+2.521 × SD.
- The sample complies if all values are within 75.0–125.0 %LC and AV50/95 ≤ 15.0.
- If AV50/95 > 15.0, the batch does not meet the acceptance criteria.
Reference: ISPE
Related Articles
- Blend Uniformity Analysis
- Weight Variation Measurement
- Measurement of Loss on Drying (LOD)
- Differences Among Blend Uniformity, Weight Variation and Content Uniformity