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

This maylead to increased retention and increased selectivity between critical pairs ofcomponents.Moreover, if the laboratory has an automated method development system,then this could b

the flow rate (mL/min) to determine the time (in minutes) needed Therefore,the lower the flow rate, the longer the equilibration time.

Some typical equilibration times for various column dimensions are shown

in Table 8-7; however, these should only be used as a guide If complete libration is not achieved, early eluting components may show differences inretention from run to run An experiment could be run such that three dif-ferent methods could be run with different equilibration times For example,

equi-if a 15-cm × 4.6-mm i.d column and a flow rate of 1 mL/min was used, thenthe equilibration times for the three methods would be 5 min (3 CV), 9 min (5CV), and 11 min (6 CV) equilibration times, respectively If the retention ofthe early eluting components are consistent (less than 1% variation in reten-tion time) in all three methods, then the lowest equilibration time could beused However, if the early eluting components show greater variation in theirretention time with the 5-min equilibration time compared to the methodswith the 9- and 11-min equilibration time, then an equilibration time of greaterthan 5 min is warranted Optimization of the optimal equilibration time isrequired for reproducible methods

Other considerations include differences in dwell volumes from the ent HPLC systems The dwell volume should be determined for all the systems

differ-in the laboratory and based on these determdiffer-inations, this should be factoredinto the calculation of the equilibration time For example, if the maximumdwell volume of all the systems in a particular laboratory to which the method

is transferred to is 2 mL and you are running on an instrument at 1 mL/minthat has a dwell volume of 1 mL, then you should add an extra minute of equi-libration time This becomes extremely important during method transferswhere the instruments in the receiving laboratory may be different

8.5.1 If Analyte Structure Is Known

Determine if analytes are acidic, basic, or neutral This will allow the chromatographer to choose a pH such that the analyte is being analyzed

TABLE 8-7 Equilibration Times for Columns of Different Dimensions

Length (cm) Diameter (cm) Column Volume (CV) (mL) 3 CV (mL) 5 CV (mL)

predominately in one ionization state Use the rules for pH shift and pKashift

to ensure that the analyte is one predominant ionization state and choose theappropriate mobile-phase pH (see Chapter 4)

Some general guidelines are as follows: If your target analyte is an acidic

analyte (pKa ≥ 3), use a 0.2 v/v% phosphoric acid mobile phase If target

analyte is a basic analyte (pKa≥ 7–9), use an ammonium acetate buffer (pH5.8) to analyze in its ionized form or use a 25 mM ammonium hydroxide buffer

(pH 10.5) or 25 mM N-methyl pyrrolidine buffer (pH 10.5) to analyze in its

neutral form Use a 10-cm × 3.0-mm column packed with 3-µm particles andintermediate polarity phase such as a C8 column that is stable for the pH atwhich you may be running the probe separation Run in the gradient mode using an acidic buffer or a basic buffer from 5% to 95% of organic component, or up to the buffer solubility limit over 10 min, and use an isocratic hold for 10 min to ensure elution of hydrophobic components Use aflow rate of 0.8–1 mL/min flow rate and 40°C temperature Injection volumeshould be on the order of 5–20µL and the concentration of the analyte should

be 0.5–1 mg/mL This corresponds to approximately 5–20µg injected oncolumn On the other hand, for neutral analytes higher analyte loading such as 50–100µg maybe used since nonideal interactions with the stationaryphase are less prevalent Note that for ionizable compounds, especially basic compounds when analyzed in their ionized state, higher mass on columnmay lead to mass overload of “hot spots” on the bonded phase and poor peakefficiencies may be observed Try not to load more than 10µg on column for basic compounds Usually greater loading capacity is obtained for basiccompounds when they are analyzed in their neutral state Note that forcolumns with larger inner diameters such as 4.6 mm, larger sample loads may

be acceptable

Once the probe gradient is run, check the diode array purity; and if LC-MS

is available, run as well to check for peak homogeneity If you have any knownprecursors or impurities, run them as well to ensure resolution from the maincomponent and to make sure they are adequately retained The main analyte

should elute between k 2–5 If the main component elutes at k = 2–5 and isspectrally pure and the impurities all elute k > 1, the method is complete Ifthe retention factor of the impurities is below 1, then an isocratic hold at theinitial organic composition should be implemented until the minor component(impurity) elutes k > 1 and then a linear gradient can be implemented Themethod could be further optimized by increasing the flow rate as long as thebackpressure limitation of the system has not been reached A general rule ofthumb is that the backpressure should not exceed 85% of the maximum back-pressure for a particular HPLC system

