Tài liệu HPLC for Pharmaceutical Scientists 2007 (Part 8D) pdf

8.5.5.5 Case Study 2: Concluding Remarks. Extreme changes in selectivity and reversal of elution order of phenolic isomeric compounds were obtained after changing either the pH of the mobile phase, the temperature of the system, or the type of organic eluent employed. Changes in the analyte ion- ization state were observed upon increasing the acetonitrile composition as well as the temperature. Method development for ionizable analytes requires a judicious choice of the mobile-phase conditions and system parameters in order to perform the analysis of the compounds in their desired ionization state. Choosing the optimal parameters in the “chromatographer’s toolbox” allows for the development of rugged and reproducible methods. 8.5.6 Case Study 3: Method Development for a Diprotic Basic Compound A case study is presented for the method development of a diprotic base com- pound. The first step in method development is to look at the chemical struc- ture of the analyte and to determine if there are any ionizable sites on the molecule. If there are ionizable sites then their respective pK a values should be determined. The pK a values may have been already determined by the pre- formulation group and close communication with that group would avoid duplication of work. However, commercially available programs such as ACD Labs (Advanced Chemistry Development,Toronto, Canada) are also available to allow for in-silico prediction of the analyte pK a . Also, using selected frag- ments of the molecule can also be helpful for pK a determination of the desired molecule because the pK a values for each of these fragments of the molecule may be readily available from the literature. This is only an estimate at best, but can guide the chromatographer down the right path for initial mobile- phase pH selection. In the following case study for this pharmaceutical compound M, the method development scenario and rationale for each iteration in the method development process is highlighted. Also, a method development flow chart for gradient separations is included which can be used as a general strategy for method development (Figure 8-37). References are made in this case study to the flow chart in Figure 8-37. 8.5.6.1 Step 1: Analyze the Molecule. In Step 1 of the flow chart (Figure 8-37) it is recommended to analyze the molecule from a physicochemical point of view (knowledge of the pK a , log P, log D). The structure of the compound must be analyzed to determine the ionizable functionalities, and the pK a values of each ionizable group should be determined. In this case, Product M is a diprotic base with two pK a values 3.3 and 5.3 estimated by ACD. This com- pound contains an aromatic basic functional group (pyridinal nitrogen) with an electron-withdrawing group, chlorine, in the meta position and also contains electron-donating groups on the same aromatic ring. Electron- withdrawing groups such as chlorine tend to intensify the positive charge of the anilinium/pyridinum ion; this destabilizes the ion relative to the free METHOD DEVELOPMENT APPROACHES 405 amine/pyridine, therefore decreasing the basicity compared to pyridine (pK a 5.2). If one does not have a program to predict pK a , the pK a of this analyte could be estimated to be close to that of meta chloro pyridine, w w pK a 2.95. The other basic functionality contains a phenyl group attached to a morpholine group. The pK a of morpholine is 8.8; but because the phenyl ring is attached to the nitrogen group, this leads to resonance stabilization and consequently leads to a reduction of the analyte pK a . Note that because this compound does have multiple basic functionalities, two ionization equilibria could be written for this amphoteric species. At mobile w w pH values between 3 and 5 the existence of multiple species is expected. Since the two pK a values are close to one another (~2 pK a units apart), the inflection points overlap, making titration and/or chromatographic 406 METHOD DEVELOPMENT Figure 8-37. Method development flow chart for gradient separations. pK a prediction for each ionizable functionality difficult. At w w pH 4.3, the basic site (pyrdinal nitrogen) is predominately neutral (90%) and the other basic site (morpholine nitrogen) is predominately ionized (90%). In turn, it would be expected to observe