Caffeine Precison problems

[michaelcarolus]

Hi All

I have a tablet formulation containing 4 actives, one of them being Caffeine. During our validation we were unable to achieve acceptable method precision for the caffeine. The combined %RSD for the method precision (6 assays from each analysts) is greater than 2% for caffeine. The %RSD for the other 3 actives are all below 1%. We have tried various other methods e.g. USP and BP methods to improve the method precision but have had no success. The formulators are insisting that the analytical method is the problem? The lab analysts think it is the product because 3 of 4 actives are well within the method precision requirements. Could the particle size of the caffeine have an influence on the method precision? Has anyone had similar problems when analysing caffeine?

Thanks in advance

Mike

[Don_Hilton]

Particle size can always be an issue in any sampling procedure - and manufacture of individual doses is a matter or sampling the batch of mixed ingreedients.

Would it be possible to take enough material for multiple analyses and homogonize them very well - grind to a very find powder and then cut the sample by a method that evenly distributes all components of the sample? If you do this and the RSD of the method drops for caffeine - this points to the product. If the variability stays - it is the method.

And if you do this and the RSD drops on handling well homoenized material - the next step would be to take individual dosses and powder them separately and analyze them. They you are taking a procedure which has been demonstrated to have low RSD - and any addition to RSD has to come from the samples or some factor associated with handling individual doses (which you have addressed with the other components).

If you can take questions about the method off the table, they drop out of the discussion.

[Consumer Products Guy]

The formulators are insisting that the analytical method is the problem...

A claim I've heard many times !!! If one has to tell them their formulation is the issue, that's like calling their baby ugly.

My company recently rushed into a formulation with multiple actives. Upon 40C testing for one month, one component had fallen low out-of-specification. It turned out that the component reacted with the preservative, why one does accelerated storage stabilities.

Years ago, we had a product that had higher active in the smaller particles/beads. We seived as to size and then assayed individually to demonstrate/prove to the formulator.

Do what Don suggested, to prove/disprove that it's a sample preparation issue. Then you'll have ammunition.

[michaelcarolus]

Hi

Thanks for the repsonse thus far. Our normal sample prep takes 20 tabs, which we grind to a fine powder. We then take an equivalent of 1 tab to prepare the sample. We have also tried milling 20 tabs to obtain a "finer" powder without success. When we do a UOC on 10 tabs, the %RSd for the caffeine is +- 5%, compared with %RSD of less than 1% for all other actives. Our next step will be to prepare a "dry powder mix" with all the excipients and actives (e.g. equivalent to 20 tabs). We will then assay the "dry powder mix" and see if we can get acceptable RSD's. If the RSD is acceptable then the formulation/manaufacturing process is the problem.

Thanks again
Mike

[Don_Hilton]

I assume that when you say that you have had no success with milling 20 tabs to a finer poweder that the RSD remains high. A couple of questions come to mind:

How are you subsampling the milled powder to come up with your subsample equivalent to one tab? I would want to rule out segregation - which happens fairly easily in particulate solid samples. Are you using a riffle samle cutter or a similar technique to avoid this? (Sampling with a spatula from a container with all of the ground material in it can agrivate the problem - as will attempting to mix the solid material by shaking or tumbling the sample container. Smooth particles flow differently from jagged particles and dense particles flow differently from less dense particles. And if you have a component that tends to segregate to the top of the pile, the bottom of the pile, or to be the particles that roll off the lump in the spatula more easliy -- that component offers to be a problem.)

When you take the material equivalent to one tab and generate an extract - how do the RSD's look for multiple injections of the same extract?

[aceto_81]

Have your tried analysing 10 individual tablets (aka content uniformity)?
You already sample the equivalent of 1 tablet, so instead of milling/grinding, allow the tablet to desintegrate in your solvent (or first add some water) and do your sample prep.

If you still have the high RSD's, it is clearly the tablet formulation.

Other thing to try: add a known amount of caffeine to some placebo tablets (tablet formulation without caffeine), and analyse. See how the RSD is here.

gl

Ace

[michaelcarolus]

Thanks Don and aceto_81 for your comments.

Don - we are milling our sample using an IKA grinder. Once milled, the contents are transfered to a mortar and pestle. We then take an equivalent of 1 table for the sample prep. This method of sample prep also gave high %RSD's. The %RSD of stds (5 inj's) as well as %RSD of each sample (2 inj's) are all below 1%.

aceto_81 - we have done UOC and the %RSD for the caffeine is +-5% compared with +-1 % for all other actives. We have also spike the placebo with the actives and the %recovery for all actives was 98.0 - 102%.

We will do a few more experiments this week and I'll let you know if have been successful.

Thanks
Mike

[Don_Hilton]

It begins to sound like there is a problem between the grinder and the liquid sample. I assume the sample prep beyond the grinding is dissolution, filteration and injection?

While I doubt caffeine would be a problem going through a filter - alequotts of a single solution made from a number of tablets can be each passed through a different filter and then the RSD of the resulting chromatorams computed.

If I were to bet, I would bet on segregation of particles. If you do not have a riffle cutter, take powdered tablets and divide the lot of powder by taking many small scoops from the edge of the pile with a spatula. Alternaltely place the scoops, on new piles, one to the left and one to the right. This will divide the lot of powder into two sub lots. Keep the scoops as even as you can make then - and be sure that there are may scoops to a pile - on the order fo 100. Cut a sub-lot into halves again using the same technique. Continue cutting until you have fractions the size of a tablet. (If you start with 8 or 16 tablets in the grinder you will come closer to the original weight of a tablet in a final cut.) This is a pain, but by taking many small samples the same way from the pile, you will reduce the effect from segregation of the powder.

I have a friend who, back in his student days, was asked to go out to the siding and get a few hundered grams of coal to assay a coal car. It seemed easy enough - and then they dumped the coal car onto a concrete pad and handed him a shovel. The coal was properly sampled. And the young man build some muscle by doing it.

[Don_Hilton]

And by the way - if particle segregation is a problem in handling samples in the lab - it is also a problem in manufacturing. And they do not quarter down drug lots to make individual tablets...

[michaelcarolus]

Thanks Don and aceto_81 for your all your help. I think we've sorted out the problem.
Normal sample prep - equivalent of one tablet to 100 ml -resulted in high %RSD (more than 2% for caffeine)
New sample prep - 10 whole tabs to 1000 ml - %RSD of all 4 actives below 1%.

It seems that taking whole tablets gave us better precision. The overall %RSD of two analyst (6 assays each) are all within our validation requirementrs of NMT 2%. We will confirm this by sending the samples to a another lab to perform reproducibility testing.

Thanks again for all the help.
Mike