A systematic approach to HPLC method development for small molecules: where do I start?

This question was asked many times. Where do I start the development of HPLC-based analytical methods for small molecules? This question is probably irrelevant to any experienced person, but it may be relevant to recent graduates who are involving themselves developing analytical methods at the start of their careers. In other words, regardless of experience, this question is critical for establishing a systematic approach to the development of analytical methods for small molecules. The easiest way to get the answer is to pick up any suitable method around us quickly, changing parameters, and incorporate a new method into the target molecules being considered for method development because nothing is, in fact, "development from scratch" until the technology is new.

The set of tools that barely differs from method to method, molecule to molecule, or when detection techniques are even different, is the selection of stationary phases, mobile phases, the detection principle, and the optimization of chromatographic parameters. At some points, the differences are insignificant. A chromatographer has to select buffers based on the molecule’s nature, has to select organic solvents or compositions of solvents to discriminate their inherent polarity and to get advantages during separation, and has to select a suitable mobile phase modifier to improve peak shape, keeping in mind the solvents' and modifiers' UV-cutoff values, which are supposed to have impacts on the quality of prospecting chromatography, including compatibility or miscibility with the mobile phase composition, pH of the buffer being considered for the target method, or even compatibility with the hardware, environmental safety, or an unwanted maintenance cost in the long run. The use of an ordinary stationary phase with unattended free hydroxyl groups (no encapped or insufficiently encapped) leaves specific tasks for the method developer to adjust during method development, which is an interesting part of method development that must be handled correctly. However, HPLC stationary phase manufacturers are smarter than ever before, and a plethora of improved stationary phases covered with the inclusion of intentionally designed target interacting functional groups, reduced particle size columns are now available to use without major adjustments, similar to driving cars with auto gear.

The area of scope of this article is to discuss the HPLC method development for small molecules using PDA or DAD detectors with limited resources, how to get advantage from the gradient scouting to qualify the intended method to be used (either qualifying for isocratic separation or gradient separation is mandatory), and how to reduce HPLC run time for both the isocratic and gradient methods without impacting the core chromatography parameters, e.g., without compromising the theoretical plate counts, resolution between closely adjacent critical peak pairs, co-elution, peak purity, specificity, selectivity, and the retention factors (K’).

Developing a gradient method can be the best choice since it helps separating analytes from the complex sample matrix based on the analyte’s polarity at the time of partitioning between the stationary and mobile phases and discriminating them from each other, which is the key to separation. However, without assessing whether the target method can be used as "isocratic " mode ?instead of direct selection of the gradient option, means allowing unnecessary risks, such as baseline drift, poor pump functioning at ?solvent mixing, the appearance of spurious or ghost peaks related to the gradient changes, incorrect peak labeling by an inexperienced analyst, variation between HPLC systems due to dwell volume differences, uncertainty during method transfer between labs, and so on.

Moreover, stability of the target analyte is one of the great concerns about the number of samples, either from process validation, content uniformity, stability sample testing, or dissolution profile testing for slow or moderately released drug products. Regardless of analyte stability, many samples need to be analyzed in a short period of time. As a result of that, a short-run-time isocratic method can be the best choice, especially for limited resources.

We will go over how to optimize a short isocratic method using gradient scouting, how to choose the right mobile phase composition, and how to apply a systematic approach to analytical method development. The option to use the UPLC technique with reduced particle size packing materials, in which analytes of interest are allowed to run (partitioned) inside an imaginary tiny space, burning the calories equivalent to the analytes running inside a giant space, resulting the same calory level burn, which is eventually separated in the same fashion in the UPLC and conventional chromatography, respectively, and contributes to a shorter run time compared to the conventional chromatography is outside the scope of this discussion.