only one inflection point from both the potentiomet- ric titration and chromatographic determination of the pK a . Alternatively, if the pK a of the analyte is not known and software is not available for in-silico prediction of the pK a , the chromatographer can go directly to step 2 (determine the isocratic conditions for pH scouting studies) and step 3 (determine pK a chromatographically) in the flow chart in Figure 8- 37. The analyte could be analyzed at six different pH values at a particular organic composition (isocratic mode) for an estimate of the analyte s s pH and for determination of the suitable chromatographic conditions to analyze the analyte. If the pK a is known, the chromatographer can go directly to step 4. 8.5.6.2 Step 2: Determine the Isocratic Conditions for pH Scouting Experiments. If the chromatographer intends to determine the chromato- graphic pK a and understand the influence of mobile-phase pH on the target analyte retention, pH scouting studies need to be performed in isocratic mode. In order to begin this process, the appropriate set of isocratic conditions to adequately retain the analyte in its fully ionized form and to elute the analyte in its fully neutral state needs to be determined. Usually a steep gradient run is used to estimate the initial isocratic elution conditions. In an example shown in Figure 8-38 a probe linear gradient from 5 to 95 v/v% acetonitrile METHOD DEVELOPMENT APPROACHES 407 Figure 8-38. Mobile phase A: 0.2 v/v% H 3 PO 4 . Mobile phase B: Acetonitrile, linear gradient from 5% B to 95% B over 10 min. Column: Luna C8(2) 150 × 4.6 mm. Injection volume, 10 µL; flow, 1.0 mL/min; wavelength, 300 nm, column temperature, 35°C. from 0–10 min was run and the target basic analyte (in its predominately ionized form) eluted at 6.0 min. The w w pH of the mobile phase was 1.9 and the flow rate was 1.0 mL/min. Note that the selection of the UV wavelength of detection should have already been performed using off-line UV, or alterna- tively a diode array detector can be used and the proper wavelength for detec- tion would consequently be extracted during the data processing. The concentration of organic in isocratic mode, which is necessary for the adequate retention of the analyte in its ionized form, now can be estimated. The w w pH of the aqueous portion of the mobile phase was chosen to be 1.9 (0.2 v/v% H 3 PO 4 ) and the flow rate was 1.0 mL/min. Using the linear gradient from 5 to 95 v/v% acetonitrile over 10 min, the v/v% organic per minute was calculated as 9 v/v% ACN/min (Scheme 1). However, the dwell volume (V D ) of the instru- ment must be accounted for because the actual gradient does commence within the column until about 1–2 min (depending on the instrument and instrumental setup). The dwell volume of most common HPLC systems is 1–2 mL and can be easily determined (see reference 19 for details). However, at 1 mL/min flow rate (F), an estimate of a 1.5-min dwell time (t D ) was used for this purpose (t D = V D /F). Note that the velocity of the analyte moving through the column under gradient conditions is not constant and follows a pseudo- exponential profile (see Chapter 2, Section 2.17). The estimation given in Scheme 1 serves as an approximation to determine the starting isocratic elution conditions from the probe gradient run. By taking into account the gradient slope of 9% ACN/min and accounting for the dwell time (1.5 min) and elution time (6 min from Figure 8-38) of the analyte from the probe gra- dient run, the estimated isocratic composition to elute the analyte at the same retention as in the gradient probe run can be calculated as shown. The esti- mated isocratic composition in which the analyte would elute at 6 min (k ~3.5) is estimated as 41 v/v% acetonitrile ±10% acetonitrile using 0.2 v/v% H 3 PO 4 ( w w pH 1.9) mobile phase. The isocratic conditions chosen to perform the pH scouting study was 30 v/v% acetonitrile. 