System Suitability

[krickos]

Hi

Well input is the following.

First note that both FDA, USP and Ph Eur consider that SSTs needs to cover the whole sequence (not necessary all SSTs), consequently only SST in begining is not good enough.

As a result some SSTs in beginning is ideal to check that stuff is good, but in end whole sequence needs evaluting to make final call.

So the risk taken is the following: If you check SST initially less risk. If SSTs turns out that the have failed you have a higher level of deviation as you actually have generated batch related data. If no batch generated data is generated due to you stop sequence as SSTs looks bad, you usually have a less complex QA related deviation to handle.

[Peter Skelton]

Another facet, if you are running good methods on well maintained systems with sample and standard prep being done correctly, you shouldn't be failing SST on any regular basis.

[joeburge]

Hi Adam

System suitability in itself says, the suitability of the system for the samples ur going to prepare. If that system is not suitable then there is no use of making samples or running those. Ultimately those samples are going to go in waste. And thats why our bracking std's also comes under system suit. so its better to chek sys suit first and then go for sample preparation.

If your lab is under GLP. then you have to inject your standard first, check the sys suit and then go for your samples.
but if you are working in non-GLP premises then you can check method's vital checks points


If incase in future, auditor ask for sys suit and that time if u dnt have any justification to your failed set then u'll be in a big trouble. U cannot say give the justification of office hours.

"Does system suitability have to be demonstrated before you run your samples."
Excellent reply above, and in answer to the original question, no SS does not "have" to pass, but if it does not, then why bother injecting samples? If your SS is not going to pass, injections are useless.

We use Chromeleon, and work under GLP. Chromeleon has the capability to test your SS within the sequence, and if it fails, to halt the sequence. This can include failure of bracketing Ccals also.

[Consumer Products Guy]

We calculate both initial and on-going SST for regulated assays. If SST does not pass, then there is no data, so at least we wouldn't need to do an out of specifications investigation.

[JGK]

Hi

Well input is the following.

First note that both FDA, USP and Ph Eur consider that SSTs needs to cover the whole sequence (not necessary all SSTs), consequently only SST in begining is not good enough.

As a result some SSTs in beginning is ideal to check that stuff is good, but in end whole sequence needs evaluting to make final call.

So the risk taken is the following: If you check SST initially less risk. If SSTs turns out that the have failed you have a higher level of deviation as you actually have generated batch related data. If no batch generated data is generated due to you stop sequence as SSTs looks bad, you usually have a less complex QA related deviation to handle.

Most labs will run an initial SST prior to sequence start then utilize the within run/post run QC samples and/or check samples as the monitors for the actual sequence.

If you want to take the logic to its nth degree, any SST is meaning once the next injection is made.

[LCFan]

Hi,

we have some really old-fashioned HPLC instruments with a classical integrator and plotting device. On these instruments, the evaluation of system suitability is sometimes complicated:
- replot of SST chromatogram with higher plot speed
- replot of SST chromatogram with higher attenuation
- manual measurements of peaks widths and heights and so on...
- manual calculation of R, N, ...

When it is late in the evening and the system is equilibrated, the SST solution is injected and preliminary assessed:

- symmetric peaks?
- peaks well resolved (R ok)
- ...

This assessment is based on a rough look on the chromatogram. When it seems ok: sequence is started (incl. multiple SST runs at the start and end of the sequence) and the lab assistant goes home.

Next day: the sequence is terminated. The lab assistant calculates the %RSD of the six SST inj's at the beginning of the sequence and the %RSD of the six SST inj's at the end of the sequence. Then, he replots the SST chromatograms for R/N/... calculation that has also been injected at the start and at the end of the sequence in order to measure the peak widths,... and finally to calculate R, N, ...

Then, he compares % RSD(start)/% RSD(end)/R(start)/R(end)/N(start)/N(end)/... with the specifications for % RSD, R, N, ...

Case 1: all values comply with the specification: sequence is valid; results can be evaluated
Case 2: any values don't comply with the specification: sequence is invalid (records will be kept)

For case 2: sequence must be re-run (with necessary adjustments, such as adaptation of the mobile phase or new column or...); in such a case, the lab assistant will try to do the first SST injections early enough to be sure to find the time to calculate the SST parameters before starting the sequence.