If resolution is not achieved between a critical pair, the use of a shallowergradient can be investigated If that does not increase the resolution, then alonger column (15-cm column, packed with 3-µm particles of the same sta-tionary phase type) should be used with a reduced flow rate of 0.7 mL/min(due to backpressure limitations)

If separation is still not achieved, consider using a different organic fier such as mixture of MeCN and methanol to possibly induce changes in theselectivity Also, the wavelength of detection must be considered, especially ifMeOH is used due to its UV cutoff (absorbs <210 nm) If methanol does giveyou desired selectivity, then an analyst needs to determine if sufficientresponse (S/N > 10 : 1) is obtained at desired LOQ (i.e., 0.05% solution oftarget), especially if the wavelength for detection is <210 nm.

modi-If changing the organic modifier does not work, consider changing themobile-phase pH (analyze the molecule in a different ionization state) Forexample, if a basic compound was originally analyzed under basic conditions(pH>> pKa), try to use acidic conditions (pH << pKa) with the acetonitrile in

the initial gradient If that still does not work, then consider using a differentstationary phase (phenyl or polar embedded) employing the initial gradient,with initial aqueous mobile phase and acetonitrile organic modifier, and repeatthe process that was performed on the original column used for initial methoddevelopment The final method optimization may include varying the gradi-ent slope, column temperature, and flow rate

Note that multiple pH values and columns can be screened in gradientmode at the same time as well This will increase the efficiency/probability ofobtaining the best column/conditions and the best demonstrated chromato-graphic selectivity Note that the aqueous phase pH values that would bechosen for these pH/column screening studies should be based on knowledge

of the physicochemical properties of the molecule, taking into consideration

the mobile-phase pH and analyte pKashifts in the hydro-organic media

8.5.2 If Method Is Being Developed for Separation of Active and

Unknown Component

Define the criteria for the method such as the LOQ, maximum run time, length detection, and so on Look at the structure of the target analyte (esti-

wave-mate pKa) or use ACD (advanced chemistry development) and determine the

best pH to run the method Try to use shorter columns for gradient scoutingexperiments (5 cm × 4.6 mm ) packed with 3-µm columns or use a high-pressure system (max pressure 15,000 psi) with 10-cm × 2.1-mm, 1.7-µm parti-cles Use 35–45°C as starting temperature If pH scouting studies are needed,run a probe linear gradient using 0.2 v/v% phosphoric acid on a short column (5-cm × 4.6-cm column) to determine the isocratic conditions for the

pH studies Run pH studies isocratically to determine the desired pH region

to understand the behavior of the impurities in the analyte mixture Thedesired pH region of the aqueous phase is the pH region where the retention

of the components in the mixture do not significantly change their retention

as a function of the pH of the aqueous phase Track impurities using diodearray if possible Run a linear gradient at a pH within the desired pH regionand hold at high organic concentration on 5-cm × 4.6-mm column If you obtain sufficient resolution, then you are finished If you need more

resolution, then use a 15-cm × 3-mm i.d column If resolution is obtained,then you are finished If desired resolution/selectivity is not obtained, thenscreen different organic modifiers/different stationary phase types Note that the separation of the critical pair may be obtained on an alternate stationary phase that offers additional selectivity In addition to the weak dispersive types of interaction that are available on a C8 or C18 phase,phenyl phases may provide additional interactions such as π–π-type interac-tions and may assist in providing additional selectivity If the impurities/activeare very polar, the use of polar embedded phases may provide additional selec-tivity by introduction of a secondary type of interaction such as hydrogenbonding close to the surface in the organic-enriched layer Alternatively, forbasic compounds, different counteranions could be introduced in the mobilephase in order to increase retention of protonated basic amines These areknown as chaotropic reagents and were discussed in Section 4.10 This maylead to increased retention and increased selectivity between critical pairs ofcomponents.

