Chemistry Reference
In-Depth Information
tAble 8.2
experimental design and Acceptance criteria for Amt
type of method
# Analysts
# lots or units
Acceptance criteria
notes
Assay
2
3 lots in triplicate
A two one-sided t-test with intersite
differences of ≤2% at 95% CI
Each analyst should use different instrumentation and columns,
if available, and independently prepare all solutions. All
applicable system suitability criteria must be met.
Content
uniformity
2
1
Include a direct comparison of the mean,
±3% and variability of the results, (%RSD),
that is, a two one-sided t-test with intersite
differences of ≤3% at 95% CI.
If the method for content uniformity is equivalent (e.g., same
standard and sample concentrations, LC conditions, and
system suitability criteria) to the assay method, then a separate
AMT is not required.
Impurities,
degradation
products
2
3 Lots in
duplicate
(triplicate if
done together
with the assay)
For high levels, a two one-sided t-test with
intersite differences of ≤10% at 95% CI.
For low levels, criteria are based on the
absolute difference of the means, ±25%.
All applicable system suitability criteria should be met. The LOQ
should be confirmed in the receiving laboratory, and
chromatograms should be compared for the impurity profile. All
samples should be similar with respect to age, homogeneity,
packaging, and storage. If samples do not contain impurities
above the reporting limit, then spiked samples are recommended.
Dissolution
NA
6 units for
immediate
release, 12 units
for extended
Meet dissolution specifications in both
laboratories, and the two profiles should be
comparable, or based on the absolute
difference of the means, ±5%.
A statistical comparison of the profiles (e.g., F2) or the data at
the Q time points similar to that performed for the assay may
be performed.
ID
1 unit
Chromatography: confirm retention time.
Spectral identification and chemical testing
can also be used, assuming operators are
sufficiently trained and the instrumentation
can provide equivalent results.
Cleaning
validation
2 spiked samples,
one above, one
below spec.
Spiked levels should not deviate from the
spec by an amount 3X the validated
standard deviation of the method, or 10%
of the spec, whichever is greater.
Essentially a limit test. Low and high samples to confirm both
positive and negative outcomes are required.
