There can be potential drawbacks with filtration as a preparation for HPLC analysis in dissolution testing. The first is that the filter may adsorb active pharmaceutical ingredients (API) from the drug mixture leaving the concentration in the filtrate too low and out-of-specification (OOS).

Unwanted drug adsorption as well as the presence of possible extractables eluted from the filter during routine pharmaceutical sample analysis can be a serious problem and cause OOS results. No single analytical method can provide reliable information on comparative filter properties and the full range of extractables for all filters. Extractables have already been discussed, therefore, this study will evaluate filters for adsorption of API.

The drug product selection and product formulations in this study represent a wide variety of compounds that differ in chemical structures, ionization properties, and molecular weights and therefore differ in binding propensity. Also, a broad range of medium for the sample preparation is matrixed into the study to evaluate elution profiles of each filter. All experiments are designed based on well-characterized (validated) United States Pharmacopoeia (USP) methods.

We will demonstrate that correct selection and use of syringe filters makes the amount of API removed by the syringe filters too small to affect the HPLC determination. To do this, five drug products are each evaluated with four different syringe filters. The four different syringe filters contain three different hydrophilic polymeric membranes, plus one without glass prefilter. The five drug products represent a variety of molecular structures and chemistries and therefore are expected to demonstrate a range in adsorption to the different filter membranes.

Table 1
Filters Tested

ID #	Manufacturer	Product Name	Cat #	Lot #	Membrane Type	Glass Fiber Prefilter
05-04-0302	Pall Life Sciences	Acrodisc PSF	AP-4424	A224229511	Supor (PES)	Yes
05-04-0303	Pall Life Sciences	Acrodisc PSF	AP-4557	A10531713	GHP	Yes
05-04-0304	Pall Life Sciences	Acrodisc PSF	AP-4310	A10531121	PVDF	Yes
05-04-0305	Pall Life Sciences	Acrodisc PSF	AP-4560	A10531858	GHP	No

Membrane types:
Supor (PES) - Polyethersulfone
GHP - Hydrophilic polypropylene
PVDF - Hydrophilic polyvinylidene fluoride

The drugs (API’s) used in this evaluation represent a range of different functionalities and structures and should therefore demonstrate a range of adsorption to membrane filters. As seen in Table 2, the chemical structures vary from single aromatic rings to multiple aromatic rings to a non-aromatic, polycyclic structure. Included in the study are an acid, a base, an amide, a urethane, an ester, and a lactone structure. The physical structures vary from a more flat and planar structure like that of acetaminophen to the flat but flexible structure of ibuprofen and ranitidine HCl, to the more rigid and distinct three-dimensional structure of simvastatin.

Table 2
Pharmaceutical Products

Results are obtained by HPLC analysis with UV detector at 238 nm for Simvastatin, 243 nm for acetaminophen, 254 nm for Ibuprofen and Loratadine and 322 nm for ranitidine HCl. All calculations are performed according to each specific USP procedure against the appropriate, well-characterized (certified) corresponding USP reference standard. Label claim percentage (%LC) of each drug is calculated as ratio of the amount of drug that is found during analysis in each filtrate to the amount known (or claimed) to be present in the tested solution, and expressed as percentage. Recovery of each drug upon filtration (i.e., %LC to centrifuged) is calculated as ratio of the amount that is found during analysis in each filtrate to the amount that is found in the centrifuged sample, and expressed as percentage.

For the chosen well characterized pharmaceutical products, the value of the label claim percentage (%LC) is between 98-102% for the accurately carried testing procedure. Any additional handling of the samples (e.g. filtering or centrifugation) is a source of additional error or inaccuracy. Therefore, specifications for the filtered samples are set to meet wider 97-103% interval of acceptability. This criterion is set based on the assumption that handling of the filtering process should not add more than 1% error to the sample analysis regardless of the individual filter compatibility. This assumption was validated in the filter study. The data spread [% relative standard deviation (RSD)], which is caused solely by filtration, is less than 1% for all filters (Table 10). Therefore, results outside the 97-103% interval are indicating out-of-specification results and signaling potential filter incompatibility. Each out-of-specification result (highlighted in yellow) is addressed individually in the discussion section of this report.

All solvent compositions are given in Table 2. Drugs in easily soluble salt form are less adsorbed by a generally hydrophobic filter membrane and more likely to stay in solution than be attracted to a membrane. When the amount of organic component is high enough to suppress ionization (and hence solubility) of such compound, it might result in precipitation. Therefore Ranitidine Hydrochloride adsorption is studied in aqueous solutions with a low amount of organic component, to study pure adsorption effect rather than solid particle retention. Ibuprofen and two binary mixtures with high and low Acetonitrile content are chosen to demonstrate binding of a non-polar drug in the free form to a filter membrane in different media.

In step two; 20 mL of the sample solution are run through each filter. The 1st, 2nd, 3rd, 5th, 10th, 15th and 20th 1 mL aliquots are collected and analyzed. The drug concentration is measured after filtration. Duplicate HPLC injections of the seven 1 mL aliquots are performed for each filter, with each drug evaluated (280 samples total). The flush volume evaluation from step two is determined as sufficient when the recovery value for the filtered sample is within 97-103% of the centrifuged sample.

In step three, filtered aliquots (from step two) are compared with centrifuged samples (from step one). The recovery of each drug preparation is determined as a percentage of label claim and as a ratio of percentage of label claim to the centrifuged sample, according to USP methodologies. The average recovery of each drug with each filter is given in Tables 5-9.

In step four, triplicate sample filtrations of each drug solution with each filter are performed. The first 3 mL flush volume is discarded and subsequent 1 mL samples are collected and analyzed by HPLC in duplicate (120 samples total). The triplicate average of each filter with each drug is itemized in Table 10. Results are reported as a percentage of label claim for each drug along with the relative standard deviation found within the sample groups.

