Also the spectral homogeneity of the peak of interest must be taken into consideration. Diode array spectra at least three points across the peak should be taken to ensure the peak is spectrally homogenous see Figure 8-14. If the peak is not spectrally homogenous, the overlay of the spectra will show 368 METHOD DEVELOPMENT Figure 8-13. Optimal wavelength selection for API and related impurities. Figure 8-14. Determination of peak homogeneity: Diode array detection (DAD). (Reprinted from reference 10, with permission.) distinct differences (see Figure 8-15). However, even if the diode array spectra do overlay, this does not absolutely ensure that the peak does not contain any coeluting impurities, because the impurities could have similar diode array spectra and/or if there is a low level of a coeluting species with a different diode array spectrum, it may not be determined by this approach. In these cases, MS detection needs to be employed to ensure MS spectral homogene- ity. MS spectra are taken across the peak and the MS spectra across the peak should not show the presence of any other coeluting species of different masses. This does not absolutely ascertain that the peak is homogeneous since isomers of the same compound will have the same [M + H] and is indistin- guishable from the parent compound. Also, the impurity that may be coelut- ing may not have an appreciable ionization efficiency at the particular mobile-phase and mass spectrometric conditions. An example of where using diode array detection may not be helpful is shown in Figure 8-16. Note that for this reaction mixture (convergent synthe- sis) the desired product 1 has the same diode array spectra as synthetic pre- cursors 2 and 3. If these two synthetic precursors had coeluted with 1, they would not have been able to be deconvoluted. This stresses the importance of running LC-MS in a parallel to diode array studies during method development. 8.4.3 Solution Stability and Sample Preparation It should be determined if the drug substance being analyzed is stable in solu- tion (diluent). During initial method development an autosampler tray cooler METHOD DEVELOPMENT CONSIDERATIONS 369 Figure 8-15. Diode array detection for elucidation of coeluting species. (Reprinted from reference 10, with permission.) and preparations of the solutions in amber flasks should be performed until it is determined that the active component is stable at room temperature and does not degrade under normal laboratory conditions. Also, since it is not known if dimeric species or more hydrophobic compounds are present in the sample solutions during the initial method development or are formed in sta- bility studies, gradient elution should always be performed with a hold at higher organic conditions (or up to the buffer stability limit). The reduction of downtime of the instrument (i.e., operations of pump com- ponents, injectors, and detectors) can be controlled to some degree if sample solutions are filtered and/or centrifuged; the use of a 0.2- or 0.45-µm-pore-size filter is generally recommend for removal of particulates [15]. Filtration as a preventive maintenance tool for HPLC analyses is well-documented in the lit- erature [16–18]. Sample preparation is a critical step of method development that the analyst must investigate. For example, the analyst should investigate if cen- trifugation (determining the optimal rpm and time) shaking and/or filtration of the sample is needed, especially if there are insoluble components in the sample. This is usually more prevalent with excipient/DS mixtures and with slurry solutions obtained during the synthesis steps of the API. Syringe filters 370 METHOD DEVELOPMENT Figure 8-16. Reaction conversion of a convergent synthesis (2 + 3) to 1 and overlay of diode array profiles. are routinely used to remove particulate contamination/insoluble components from samples prior to chromatographic analysis. The objective is to demonstrate that the sample filtration does not affect the analytical result due to adsorption and/or extraction of leachables. A word of caution here is that filter studies should be performed to ensure that no adsorption of the compound on the filter is observed. This is particularly the case with protein and peptide samples. Note that for proteins and peptides the impact of centrifugation (speed and time) must be investigated because this may lead to increased aggregate formation. Also, for protein and peptides the initial concentration of the sample could also have an impact on the concen- tration gradient of the sample in the centrifuge tube, and the concentration of the top, middle, and bottom portions should be assessed. The effectiveness of the syringe filters is largely determined by their ability to remove contaminants/insoluble components without leaching undesirable artifacts (i.e., extractables) into the filtrate. Extractables