Distributed Pharmaceutical Analysis Laboratory
Draft of procedures for method validation
4 November, 2014
M. Lieberman
1.
VERSION: THIS MANUAL WAS DRAFTED 11/4/2014
2
2.
CAVEATS
2
3.
LIMITATIONS OF THE ANALYTICAL METHODOLOGY
3
4.
SAMPLE PREPARATION
4
SAMPLE STORAGE AND TRACKING:
EXTERNAL STANDARDS:
PHARMACEUTICAL DOSAGE FORMS:
4
4
5
5.
METHOD VALIDATION
7
6.
SUGGESTED INSTRUMENT PARAMETERS, SOLVENTS, AND GRADIENTS
8
A)
B)
C)
D)
AMOXICILLIN
AMPICILLIN
AMOXICILLIN-CLAVULANATE (OR AMOXICILLIN-CLAVULANIC ACID)
CIPROFLOXACIN
8
11
12
14
7.
ANALYTICAL METRICS (HOW TO TELL IF YOUR HPLC IS BEHAVING PROPERLY):15
8.
ASSAY AND QUALITY CONTROL
16
1. Version: This manual was drafted 11/4/2014
2. Caveats
Participants in the DPAL must be aware of some legal issues related to
pharmaceutical analysis. We report poor quality drugs to the medical regulatory
authority (MRA) in the country where the drugs originated, and also to the WHO
Rapid Alert system. Since we are doing single tablet analysis, analytical results must
be replicated on several samples before triggering a report to the MRA or WHO. If
your laboratory turns up a poor quality product that meets these criteria, Prof.
Lieberman will work with you to report to the appropriate authorities. Please
preserve such samples carefully; in some cases, we may want to subject the sample
to LC-MS, or the country MRAs may want to analyze the samples themselves (eg to
support legal action).
Second, due to the prevalence of counterfeit and improperly labeled products in
developing world markets, it is possible that information such as manufacturer name,
lot numbers, or expiration dates on the packaging materials are falsified or missing.
Legal counsel at UND recommends that we all use the term "stated to be
manufactured by thus and such company" in communications about specific
pharmaceutical products. This information must be conveyed to students who work
on the analyses.
Third, students must obtain written permission from their instructor before posting
pictures of products, analytical data, or other product-specific information in a public
forum such as a poster session, news article, web site, or social media posting. Please
check that the students adhere to the necessary wording describing the origin of the
drugs and that their comments about the products are factual and non-inflammatory.
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
2
3. Limitations of the analytical methodology
Our procedures for quantifying the active ingredients in pharmaceutical dosage forms
are based on monographs published in the United States Pharmacopeia (USP) or
British Pharmacopeia (BP), but they are not pharmacopeia methods. Modifications
must be made to the assay methodology to accommodate the circumstances of sample
collection. For example, we analyze single drug tablets, rather than pooled samples of
20-50 tablets, simply because the samples available are single packs that do not
include large numbers of tablets, and we may not have the level of quality assurance,
maintenance, and record keeping in our HPLC facilities that a commercial lab must
maintain.
In order to ensure that the strengths and limitations of the analytical methods are well
understood by each user, we require two types of proof. First, before any
pharmaceutical products can be assayed, the user must demonstrate the accuracy,
precision, and linearity of their method and instrumentation according to standards
laid out in USP <1226> and detailed below. We call this step method verification.
Second, specific quality control samples must be run during the assay of
pharmaceutical products. If these control samples do not assay correctly, as described
in the directions for each analyte, the results from the pharmaceutical products must
be thrown out and the samples re-run. We call this step quality control. Data from
method verification and quality control must be submitted to the DPAL database to
show proper performance of the method and assays
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
3
4. Sample preparation
Sample storage and tracking:
The pharmaceuticals sent for analysis are forensic samples. They must be stored in
a way that does not promote degradation or contamination, and you must keep
good records of each sample so you can track who performed what tests on it and
what the results were.
Samples must be stored in a cool, dark environment. We recommend a plastic bin
with a snug lid in a refrigerator. Allow the sample to come to room temperature
before you work with it, so water does not condense on the cold surfaces.
Each sample consists of at least 3 tablets or capsules which will be shipped in a
plastic bag labeled with a UND tracking number. The tracking number will be of the
form "14-xxxx" for a sample collected in 2014. When you run tests on individual
tablets or capsules, label them as "14-xxxxa, b, c, etc" Have the students sign out
samples for analysis and make sure they use the proper sample tracking number in
their records and chromatogram labels.