408 METHOD DEVELOPMENT Gradient slope × (Elution time from probe gradient run − Dwell time) = Isocratic % organic composition ± 10% organic composition 9% ACN/min × (6 min − 1.5 min) = 41% ACN ±10% ACN Gradient Slope = 90% ACN min ACN min Dwell Time = 1.5 mL mL min 10 9 10 15 = = % . . min Scheme 1. Estimation of isocratic conditions from gradient probe run. 8.5.6.3 Step 3: Determine pK a Chromatographically. The retention of the analyte can be determined at different eluent pH values and can be used to determine its pK a in a particular hydro-organic mixture and assist the chro- matographer in proper pH selection for the aqueous portion of the mobile phase. The goal is to avoid a pH region where minor changes in pH can adversely affect the retention of the target analyte. If the pK a values are not known, it is suggested to perform the pH scouting experiments from pH 1.5 to 10 (at least five w w pH values should be investigated). In this example, the pK a values of the two ionization centers were predicted by ACD, and the s s pH values chosen were at least 1 unit less than the lowest s s pK a in the molecule and at least 1 unit greater than the highest s s pK a of the molecule. Generally, the pK a of a basic compound decreases by about 0.2 pK a units per 10 v/v% ace- tonitrile (see Chapter 4, Sections 4.5 and 4.6) and if 30 v/v% acetonitrile is used, it is expected to lead to a reduction of 0.6 pK a units (0.2*3 = 0.6 pK a unit basic analyte pK a shift) for both basic ionization centers. Therefore this cor- relates with a s s pK a of 2.7 for pyridinal nitrogen (i.e., 3.3 − 0.6 = 2.7) and with a s s pK a of 4.7 for the morpholinal nitrogen (i.e., 5.3 − 0.6 = 4.7). The expected pK a determined by chromatography is to be midpoint of these two s s pK a values (i.e., 3.7). In step 3, for this study the upper s s pH for the mobile phase to be prepared was determined to be s s pH 6.7 (at least two units greater than the highest s s pK a of the molecule). The lower s s pH for the mobile phase (containing 30 v/v% MeCN) that should be prepared for this study should be 1.7, but this would mean that an aqueous mobile-phase w w pH of 1.1 would have to be prepared to obtain a s s pH of 1.7 (see Chapter 4, Section 4.5 for pH shift). Remember that the pH shift of the mobile phase for a phosphate buffer is approximately 0.2 pH units in the upward direction for every 10 v/v% acetonitrile. In this case, not to compromise the stability of the packing material (column chosen has recommended a lower pH limit of w w pH 1.5), a pH of w w pH 1.6 was chosen to be prepared which correlates to a s s pH of 2.2 ( w w pH 1.6 + 0.6 units upward pH shift upon addition of 30 v/v% acetonitrile). Most definitely the final method will not be set at this low pH, since the analyte would exist in multiple ionization states; however, the experiment was performed at this low pH to elucidate the effect of the pH on the analyte retention in this low-pH region. Five to six s s pH values would then be chosen between s s pH 2.2 and 6.7 to run the pH study at isocratic conditions (30 v/v% MeCN). In Figure 8-39 the chro- matographic retention as a function of the s s pH are shown. Note that blanks do not need to be run, just multiple injections of the same analyte. Once the retention of multiple injections of the target analyte is achieved, the column is deemed to be equilibrated with the mobile phase. Usually after eluting 25 column volumes through the column, the column is assumed to be equilibrated (this may not be the case if an ion-pairing reagent is used). In this study, three injections at each pH were run. Only the last injection is shown in Figure 8-39 and Table 8-8. One recommendation is to perform the pH study either METHOD DEVELOPMENT APPROACHES 409 from low to high pH or from high to low pH for faster equilibration between each successive pH experiments. The k (retention factor) values are then plotted versus the s s pH values, and the inflection point of this sigmoidal relationship could be taken as the s s pK a of that particular compound at particular hydro-organic mixture. The s s pK a determined at 30 v/v% MeCN was determined to be 3.9 (using nonlinear regression analysis program MathCad 8). This corresponds well to our origi- nal estimation of s s pK a 3.7. The retention of this analyte leveled off between pH values w w pH 4.6 ( s s pH ~5.2) and w w pH 6.1 ( s s pH ~6.7), where the analyte is in its neutral form. No lower 410 METHOD DEVELOPMENT Figure 8-39. Phenomenex Luna 3u C8(2) Column. [150 × 4.6 mm, 3 µm]. Mobile phase: 10 mM K 2 HPO 4 , acetonitrile (70 : 30, v/v), pH adjusted with w/ H 3 PO 4 ; flow rate, 1.0 mL/min; injection volume, 10 µL; wavelength, 247 nm., column temperature, 35°C. TABLE 8-8. Retention of Diprotic Compound at w w pH and s s pH w w pH s s pH t R 6.1 