Identification of an unstable degradation product

[Mattias]

This is not a pure chromatographic question, but anyway:

In one of our API's, there is a small impurity peak present at about 0.05-0.07%. The daily dose of the drug is high (>2g /day)

The LC methods for the API and the finished product are very different, and we have never seen this impurity in the finished product.

Question 1: if we can assure that we cannot detect the impurity in the product, do we still need to identify and qualify the impurity?

I have tried to identify the impurity via LC-MS, and I have got the exact mass, and MS/MS fragmentation of the impurity. This is however not enough for identification, and I guess the next step would be NMR analysis. The problem is that the impurity, after fraction collection, degrades totally within 30 minutes. It seems to be protected in the sample solution, where it doesn't decrease (or increase) over time. I would need many fraction collections to get enough material for an NMR analysis (acc. to the external NMR lab)

Question 2: Any ideas of how to proceed?

Thanks!

[DR]

1) Tough to do as the impurity is unstable, but you need to see to it that your final product method can show this impurity - this will allow you to demonstrate that said impurity is absent from the final product.

2) Sounds like it's time for a LC-NMR demo, or you have to review the forced degradation study data and see what drives production of this undesirable. It could be that something in the final product interdicts the reaction that forms the impurity. Knowing this will be a big plus when the FDA asks about the impurity profiles of the API & dosage form.

3) Doing the right thing... You also need to determine what happens to the unstable degradent. Is it further degrading into potentially toxic byproducts that either have no chromophore or a very low k with your method?

[Mattias]

Thanks for your answer!

I have not been able to show that the impurity is present in the product. It probably breaks down during the manufacturing of the tablet. It was not formed in the stress study (actually it disappeared when the substance was stressed)

I have analysed fraction collections on LC-MS to follow the degradation of the impurity. As said before it breaks down in about 30 minutes, and the API is one of the degradation products (!). There is one more peak forming, but I have not been able to get any MS signal of that one.

I guess it could be possible to do one more fraction collection of this degradation product of the impurity. But it will be tough as the peak is diluted in the fraction collection (and is just 0.05% in the original solution). The NMR lab says they need 5 µg to perform the measurement....

LC-NMR could maybe be the solution, but is the sensitivity higher on those instruments compared to the ordinary ones?

[DR]

I think it is higher - but you will need decent resolution.

[JM]

If i understand correctly your impurity degrades in to API molecule on isolation. This indicates either your impurity is a dimer of the API or a adduct with other content of your formulation. I have experienced this kind of conversion. I would rather look at the MS and MS/MS data carefully to see the mass difference of your impurity with API.

Have you tried analysing fraction immediatly after isolation by MS? or try to lyophilize?

JM

[Mattias]

JM> I have the mass of the impurity and it is not a dimer of the API (7 mass units difference). Since the impurity is present in the substance, it must be an adduct with some other impurity in the substance (?).

The impurity breaks down to the API and an unknown peak upon isolation. The key would be to get MS/MS data of this unknown peak, but I have tried electrospray in both positive and negative mode without getting ionisation.

I have good separation of the API and the impurity and I can see that the UV spectra are almost identical. However, in the impurity all three UV max are shifted c:a 3 nm (longer). After isolation and degradation of the impurity peak I see the API and another peak with an entirely different UV spectrum (one UV max at 245 nm).

I'll go to NMR from here and hope for more clues!

[ym3142]

it is really interesting.
FW(impurity)=FW(API) +/_ 7. when you interprete a MS you will believe that the impurity and the API are not related since there is no fragments with a size of 7 atomic units. So it has to be:

impurity + unknown A--->API + Unknown B

Do I understand correctly?

[ARG R&D]

From my experience with elusive unknowns and API vs DP profiles comparison, I found in many cases reactions between api + formaldehyde, trichloroacetic, acetaldehyde. They are present as inpurities in the reagents and usually give adducts, schiff bases, unstable cpds that upon work up they dissapear or transform in some other things, but usually with low MW ratio with the parent cpd. Try reacting your API with formol, etc if you have not yet during the forced deg experiments. In addition try expanding the oxidation with rose bengal + light or hydrogen peroxide + copper or AIBN

I hope it helps

[JMB]

Mattias,

1. Are you running +ve or -ve ESI ??

2. A 7 amu shift in MW will give you ODD (and odd N count) if API MW is even, and EVEN (with even N count) if API MW is odd; if the N count is changing, then you have some very funny business going on !!

3. Note that Li = 7 amu, but the difference between API and impurity would then be [M-H+Li] i.e. 6

4. Is the impurity +7 or -7 amu versus the API ??

JMB

[Mattias]

Thanks for all ideas!

I realise that I have not been clear about the masses of my API and the impurity.

The API has a mass of 200 [M+H]
The impurity has a mass of 407 [M+H]. If is was a dimer I guess the mass would be 399 [M+H], so the difference in since is rather 8 than 7... Sorry for the confusion!

Most of my runs I have run with the ESI+ (the mobile phase is very acidic). However I have tested ESI- after post-column infusion of ammonia, but I didn't even see my main peak. I have not so much experience with ESI-, though.

[alealbor]

You must to used an ESI source, because ESI is less "aggressive" to the analyte. Then, when you work in Full Scan MS/MS (MS/MS with all ions products), you must to identify somehow the ions products.
If you have LC-MS/MS from Thermo Electron (TSQ Quantum), you can used a MassFrontier software to identify this ions products, and then, the same software teel you witch is your impurity.

If you need more info write to me: aa@eidomet.com.ar


Alejandro from Argentina

[JMB]

Mattias,

From the information that you provided,

API MW 199 (N = ODD)
IMP MW 406 (N = EVEN)

so the N count is changing. You should consider your excipients, residual synthetic reagents etc.; assuming that

API + Unknown (X) ---> IMP + a small molecule (e.g. H2O)

If it is water that is eliminated in the condensation, then X = 406+18-199= 225 amu (N = ODD).

As mentioned in an earlier post, there are many adducts that can be formed.
Since you have accurate mass data, you can get a reasonably small number of possible molecular formulae that must have N = EVEN.

Good luck with this,

JMB