The scope of this article includes how to separate a mixture of multiple specified impurities, including all other analytes of interest (for example, the drug substance, preservatives, and stabilizing agents, if any) in the composite mixture from gradient scouting, how to optimize the gradient run time associated with the HPLC column stationary phase particle size, flow rate, and the dimension of the HPLC column being used. We will also discuss how to evaluate the impact of the late-eluting peak(s) supposed to be retained inside the stationary phase and how to enhance their elution through modification of the gradient scouting program.

This article will also discuss one key component of a gradient scouting study, which includes the selection of the initial gradient composition (IGC) and other associated factors. Setting up the IGC precisely is basically the "Key Starting Point" of gradient separation and is expected to contribute to the perfect separation over the gradient program being set. Though Quality by Design (QbD) is an integral part of method development activities, this article is outside the scope of QbD components.

To make the discussion simple about where to start the method development and how to optimize the initial gradient composition (IGC) for gradient separation and isocratic composition from gradient scouting, let’s imagine a drug substance named Khilgaon-X having six specified impurities, Khl-1 to Khl-6, and we are trying to develop an analytical method for the formulated drug product of Khilgaon-X drug substance. Our target is to separate all six known impurities, i.e., Khl-1 to Khl-6, from the Khilgaon-X drug substance and the complex sample composite of excipients, preservatives, and stabilizing agents, if used any during formulation.

The acceptance criteria for the intended method to be able to comply with the FDA and ICH guidelines has primarily been set. A few of them but not limited to are as follows: all the target analytes should have reasonable retention times or “stay period” inside the stationary phase so that a reasonable number of retention factors (K’) are achieved, which contributes to passing the resolution requirements; there should not be any co-elution; and peak purity criteria should be achieved. We have gathered all the information about the chemistry of the Khilgaon-X drug substance, including associated known impurities, stationary phase’s chemistry information, particle size of packing materials, column dimension, etc. We will apply the principles of method development from scratch, and a systematic approach which will be driven by chromatographic basic separation principles.

Let’s prepare a composite solution of the drug substance and all the associated impurities using a suitable diluent. We need to confirm that all the analytes are in solution. Gradient scouting starts from 5% of B to 95% B, where "B" is the organic phase (acetonitrile) and the associate composition of 95% A to 5% A where "A" is the aqueous phase (water), with gradient curve 6, and this is simply a pre-screening step to see the positions of all the analytes of interest. We are at the very early stage of deciding the wavelength, flow rate, or other chromatographic parameters to optimize, and since the detector is a PDA or DAD, wavelength optimization is a little bit easier. Since we are using only water and the tentative pH of water is around 5.5–6.3, irrespective of the analyte’s nature and their respective retention times or separation patterns, the adjustment of the aqueous phase pH can be optimized at the later phases.

For gradient scouting, it involves the total percent of aqueous and organic phases being delivered through the stationary phase, and time has an important factor on the whole gradient screening, and we need to keep in mind in terms of discriminating the analytes of interest based on their respective polarity and how much quantity of the aqueous and organic phases are inside the column stationary phase to interact with the analytes. For isocratic mode, composition remains the same irrespective of time; however, for gradient separation, time has an impact, and we need to keep in mind the impact of a “shallow” and “steep” gradient and how it works. If the gradient time is shorter for the same flow rate and column dimension, it is a "steep" gradient; if it is longer, it is a "shallow" gradient. A sharp and tall peak can be obtained for steep gradient; however, resolution may decrease. On the other hand, peak broadening occurs with a shallow gradient, however, resolution is expected to increase. A perfect balance is important. ?This option can be optimized in the later phases.

For this screening phase, it is important to select a systematic gradient time based on the flow rate and other chromatographic factors involved. We need to calculate the default gradient time (tG) for the gradient scouting using the equation mentioned below:

Gradient time (tG) = (K* x d%B x Vm x 5)/ F

where K* is the retention factor (minimum 2, maximum 10, and ideal 5), ?%B is the difference between the final percent B and the initial percent B expressed as percentage (0.90), Vm is the void volume, F is the flow rate (mL/min), and 5 is the constant factor for small molecules (1000 Da). Gradient time (tG) depends on the characteristics of the packing materials and the column dimension, including the flow rate being used. For example, if the flow rate is cut in half or the column length is doubled, the gradient time simply doubles, and vice versa.