Moreover, if the laboratory has an automated method development system,then this could be used to determine the best set of gradient conditions to givethe best resolution between the critical pair or pairs on multiple columns.When using an automated method development system such as AMDS(Waters, MA) for gradient optimization, generally two types of organic mod-ifiers are used at two different temperatures employing a steep/shallow gra-dient on two to six columns Based on these scouting runs and the users’acceptance criteria for the method, a resolution map is generated by input ofthe data into Drylab, Chromsword, ACD, or another program From this res-olution map the best conditions are chosen and optimized (change in flow rate,multistep gradient ramps, etc.), and these conditions are run to confirm thatthe method that was predicted is indeed representative of the actual separa-tion This is typically called a verification run AMDS relies on constraints ofthe DryLab model Note that Drylab is not suitable for the following types ofcompounds:

• Chiral compounds

• Achiral isomers or diastereomers

• Inorganic ions

• Carbohydrates

• Proteins and peptides

The DryLab model utilized in Waters AMDS has additional requirements: Thenumber of sample components should not exceed 12; peak area% should begreater than 1% These requirements are necessary to achieve greater predic-tion accuracy only Any discrepancies could be corrected manually in DryLabusing the data entry screen by manually entering the retention of the compo-nents from the scouting runs (to assign the peaks with a certain number).DryLab has been used for the method development of model drug candidates

and their degradation products, by optimization of temperature and gradientslope, and the historical review on the milestones and concepts in the devel-opment of DryLab software is given in references 20–23.

8.5.3 Defining System Suitability

System suitability parameters with their respective acceptance criteria should be a requirement for any method This will provide an added level

of confidence that the correct mobile phase, temperature, flow rate, andcolumn were used and will ensure the system performance (pump and detector) This usually includes (at a minimum) a requirement for injectionprecision, sensitivity, standard accuracy (if for an assay method), and retentiontime of the target analyte Sometimes, a resolution requirement is added for

a critical pair, along with criteria for efficiency and tailing factor (especially

if a known impurity elutes on the tail of the target analyte) This is added

to ensure that the column performance is adequate to achieve the desired separation

System suitability requirements for retention time, efficiency, resolution,and tailing factor are set based on prior method challenging experiments andprior method development experience This is a dynamic process; and as theuser gains more experience with the method, the breadth of the acceptancecriteria is further expanded until the method is finally validated for theintended purpose

Two examples are given for setting system suitability requirements for challenging separations In the first example, if a separation is to be carriedout where the retention of the target analyte may have a greater propensity

to vary with slight changes in pH, tighter controls for the pH requirementshould be implemented, where the pH of the aqueous phase should be controlled to ±0.05 units Moreover, some preliminary experiments should

be performed using an aqueous mobile-phase pH ±0.2 units from the desired

pH to determine if this will have an effect on the critical pairs in the tion and what the desirable retention time window is This information is useful to define the system suitability criteria for the method Also, it is rec-ommended to run the separation on different lots of columns to see if there

separa-is any lot-to-lot variability Preferably, running the separation on columns thatwere made from different batches of base silicas is desirable Also, obtainingcolumns from different synthetic bonding batches made on the same batch ofsilica is also desirable In the example shown in Figure 8-22 for a drug productthat contains two actives, three different columns from three different lots ofbase silica were used and the pH of the aqueous mobile phase was varied from5.7 to 6.1, with the target pH being 5.9 Some of the specific system suitabilityparameters and acceptance criteria that were set included tailing factors (5%peak height), retention time windows for peaks A and B, and sensitivityrequirement Some of the selected system suitability parameters were set tothe following:

System Suitability Parameters

• Tailing factor (5% peak height) for peak B ≤ 1.5

• Tailing factor (5% peak height) for peak A ≤ 1.5

• Rt for peak A must be 12.0 ± 1.3 min

• Rt for peak B must be 21 ± 1.0 min

• The S/N of the LOQ solution (0.05%) for both actives A and B must be

≥10 : 1

In the second example, if it is known that a potential degradation product canoccur and will elute close to the active, a resolution requirement should be setfor this critical pair When trying to set a resolution requirement between crit-ical pairs of impurities, standard samples containing the critical pair should bereadily available However, standard samples may not be available with allcritical impurities so the standard may be spiked with authentic impurities Ifauthentic impurities are not available or are in limited quantity, then the drugsubstance may be degraded in solution using mild stress conditions to produce

a decomposition product or products that can be used to define a resolutionrequirement for a critical pair The mild stress conditions should produce

decomposition products in situ in a fast time scale In the following example

in Figure 8-23, the drug substance was stressed with 3% hydrogen peroxidefor 1 hr at 25°C and 80°C to generate impurity A At 80°C, suitable degrada-tion was obtained to determine the resolution requirement between impurity