are often the result of inappropriate material construction and improper handling of the device during the manufacturing process. Particular attention should be paid to potential extractables from the membrane and housing material. The sample preparation procedure should be adequately described in the respective ana- lytical method that is applied to a real in-process sample or a dosage form for subsequent HPLC analysis. The analytical procedure must specify the manu- facturer, type of filter, and pore size of the filter media. Also, it must be known if the particular filter type is compatible with the type of analyte, organic solvents, and pH of the solution to be filtered. The following procedure may be used to determine if there is any absorp- tion on the filter.A stock solution is prepared at the target concentration. One aliquot of the stock solution is centrifuged, and other aliquots from the cen- trifuged stock solution are filtered through the desired filters (pre-wet with 5 mL of diluent) and the results compared. If any additional peaks are observed in the filtered samples, then the diluent must be filtered to determine if a leach- able component is coming from the syringe filter housing/filter. In Figure 8-17 a solid oral dosage form was prepared at 1 mg/mL concentration. The initial stock solution was centrifuged (no filter) and two additional samples from the centrifuged solution were filtered with a nylon filter and a cellulose filter. The area counts (Table 8-4) of all three solutions were compared, and it was shown that significant absorption was observed on the nylon 66 filter. Further opti- mization of the sample preparation would include removing the centrifugation step and just filtering the supernatant (solution above the undissolved excipi- ents) with the cellulose acetate filter. Another example includes the recovery (mass %) of API and degradation products of API from two 100-mg tablet (5 tablets) sample solution clarified by filtration and clarified by means of centrifugation. The data in the Table 8-5 demonstrates that the two methods of sample clarification are equivalent and that the filtration procedure (0.2-µm Nylon filter, with 5mL pre-wet) is METHOD DEVELOPMENT CONSIDERATIONS 371 372 METHOD DEVELOPMENT Figure 8-17. Comparison of filtered (nylon filter versus cellulose filter) versus no filter (centrifuged). Column: Luna C18 (2). Mobile phase: (A) 10 mM ammonium bicarbon- ate, pH 7.5; (B) MeCN, linear gradient from 0 to 15 minutes, 20—70% of B. Sample concentration: 1 mg/mL. TABLE 8-4. Area Counts for Centrifuged/Filtered Solutions Type of Sample Preparation Area Counts 30-mm Nylon filter, 0.2 µm 484,155 No filter (supernatant solution) 5,612,755 13-mm Cellulose acetate filter, 0.45 µm 5,633,064 TABLE 8-5. Filter Evaluation Results for API Assay-Related Substance Samples API Impurity 1 Impurity 2 Impurity 3 Impurity 4 Sample Name (%) (%) (%) (%) (%) Filtered samples 1 99.0 0.079 0.082 0.025 0.042 2 99.4 0.079 0.082 0.025 0.042 3 99.5 0.079 0.084 0.025 0.043 Mean 99.3 0.079 0.083 0.025 0.042 % S rel 0.3% 0.0% 1.4% 0.0% 1.4% Centrifuged samples 1 99.1 0.080 0.080 0.025 0.042 2 98.9 0.081 0.080 0.025 0.042 3 100.0 0.078 0.078 0.025 0.042 Mean 99.3 0.080 0.079 0.025 0.042 %S rel 0.6% 1.9% 1.5% 0.0% 0.0% adequate and does not cause any specific absorption of the active and/or impurities. Other considerations for sample preparation include incorporation of methanol in the sample preparation scheme, especially if a second dilution is used (check for sample reactivity). The impact on peak shape (diluent/mobile phase mismatch for components with k < 2) should also be considered. Sample preparation usually constitutes approximately 70% of solvent usage, and incorporating methanol for routine sample preparation can lead to reduction in solvent costs. 8.4.4 Choice of Stationary Phase Ideally for a reversed-phase separations, the retention factors (k) for all com- ponents in a sample should lie between 1 and 10 to achieve separation in a reasonable time. For a given stationary phase the k of a particular component can be controlled by changing the solvent composition of the mobile phase. However, the impact of eluent composition will depend on the type of sta- tionary phase and the nature of the components in the mixture. In reversed- phase HPLC the most common solvent mixtures are: water and acetonitrile, water and methanol, and water and THF. The elution strength increases as the organic portion of the modifier increases. Thus, to optimize a chromatographic separation, the concentration of the organic modifier is adjusted so that the k of the components in the sample are in the range of 1 to 10. However, some- times due to the hydrophobic nature of the compound, even high concentra- tions of organic modifier will not allow elution of all components in a single run and the chromatographer can try one or a combination