Introduction

The reversed phase of proteins and peptides in TFA is the standard method of choice for initial method development. Too often no other buffers are ever tried. If the separation of closely related molecules is the goal, the buffer systems listed below should be tried with a single column and solvent gradient before further method development is started. This holds true for analytical and preparative method development. Also, if the final method needs to be validated and subsequently pass FDA inspection and only TFA has been investigated, then the FDA inspector may decide that the validation package will not be acceptable until other buffer systems have been tried. If only a few lot numbers of HPLC columns are acceptable and only TFA has been tried, then this is another situation where trying different buffers can give one a more rugged method with higher resolution that will work on all or most lot numbers of columns. It is helpful to develop the methods with a sample from stability studies and/or side fractions from the preparative isolation because the presence of more peaks than the main peak is invaluable for comparison and confidence that the final method of choice will separate all the important impurities and breakdown products.

NaH2PO4

Make a 100mM concentrate of NaH2PO4, the pH will be ~ 4.4. Then adjust the pH in 1/3 of the concentrate to 2.0 with H3PO4. Adjust the pH in another 1/3 of the concentrate to 6.5 with NaOH. Now one can dilute 1/5 or 1/10 and make three different working buffers of 10-20 mM NaH2PO4 at three different pH levels. It is important that one does not try to blend the concentrated phosphate buffers with more than 25-30% ACN because of precipitation problems. If the molecule of interest elutes at higher than 30% ACN do not put phosphate in the final eluting solvent. Also do not blend the phosphate buffer with 100% ACN, instead have the eluting solvent concentration be 80% ACN. During the gradient the 10-20 mM phosphate buffer will decrease in concentration as the concentration of solvent increases and there will be no precipitation problems. If phosphate at a particular pH looks better than all other buffers tried then one can try different pH levels for further optimization of the resolution. After first optimizing the pH of sodium or potassium phosphate buffer systems then one can use triethyl amine phosphate (TEAP) at the optimal pH to check for further improvements in resolution. If TEAP only gives marginal improvement of resolution then it is better to stay with the sodium or potassium salts because of problems with the purity and stability of TEA.

Acetic Acid and Sodium and/or Ammonium Acetate

Make a starting buffer with 20 mM acetic acid, pH ca. 3.2 and make a starting buffer with ca. 20 mM sodium or ammonium acetate pH 6.5.

5 mM HCl

Make a starting buffer with 5 mM HCl. Do not put HCl in the ACN - it will turn yellow. There will be enough HCl left during the gradient to show the effect of this buffer.

Other Buffers

One source of ideas for different buffers than those listed above is from stability studies on the molecule of interest. If the stability studies indicate a buffer that gives less stability at the same pH that other buffers show good stability then try that counter ion or buffer for the analytical method. A loss of stability suggests a strong and specific interaction between the analyte and the buffer which might give very high resolution when used with the reversed phase analysis system. Citrate buffer systems are a possibility. A corollary to this is if a buffer gives very good stability then this may not be a good analytical buffer but would be an excellent process purification buffer system to investigate.

Theory: Why are peaks sharper and show more resolution in only some buffers? The best answer is to ask why peaks are broad in the first place?

The more flexible the molecule, the more conformations and the wider the peak. 1) Counter ions like phosphate can ion pair with Arg, Lys and His and impart greater rigidity to the molecule and in addition exclude these basic groups from the hydrophobic surface of the column. Whereas TFA will ion pair with the same amino acids but then the trifluoro carbon of TFA brings these charged amino acids to the hydrophobic surface leading to broader peaks and longer retention times. 2) Certain pH levels can also lead to a more rigid structure. Different pH levels are also useful for the separation of de-amidations. Both the amide and carboxyl will have no charge at pH 2 but the free carboxyl will have a negative charge at pH >4 while the amide would remain uncharged and this difference can lead to more resolution.

Technical Details

The starting buffer is the low solvent buffer used to equilibrate the column. Use a gradient slope of ~1% solvent change / minute at 1ml/min for 4.6 mm i.d. columns with all of the above buffers. After the runs are done it is usually obvious which buffer system should be used for the rest of the method development because peaks will be sharper and possibly more peaks will appear with the best system. If TFA is not the best system, then further work may be needed to test pH levels that are different then the above buffers. This work will frequently show that the best buffer system for preparative isolation is different than the best buffer for analytical separations. Using a different buffers for prep and analysis avoids the circular logic inherent in analyzing preparative fractions with the same separation system used for purification.

Further optimization can be done using different columns and solvent systems. Even if the above work indicates that TFA is the best buffer system the time is not wasted because the work will form the foundation of the validation package of the final method. Also there is nothing wrong with using two different reversed phase analysis methods for an injectable pharmaceutical. If there is a lot of history with the TFA system and a new system is better for analysis of closely related species, then use the TFA system with a broad range solvent gradient and the high resolution system with a gradient that concentrates on the resolution of the closely related molecules. By doing this one can have two methods with ~45 minutes between injections instead of one method which takes 90 minutes between injections.

After the above has been done and the final buffer system has been chosen, test the method out with different flow rates, different gradient slopes and on different types of HPLCs such as high pressure mixing versus low pressure mixing. If only one type of machine is available try to imitate the other type by varying the gradient. If you have a low pressure mixing HPLC then find out the dead volume in the system. Start the gradient early and after the dead volume has been delivered inject the sample and start the data system. If the results are acceptable then add this data to the validation report. If the resolution is not acceptable then show how the conditions can be varied to achieve the desired resolution and include these results in the validation report. If you have a high pressure mixing HPLC then program a 5 minute initial hold step at the starting conditions, if there is a hold step already then add an additional 5 minutes to it. Again if the separation is acceptable include the data in the validation report otherwise adjust conditions and add these adjustments to the validation report.

Conclusion

The validation report and the method SOP should be written in a way that accentuates achieving the critical resolution between the important peaks.

The USP methods have allowances built into the method to allow for different HPLC systems, different HPLC columns and day to day variation in conditions but these methods require a critical resolution to show system suitability.

One can write and validate a method so that only one type of HPLC and only one lot of columns will be acceptable but this is neither advisable nor wise. The best method will have the parameters investigated so that the maximum ruggedness is built in before the final validation and SOP are finished.