A and the active B (target analyte) This requirement was set because it waspostulated that this drug substance could be readily oxidized Indeed in solid state stability studies, minor amounts of the impurity A (oxidized impu-rity) were observed under accelerated conditions (40°C/75% RH, 3 months)

Figure 8-22 Waters XBridge 150- × 3.0-mm, 3.5-µm C18 column Column temperature40°C [(A) 90%: 20 mM ammonium phosphate buffer: 10% MeCN, (B) 100% MeCN].Gradient: 10% A to 85% B over 38 min Flow: 0.6 mL/min

Another example in regard to in situ degradation for generation of a system

suitability sample is given in the literature [24]

8.5.4 Case Study 1: Method Development for a Zwitterionic Compound

Method development for the analysis of a zwitterionic drug substance byreversed-phase HPLC was undertaken.The zwitterionic compound A contains

an acidic functionality, (wpKa4.0) and a basic functionality (wpKa3.0) Both of

these pKa values were determined using ACD Labs (Advanced ChemistryDevelopment, Toronto, Canada) software Given this information, the chro-

matographer could apply the pH and pKa selection rules (including pH and

pKashifts) outlined in Sections 4.5 and 4.6 in Chapter 4 to select the optimal

pH to work at in order to avoid working near the pKa values of either of the ionizable functionalities The following case study will illustrate (a) why

working at pH values at or near the pKavalues of the API will lead to rations that may not be robust and (b) what influence the pH has on the inher-ent retention of intermediate compound A and related synthetic by-products.These experiments could be conducted as an exercise to further understandthe effect of pH on the retention of the species in the sample of interest sincethe synthetic by-products may have different ionizable functionalities then theparent compound (intermediate)

sepa-Figure 8-23 In situ degradation for generation of system suitability solution.

8.5.4.1 Gradient Screening. An initial method development was performedusing a Phenomenex Luna C18 (2) column with acetonitrile as the organicmobile-phase component, and the aqueous portion was a 10 mM ammoniummonohydrogen phosphate buffer adjusted to pH 2 with phosphoric acid Ini-tially, a linear gradient was used from 60% to 80% MeCN with a hold at 80%MeCN for 10 minutes An early eluting component was observed close to thevoid volume using this probe gradient Also, no peaks were seen to eluteduring the 80% MeCN isocratic hold Therefore, a new gradient method(shown in Figure 8-24) with an initial isocratic hold to retain the more polarspecies and removal of the latter isocratic hold at 80% MeCN was used Thenew method employed an isocratic hold at 50% MeCN for 5 min, and then alinear gradient was run from 50% MeCN to 80% MeCN from 5 to 25 minutes.Note that a 150- × 4.6-mm column was used, but a 150- × 3.0-mm could havebeen easily used with proper adjustment of the flow rate.

pH study can be conducted The pH study in gradient mode was carried outusing 10 mM ammonium monohydrogenphosphate as a buffer The wpH of theaqueous portion of the eluent was adjusted to 2, 3, 4, 5, 6, and 7 with phos-phoric acid Phosphate is not a buffer at pH 4 and 5, but this is only used forthe pH screening experiment In the event that pH 4 or 5 was deemed accept-able for the separation, a suitable buffer that has buffering capacity in that

Figure 8-24. wpH study on zwitterionic compound A on a Phenomenex Luna C18 (2)column Method conditions are indicated in the figure

region would be chosen Note that upon changing the wpH of the mobile phase,

at least 25 column volumes of the new mobile phase were passed through thecolumn prior to sample analysis This step should be a general requirementwhen performing wpH scouting studies An alternative approach is to performrepeat injections of the intermediate after changing the mobile phase wpHuntil consistent retention times are obtained for all components in the mixture,which would deem that the column is adequately equilibrated

As can be seen from Figure 8-24, the retention of all of the impurities andthe API are dependent on the pH of the mobile phase The intermediate andthe related impurities exhibited lower retention at low pH Retentionincreased initially with increasing pH, where it reached a maximum and thendecreased as the pH was further increased The optimal pH range to carry outfurther method development was determined to be wpH 6–7, where the reten-tion of the API and related impurities did not change as a function of the wpH.The intermediate and its related impurities are zwitterionic in nature andcontain both acidic and basic functionalities; this is confirmed by its bell-shaped dependence on the mobile-phase pH (Section 4.5 in Chapter 4) Peak