of the following approaches: (1) Use a stronger modifier; (2) apply a steeper gradient; (3) use a less hydrophobic stationary phase. Detailed discussion of the reversed-phase separation principles and separation optimization is given in Chapter 4. The type of column chosen for a particular separation depends on the com- pound and the aim of analysis. Pharmaceutical companies may have a pre- ferred list of columns that have good demonstrated performance in regard to pH/temperature stability. These columns that have been selected by a specific laboratory are known to be stable within predefined pH and temperature regions in which method development/column screening are employed. A good understanding of the chemical stability of the stationary phases is needed, and some examples are shown in Section 8.10. Screening columns from each of the following various column classes should provide for the desired chromatographic selectivity, even for the most challenging separations: (1–3) C8 or C18 stable at pH < 2, pH 2–8, and pH > 8–11; (4) phenyl; (5) pentafluorphenyl; (6) polar embedded and stationary phases that could be run in 100% aqueous. A certain number of columns in each of the six column classes and subclasses could be chosen as standard columns that the chromatographers choose as a first choice for performing method development.These standard columns could be chosen based on some METHOD DEVELOPMENT CONSIDERATIONS 373 set of internal criteria (ie., chromatographic selectivity for a set of compounds, bonded phase stability, and lot-to-lot reproducibility). The criteria for selec- tion may include that the column is stable for a certain number of column volumes (efficiency, tailing factor, retention time criteria for predefined probe analytes) at the recommended max and min pH at a particular maximum tem- perature. By tracking the column usage (number of column volumes run at a particular pH/temperature), this will reduce the number of system suitability failures and decrease the cost of the consumables for a particular laboratory. Moreover, this information should be shared among the analytical chemists in the different line functions (DMPK, Drug Substance, Drug Product, Prefor- mulation, TechOps, or PharmOps) to ensure that these columns are readily available and that the practical experience can be shared for the selected columns within a particular company. Also, it is generally recommended not to use the same column for multiple projects, especially when performing release and stability testing. For more hydrophobic compounds, a stationary phase that has a lower surface area should be used. For very polar compounds that cannot be retained on traditional C18 phases, less hydrophobic columns such as C4 and polar embedded stationary phases could be used. However, all this is also depen- dent on the pH of the analysis since some columns are not stable at low pH (<2) and higher pH (>7) for extended periods of time. This should be taken into careful consideration when defining a column(s) during the development of a method. Moreover, the effect of pH on a particular compound’s retention needs to be determined first before exploring the retentivity and selectivity of differ- ent columns. The strategy and choice of the optimal pH for analysis was dis- cussed in Chapter 4 and is further reinforced in the case studies within this chapter. After the optimal pH is chosen for the separation and the gradient has been optimized on a particular column and the optimal selectivity still has not been achieved between critical pairs, then a column screening can be per- formed. For method column screening, generally columns with 10-cm or 5-cm × 3.0-mm i.d. could be used that are packed with 3-µm particles. Implementa- tion of a column switcher that can use six different types of stationary phases such as two types of C18 from different vendors, phenyl, two polar embedded, and pentafluorphenyl is suggested. In Figure 8-18, a mixture of acids and bases was analyzed on three types of columns: phenyl, polar embedded, and C18 column. Significant differences in selectivity were obtained. The separation could be further optimized by modifying the gradient slope and employing off-line method development tools such as Drylab for further optimization and resolution of the critical pairs. Moreover, once a particular column or columns that have provided the best selectivity are chosen, an automated method optimization may be performed. This would include employment of an integrated HPLC method development system such as AMDS/Drylab such that the gradient slope/temperature 374 METHOD DEVELOPMENT can be further optimized on multiple columns that had shown the best selectivity. 