accuracy validation failure

[fandaga]

I need advice to troubleshoot an accuracy validation failure of a reversed-phase HPLC assay/degs method for a tablet dosage form. The accuracy experiment design was 6 preparations of drug substance (DS) and known degradants spiked into the excipient matrix (investigation phase validation). The DS was spiked at the 100% level, and the degradants were spiked at the 0.1% level. Accuracy passed for the degs but failed for the DS with two high recoveries of 103.5% and 103.1% (acceptance criteria of 97.0% to 103.0% for DS). The other recoveries were 99.2%, 100.6%, 101.4%, and 102.3%.

I executed the validation several times on a non-GMP instrument and each time with very consistent, passing recoveries between 99.0% to 101.0% before I handed it off to the validation chemist. I also did full characterization of the method consistent with validation requirements of ICH Q2. This included full robustness testing. So far in the investigation, we've ruled out the instrument (including the injector) and any obvious analyst errors (such as pipetting, QS to volume, and weighing).

The method uses quantitation by external standard, and one thing that concerns me is that the standard preparation is not the same as the sample preparation. Standards are prepared by dissolving 30 mg of DS with 50 mL of MeOH / H2O (80:20, v/v), then QS to 100 mL. Samples are prepared by extracting the DS from crushed tablets (100:524 DS/Excipients, w/w) using 50 mL of MeOH. Sonication for 20 minutes, followed by shaking for 40 minutes. QS to 100 mL with MeOH, then pipet 15 mL into 50 mL and QS with MeOH / H2O (80:20, v/v).

The final concentrations of standards and samples are both 0.30 mg/mL, but the solvent composition of each solution is slightly different. The composition of standards is MeOH / H2O (80:20, v/v) and the composition of samples is MeOH / H2O (86:24, v/v). I'm wondering if the method may not be as user friendly as I thought because of these preparation differences.

Any advice is much appreciated.

[mohan_2008]

Your diluent composition for your:

Sample Diluent: (86:14 MeOH:H2O)
Standard Diluent: (80:20 MeOH:H2O)

Potential problems with the method:

1. Standard preparation (my assumption): You may have achieved the standard solubility - readily with less sonication or shaking as compared to your sample preparation.

2. Sample Preparation (my assumption): It appears considerable time was spent on sonication/shaking to obtain a possible complete sample extraction into the sample diluent.

In step 1 (standard prep). since you used 80:20 Methanol:Water which has 20% water in it - should not introduce much error while QS to volume.

In step 2 (sample prep). since you used 100% Methanol (no water) to extract the sample - followed by sonication and QSing - should introduce considerable error if the methanol is not allowed to equilibrate to room temperature prior to QS ing. It takes several hours for the methanol to contract and specially if sonication generates heat. I think this is what happened.

You should not have any problem using different diluents unless your mobile phase is considerably weak and you are generating a strong solvent interaction due to the higher organic.

[grzesiek]

"The other recoveries were 99.2%, 100.6%, 101.4%, and 102.3%." - any pattern in the recoveries?

"So far in the investigation, we've ruled out the instrument (including the injector) and any obvious analyst errors (such as pipetting, QS to volume, and weighing). " - how?

Have you tried preparing samples simultaneously and checking recoveries in the same run?

[Consumer Products Guy]

While not exactly the same, your diluents are in reality pretty close, I would doubt that is an issue.

When I do accuracy/recovery studies I use a concentrated spiking solution and then dilute that concentrate down to make my standard for the accuracy study, eliminates a source of error. We do triplicate preps, and also at 80, 100, and 120% of target (plus a placebo).

Also, doesn't FDA ORA 4.5 document list in its Appendix accuracy limits of 95.0 to 105.0%?

[fandaga]

Answering some of the questions:

1. The DS is soluble in the standard diluent at greater than 2X the concentration.
2. The extraction time is due to the product being a solid dispersion of the DS in a polymer. The polymer balls up and is sticky in high MeOH concentration. Without the high MeOH concentration, the DS isn't soluble. The sonication breaks up the polymer and provides acceptable extraction recoveries.
3. Not having the samples properly equilibrated was the first thing I thought of. Unfortunately, this wasn't the issue as the samples were reprepared from standards after waiting for several hours to equilibrate. This still resulted in high and variable recoveries.
4. There wasn't any pattern in the recoveries.
5. Instrument was ruled out due to passing system suitability, and I was able to get good recoveries on their instrument. Also, our instrument methods were exactly the same.
6. I watched the validation chemist re-prepare samples that still failed (variable and high). There was no issue with pipetting and QS to volume. Although, I didn't see the weighing, the balance is calibrated everyday, and this chemist has successfully completed many other validations. Also, there is nothing especially tricky about weighing this DS.
7. The acceptance criteria is from our SOP. Actually, I wasn't aware that it was tighter than the FDA requirement.

Thanks very much for all the comments. Perhaps, there was a problem with weighing...what weighing problem could cause a high result?

[mohan_2008]

In order to more accurately understand the problem -

Can you give me info:

1. Name of the Polymeric matrix used to encapsulate the DS.
2. Solubility of the Polymeric matrix in Methanol (is it more soluble in Methanol or in Water).

[JGK]

Do you use a tablet for each individual sample or a "pool" of crushed table material.

what are the tolerances ±% for the DS in the tablets for this batch?


for example, if you use an individual tablet/sample is the 103.5 and 103.1% values within the tolerance range for DS in the tablet?