6.7 49.8 4.6 5.2 49.1 3.6 4.2 34.3 2.6 3.2 11.2 2.1 2.7 6.0 1.6 2.2 3.8 limit plateau is observed where the analyte would be in its fully ionized form. Therefore, further analysis for this compound should be carried out at pH values > s s pH 6.1 using no less than 30 v/v% acetonitrile in the mobile phase. Note that at even higher organic compositions the basic analyte pK a is further reduced (up to 60 v/v% acetonitrile) and the pH of the aqueous portion (adjusted with acidic modifier) of the mobile phase is further being shifted upwards. This is a favorable situation because the analyte is being analyzed in a more neutral state with an increase in the organic concentration. Hence this is the situation when gradient elution is used and higher organic content is employed. All these analyses were conducted on a 15- × 4.6-mm, 3-µm column. However, the pH scouting analysis could have been performed faster. A 5-cm × 4.6-mm column on a conventional low-pressure HPLC system could have been used, further reducing the analysis times of each run by at least three times. Also, a 5-cm or 10-cm × 2.1-mm column, packed with 1.7-µm particles, could have been used in conjunction with an ultrahigh-pressure HPLC system, allowing for further reduction of the analysis time for the pH scouting exper- iments (see Figure 4-29 in Chapter 4). 8.5.6.4 Step 4A: Determine the w w pH of the Aqueous Portion of the Mobile Phase for Gradient Screening Studies (Paper based evaluation). After gaining confidence with the approaches described in Step 2 and 3 in Figure 8- 37 or if the analyte pK a , is known, then the chromatographer can go directly from Step 1 to Step 4A in Figure 8-37 to determine what the starting mobile- phase w w pH should be in order to perform the gradient screening studies. There are three items that need to be considered: 1. In what form will the molecule be analyzed (neutral or ionized)? For this particular molecule we want to analyze the molecule in its neutral form. 2. The pK a shift of the ionizable analyte. For this example, since the analyte is basic, the downward pK a shift for basic analytes must be accounted for.The working pH should be at least 2 pH units above the basic analyte pK a to be fully neutral. One pH unit could also be used (analyte is approximately 90% in neutral form). 3. The pH shift of the mobile phase. In this example, an acetate buffer was chosen. The upward pH shift (acidic) of the acetate buffer upon addi- tion of the organic must be accounted for. If the buffer contains both acidic and basic functionalities (i.e., ammonium acetate), the pH shift is dependent on the pH that is chosen. This is based on the respective pK a of the counteranion and countercation of the buffer species employed. For example, at pH values below 7 the acidic pH shift rule would apply for the acetate counteranion, and at pH values greater than 7 the basic pH shift rule would apply for the ammonium countercation (see Section 4.5). METHOD DEVELOPMENT APPROACHES 411 If Product M, a diprotic base, is to be analyzed in its neutral form, the higher w w pK a of Product M (which is 5.3) needs to be considered because the other w w pK a of 3.3 is less basic. Let us use to try to determine at what w w pH the analyte would be in its neutral form at eluent conditions of 30 v/v% MeCN and 70 v/v% acidic buffer. The goal is to calculate w w pH of the buffer in order to obtain the basic analyte in its fully neutral form. Step A. First, account for the downward pK a shift for the basic analyte upon addition of organic. For every 10 v/v% increase in acetonitrile, the s s pK a of the analyte decreases by 0.2 pK a units. (Highest analyte pK a is 5.3 and considering 30 v/v% of acetonitrile in the mobile phase the pK a shift is (0.2 * 3 = 0.6) 5.3 − (3 * 0.2) = 4.7 ( s s pK a ) Step B. Once s s pK a is determined, the s s pH at which the analyte would be in its fully neutral form (>99%) needs to be determined. This corresponds to s s pH that is 2 pH units greater than the s s pK a of 4.7 (calculated above). Note that if one wanted to determine the s s pH in which the analyte would be ≥90% of its neutral form, this would correspond to working s s pH 1 unit greater than the s s pK a of 4.7. (This is also acceptable from a method robustness point of view.) 4.7 + 2 = 6.7 ( s s pH) Step C. Then account for the pH shift of the acetate buffer (acidic buffer) upon addition of acetonitrile. For every 10v/v% increase in acetonitrile, the pH of the acidic buffer increases by approximately 0.2 pH units. This would correspond to a 0.6 pH unit increase: 3 * 0.2 = 0.6 Step D. Then determine what the minimum w w pH of the aqueous portion of the buffer should be by taking into account the upward pH shift of the aqueous portion of the mobile phase upon addition of organic. Therefore the optimal pH to analyze this compound would be at an aqueous mobile phase pH of ≥6.1. 