Chromatography Forum

Hi everyone,

I'm facing a real big big challenge in method re-development for impurity determination.

The original method conditons are the following:
- ACE 3 C18 3.0 µm 150 mm x 4.6 mm
- Detector Wavelength : 220 nm
- Column temperature : 20 ºC
- Autosampler temperature : 10 ºC
- Flow : 0.9 mL/min
- Mobile phase A : MeOH:ACN:20mM K2HPO4 pH 7.0 (50:17:33, vv)
- Mobile phase B : MeOH:ACN:20mM K2HPO4 pH 7.0 (55:35:10, vv)
- Dissolution mixture : 0.01% TFA in MeOH
- Injection volume : 15µL
- Run time : 45 min
- Gradient Program : Time (min) % Mobile phase A %Mobile phase B
0 100 0
2.0 100 0
20.0 60 40
30.0 0 100
38.0 0 100
38.1 100 0
45.0 100 0

This original method was scientifically sound validated (not fully), but we are having increasing problems in stability (example attached):
- Variability in the individual content of impurity peaks
- Co-elution of impurity peaks / peak distortion
- Poor reproducibility in relative retention times of impurity peaks

The drug product is the worst case in terms of matrix complexity, since this product is manufactured by blending the API with additional formulation excipients, originating a more complex excipient matrix which is finally compressed into film coated tablets.

A significant matrix effect caused by the excipient portion of the formulation can be recognized when chromatograms of drug substance versus drug product (tablets) solutions are compared. A marked baseline rise is observed.

At different wavelengths, baseline rise is attenuated but we loose sensitivity for the main peak and impurities.

I would like to ask some help or recommendations based on your experience in order to optimize chromatographic solutions and ensure sufficient resolution as a form of anticipating a future regulatory audit.

Anyone interesting in discuss this in detail? Could you please indicate an email to go through the steps and tests already performed with gradient changes, etc.

Thank you in advance,
Pedro Serôdio

Immediately I would point on several things:
Temperature difference column to sample is 10°C
Mobile phase shows not a significant difference from A to B. Did you try an isocratic run?
45 min runtime with that 150 x 4,6 column is a lot.
What sample preparation do you use to get rid of the matrix?
I'm missing your example, nothing attached.
Did you try any other column modification beside C18? I would test also HILIC.

Try dissolving the samples in mobile phase and not in methanol - the methanol is a strong eluent that will shift peaks around and distort them.

Peter

Immediately I would point on several things:
Temperature difference column to sample is 10°C
Mobile phase shows not a significant difference from A to B. Did you try an isocratic run?
45 min runtime with that 150 x 4,6 column is a lot.
What sample preparation do you use to get rid of the matrix?
I'm missing your example, nothing attached.
Did you try any other column modification beside C18? I would test also HILIC.

Thank you very very much for your interest and suggestions.
In fact, we have tested isocratic (MeOH:ACN:Buffer 55:20:25 ; the main peak elutes at a total organic content/aqueous buffer of 70:30) and several other gradient approaches but we are not getting any considerable improvement in separation/resolution of pre and post-main peak co-eluted impurities. I have a file where I summarize all these tests with chromatograms but I'm not finding how to upload it here. Can you help on this?

The sample prep is:
- transfer 10 tablets to a 100mL low-actinic volumetric flask
- add about 50mL of dissolution mixture and vigorously shake using a wrist-action shaker for 70min. If after 70min the tablet has not disintegrated, continue shaking until complete disintegration.
- add dissolution mixture up to the neck of the flask and sonicate for 10 min
- allow to equilibrate to room temperature
- dilute to volume with dissolution mixture and mix well
- clarify the solution by centrifuging 30mL of the stock for 10 min at 2000 rpm
- use the supernatant as the sample solution

No, we only have changed within C18 with other equivalent columns (Zorbax, YMC, Xbridge) unsuccessfully. I think HILIC will change dramatically the polarity and affinity molecule/column and consequently elution order. Thanks for the reccomendation.

It would be nice to know how to upload the file.
I would appreciate any other suggestions and directions to improve the chromatographic profile.

Thank you again and I expect additional feedback from your experience.

Try dissolving the samples in mobile phase and not in methanol - the methanol is a strong eluent that will shift peaks around and distort them.

Peter

Thank you for your interest.

We have already tested to "disperse" (at the end this is like "a suspension" not a solution) in mobile phase and inject it in HPLC.
The profile was similar to the one obtained with methanol: no improvement in peak shape (separation/resolution) of the pre and post-main peak impurities.

The sample is a tablet with 32% API and the remaining portion corresponding to several excipients. The solvent needs to be compatible with mobile phase (not too strong) and must be able to desintegrate the tablets first and then disperse the powder to dissolve the API. We need centrifugation to clarify "the solution".

