Placebo peaks in related substances test

[Oliver]

In the lab where I work we were having a discussion about the validity or not of subtracting the placebo peak area from an unknown peak area that we observe in a sample.
Our normal procedure is that if the sample has an unknown peak at a particular RT and the placebo chromatogram also shows a peak at that same RT±10%, we subtract the area due to the placebo from the area of the unknown peak in the sample chromatogram.
Often we run a placebo at the begining and end of a sequence of runs and work with the mean area of both chromatograms.
If the placebo peak is larger than the coinciding peak in the sample, then we eliminate the unknown peak in the sample.

Now we got a sugestion that if the method is validated then it is guarantie enough that the peaks of interest (known and unknown) will never coincide with the placebo peaks.
So the argument was that we should ignore any unknown peak whose RT coincides with a placebo peak, irrespective of whether the area is larger or smaller than that observed in the placebo chromatogram.

Personally I am doubting this argument becasue since we have a specification on unknown peaks, then as their name says they were not included in the validation! hence it is our objective to quantify them.

Any opinions

[adam]

Most of our related substances methods state "integrate all peaks that do not appear in the blank" (or the placebo, in your case).

So there is no subtraction. We simply interpret any peak that is not in the blank/placebo as not being a related substance.

[adam]

I mistyped the last sentence.

We interpret any peak that IS in the blank or placebo as not being a related substance.

[Oliver]

But what if there is some compound in the sample that elutes at the same time as a placebo/blank peak, wouldn't you be missing it out?

I have observed chromatograms where a peak in the sample coincides with a peak in the placebo but it is 3 times larger, sometimes more.
The placebo was an exact preparation (concentration of constituents) as the sample less the active pharmaceutical compound.

[adam]

When the method was validated for specificity, this is the type of thing that should have been caught.

Technically, if you have an interference that elutes at the same time as a peak of interest, the method is not appropriate for its application.

[pawan ratra]

Ideally the area counts of the peak in placebo and sample should be the similar. But practically there are may constraints.
First, getting a representative placebo is the first criterion. This means that all the excipients used in the manufacturing of the product should be same for placebo including the same lot nos., same manufacturer etc. Thus every batch of a product may have different placebo.

Secondly, the placebo preparation should undergo same steps of manufacturing with the same type of equipment used for the manufacturing of the product, which is practically not feasible. Therefore a placebo is made in lab scale rather than manufacturing scale with different types of equipment and may be with different preparation steps (e.g. speed and time of mixing). Due to this the number of peaks and the area counts of a peak in the placebo may be different than that in the sample.

Third, if the sample is kept for the stability testing, whether the placebo also gets the same treatment as sample. This is important as changes in the placebo may take place during stabilit testing.

Thus all of the above points should be considered in making the decision.

And even if you feel that one of the degradation product is merging with the placebo peak, then the method is not specific for that particular peak. Thus either the method should be modified to separate that peak or additional test method should be developed to quantify that unknown peak. If the unknown peak is greater than 0.1%, it should also be identified.

[gtma]

How large does your placebo peak gets?

[unmgvar]

do you spike the samples with placebo as well? it's will give a first indication on a way to proceed.

we had a syrop product with a placebo occuring very close sometimes together with one of the degrdation products. as part of the system suitability test we included a spiked injection that clearly showed that we were able to differetiate between both peaks.
in order to achieve the SST we simply adjusted mobile phase composition according to USP and EP

[Oliver]

well, Pawan you made me realise a couple of things;
1) quite possibly not all excipients used for the placebo are of the same lot as were used in the actual batch.
2) the process of placebo preparation is not identical to that of the sample. In the lab we weigh out the constituents, where as in production they granulate the mix.(wet it then pass it in a fluid bed dryer)
3) We are thinking of periodically asking production to do a very small placebo granulation, still the constituent batch numbers might not be identical to the actual product batches.

gtma: To get exact values I have to go back to work to find some data but I remember a range of cases where a peak in the placebo was just slightly larger than the sample 1.1 times to 2 or 3 times the peak size in the sample.

unmgvar: yes thanks, spiking would help in identifying the peaks with ab overlay of the spiked and unspiked chromatograms.

In essence, it seems that without having a placebo made from the same lot numbers used in procdution of the sample batch it might be difficult to make conclusions. Also ideally this has to be processed in the same way of the sample...
=================
still just one last question, if the placebo is made properly and the follwoing conditions are met:
a) no placebo peak is larger than a sample peak
b) all placebo peaks are present in the sample chromatogram

would it be justifyable to estimate an unknown substance peak in the sample by subtracting the area of the placebo peak with the same RT? (then i guess this would only be permissible for unknown peaks with a net result of less than 0.1%; otherwise the method has to be optimised to identify and seperate this impurity.

[pawan ratra]

Polymers generally give this type of problem due to varying amount of chain opening and hence difference in absorbance. You have to identify whether the peak in sample is due to placebo or not. The easiest way is to check the peak purity and/or compare the spectra at each point of the peak in the sample and placebo. If the peak is due to one of the excipient in the placebo, try preparing the excipient solution in different ways to increase the peak area and hence proving that th peak is due to placebo. No change in the method is required.
If the results indicate that the peak in the sample is the sum of placebo and impurity area, you have to change your method accordingly. Subtracting the placebo area counts can give you an idea about the amount of impurity but is not accurate. Thus it is not the solution.

Another way of dealing this problem is to optimze sample preparation to get miimum peaks from the placebo or if possible change in wavelength for minimum interference.