The chromatographic conditions and standards are given below and in Table 3.

Instrumentation

1. Hitachi (San Jose, CA, USA) HPLC systems with the following components:
   - Hitachi L7200 Autosampler.
   - Hitachi L7400 UV Detector
   - Hitachi L7100 Gradient Pump.
   - Hitachi L7300 Column Oven.
   - Perkin-Elmer (Wellesley, MA, USA) 970A Interface.
   - Perkin-Elmer TotalChrom Data acquisition system and processing software.
2. HPLC columns (as directed in each applied USP method)
3. General laboratory equipment and Class A analytical glassware

Table 3
Reagents and Reference Standards Used

Acetaminophen USP RS (lot J2C423)	Valerophenone (Reagent grade)
Ibuprofen USP RS (lot 1335508)	4-Isobutylacetophenone (Reagent grade)
Loratadine USP RS (lot G0D344)	Chloroacetic Acid (Reagent grade)
Ranitidine USP RS (lot H0B268)	Ammonium Acetate (ACS-certified Reagent)
Simvastatin USP RS (lot I0D382)	Glacial Acetic Acid (ACS-certified Reagent)
Acetonitrile (HPLC grade)	Sodium Hydroxide (ACS-certified Reagent)
Methanol (HPLC grade)	Sodium Phosphate Monobasic (ACS-certified Reagent)
DI Water suitable for HPLC	85% Phosphoric Acid (ACS-certified Reagent)
Ammonium Hydroxide (ACS certified Reagent)	Potassium Phosphate Dibasic (ACS-certified Reagent)

Acetaminophen (Tylenol*)	Ibuprofen (Motrin*)	Ranitidine (Zantac*)	Loratadine (Claritin*)	Simvastatin (Zocor*)
Average % LC	Average % LC	Average % LC	Average % LC	Average % LC
101.2	100.9	100.3	97.6	100.7

Flush Study (Optimization of the filtration parameters — flush volume determination). Data from all flush studies are summarized in Tables 5-9.

Table 5
Amount of Acetaminophen in Filtered Samples

Table 6
Amount of Ibuprofen in Filtered Samples

Table 7
Amount of Ranitidine HCl in Filtered Samples

Table 8
Amount of Simvastatin in Filtered Samples

Table 9
Amount of Loratadine in Filtered Samples

Table 10
Filter Study Summary with a 3 mL Preflush

The flush study demonstrates greater variations in the label claim percentage results in the first 1 mL effluent. Four out-of-specification results occur in the first 1 mL of filter eluate. The filter study demonstrates that all four Acrodisc® PSF syringe filters are appropriate for each of these drugs based on acceptable results after a 3 mL flush.

The out-of-specification result for the Ibuprofen (shown in bold in Table 6) is attributed to possible initial adsorption of the organic solvent by the filter membrane. This initial adsorption will cause the first aliquot to have an apparent concentration above the starting concentration, which causes the initial reading to be OOS, on the high side. Disposal of the first 3 mL of the effluent is the recommended resolution to allow use of this filter with this solvent system.

The out-of-specification results for the Loratadine (shown in bold in Table 9) are attributed to the inconsistency incorporated into the study design. A flush size of 3 mL is found sufficient to avoid inherent inaccuracy of the human handling and remove any potential API binding concerns.

The 3 mL discard volume is chosen as an amount that is sufficient to ensure accuracy and consistency of experimentation and environmentally responsible solvent usage. We believe that if more solvent is necessary for the analysis, the filter is not suitable for that particular application. Results of the triplicate determination of a label claim and %RSD for each drug and each filter are reported in Table 10. The results confirm that filtration does not affect the finished drug product assay numerically and all tested filters are suitable for achieving 97-103% accuracy with the data spread (precision) less than 1%.

• Azopharma Contract Pharmaceutical Services, Filter Evaluation Study Report Number R050T18B, (October 2005) (Internal Pall Report)
• Ding W. and Scheer L., Syringe Filter Efficiency and the Effect of Filtration on HPLC Column Life, European Biopharmaceutical Review, p. 60, (2004).
• Lindenberg M., Wiegand C., Dressman B., Comparison of the Adsorption of Several Drugs to Typical Filter Materials, Dissolution Technologies, (2005).
• R.W. Yost, L.S. Ettre, and R.D. Conlon, Practical Liquid Chromatography. An Introduction, Perkin-Elmir, Norwalk, (1980).
• W.W. Yau, J.J. Kirkland, and D.D. Bly, Modern Size-Exclusion Liquid Chromatography, Wiley, New York, (1979).
• Snyder L.R., Kirkland J.J. and Glajch J.L., Practical HPLC Method Development, 2nd Edition, John Wiley & Sons, Inc, (1997).
• USP 28 — NF 23, page 1702, 1704 (Ranitidine Hydrochloride Monograph). Pharmacopeial Forum: Volume Number 30(6), page 2033.
• USP 28 — NF 23, page 1770, 1771 (Simvastatin Monograph). Pharmacopeial Forum: Volume Number 30(5), page 1647. Pharmacopeial Forum: Volume Number 29(2), page 437.
• USP 28 — NF 23, page 16, 17, 19 (Acetaminophen Monograph). Pharmacopeial Forum: Volume Number 27(3), page 2494, 2495.
• USP 28 — NF 23, page 991, 993 (Ibuprofen Monograph). Pharmacopeial Forum: Volume Number 27(4), page 2740.
• USP 28 — NF 23, page 1149, 1151 (Loratadine Monograph)
• USP 28 — NF 23 Supplement: No. 2, page 3500. Pharmacopeial Forum: Volume Number 30(6), page 2011. Pharmacopeial Forum: Volume Number 29(4), page 1045.