6.7 − 0.6 = 6.1 Max pH of the aqueous portion of the mobile phase in order to have analyte in fully neutral form (>99%) at 30 v/v% MeCN. The prediction of w w pH 6.1 for the aqueous portion of the mobile phase using Steps A–D agrees well with the actual experiments that were performed in Step 3, where the retention of Product M was independent of the pH when the w w pH is greater than 4.6. w w pH, greater than 4.6 were used for further studies in Step 5. 412 METHOD DEVELOPMENT 8.5.6.4.1 Step 4B: Determination of the Optimal w w pH of the Aqueous Portion of the Mobile Phase in Gradient Mode. Alternatively, the chromatographer may proceed directly from Step 1 to Step 4B, to determine the optimal w w pH of the aqueous portion of the mobile phase under gradient conditions. This can be accomplished by running six linear gradient experiments (same gradient profile, e.g.from 5% to 85% of organic over 20 minutes) using different aqueous portion w w pH in each run ( w w pH: 2—10.5). This would allow for the determination of the desired w w pH region under gradient conditions: a pH region in which the reten- tion of the components in the sample does not change significantly as a function of the w w pH. An optimal w w pH (within the desired pH region) of the aqueous portion of the mobile phase can then be selected and used in Step 5. 8.5.6.5 Step 5: Gradient Scouting Studies with the Optimized pH of the Aqueous Phase. Once the optimal w w pH is known, the gradient conditions can be optimized for obtaining the best selectivity and resolution of all critical pairs. Multiple samples could be run which include a crude sample (better to use a sample that has an elevated amount of impurities) and forced degraded samples. Two gradients can be run at two temperature (35°C and 50°C)—one with a shallow slope (i.e., 5% MeCN to 95% MeCN over 20 min) and one with a steep slope (i.e., 5% MeCN to 95% MeCN over 8 min)—and then the gra- dient can be modified accordingly if needed. If the optimal selectivity and res- olution of all critical pairs cannot be obtained and/or the target analyte is not spectrally homogeneous, go to Step 6, Figure 8-37. (Screen different columns/ mobile phases with the optimized w w pH of the aqueous phase using an auto- mated method development system or column switcher.) Alternatively, the results from the gradient runs for each sample can be inputted into Drylab, ACD, or Chromsword for further optimization (see Sec- tions 8.5.6.11). For the predicted experimental conditions (i.e., gradient slope, temperature, flow rate), if desired selectivity and resolution can be obtained, an experiment can be run for verification.The peak purity for the main analyte (MS and DAD detection) should be checked in the verification run. If the desired selectivity and/or the target analyte are not spectrally homogeneous, go to Step 6, Figure 8-37. In this case study, further method development was carried out on a crude sample using a 10 mM ammonium acetate buffer that has a w w pH 5.8 (note acetate has suitable buffering capacity from pH 3.8 to 5.8). Two gradient runs (shallow/steep gradient slope) were performed. The best chromatography was obtained in gradient mode with a linear gradient from 5% acetonitrile to 95% acetonitrile over 20 min (shallow slope), with a 3-min hold at 95% acetonitrile. A hold at higher organic is usually recommended in the early stages of devel- opment to ensure the elution of very hydrophobic components. It is also rec- ommended to employ a high organic hold for stability-indicating methods for the API and drug product in the event that higher molecular weight species (i.e., hetero or homo dimers of the API, API containing hydroxyl/amino group which could react with stearic acid to form a more hydrophobic degradation product) are formed in the solid state upon storage. METHOD DEVELOPMENT APPROACHES 413 