protein RPHPLC nonlinear dose response calibration curve..

[vickig]

yes, that is what I mean...use the lower range to demonstrate linearity of impurities and the upper range to demonstrate linearity of the main protein at the assay concentration.
The only way I can see to use this data to prove accuracy is (assuming this is a drug product method) if your linearity is performed in the sample matrix...AND the method quantitation is vs a standard curve...then you could conceivably back calculate the standards. This is common in MS assays, where the stds are prepared in blank matrix.
Same is true for %impurities...if both matrix and main protein are present in the linearity samples, the data can be used to demonstrate accuracy. Make sure, however, that you are calculating as is described in the method for impurities (%w/w vs main peak, conversion factor, or area%, etc.).

I don't know what you mean by log log conversion of data. Log scales are still used...

[grzesiek]

could you provide some experimental data from which linearity was calculated?
and maybe chromatograms of peak at 100% and in the lower range?

[vickig]

Try "Development and Validation of Analytical Methods" Edited by Reilly and Rosanske. There is an entire section devoted to specific methods and applications.

[danko]

If I was in alicee’s shoes, I would rather have troubleshot the method and eventually found the reason for the non-linearity. Because the relationship in question should be a linear one!
I can see that the topic started off a couple of months ago. So there has been time enough to solve even the most mysterious problem.

Best Regards

[alicee!]

If I was in alicee’s shoes, I would rather have troubleshot the method and eventually found the reason for the non-linearity. Because the relationship in question should be a linear one!
I can see that the topic started off a couple of months ago. So there has been time enough to solve even the most mysterious problem.

Best Regards

the reason for this nonlinearity to my understanding is the low protein recovery from the column (lower than expected area response) at the low amounts..
I havent been able to solve this so far.. and the same kind of behavior is observed for an RP method as well as an SEC or an IEC method.. in my case. Low recovery when lower amounts are injected.

Say for example, 15 mcg protein gives an area of 15000.
80-120% of this amount.. (i.e 18 mcg gives 18000 and 12 mcg gives 12000) give expected area and hence show linear response in this range.

But when low amounts like 2% amount (0.3 mcg should give area of 300 but consistently gives area around 240 mAus ie. 80% recovery), further 1% (1% i.e. 0.15 mcg, expected area 150, but area obtained is near 105 mAus i.e. 70% recovery) and (0.5% {0.075 mcg} expected area 75 but consistently obtained as 45 mAus i.e. 60% recovery..) as i go low on the amounts.. the recovery is consistently lower.

when the linear plot of 0.2-120% is evaluated.. because of the high values near 80-120% the line of the curve falls above the observed points at these lower amounts.. and hence the curve shows higher expected values for these low amounts.
all the points in the lower range 0.2-4% are not on the line.. are much lower on the line.
the intercept is highly negative.

I am still looking at answers to why this recovery is obtained..
is it usually observed or is with my methods that this is happening.
but i have observed this with reported/pharmacoepial methods also when reproduced in my lab.

in sucha case, i m not sure how to validate accuracy for the method, the purpose of the method being:
1. impurity quantitation (which fall near 0.2-2%)
and 2. total protein quantitation

i donot have purified impurities.

i'm hoping to solve these with further discussions.. thanks!

[grzesiek]

"0.5% {0.075 mcg} expected area 75 but consistently obtained as 45 mAus i.e. 60% recovery.." - 75 is very small indeed, I would love to see the chromatogram

[alicee!]

"0.5% {0.075 mcg} expected area 75 but consistently obtained as 45 mAus i.e. 60% recovery.." - 75 is very small indeed, I would love to see the chromatogram

75 was just an example, for my impurity methods i m compelled to look at even area values like 20.
what are the minimum reliable values that you consider.
the amounts and the ranges please?

thanks.

[grzesiek]

i don't go under a hundred because of noise

[vickig]

It is not unusual for proteins to stick to surfaces at low concentrations. Perhaps if you switch to polypropylene the recovery will improve.

[HW Mueller]

vickig, what are you trying to say with "It is not unusual for proteins to stick to surfaces at low concentrations."?

Unless I missed it, alicee! never told us what solutions/mobile phases she has used, but still I suspect that a good book on handling proteins would help here

[vickig]

HW,
What am referring to is the fact that is has been well documented that some proteins adhere to untreated glass surfaces. This loss of protein from a solution due to surface adherence will have a greater effect on the total amount of protein at lower protein concentrations.
I was thinking that Alicee!'s poor recovery at low concentrations could be a result of the protein sticking to untreated glass vials. The sticking would also be occurring at the higher concentrations, but would have a smaller effect.

[danko]

But the recovery in this case is higher than expected - in the region of 200 % at lower concentrations i.e. loads.

Best Regards

[vickig]

you're right and I am confused, as alicee! also refers to losses at lower levels, as in...

"my problem is.. which is pretty much similar with my sec and iec methods also.. that there is loss at the lower levels.. i.e. my higher range shows good recovery and linearity.. i.e. near 80-120% of 15 mcg..
but when i evaluate the recovery (in terms of area expected for say 1%, 0.5% of 15 mcg based on results obtained for 15 mcg).. the area values r much lower than expected. there is some protein loss at the column?"

[danko]

There could be some problems in connection with the peak integration. Also, there could be some procedural errors. How is the study designed? Finally, even simpler things like the calculation procedure could result in this kind of mysteries.

So, maybe it’s time for more thorough description of the case as well as posting some data, if the problem is to be solved.

Best Regards

[HW Mueller]

vickig, ok, can follow you on "sticky" now. But if you know this is well documented you probably also know that methods to overcome it are also well documented. For instance, one of the older methods was the addition of albumin (usually for handling other than HPLC). This showed that the attachment of proteins to surfaces is usually reversible. The closest to irreversible attachment that I ever experienced are proteins that stick to column frits. Here some drastic measures were necessary to clean those (LiDS + dithiothreitol, described before).