A complete separation of all the six known impurities and the drug substance with the initial gradient scouting (5% B to 95% B) during the screening phase may or may not be successful; some impurities may be co-eluted, some impurities may be merged with other impurities, and some may be retained inside the column. After the initial gradient scouting, all the six impurities and the drug substance may not be seen even in the chromatogram; however, it will give us an idea about the initial separation profile, which is the “key component” for the subsequent method development steps.

Using a multi column screening technique and taking advantage of variable solvent compositions with the aid of specific software, e.g., Fusion, is outside the scope of this article. The gradient time calculated and applied using the above-mentioned equation may be increased to allow for the elution of retained analytes if necessary or other associated factors. e.g., column temperature can be increased with the same set up, or the pH of mobile phase A (the aqueous phase) can be modified depending on the analytes of interest polarity profile, though all these steps are not part of the initial gradient scouting phase.

Let’s assume all the six known impurities and the drug substance have been found to be separated with our gradient scouting. Here, three possible situations may appear during gradient scouting: all the peaks may either appear within the first half of the total gradient time, may be in the last half, or may be spread out across the beginning to the last minute of the total gradient run time. This is the another “key point” in evaluating whether the target method being developed will be qualified for isocratic separation or, if there is no other choice, whether gradient separation is mandatory at this situation.

In our example, since there are six known impurities and one drug substance, all the analytes of interest are within a wider range of polarity, and all of them possess unique chemical entities in terms of their chemical and physical properties, even though known impurities are structurally similar to the parent molecule of the drug substance, they are still distinct, and a gradient separation is clearly required no matter method is qualified for isocratic option.

Whether the method will be used to determine major analytes such as API, preservatives, or stabilizing agents (USP method category I) or quantitative determination of trace components such as impurities (USP general method category II), we must decide first whether the scope will be isocratic or gradient from the scouting screening. The difference in retention times between the first and last peaks in gradient scouting can be critical information in making that assessment. If the difference is less than 25% of the total gradient time, then it is assumed that the target method can be isocratic separation; if the difference is above 25%, then isocratic separation is not possible and gradient mode will be the best choice and it is “mandatory”.

Let’s assume the HPLC column being used in our example has a length of 150 mm and an internal diameter of 4.6 mm. If the surface area of the packing materials (to be found in the column manufacturer's certificate of analysis) is 325 g/m2 and the particle pore diameters are around 100A°, the calculated gradient time using the above equation is around 35 minutes if the final flow rate is 1.0 mL/min. During our gradient scouting, let’s assume our first peak, Khl-1, elutes at 4.2 min and the last peak, Khl-6, elutes at 17.5 min. The difference between the first and last peaks is 12.8 minutes, which is 36% of the total gradient time of 35 minutes. Since the difference is greater than 25%, isocratic separation is not possible, and the gradient method is mandatory.

Let’s imagine two other possibilities: the first peak appearing at 2 min (obviously after the void time of approximately 1.5 min) and the last peak appearing at 10 min, or for the second scenario, the first peak appearing at 22 min and the last peak appearing at 30 min. The difference between the first and last peaks in both cases is 8 minutes, which is 23% of the total gradient time of 35 minutes, primarily the method to be qualified for the isocratic separation rather than the gradient run. In our example, since the number of analytes is seven, including the drug substance, in such a situation the preference of the gradient option rules out the isocratic option even though it's been qualified, and that’s just because of the wider number of analytes being separated and the degree of complexity of the method. If the analytes of interest are few, then, the isocratic option should be the best choice.