8.4.5 Mobile-Phase Considerations 8.4.5.1 Choice of pH. If analytes are ionizable, the proper mobile-phase pH must be chosen based on the analyte pK a so the target analyte is in one pre- dominate ionization state ionized or neutral. If possible, method development at both of these defined mobile-phase pH values is encouraged to maximize the potential gains that may be obtained in regard to selectivity (for the neutral and ionized forms of the target analyte and related substances). Alteration of the mobile-phase pH is one of the greatest tools in the “chro- matographers toolbox” allowing simultaneous change in retention and selec- tivity between critical pair of components. Analytes may be analyzed in their ionic form or neutral form. This may be dependent on the type of analysis that is required. If fast analysis is required, then analysis of the component in its ionized form may be acceptable if the desired resolution from the matrix com- ponents is achieved. However, if adequate resolution of the active from its process-related impurities/degradation products/excipients are not obtained, then mobile-phase additives may be added to the mobile phase or the mobile- phase pH may be adjusted so the analyte may be analyzed in its neutral form in order to potentially enhance the selectivity/resolution between critical pairs of components. Increasing flow rate, increasing temperature (up to column sta- bility limit at a particular pH), increasing the concentration of the organic METHOD DEVELOPMENT CONSIDERATIONS 375 Figure 8-18. Effect of type of bonded phase on the separation selectivity. eluent, and using shorter columns with narrower dimensions may be used to obtain more desirable run times. However, speed does not come without a price, and the influence of the aforementioned parameters on the resolution of the critical pairs in a mixture/sample needs to be evaluated. 8.4.5.2 Buffers. In order to develop rugged HPLC methods, knowledge of choosing the right buffer is very important. Buffers that are selected should have a good buffering capacity at the specified mobile-phase pH.Also, the con- centration of the buffer should be at least 10 mM to provide the needed ionic strength to suppress any undesired analyte solvation effects that may lead to poor peak shapes. Methods that specify a phosphate buffer in the pH range of 4 to 6, or an acetate buffer in the range of 6 to 7, are, unfortunately, not good buffers. These buffers are not just useless in these pH ranges, they com- plicate the preparation of mobile phase unnecessarily and give the analyst a false sense of controlling the reproducibility of the separation. Optimum buffering capacity occurs at a pH equal to the pK a of the buffer. In general, you can expect most buffers to provide adequate buffering capac- ity for controlling mobile-phase pH only within ±1 unit of their respective pK a values. Beyond that, buffering capacity may be inadequate. Also, buffers are great media for growing bacteria. It is recommended to have at least 10 v/v% of organic in the aqueous phase to prevent bacterial growth. Table 4-3 in Chapter 4 lists some commonly used buffers for reversed-phase HPLC. In this table the buffers and their respective pK a values, and UV cutoffs are listed. Since it is becoming more common to find HPLC interfaced to mass spectrometers, volatile buffers for LC/MS applications are also indicated. 8.4.5.3 General Considerations for Buffers. The type of buffer that is chosen will depend on the wavelength of the method and the concentration of organic in the mobile phase. A judicious choice of type and concentration of buffer must be made to ensure mobile-phase compatibility. • Phosphate is more soluble in methanol/water than in acetonitrile/water or THF/water. • Some salt buffers are hygroscopic. If an analyst makes a 20 mM buffer and the original buffer salt contains 20 w/w% water, then the buffer con- centration would be 16 mM. This may lead to changes in the chromato- graphy (increased tailing of basic compounds, and possibly selectivity differences). • Ammonium salts are generally more soluble in organic/water mobile phases than potassium salts, and potassium salts are more soluble than sodium salts. • TFA can degrade with time, is volatile, absorbs at low UV wavelengths, and is not a buffer at pH > 1.5. 376 METHOD DEVELOPMENT • Citrate buffers can attack stainless steel.When using these buffers, be sure to flush them out of the system as soon as the analysis is completed, but this is a recommendation for any buffer system. • Microbial growth can quickly occur in buffered mobile phases that contain little or no organic modifier. This growth will accumulate on column inlets and can damage chromatographic performance. • At pH greater than 7, phosphate buffers accelerates the dissolution of silica and severely shortens the lifetime of silica-based HPLC columns. If possible, organic buffers should be used at pH greater than 7. • Ammonium bicarbonate buffers usually are prone to pH changes and are usually stable for only 24 to 48 hours. The pH of this