It would be nice if you would be able to see my file with chromatograms. Can you indicate me how to upload the file here?

Thank you again.

Dear Pedro,

am I right: you place ten tablets in the same 100 mL flask?

Each of these tablets contain 32% of API.

Could you outline the final concentration of the sample solution that is finally injected?

Thank you.

Regards

Florian

Dear Pedro,

am I right: you place ten tablets in the same 100 mL flask?

Each of these tablets contain 32% of API.

Could you outline the final concentration of the sample solution that is finally injected?

Thank you.

Regards

Florian

Thank you very much Florian for your interest.

10 tablets of 100 mg API in 100 mL generates a final concentration of 10 mg/mL for impurity evaluation, then we dilute this solution to 5 mg/mL to determine the assay. LOQ is 0.05% in relation to 10 mg/mL, to ensure a signal-noise ratio of 15.

One of our first thoughts was to use only 5 mg/mL for assay and impurities simultaneously, ensuring that we are able to reduce also the LOQ to half the actual value (to maintain the threshold as 0.05%). Even doing this we will face some issues with resolution/separation of pre-main peak impurities.

I need to upload the file with chromatograms to provide a better understanding of this issue.

I would appreciate all your comments questions as an added value.

Thank you

Pedro

Dear Pserodio,

When you are on the side "View all topics" on Liquid Chromatography in this forum you will find a topic from Tom Jupillee, How to embed chromatograms and reports. Hope you will be successful to upload your report and chromatograms.
Depending on the substances used to cover the tablets I guess there are used some sugar compounds. What do you do to get rid of matrix in the sample. Did you try SPE already?

Dear Pserodio,

When you are on the side "View all topics" on Liquid Chromatography in this forum you will find a topic from Tom Jupillee, How to embed chromatograms and reports. Hope you will be successful to upload your report and chromatograms.
Depending on the substances used to cover the tablets I guess there are used some sugar compounds. What do you do to get rid of matrix in the sample. Did you try SPE already?

Thank you for the info.
I'll upload the chromatograms as images, yes in a next topic. I really need the input of people with experience.
We do not have SPE in-house. We only do mechanical shake and centrifugation in an attempt to clarify the solution.
Yes, but it could be an option.

This is the overlay of profiles with original gradient using MeOH and Mobile phase A as sample solvent.
No improvement in peak shape is observed.

Original gradient (red) vs. extended gradient with column at 30ºC.

Dear Pedro,
it seems that you are the owner of a very interesting question.

I would try to describe my first thought on the described problem. Nevertheless without hands-on experience and without seeing the complete instrument-setup and substance it is always difficult to guess.

Flow is to low (when the column is filled with HALO particles, flow is extremely low).

Injection volume is maybe too high. Changing to mobile phase will not help much. My recommendation is to reduce the injection volume to 5 µL (better will be 1 µL) to be sure there is no overloading effect during the experimental runs.

Is TFA in sample solvent really necessary?, TFA will give you an addition signal in blank-runs. The TFA may also influence the pH of your chromatographic system on short-time.

Using a cooled autosampler is usually a good idea, except when you have problems with the solubility/precipitation. Is your centrifuge also cooled down to a temperature below the temperature of the autosampler? Nevertheless 2000 rpm sound quite low value for using a centrifuge.

The baseline of the provided chromatogram looks strange on my first thought. Therefore I’m also not sure if re-equilibration was sufficient. Maybe we should have a closer look on it. Was a DAD or a simply UV-Detector used for this chromatograms?

Some of your peaks (but not all) have a observable fronting. The first conclusion is that the chromatographic system is basically not suitably for your compound. Of course there are reasons that one judge to use such a system is at the end good.

Other aspects to get doubts on the robustness of the described method are:
Ternary mixture (e.g. ACN and Methanol) is used
Salt-concentration gradient with usual (?) C18 column at highly organic conditions
Using such a small organic content change during the gradient, will combine all disadvantages of gradient and isocratic mode for your analysis
Is the using of phosphate buffer especially at pH 7 really necessary? Was the necessary buffer-strength really evaluated? Is there are ratio to choose pH 7?
Was there are trial about using 0.1 % TFA in mobile Phase?

Your chromatographic system have low performance and separation capabilities (peaks of interest elutes during 10 minutes of a >40 minutes chromatogram even if many advanced (and therefore non-robust) techniques (e.g. salt gradient, ternary mixtures) were used.

I think there is no simple solution to fix it. It is the time to make a complete new development with consideration of the latest developments in the hplc-sector.

Good-luck
Klaus

Now I have seen your most recent chromatogram. Here are my first thoughts:
You have problems with re-equilibration. Problems with the pump are also possible. Or chemistry interactions are absolutely not sufficient for your substance.

