(T1130-11-61) The Role of Filtration in Maintaining HPLC Data Quality

Purpose: High pressure liquid chromatography (HPLC), ultra-high pressure liquid chromatography (UHPLC) and liquid chromatography coupled to mass spectrometry (LC-MS) are common workhorses in manufacturing and development for characterization of both synthetic small molecules and biopharmaceutical large molecules. Sample preparation, such as syringe filtration, is critical to achieving high-quality data^1,2; however, in many cases, even simple sample preparation steps are avoided in favor of direct injection methods. In a previous study, it was shown that omitting filtration of both samples and mobile phase containing particulates could cause clogging of an HPLC column (indicated by increased backpressure) decreasing column lifetime^3-5. The hypothesis of this study was that omitting filtration, thus introducing particles into the HPLC would cause detrimental changes in data quality, evident through changes in peak height, area, and retention time for small molecule drugs, dependent filter type and particle retention.

Methods: Data quality was assessed by collecting chromatograms at 275 nm after passing filtered and unfiltered samples of a “Drug Mixture” (0.1 mg/mL of each caffeine and acetaminophen reference standards spiked with 0.05% wt fluorophore-tagged particles (0.513 µm diameter) in mobile phase (MP)), through a UHPLC system equipped with an HPLC column. Column backpressure (PSI) was monitored as 10 µL samples were injected up to 500 times or until the pressure reached a cutoff of 8000 psi, whichever was achieved first. Method details are provided in Table 1. Three cases were tested: Filtered Drug Mixtures through 0.45 µm pore size syringe filters of two filter materials from two manufacturers (MFR-1, hydrophilic polytetrafluoroethylene (PTFE) and MFR-2 regenerated cellulose (RC) of n=3 devices per two lots each, and an unfiltered Drug Mixture. Prior to each run, 200 µL from each filtered or unfiltered sample was withdrawn, read at Ex/Em 480/520 and compared to a six-point calibration curve to assess retention.

Results: Similar to previous findings^4,5, backpressure increased as particles were introduced into the HPLC. MFR-1 showed the highest retention ( >98%) and lasted >500 injections. However, MFR-2 showed lower retention (87% particle retention, thus allowed 13% into the detector) only lasted 250 injections before reaching the cutoff pressure and unfiltered samples (no retention), lasted only 42 injections before reaching the cutoff. This indicated that the presence of particles, even at 13%, caused significant impact to column lifetime Data quality correlated with retention and pressure. Full chromatograms (275 nm) were collected, (Fig. 1A-B), and peaks for caffeine (Fig. 1D-F) and acetaminophen (Fig. 1G-I) were analyzed for filtered MFR-1 (Fig. 1, col1) and MFR-2 (Fig. 1, col2), and unfiltered (Fig. 1, col3) samples. For MFR-2 filtered and unfiltered samples, caffeine peaks appeared to decrease in intensity (indicating a possible change in signal-to-noise), and retention shifted for both drug peaks, with acetaminophen demonstrating the most pronounced shift. Overall, shifts were more severe with more particles present over fewer injections. Meanwhile, there were minimal shifts for highly retentive MFR-1 filtered samples. Factors impacted by the presence of particles—peak height Fig. 2A,D), retention time (Fig. 2B,E), and peak area (Fig. 2C,F)—were investigated separately for caffeine (Fig. 2A-C) and acetaminophen (Fig. 2D-F). For samples with particulates (MFR-2; unfiltered), significant shifts were observed. For MFR-2 filtered samples, particles caused consistent decreases in height, retention and area for caffeine, while area and height did not change until close to the cutoff for acetaminophen. For unfiltered samples, changes were sharp compared to filtered samples. In the case of MFR-1 filtered samples, there were smaller overall shifts in all factors for both drugs across all 500 injections. Data collected for the diode array demonstrated similar results. This indicates that reproducibility and reliability of data from samples clarified with MFR-1 syringe filters were improved over MFR-2 syringe filters and unfiltered samples. Further, this correlates to the particle retention values (11% difference) observed between MFR-1 versus MFR-2 filters. Different properties of filter materials can contribute to effective or ineffective particle clarification, such as structure arising from varied casting methods (including shape, size, frequency, distribution and symmetry of pores)⁴. Other properties such as filter chemistry, filter diameter, thickness and housing can also play a role in non-specific binding of particulate matter. Thus, it is imperative to carefully consider filter material choice^4,5. Importantly, this method was isocratic and 10 µL of sample was injected. More severe changes would likely occur with gradient methods, complex drug mixtures, and increased sample injections.

Conclusion: Syringe filtration is the simplest and most cost-effective way to clear particles from samples. This study showed that filtration using highly retentive membranes produced reliable data quality of a two-component drug mixture compared to low retention filtered and unfiltered samples, which showed relatively poorer quality and peak shifts. Thus, efficient syringe filtration of samples is one key factor to consider for extending HPLC column lifetime, but also collecting the highest quality data.

References: [1] Majors RE. Column Pressure Considerations in Analytical HPLC. LCGC North America. 2007, 25(11): 1074–1092.
[2] Joshi V and Chernokalskaya E. Filtration as a Sample Preparation Technique Prior to Mass Spectrometry: Selecting the Right Filtration Device in Sample Preparation in Biological Mass Spectrometry, 2011.
[3] Joshi V. More Success in LC-MS – Tips and Tricks for Sample & Mobile Phase Preparation and Column Selection. Sigma Aldrich Technical Content, 2018.
[4] Lozeau, L.D., Dube, M. Comparison of syringe filters for maintaining optimal UHPLC backpressure. Analytix Reporter, 2024, 16, 28-31.
[5] Lozeau, L.D., Joshi, V., Laws, A. Factors to consider for increasing HPLC column lifetime. AAPS PharmSci360 Conference Proceedings, 2022.

Acknowledgements: The authors would like to acknowledge Cory Muraco and Dave Brewster. The authors would also like to disclose that this abstract refers to work published in AAPS PharmSci360 as a poster in 2022. There are no conflicts of interest.


Table 1: HPLC conditions and method notes.


Figure 1: Chromatograms of intensity versus retention time for the drug mixture at 275 nm of MFR-1 (A, D, G, col1), MFR-2 (B, E, H, col2) and unfiltered (C, F, I, col3) samples. Full chromatograms are depicted (A-C, row1), while individual drug peaks are also highlighted for caffeine (D-F, row2) and acetaminophen (G-I, row3). For ease of viewing, chromatograms were constructed for every 50 injections for MFR-1 and MFR-2, and every 10 injections for unfiltered up to the last injection.


Figure 2: The shift in caffeine (A-C) or acetaminophen (D-F) peak height (A, D), retention time, min (B, E), and peak area (C, F) at 275 nm versus injection number for MFR-1 (purple circle), MFR-2 (pink triangle) and unfiltered (green diamond) samples.