External standards:
External calibration standards are created from analytical grade reagents that are
traceable to USP or BP standards. The standard should include a certificate of
analysis, and you must take the reagent purity into account when calculating its final
concentration. Store dry standards as directed on the bottle (most must be kept
cold).
amoxicillin: Sigma-Aldrich PHR1127-1g is traceable to USP and costs $51 per gram.
The "known" API standard should contain about 0.5 mg/ml of ampicillin,
amoxicillin, or ciprofloxacin, and 0.5 mg/ml of amoxycillin plus 0.2 mg/ml of
clavulanate if you're analyzing amoxi-clav. You will use this standard to determine
all the other concentrations.
Calibration standards should span the range from 5% to 200% of the expected API
concentration in the experimental samples and at least 5 standards should be used
to construct the calibration curve. You will use these standards to establish
linearity over the range of interest. It is OK to use the calibration curve if the
samples are run at the same time as the unknown. A calibration curve generated on
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
4
one day cannot be used to assay concentrations of samples run on another day.
Since it takes 5 runs to do the calibration curve we prefer to establish linearity and
then use a single-point external standard to assay concentrations of unknown.
Prepare a "normal" unknown sample in the 95-105% range, an "overdosed" sample
in the 140-160% range, and a "deficient" sample in the 20-50% range. Also prepare
a method blank which is nominally 0%.
Pharmaceutical dosage forms:
Samples for analysis should contain about 0.5 mg/ml of ampicillin, amoxicillin, or
ciprofloxacin, and 0.5 mg/ml of amoxycillin plus 0.2 mg/ml of clavulanate if you're
analyzing amoxi-clav. Accurately weigh a tablet or the contents of a gel capsule and
take a portion of the powder that will give a 0.5 mg/ml solution when diluted to
volume. Tablets should be crushed to fine powder in a mortar and pestle and well
mixed, and contents of capsules should be well mixed. For preparation of the
analytical sample, weigh out at least 25 mg of powder on an analytical balance;
remaining powder should be labeled and frozen for storage.
For example, the total contents of an amoxicillin capsule with a nominal dose of 500
mg amoxicillin might weigh 627 mg due to excipients. To prepare a 0.5 mg/ml
solution, a portion of roughly 63 mg would be accurately weighed and dissolved in
100 ml of solvent. Samples should be thoroughly mixed (eg by stirring on a
magnetic stirrer for 5 min, or 5 min sonication) plus 2 min of hand shaking and
inversion of the volumetric flask. All samples must be filtered through a fresh 0.45
micron syringe filter to remove particulates that might clog the HPLC column.
Typically we will filter about 1 ml of the sample into an autosampler vial, discarding
the first drops of filtrate.
Amoxicillin: Use 20 mM monobasic sodium phosphate at pH of 4.4 to make
the samples. Keep standards and standard solutions refrigerated. Use
standard solutions within 3 weeks. Alternatively, freeze for storage up to 2
months.
Ampicillin: Use 20 mM monobasic sodium phosphate at pH of 6.7 to make
the samples. Keep standards and standard solutions refrigerated. Use
standard solutions within 3 weeks. Alternatively, freeze for storage up to 2
months.
Amoxi-clav: Use DI water to make the samples. Clavulanate is thermally
unstable and should be used within 6 hours of preparation.
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
5
Ciprofloxacin: Use 25 mM phosphoric acid, adjusted to a pH value of 3, to
make the samples. Keep standards and standard solutions refrigerated. Use
standard solutions within 3 weeks. Alternatively, freeze for storage up to 2
months.
It would be a good mini-experiment to run periodic chromatograms for a
standard sample that is left out at room temperature for several days.
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
6
For ampicillin, amoxicillin and ciprofloxacin, the solution is made using the buffer used
in the mobile phase. For amoxicillin we use For ampicillin we use 20 mM monobasic
sodium phosphate at pH of 6.7. For ciprofloxacin we use 25 mM phosphoric acid,
adjusted to pH of 3. For amoxicillin clavulanate the solution is made using water.
Solutions are sonicated for around 5 minutes and then filtered using a 0.45 mircon
syringe filter.
5. Method Validation
The goal is to establish that a standard USP or BP method, when run on your
instrument and with your reagents, satisfies a set of analytical metrics. These
directions are specific for HPLC methods using an external standard with a fixedwavelength UV-Vis or diode-array detector.
 Establish control chart: Set up a record keeping mechanism so you can track metrics such as
the integrated intensity of the known standard for different analysts and days of operation. A
page in the instrument log book can be used, or a paper chart, or a computer spreadsheet.
