How to set system suitability?

[zee]

i am in the process of writing a system suitability SOP. I have been through the pharmacopeia but it was very confusing. Has any one got any links where I can see what really are the requirements for testing the system suitability.

[krickos]

Hi

Well not an entirely easy topic, especially if you want to cover several pharmacopieas and regulatory zones (US, EU etc).

The LC GC journal (link to LC GC on to right corner of this forum) has a decent recent article I think that you can study.

Then it is pretty hard work to go through the requirements for your area.

Some hints:

USP: <621> Mainly recommendation but few mandatory apart from RSD. <467> Residual solvents, usually with headspace GC, RSD 15% for class 1 solvents, most tend to go for stricter requirements for class 2-3 solvents, likewise if liquid injector is used.
Individual technical chapters like CE may have different requirements, seem to recall RSD 3,0% for CE but unsure if it was USP.

FDA: How they inspect and what is written in a valid guide seems to differ, but check out the reviewer Guide for analytical validation from 1994. The new draft on validation from 2000 is still a draft but they do inspect based one some SST requirements mentioned there, like blank titrations (seen 483s cover that). Likewise FDA want to see bracketing of standards covering the whole sequence especially in assays (also seen in 483s).

Ph Eur: Slightly easier. Like USP any N or resoultion requirement is listed in the individual monographs.
Chapter 2.2.48 list a number of mandatory requirements, RSD is a bit special but related to spec limits and assays. Mandatory tailing factor 0,8-1,5 and signal to noise test based on listed disregard limit in monograph(only related substances).
NOTE: There is a writing that SST should cover the whole sequence but is up to the analyst to chose an appropiate verfication scheme depending on analysis etc...

Ph Eur also got a part around residual solvent analysis 2.4.24 but it is pretty much like USP.

Good luck you will likely face some internal discussions.

[tom jupille]

All the regulatory documents have "suggestions", but I don't think any have "requirements". Conceptually, it's straighforward: a set of criteria that demonstrate that the method works on this system on this day.

The details are open to the developer, but should be supported by robustness data from the validation. Obvious possibilities:

- retention time (or k', or relative retention) for peaks of interest. How tight this is depends on how many peaks, and on the possibility of specificity problems (e.g., a potency assay with only one peak can have very loose standards, whereas an impurities assay with many peaks will have to be commensurately tighter)

- resolution values for critical pairs of peaks (especially if there is robustness data to show that resolution can vary significantly with changes in conditions)

- repeatability for a reasonable number of replicates. The actual spec should be commensurate with the product or process being controlled (e.g., if your manufacturing process calls for a tablet to have 100 mg +/- 5 mg of API, then you must have significantly better precision in your measurement!)

- plate count (if you have a resolution spec, per above, this may be a throwaway to keep regulators happy; you should fail the resolution spec before you fail a plate count spec).

- baseline noise (if LOD and LOQ are of concern)

- signal/noise ratio at LOD or LOQ (again, if those are of concern)

- linearity and y-intercept of the calibration line (more important for wide-range assays than for narrow-range assays like potency or content uniformity).

There's no real mystery here: it should all be based on good science, and intended to answer the question "how can I know the method is working properly?".

[zee]

Thank you for your help guys