Gradient scouting helps us selecting the isocratic composition of mobile phases A and B, and this can be obtained from the composition at the average retention time of the first peak and the last peak of the scouting run. For example, our first peak appeared at 22 min and the last peak at 30 min, and we intend to develop the method in the isocratic option, so, first, we need to estimate the mobile phase composition at the average retention time of the first and last peaks, which is 26 min (22+30/2). A gradient scouting run from 5% B to 95% B for 35 min estimates a gradient composition of 40% A and 60% B at 26 min, which is the target isocratic mobile phase composition for the intended isocratic separation.

The separation profile of analytes obtained during the scouting run, with the first peak appearing at 22 minutes and the last peak appearing at 30 minutes, should have a similar pattern to the isocratic separation of mobile phase A at 40% and mobile phase B at 60%. Since initial isocratic composition starts at 40% aqueous phase and 60% organic phase instead of 95% aqueous phase and 5% organic phase (gradient scouting run), our first peak should appear somewhere after the void time (after 1.5 min), and all the subsequent peaks should appear immediately after the first peak (preferably within 8 minutes of the first peak coming out).

Let’s go back to our initial example where Khilgaon-X was our imaginary drug substance and we intended to develop the analytical method to be able to quantify the main analyte within the shortest possible run time so that many samples could be analyzed within the shortest possible time. Developing an analytical method for the quantification of the main component, e.g., the API, can be done on a "trial and error" basis, and selection of the isocratic mobile phase composition of the aqueous and organic phases in the 40:60 ratio can also be obtained by applying ?just a few “trial and errors” runs instead of lengthy and tedious gradient scouting; however, the separation of associated analytes, e.g., the six known impurities supposed to be present with the main component, remains unattended, which may be the concern with the method being developed. In such a situation, forced degradation can be a useful tool to confirm the peak purity of main components, and if the main peak is spectrally homogeneous and pure, there should not be any concern for the method of quantification of the main component. Moreover, an extended run time of 2.5x the main peak retention time is usually applied to confirm all the possible peaks are completely eluted from the HPLC column so that they can’t interfere with the next run, and nonetheless, all these steps may not be able to provide sufficient information about all the six impurities of interest until all of them ?are injected and their respective positions are confirmed ?in the chromatogram.

Establishing the isocratic composition from the gradient scouting is unique in all these aspects and can make answers all the uncertainties since the separation profiles of all the known impurities exist there and their respective retention times are known. The flow rate can be adjusted to optimize the target run time without impacting the relative retention time (RRT) of known impurities, or even if impacted, the consequences are known and can be adjusted as appropriate. Moreover, there is no additional burden to keep the total run time up to 2.5X the retention time of the main component of interest.

For the selection of a gradient option from gradient scouting, the initial gradient composition needs to be selected using the same approach as the isocratic mobile phase composition, which has been selected based on the average run times of the first and last peaks. For the isocratic run, the composition is fixed throughout the run time, however, for the gradient run, this is the initial gradient composition, and the subsequent gradient program can either be the same pattern of 5% B to 95% B or can be modified based on the separation requirements. The choice of preference is to continue the same trend of 5%B to 95%B.

Let’s go back to our original example. Our first peak, Khl-1, appeared at 4.2 min and the last peak, Khl-6, appeared at 17.5 min; their difference was 12.8 min, which is more than 25% with respect to the 35 min gradient time, and as per the assessment, isocratic separation is not possible; the gradient option is the only option for such a situation. So, it needs to estimate the initial gradient composition based on the average retention time of the first peak and the last peak, which is 11 min, and the estimated composition is 67% aqueous phase and 33% organic phase. This systematic approach allows us to repeat the same peak separation profile as obtained during gradient scouting. The only difference is, at this step, the initial gradient composition is 67% aqueous phase and 33% organic phase instead of 95% aqueous phase and 5% organic phase in the gradient scouting initial composition. It is important to remember that, once the initial gradient composition has been estimated and implemented, the HPLC column must be equilibrated with this new initial gradient composition for a period, for example, 3-5 minutes in isocratic mode, before gradient to the target level begins.