 Precision: The relative standard deviation (RSD) for the integrated intensities of 6 injections
of the known standard should be below 2%.
 Linearity: Prepare and run at least five calibration standards over the concentration range of
5% to 200%. Calculate a regression line for the calibration data, including correlation
coefficient, y-intercept, slope, and residual sum of squares. The correlation coefficient should
be 0.98 or better and the y intercept should be zero within error of measurement.
 Accuracy and range: Perform three replicate assays each of the overdosed sample, the normal
sample, the deficient sample, and a solvent blank (total of 12 determinations). Run the
external standard after every 5 runs and check that the integrated intensity for all the external
standard runs falls within 2% RSD. Use the average external standard signal to determine the
concentrations of the overdosed, normal, deficient, and blank samples. The measured
concentration of each sample should be within 2 standard deviations of its true concentration.
 Accuracy via spike recovery: If available, a sample of a pharmaceutical dosage form (tablet or
capsule) of the target drug should be prepared for analysis and a portion spiked with an extra
30% of the API. Calculate the % recovery of the spike; it should be within 90-110%.
 Specificity: This can be demonstrated by showing that a spike can accurately be recovered
from a dosage form matrix (the test in the "accuracy via spike recovery" section). A more
robust demonstration which provides a higher level of impurities and degradation products is
to stress the dosage form (eg by baking the tablet or powder for an hour at 100C), then use
that as the matrix for a spike recovery experiment.
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
7
 Optional: LOD and LLOQ: LOD or LLOQ determination is carried out using the slope of the
calibration curve and the SD of low concentration samples. Best practice is to prepare a
sample at about 2-3 times the expected LOD; you could also use the SD for the blank runs.
However you measure the LOD and LLOQ, samples near the LOD or LLOQ limit should be
run and their chromatograms shown in the report.
6. Suggested instrument parameters, solvents, and gradients
a) Amoxicillin
Instrument: Waters e2695 High Performance Liquid Chromatograph
Column: Kinetex 250 x 4.6 mm C18 column, 5 μm particle size and 100 Å
Temperature: 30°C
Detector: Waters 2998 Photodiode Array Detector
Analytical Wavelength: 220 nm
Amoxicillin and Amoxicillin/clavulanate Method I:
high flow rate, fast runs
Mobile Phase
Buffer
Monosodium
Phosphate
Time (min)
0.0
7.0
7.1
8.5
10.0
Concentration of
Buffer (mM)
20
Methanol (%)
pH of Buffer
4.4
Sample Injection
Volume (μL)
20
Buffer (%)
Flow
Ramp
(mL/min)
5
95
1.5
None
5
95
1.5
None
90
10
1.5
Linear
5
95
1.5
Linear
5
95
1.5
None
Table 1. Amoxicillin and Amoxicillin/Clavulanate Method I
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
8
13-0131 6-5-14 Sm (SG, 10x10)
Diode Array
220
Range: 4.883e-1
Area
3.45
38724
4.0e-1
3.5e-1
AU
3.0e-1
2.5e-1
2.0e-1
2.46
12102
1.5e-1
1.0e-1
5.0e-2
0.0
0.00
Time
2.00
4.00
6.00
8.00
10.00
12.00
14.00
Method 1 chromatogram (clavulanate at 2.46 min, amoxi at 3.45 min)
Amoxicillin and Amoxicillin-clavulanate Method II:
lower flow rate, better if your column pressure is too high.
Mobile Phase
Buffer
Monosodium
Phosphate
Concentration of
Buffer (mM)
20
Time (min)
Methanol (%)
Buffer (%)
0.0
5
1.0
pH of Buffer
4.4
Sample Injection
Volume (μL)
10
Ramp
95
Flow
(mL/min)
0.5
5
95
0.5
None
6.0
80
20
0.5
Linear
8.0
80
20
0.5
Linear
13.0
5
95
0.5
Linear
18.0
5
95
0.5
None
None
Table 2. Amoxicillin or Amoxicillin/Clavulanate Method II
@@what are typical retention times?
Analytical metrics for amoxicillin:
Column capacity factor 1.1-2.8 (if isocratic program is used),
column efficiency >1700 theoretical plates,
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
9
tailing factor <2.5 ,
RSD for replicate injections <2.0%.