mobile phase tends to become more basic due to the release of carbon dioxide. • After buffers are prepared, they should be filtered through a 0.2-µm filter. • A “test tube test” should be conducted to determine if the buffer at the concentration it is prepared will precipitate in the column/system when it is exposed to the highest organic concentration in the gradient. The tem- perature should also be considered as well. Buffers generally will have a higher solubility at higher temperatures. The test tube test can be per- formed by preparing the mobile phase in a 10-mL test tube and then putting the test tube in the refrigerator and/or water bath (to mimic higher temperatures) to determine if any precipitation occurs. The results of “test tube tests” of phosphate buffers (10 and 25 mM) in various ace- tonitrile/water compositions at room temperature and 5°C are given in Table 8-6. • Mobile phases should be degassed if an on-line degasser is not available on the HPLC system. Also, the purity of the buffer should be taken into consideration. Small amounts of trace impurities can absorb in the UV wavelength of interest and cause a high background absorbance, thus suppressing the limit of detection for a particular analysis. One such case is with N-methyl pyrrolidine, although it does not absorb above 210 nm; sometimes the use of this reagent is not fea- sible unless the wavelength of detection is greater than 225 or 254 nm due to the presence impurities with chromophores that absorb in that region. If mobile phase A had 50 mM N-methyl pyrrolidine that was contaminated with some low-level impurities and mobile phase B had MeCN and a linear gradi- ent was run from 5% MeCN to 95% MeCN and the wavelength that was being monitored was 210 nm, a decrease in the baseline would be observed due to a dilution effect of the buffer impurity background absorption.The same behav- ior is usually observed when TFA, acetic acid, and/or formic acid are used in the aqueous portion of the mobile phases and a wavelength of <220 nm is used. Note this is also dependent on the concentration of the acidic/basic modifier employed. It is generally recommended to use the same concentration of TFA or other acid (UV absorbing) in both the aqueous and organic portions of the METHOD DEVELOPMENT CONSIDERATIONS 377 [...]... not affect the selectivity and result in the same retention factors However, increasing flow rate is limited by the pressure limitation of the HPLC equipment For more information on gradient separation and developing fast HPLC methods, see Chapter 17 and Practical HPLC Method Development from Synder et al [19] 8.4.6.3 Linear Versus Multi Step Gradients If the gradient separation mode is selected, it is... gradients shallow or steep allows for obtaining differences in the chromatographic selectivity This would be attributed to the different slopes of the retention versus organic composition for each analyte in the mixture Therefore, changing the gradient slope or steepness of the gradient is an important variable that should be considered in reversed-phase method development for controlling the retention... challenging Most of the modern HPLC equipment can reproduce gradient profile with high accuracy, but permanent variation of the eluent composition 382 METHOD DEVELOPMENT and usually pH also stresses the column, which may decrease its lifetime Some recently developed columns can withstand repetitive stress for 10,000 to 20,000 column volumes Further description of the development of fast HPLC methods using a gradient... coefficient at that alternate wavelength, then it could be used for monitoring the analysis Depending on the molar extinction coefficient, at this higher wavelength (even though not the primary maxima) a higher S/N ratio may be obtained for the LOQ solution since the mobile-phase additives will not absorb in this higher-wavelength UV region Therefore there is less interference from the background buffer... effect 8.4.5.5 Practical Aspects of Preparing a Buffered Mobile Phase The following steps are suggested for preparing a buffered mobile phase: 380 METHOD DEVELOPMENT • Define the appropriate w pH for the separation and then select an approw priate buffer Refer to Table 4-5 to determine the appropriate buffer for your application Prepare an aqueous buffer solution of the desired concentration and w pH w Measure... stability of the mobile phase should also be considered THF is known to form peroxides However, most analytical chemists use THF that is not inhibited, since the inhibitor (BHT or cresol) absorbs . limitation of the HPLC equipment. For more information on gradient separation and developing fast HPLC methods, see Chapter 17 and Practical HPLC Method Development. lytical method that is applied to a real in-process sample or a dosage form for subsequent HPLC analysis. The analytical procedure must specify the manu- facturer,