Steroids/Hormones: transfer problem between instruments

[LCFan]

Hello,

I prepare three standard solutions (80%, 100% and 120 % level) and three solutions of my sample. I fill each of the six solutions in two vials resulting in Vial series 1 and vial series 2

I make a sequence with vial series 1 on my HPLC instrument No. 1 using Column No. 1 and Mobile Phase Batch No. 1:
r2: 1.0000
y-intercept: 0.0
Assay of sample: 100 %

6 hours later, I take column No. 1 and mobile phase No. 1 and connect both on HPLC instrument No. 2. Then, I make a sequence with the vial series 2:
r2: 1.0000
y-intercept: 0.0
Assay of sample: 93 %

What is wrong? The different instruments or the different vial series?

=> Sequence with vial series 2 on HPLC instrument No. 1 and Mobile Phase Batch No. 1 and Column No. 1:
r2: 1.0000
y-intercept: 0.0
Assay of sample: 100 %


Thus, the difference in assay is based on the different instrumentation only.

Did anybody of you ever had such problems with steroids / hormones? How did you solve the problem?

Thank you very much for every hint.

Florian

[HPLCaddict]

First of all: Is this phenomenon reproducible or did you try it only once.

If it indeed is reproducible, some thoughts you might want to consider:
- are those 2 HPLCs of the same built, i.e. can you compare peak areas directly? If yes, which ones are different, the samples or the standards.
- we're talking about "nice" chromatography here, right? No overlapping peaks, no excessive tailing, enough retention of your analyte?
- any differences in the chromatograms between the two HPLCs (concerning retention times, peak shapes, etc.)?
- what's the difference between the references and your samples, i.e. what's the matrix of your samples?
- syringe speed of the two autoinjectors?
- do you know the "true" assay value, i.e. is 93% really too low, or could it be that 100% is to high?

[HPLCaddict]

Forgot one thing: Your calibration data looks just too good to be true
Three calibrations with r2 of 1.0000 and y-intercept 0.0 each? Yes, in theory it should look like this, but in practice it won't. Especially the y-intercept is never really zero. Never! Neverever! (the question is, if it's significantly different from zero).
Looks as if you forced your calibration line through zero. What are the results if you don't do this?

[LCFan]

Dear HPLCaddict,

thanks a lot for your ideas.

The effect is reproducible (up to now over 5 sequences).
The two instruments: 1: Merck Hitachi (UV) / 2: Waters UPLC (PDA)
There are two peaks: one at 4 minutes, symmetric (assay relevant) and one at 9 minutes (a small impurity peak)
On both systems, peaks are ok and with similar retention times.
Column: LiChrospher 100 RP-18 125-4, 5 µ
Matrix: Polyethylenglycol ointment
Syringe speed: Instrument 1: ?? µL / min (level 3 of 5); Instrument 2: 100 µL / min
The analysed product is manufactured with the target of 100%. In the mentioned sequences, I always found 100% with instrument 1.
The r2 is not always 1.0000, but around 0.9995 or higher
The y-intercept is not exactly 0, but never more than +/- 1% of the area of the 100%-level peak and not significantly different from zero (and not forced through zero)

That's the additional information I can provide.

Regards

Florian

[lynzjm]

does one of your systems calculations take into account the purity of the standard and the other doesnt perhaps?

Or does one system have a line flush and the other doesnt leading to carry over "false" readings from the previous injection(s) for either your standards or samples?

Is there any differences between your 100% standard and your sample i.e. matrix or impurities which could be causing the differences bewteen the two different modes of detection?

If possible you could tie it down to the mode of detection if you could hook up the PDA detector to the Merck instrument and see what you get!

Sounds like a bit of a headache hope you get to the bottom of it!

lynz x

[Hollow]

suggestions / questions:

- are both calculations done with the same CDS or has every system their own?
- did you try to do the evaluation by peak height? Will the differences be the same?
- Are the acquisition parameters the same on both systems? (Data rate, bandwith?)
- what about the S/N? comparable on both systems?
- how many data points over the peak to you have on both systems
- what is the injection volume? Have you chosen the correct inj-modus on the Acquity? (PLNO is recommended up to 75% of the maximum loop volume only, assuming you're not using a FTN system)
- have you already tried to reduce the sample-draw speed and see if it gets better?