As we already discussed, the subsequent gradient pattern should be the same as the original 5%B to 95%B, which means our new initial starting point is 33%B. So, we need to continue the gradient run up to 35 minutes, and the final B will be 95%. We should expect the same separation profile as we have seen in the scouting run, however, retention times are expected to reduce for all the analytes' peaks since we have modified our initial gradient composition. Initially it was 5% B, now after modification it is 33%B.

We need to keep in mind that our initial gradient screening for all the subsequent steps is guided by the initially calculated total gradient time of 35 minutes and will remain the same even after adjusting the initial gradient composition. Depending on the separation requirements, a total of 35 min can be programmed from 33% B to 95% B or can be divided into three different segments: the initial and last part under the "steep gradient," where gradient composition changes are faster within the shorter run time window, and the middle part under the "shallow gradient," where gradient composition changes are slower comparatively under the longer run time window. Separation improves at a shallow gradient compared to a steep gradient; however, chances of peak widening take place at a shallow gradient and need to be optimized as excess peak widening contributes to a loss of sensitivity which has already been discussed at the beginning. Finally, whatever needs to be set is determined by the “closely adjacent peak pairs” or the needs for further adjustment or improvement discovered during gradient scouting.

At this level of method development, we are somewhat confirming that all the six known impurities and the drug substance have been completely eluted. However, to double check for any retained analyte inside the column, we need to include at least 1.5X of the total 35-min gradient run time with the final gradient composition (95% B:5% A or any other modified steep gradient at the end where %B is still in a higher composition) and continue the run, for example, up to 52 min. The unknown analyte, if any, is still being retained inside the packing material and should elute at this extra time window, and any necessary adjustments may be made in such a situation. This additional run time may also be useful in understanding the presence of spurious peaks in the form of gradient artifacts. In most cases, only the spurious peak caused by gradient changes appears as an unretained peak. This is because changing % B (organic phase) from a lower concentration to an extreme concentration (up to 95%) barely allows analytes to be retained inside the column, and eventually everything should be completely washed out at this higher organic concentration for RP chromatography.

Let’s go back to our initial total gradient time of 35 min with the initial gradient composition already set based on the average run times of the first peak and the last peak in the original gradient and after confirming that no unretained peak(s) appeared at the extra time, i.e., 35–52 min. At this step of method development, for the gradient option, the flow rate optimization step can be included.

For the same column dimension and packing composition, if the flow rate is just changed to double, the total gradient time should be reduced to half of the 35 min, i.e., 18 min, and vice versa, where the gradient program will remain unchanged and the separation profile should also be identical, except for the shifting of RRTs of six known impurities of interest, including the drug substance Khilgaon-X. The modification of flow rate to double is an extreme change that significantly reduces gradient run time while making it difficult to maintain the original separation profile obtained during the scouting phase. Because the number of total analytes is too large, the possibility of peak merging or loss of resolution is increased. In such a situation, flow rate can be optimized within a moderate range, i.e., instead of jumping directly from 1.0 ml to 2.0 ml, 1.2 and 1.3 can be obtained gradually, and it is necessary to observe optimum resolution of the critical peak pair(s).

Also, we need to keep in mind the elution of any analyte near the void time since increasing the flow rate means decreasing the void time. The flow rate can be reduced as much as the retention factors (K') for all the seven analytes of interest found within the range of not less than 2 and not more than 10, including the resolution requirements of the most closely adjacent peak pair(s) of at least 1.0, to achieve the shorter gradient time. Column temperature can be combined with flow rate optimization because elevated column temperature helps reducing column back pressure for a higher flow rate, reduced particle size, smaller columns, and higher viscosity solvents. The most important aspect is, an increased column temperature aids in peak separation, or improving resolution.