Dear Pedro,
it seems that you are the owner of a very interesting question.

I would try to describe my first thought on the described problem. Nevertheless without hands-on experience and without seeing the complete instrument-setup and substance it is always difficult to guess.

Flow is to low (when the column is filled with HALO particles, flow is extremely low).

Injection volume is maybe too high. Changing to mobile phase will not help much. My recommendation is to reduce the injection volume to 5 µL (better will be 1 µL) to be sure there is no overloading effect during the experimental runs.

Is TFA in sample solvent really necessary?, TFA will give you an addition signal in blank-runs. The TFA may also influence the pH of your chromatographic system on short-time.

Using a cooled autosampler is usually a good idea, except when you have problems with the solubility/precipitation. Is your centrifuge also cooled down to a temperature below the temperature of the autosampler? Nevertheless 2000 rpm sound quite low value for using a centrifuge.

The baseline of the provided chromatogram looks strange on my first thought. Therefore I’m also not sure if re-equilibration was sufficient. Maybe we should have a closer look on it. Was a DAD or a simply UV-Detector used for this chromatograms?

Some of your peaks (but not all) have a observable fronting. The first conclusion is that the chromatographic system is basically not suitably for your compound. Of course there are reasons that one judge to use such a system is at the end good.

Other aspects to get doubts on the robustness of the described method are:
Ternary mixture (e.g. ACN and Methanol) is used
Salt-concentration gradient with usual (?) C18 column at highly organic conditions
Using such a small organic content change during the gradient, will combine all disadvantages of gradient and isocratic mode for your analysis
Is the using of phosphate buffer especially at pH 7 really necessary? Was the necessary buffer-strength really evaluated? Is there are ratio to choose pH 7?
Was there are trial about using 0.1 % TFA in mobile Phase?

Your chromatographic system have low performance and separation capabilities (peaks of interest elutes during 10 minutes of a >40 minutes chromatogram even if many advanced (and therefore non-robust) techniques (e.g. salt gradient, ternary mixtures) were used.

I think there is no simple solution to fix it. It is the time to make a complete new development with consideration of the latest developments in the hplc-sector.

Good-luck
Klaus

Thank you very very much for your time, interest, thoughts and recommendations.

- Reducing of injection volume will impact on sensitivity. Currently LOQ is 0.05% of 10 mg/mL with an area about 37000 (S/N > 15). Obviously with other optimized condition I can compensate and increase sensitivity.

- You're right. I think TFA is not critical to be present in solvent. When we have tested mobile phase (with no TFA) as solvent, desintegration/dispersion of tablets was also able to be completed. So no need to include TFA. I belive only MeOH has the same effect on sample "dissolution".

- After centrifugation we cool de solution at room temperatura (20ºC, not less). We have already tested 5000 rpm during an extended time (20 min) but no significant improvement was observed in terms of excipient sedimentation in the tube. Solution seems more clear but not so much.

- In routine we use DAD, now for redevelopment we are using PDA (190-400nm). At 250 nm, baseline becomes flat (good look!) but we suffer a tremendous loss of sensitivity for main peak and impurities. I'll upload other examples with zoomed images for better understanding.

- We have received this method from an outsourced origin. I think they used initially only MeOH/buffer but due to high backpressure (~3500 - 4000 psi), they added ACN to "decrease" the pressure.

- The API (not the corresponding tablet) was originally analyzed with an aqueous buffer 20mM ammonium acetate / MeOH. I don't have a rational to use phosphate salt buffer in fact.

- "Using such a small organic content change during the gradient, will combine all disadvantages of gradient and isocratic mode for your analysis". With this column even when we extend the gradient starting from 55% buffer and run a organic wider amplitude , baseline noise seemed to become worst.

- "Was there are trial about using 0.1 % TFA in mobile Phase?" We did not try that yet, but it could be an alternative option to test.

Since you do not have so much time to full development, what do you think we should prioritize in order to achieved a optimized solution in a timely manner?

Thank you again for your understanding and help.

Pedro

Now I have seen your most recent chromatogram. Here are my first thoughts:
You have problems with re-equilibration. Problems with the pump are also possible. Or chemistry interactions are absolutely not sufficient for your substance.

With ACE 4.6x150 mm, 3 um using the original and the other tested gradients we always have around 3500 psi.
In these test we are using three independent channels (for ternary mixture) in a Waters Alliance equipment.

XBridge, YMC Pro C18 columns did not bring better resolution when used with original and extended gradient.
From your experience and from the examples I have posted, do you agree on using original column ACE 4.6x150 mm, 3 um, but changing mobile phase (e.g. remove phosphate salt and include TFA), and other parameters?

The API and its major impurity is similar to Boceprevir molecule.