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
10
b) Ampicillin
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
11
c) Amoxicillin-clavulanate (or amoxicillin-clavulanic acid)
Instrument: Waters e2695 High Performance Liquid Chromatograph
Column: Kinetex 250 x 4.6 mm C18 column, 5 μm particle size and 100 Å
Temperature: 30°C
Detector: Waters 2998 Photodiode Array Detector
Analytical Wavelength: 220 nm
Amoxicillin and Amoxicillin/clavulanate Method I:
high flow rate, fast runs
Mobile Phase
Buffer
Monosodium
Phosphate
Concentration of
Buffer (mM)
20
pH of Buffer
Sample Injection
Volume (μL)
20
4.4
Time (min)
Methanol (%)
Buffer (%)
Ramp
95
Flow
(mL/min)
1.5
0.0
5
7.0
5
95
1.5
None
7.1
90
10
1.5
Linear
8.5
5
95
1.5
Linear
10.0
5
95
1.5
None
None
Table 1. Amoxicillin and Amoxicillin/Clavulanate Method I
13-0131 6-5-14 Sm (SG, 10x10)
Diode Array
220
Range: 4.883e-1
Area
3.45
38724
4.0e-1
3.5e-1
AU
3.0e-1
2.5e-1
2.0e-1
2.46
12102
1.5e-1
1.0e-1
5.0e-2
0.0
0.00
Time
2.00
4.00
6.00
8.00
10.00
12.00
14.00
Method 1 chromatogram (clavulanate at 2.46 min, amoxi at 3.45 min)
Distributed Pharmaceutical Analysis Laboratory
12
DRAFT validation manual 11/04/2014
Amoxicillin and Amoxicillin-clavulanate Method II:
lower flow rate, better if your column pressure is too high.
Mobile Phase
Buffer
Monosodium
Phosphate
Concentration of
Buffer (mM)
20
Time (min)
Methanol (%)
Buffer (%)
0.0
5
1.0
pH of Buffer
4.4
Sample Injection
Volume (μL)
10
Ramp
95
Flow
(mL/min)
0.5
5
95
0.5
None
6.0
80
20
0.5
Linear
8.0
80
20
0.5
Linear
13.0
5
95
0.5
Linear
18.0
5
95
0.5
None
None
Table 2. Amoxicillin or Amoxicillin/Clavulanate Method II
@@ what are typical retention times?
Analytical metrics:
resolution between the amoxicillin and clavulanate peaks must be at least 3.5,
column efficiency for each peak at least 550 theoretical plates,
tailing factor below 1.5,
RSD for replicate injections less than 2.0%.
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
13
d) Ciprofloxacin
Instrument: Waters e2695 High Performance Liquid Chromatograph
Column: Kinetex 250 x 4.6 mm C18 column, 5 μm particle size and 100 Å
Temperature: 30°C
Detector: Waters 2998 Photodiode Array Detector
Analytical Wavelength: 220 nm
Mobile Phase
Buffer
Phosphoric acid
Time (min)
0
3
5
7
10
Concentration of
Buffer (mM)
25
Buffer %
95
70
70
95
95
pH of Buffer
3
Acetonitrile %
5
30
30
5
5
Sample Injection
Volume (μL)
???
Flow Rate (mL/min)
1
1
1
1
1
retention times?
Analytical metrics:
column efficiency >2500 theoretical plates,
tailing factor <4.0,
RSD for replicate injections <1.5%
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
14
7. Analytical metrics (how to tell if your HPLC is behaving
properly):
Measuring theoretical plates: We recommend the British Pharmacopia method
due to the simplicity of measuring peak width at 1/2 max height. tr is the
retention time of the peak. This method will slightly underestimate column
efficiency.
Measuring resolution of two peaks: Again, for peaks with tailing, it's easier to
use a formula with peak widths measured at 1/2 max height:
Measuring tailing factor (USP method)
Measuring column capacity factor (for isocratic methods only): k'=(tr - t0)/t0,
where tr is the peak retention time and t0 the dead volume of the column
(measured by the elution time for the solvent front).
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
15
8. Assay and Quality Control
External calibration standards are created from analytical grade reagents as directed
in section 4. Five injections of the external standard must show a peak area within 2%
relative standard deviation (RSD), and the range of retention times must be within 0.3
minutes. After every five unknown sample runs, the standard is injected as a quality
check and it must assay within 2% RSD of the 5 initial injections and be within the 0.3
minute time range. If a quality check fails, data after the last passed quality check is not
used.
Typically we analyze one pill from each package. If a sample fails analysis (assay
value <90% or >120% of stated API content) then a new sample is prepared from the
remaining powdered pill material and re-assayed (it would be good practice to perform
this duplicate assay routinely). If this also fails, two more tablets are assayed.
A copy of this Excel spreadsheet is posted on the DPAL site.
Distributed Pharmaceutical Analysis Laboratory
DRAFT validation manual 11/04/2014
16