The gradient scouting is a systematic approach to developing analytical methods from "scratch" both for the isocratic and gradient separations, either for single or multiple analytes of interest. Everything is guided by the basic chromatographic separation principles, e.g., selection of packing materials, particle size, particle pore volume, packing material’s surface area, initial flow rate selection, void volume, calculation of default gradient time (tG), and finally selection of scouting from an initial composition of 5%B (organic phase) to 95%B (organic phase) and the corresponding aqueous phase A composition. Let’s imagine we have increased the flow rate from 1.0 mL/min to 1.5 mL/min, and with this new flow rate, our calculated total gradient time is 24 min. At this point, we must return to our previous steps and determine the average retention times of the first and last peaks, followed by an estimation of the initial composition of mobile phase A and mobile phase B, while keeping the subsequent gradient programs consistent—basically, we must repeat the previous steps one more time. As we mentioned before, the HPLC column needs to be equilibrated for a while in isocratic mode with the estimated initial gradient composition as the starting point, the same way we need to select the gradient separation time so that the HPLC column gets ready for the next injection, and this composition should be the same as the initial mobile phase composition in isocratic mode.

Getting all the six known impurities and the drug substance separated from each other with the suitable retention factors (k’) and resolution is the primary objective of the method development, and this can be obtained using at least 1-3 percent of the target nominal concentration of the test solution to be used for impurity analysis, which is slightly higher. The purpose of this higher concentration of known impurities is to separate them from each other without losing sensitivity during the gradient run.

Once the separation profile is established, it needs to estimate the limit of detection (LOD) and the limit of quantitation (LOQ) for all the six known impurities, including the API itself, which will be used to quantify the known and/or specified impurities and the unknown and/or unspecified impurities respectively. The systematic approach to determining the LOD and LOQ is to make a series of dilutions of the above-mentioned 1-3% stock solution of all the six known impurities and API stock solution until their theoretical concentration is assumed to be close to “zero”.

Let’s assume we have a total of ten (10) dilutions of the stock solution mentioned above. To estimate the concentration of LOD and LOQ from the calibration curve, one of the best options is to inject all the 10 dilution steps into the HPLC system with triplicate injections at each dilution level using the gradient method just developed. Low concentration dilution steps may contribute areas of analytes of interest close to base line noise and isolating them from base line noise can be difficult in most cases because analyte concentration is theoretically very trace but practically close to zero level concentration and cannot be detected with the chromatographic conditions being used, including the HPLC system's amplification/ best performance. Even some dilution steps out of a total of ten (10) steps may not respond in terms of showing up any analyte, or some analytes may show up and others may not, which is quite normal.

As we mentioned at the very beginning of this discussion, all the six known impurities, e.g., Khl-1 to Khl-6, are somehow structurally and chemically similar or close to our imaginary drug substance Khilgaon-X, but eventually they are not the same chemical entity, and because of their inherent differences, they have been found to be separated from each other with the gradient method being applied. So, in that sense, their respective sensitivity in terms of detection, should also be different, irrespective of the wavelength being used.

It needs to calculate the standard error (σ) by plotting all the concentrations of all the 10 dilution steps against the area responses observed from the calibration curve. Remember that any dilution steps missing for detection response (no area detected), especially at the lower dilution steps, should still be included in the calibration curve, and standard error (σ) will correct for the lost step(s), if any, and accordingly will adjust the appropriate estimation of LOD and LOQ concentration for each analyte of interest.

Assume that, of the total of 10 dilution steps from 1% (with respect to the target nominal concentration of the impurity test solution) to 10000x dilution, the last dilution step (10?7) is practically close to zero concentration and should see no detection. However, if this low dilution still has consistent detector responses for the triplicate injections to be able to be separated from the base line noise due to either the super performance of the HPLC system electronics or the respective molar absorption capacity of the analyte of interest, then, the estimated LOD and LOQ (LOD=3.3σ/S, LOQ =10σ/S, where σ = Standard Error, S= Slope of calibration curve) concentration will be very low. On the other hand, for example, if the last three or four dilution steps having low concentration of analytes of interest are found incapable to be detected or even if detected their area responses are not consistent (% RSD of triplicate injections are significantly high), the standard error, σ will automatically be impacted and will be pushed forward to estimate the respective LOD and LOQ concentration at the higher end and the process is spontaneous.