X elutes before the main component at low wpH (2.0), but it elutes after themain component at pH values higher than 2.0 (wpH 3–7)

critical pairs are changing elution order, the use of diode array and/or LC-MSshould be employed to assist in peak tracking In this particular example com-paring the elution of impurity X and the intermediate, it was believed that theelution order had switched at wpH 2 and wpH> 3.1 Therefore the reversal ofelution order was determined by comparing the diode array spectrum of impu-rity X and was further confirmed by LC-MS (note that ammonium bicarbon-ate mobile phase was used for pH 7 LC-MS analysis and TFA was used for pH

2 LC-MS analysis, with both using ESI in the positive ion mode) Note that the diode array profiles of impurity X did not directly overlay at pH 2 and

pH 7, and an isobestic point (where two substances absorb at a certain length of light to the same extent) was observed that can be attributed tochanges in conjugation of the aromatic ring when analyzed at different pHvalues (see Section 8-6 for more information on the effect of pH on changes

wave-in UV absorbance) Peak trackwave-ing at different pH values by diode array times is a challenging task, especially if the analyst wants to compare the UVspectrum of the impurity present at different ionization states This was thedriver to perform LC-MS analysis in order to confirm the [M + H]+ion of thisimpurity species Indeed, when LC-MS analysis was performed, it was con-firmed that this impurity had shifted elution order when the pH of the mobilephase was changed from 2 to 7 An extracted ion spectrum of the [M + H]+ion

some-of impurity X at pH 7 was performed for facile identification some-of the impurity

the chromatograms in Figure 8-24 revealed that some late eluting peaks were

observed with the pH 5, 6, 7 mobile phases, and those peaks were not observedwhen the lower pH mobile phases were used In order to troubleshoot if thesepeaks are indeed present in the sample or artifacts, it should be determined ifthe late eluting peaks are (1) synthetic process impurities with different ion-izable functionalities or (2) impurities formed in the sample solvent (indicat-ing lack of solution stability) In order to make this assessment, the stability

of the intermediate in the diluent was challenged In this case study the tion was stored at room temperature under normal light conditions and thediluent was acetonitrile The experiments for wpH 2–4 were performed on day

solu-1, and those for wpH 5–7 were performed on day 2 (≈36 hr after initial ration) A further investigation was performed by preparing a fresh stock solu-tion and storing one-half of the solution in the refrigerator (4°C) for 36 hrwhile the other half of the solution was stored in a clear volumetric flask onthe bench (ambient conditions), and it was determined that these impuritiesare actually formed in the diluent at room temperature under normal lightconditions (see Section 14.8.1 for further details) The solutions were deter-mined to be light-sensitive The case study message is that fresh solutionsshould be prepared daily in amber volumetric flasks and a tray cooler should

prepa-be used when possible when the stability of the sample in solution has not yetbeen determined

method development, multiple columns at various pH values can be screened

in isocratic or gradient using a column switcher or commercially availablemethod development systems that have the ability of running five or morecolumns The reason is that different stationary-phase types may provide a dif-ferent selectivity and give the chromatographer additional confidence inresolving potential co-eluting species In this case study, the separation per-formed on a Luna C18(2) (Phenomenex, Torrance, CA) was compared to theseparation performed on a polar end-capped column, Synergy-Hydro-RP(Phenomenex, Torrance, CA) Similar trends in the retention dependence rel-ative to wpH were observed for all impurities and intermediate on both types

of columns, since the effect of pH on the analyte retention is a function of theanalyte ionization state (Figures 8-24 and Figure 8-25) However, differences

in selectivity and differences in the magnitude of the retention can be related

to stationary-phase type and surface area of the column, respectively ences in selectivity were observed between peak X and impurity Y at wpH 2when comparing these two columns The Hydro-RP column showed greaterselectivity at wpH 2 between impurity Y and impurity X Note that the lateeluting degradation products present in Figure 8-24 at wpH 5–7 were notobserved using this column, since the samples were stored protected fromlight It was determined that the storage conditions were important to mini-mize the degradation product formation Although similar retention profileswere obtained at wpH 6 and 7 (desired pH range for the separation) on bothcolumns, the pH stability of the Phenomenex Luna C18 (2) (up to wpH 10) is