Also, another driver for choosing the ammonium acetate buffer was that it is also LC-MS compatible.Also, this acetate buffer could be used since Product M has maximum wavelength for absorbance at 247 nm, and there is no back- ground absorbance from the buffer at this wavelength. With this linear gradi- ent employed, the target analyte elutes at approximately 14 min (Figure 8-40). To determine the approximate concentration of acetonitrile in which the analyte elutes, the following calculation can be performed. Using the gradient slope of 4.5% MeCN/min and the dwell time of 1.5 min (dwell volume of system is 1.5 mL and flow rate is 1.0 mL/min), the analyte elutes at approxi- mately 50% acetontrile: [14 min − 1.5 min] × 4.5%/min = 56% MeCN. Note that the velocity of the analyte movement through the column using gradient conditions is not constant and follows a pseudoexponential profile and that this estimation just serves as an approximation to determine approximate elution conditions to ensure that the analyte is being analyzed in one pre- dominate ionization state accounting for pH and analyte pK a shift. At this organic eluent composition range (50% MeCN ± 10% MeCN) the analyte is still predominately in its neutral state (since the s s pK a is further lowered upon addition of organic). Note that as the acetonitrile content is increased up to 60 v/v% acetonitrile, the pK a of basic compounds generally continues to 414 METHOD DEVELOPMENT Figure 8-40. Analysis of Product M (free base) as neutral species using two types of volatile buffers. Chromatographic conditions: Column: Luna C8(2) 150 × 4.6 mm. Mobile phase: Aqueous (see A and B for exact conditions), acetonitrile. Wavelength, 247 nm; column temperature, 40°C; flow, 1 mL/min; injection volume, 10 µL. Linear gradient from 5% acetonitrile to 95% acetonitrile over 20 min, with 3-min hold at 95% acetonitrile. [...]... in Figure 8-40, an acidic diluent containing acetonitrile was used as the sample preparation solvent In Figure 8-40A for the sample stored in the acidic diluent for 4 days at room temperature shows some degradation compared to Figure 8-40B for the sample stored in the acidic diluent for one day at room temperature The free base had limited solubility in water: acetonitrile diluents, so the addition... by diode array analysis A single quadrupole mass spectrophotometer which may be available in analytical departments of most pharmaceutical companies, can be used for this purpose Mass spectra should be taken across the peak (−5%, apex, and +5%) as with the PDA This was performed for Product M, and it was determined that the peak is not spectrally homogeneous; also, an impurity with an odd number of nitrogens... considered is the stability of the analyte in the diluent Product M was found not to be very stable in acidic diluent So going forward a sample tray cooler, 4°C, was used to enhance the stability of the sample in the diluent, and it was determined to remain stable for at least 72 h at 4°C and for no more than 24 h at room temperature 416 METHOD DEVELOPMENT Figure 8-41 Determination of peak homogeneity Chromatographic... acetonitrile 8.5.6.8 Checking for Peak Purity If the selectivity of all the components in the separation is acceptable, then the next step is to check for the peak purity of the target analyte Spectral purity of the active peak was determined using PDA (photodiode array) detector Diode array detection is used to determine if the peak is spectrally homogeneous This is performed by overlaying spectra... species In this example, there was coelution noted, and it was found that changing the gradient slope on this C8 column did not provide for any increased selectivity between the co-eluting species and the target analyte Therefore, further method optimization experiments were performed to resolve the impurity from active This warranted proceeding to Step 6, Figure 8-37: Screen different columns/mobile phases... employed (since analyte is in its fully neutral form in this pH region) The 10 mM ammonium acetate buffer has a UV cutoff . for initial mobile- phase pH selection. In the following case study for this pharmaceutical compound M, the method development scenario and rationale for. analyte in its fully neutral form. Step A. First, account for the downward pK a shift for the basic analyte upon addition of organic. For every 10 v/v% increase