Things need to keep in mind during the estimation of ?the LOD and LOQ concentration of analytes of interest, in the calculation formula, σ is the " Standard error". The difference of standard deviation and standard error is, standard deviation corresponds to the single data and measure how multiple readings of single data deviates from each other, on the other hand, standard error corresponds to multiple data instead of single data and how multiple data deviate from each other. In our example, the responses of 10 steps of series dilution samples against their respective concentrations, the variation to be estimated using the statistical tool "standard error," and the estimated concentration of analytes of interest can be determined by combining the other calibration curve tool "slope."

Once the LOD and LOQ concentrations are estimated from the calibration curve using the above-mentioned equations, it is necessary to confirm the consistency of the actual LOD and LOQ concentrations at this stage of method development. This can be done by preparing the estimated LOD and LOQ concentrations for the analytes of interest separately and injecting the solution into the HPLC system to confirm three core acceptance criteria, i.e., the signal-to-noise ratio (S/N ratio) of 3 and 10 for LOD and LOQ, respectively, and the RSD of six replicate injections of LOQ solution of no more than 15%, including the spiked placebo recovery at LOQ level for at least 70–130% or other scientifically justified ranges.

A calculated and confirmed LOQ concentration is a useful tool for calculating the nominal concentration of the test solution to be used for impurity quantification. The nominal concentration of the test solution can be optimized using the main peak shape of our fictitious drug substance, Khilgaon-X, the loading capacity of the column, and so on. Since the accuracy or recovery study at the target level (LOQ to 150% of specification level) is assumed to be part of the pre-validation prior to the formal method validation, the target level of all the analytes of interest (all six known impurities at their respective levels and the drug substance itself representing the unknown impurities at the target level) is expected to be spiked into the complex sample matrix of the formulated drug product and into the placebo mixture, respectively, and appropriate recovery will be confirmed. So, the concern about setting up a higher nominal concentration test solution to be able to detect all trace impurities is not a real concern.

However, under the assumption that the estimated LOQ concentration is 0.05%, the nominal concentration of the test solution should not be too diluted and should correspond to 100%. The method for optimizing the nominal concentration of test solution for the analytical methods to be developed for drug substances is much easier to adjust than the method for the formulated drug product with a low dose per unit, especially for potent drug substances with very trace dosage units (very low dose drug product). In such a case, matching the test solution's nominal concentration of 100% to the estimated LOQ level of 0.05% is difficult. For the solid dosage forms of a formulated drug product, several unit doses can be combined into a small volume of diluent, and the nominal concentration can be increased if the target analyte is highly soluble in the diluent and the excipient matrix has less impact. Appropriate filtering can be useful; however, for the liquid dosage forms, it is difficult to increase the nominal concentration. Solid-phase extraction (SPE) can be the alternative in such a situation to increase the analyte concentration of the nominal test solution; this is, however, outside the scope of this discussion.

Any method to be developed, either for the quantitation of the major components or for the quantitation of trace impurities, can be derived from systematic gradient scouting, which could be the "starting point." The isocratic or gradient separation must be fitted and realistically be justified based on the number of samples to be analyzed each time and the number of analytes. Since quantification of major components of an assay method is straightforward, a shorter run time should be expected for such methods. Even if the method requires a gradient, a systematic approach must be used to optimize the shortest gradient time while keeping the analytes' interest, sensitivity, and resolution requirements in mind. We must consider the consequences of gradient separation in terms of method transfer and reproducibility, as well as the expected variation from equipment to equipment. So, the systematic method development approach from gradient scouting can be a useful tool to evaluate the methods' correct pathway. This will aid in resource conservation by developing robust methods.

This article has been written based on the information available in chromatography textbooks, journals, and published articles.