Thus, from a quality point of view:
- SST INJECTIONS must be performed at start and end of a sequence
- their EVALUATION can be done at any time

From an economic (and also human) point of view:
- SST injections at start of sequence shall be evaluated as soon as possible after they have been run in order to prevent unnecessary injections in the case of a non-fulfilled SST at the start of a sequence

Florian

[jm_b2004]

Hi All,

I have just joined this family owned pharma company in New Zealand. And I find it very intriguing that the QC guys do system suitability injections and sample injections very differently from what I have been accustomed to. I have been working with hplc testing for more than 10 years now from 3 big multinational well known companies, until I joined this family business.

I am used to injecting 5 or 6 injections of standards for SS, and sometimes I do 2 injections of a second standard as a check, and sometimes I do bracketing (depending on the length of the run). After the standard injections, I usually ran the samples as soon as possible.

Now, I am intrigued on why they accept the SS that was done the previous day? The reason I heard was that the mobile phase was not changed nor the setting or the column. I find the reason a bit lame. Another thing is that they do is a "retention time check" (inject 2x of the standard the next day to check if the system is suitable). WTF?! Correct me if I am wrong, don't we need to do a SS rather than inject a standard check?

They even prepared a new standard and use it as a "retention time" check rather than do a new SS since it's a new standard.

Can someone testify if this approach is acceptable to you guys.

Cheers!
Joe

[Alfred88]

Hi.
We perform the SS check for each sequence (first six injections, plus continuous bracketing injection to the end of sequence).
If we start several sequences in a day, each sequence will have its SS statistics.

Day D can not use SS statistics from day D-1, unless all data belong to the same sequence.
Alfred.

[gtma]

Some very good pts shared. Generally it the run continued or uninterrupted into the next day, I would not recommend repeating system suitability eg 6 injections of std. However, you may want to check Rs or LOQ if applicable. This approach is often used for chrom. robustness where all the non-std conditions are bracketed by the std condition run.

[gtma]

It's a very good practice to check system suit at the start of the run before injecting samples . However, if the run time per injection is very long, it's not very practical. But you can get a good sense if SST will pass or fail after the first few injections. We also use all stds for the calibration curve/SST. I have the luxury to check my runs from home...not a good work and life balance. Some companies require the samples to be processed even if the SST fails. You certainly want to avoid analyst failing SST (later in the run) because the sample results are OOT/OOS. For problematic methods, it is strongly recommended to state in the method that SST at the run must be checked before injecting samples. This brings me to another point, does it make sense to perform SST (multiple inj of std) for impurity method using area % approach?

[gtma]

By the way, would Chromeleon inject addition stds if the first or second injections is a problem?

[astorg]

New Question:
We run a set of replicate working standard injections at the start of a run (5 or 6), for system suitability. Then a check standard injection (different weighing). Then samples, interspersed with single working standard injections, plus a single working standard at the end of the run.

When calculating the result for a sample, would one use the average of the starting working standards used for system suitability? Or would one use the average of the two bracketing standards that were interspersed during the run?

My view is that one should use the average of the starting working standards. This way all results are against a fixed standard, and problem solving is simple.

Using the bracketing standards would mean that different sets of results, within the same run, would be calculated against a different standard values. The average of 2 standards is inherently more variable then the average of the 5 or 6 starting replicates. Plus if one sample failed, then the entire run is suspect because of the use of varying standards throughout the run.

Please, I invite comments on the merits and demerits of either approach.

[HPLC chemist]

I would use the average of the bracketing standards, in case the peak response drifts and therefore the results are more accurate and valid. By using the average of the SST standards your peak response (and results) can drift (become more inaccurate) and you will have an OOS too.

[astorg]

I was wondering the same thing, but had a dilemma: If my system is drifting, then is my system still suitable for use. That's the point of having the system suitability at the start, plus bracketing standards; to catch a system moving away from suitability. It would seem to make the justification convoluted.

[HPLC chemist]

[I was wondering the same thing, but had a dilemma: If my system is drifting, then is my system still suitable for use. That's the point of having the system suitability at the start, plus bracketing standards; to catch a system moving away from suitability. It would seem to make the justification convoluted.]

Astrong,

Convoluted yes! However, with my suggestion your reported results will be more accurate in all cases. In addition if things go south (and they will) you can at least save part of the HPLC run.