Cannabis Quantitation Differences LCMS v GCMS

We are a toxicology lab that identifies and quantifies drugs of abuse in blood and urine samples. We use Agilent 6470 QQQ for our screening, and a mix of LCMS, and GCMS for confirmation. Cannabis (in the form of THCA) is one of the analytes we screen using the 6470 LCMS  and confirm using a 7000 series GCMS. In some cases we see that for THCA the screening results do not match the confirmation results. The differences can be quite significant, ranging from 50% to 2,000% differencs, with the confirmation result coming back higher each time. 

We expect the screening method to perform less well at quantifying as it's a single point calibration at the cut-off, while the confirmation is a 6 point calibration curve ranging from 50% below the cut-off to 30 times the cut-off. Even so we do not see this discrepancy arising between LCMS screening and other LCMS based confirmation methods using Sciex instruments for confirmation. Cannabis is the only analyte that is affected in this way. 

Can anyone help narrow down what might be causing this.

Some things to note:

1. Our initial thoughts were that Matrix effects in the screening method were causing this difference, but we use THCA-D9 as the internal standard in screening and it co-elutes so this should correct for any Matrix effect.

2. Our blood sample prep is SPE based and gives a clean extract. We start with 250uL of blood and end up with 50uL of 100% MeOH extract, injecting 2uL onto the column.

3. This affected samples are always blood samples, never urine. We confirm all our Urine and Blood using GCMS , but while we screen the Blood using the 6470 in ESI mode we don't see THCA in Urine unless we do a second LLE extraction on the Urine samples. We then screen those Urines on a Sciex instrument using APCI mode. So blood and urine are treated differently with regard to cannabis only. The use of APCI for the Urine may be the reason we don't see a difference between Urine screening and Urine confirmation.

4. We've investigated a little by performing some MS2 scans on the 6470 QQQ using the affected blood samples. We injected washes, extracted calibrator and extracted samples known to display a difference between screening and confirmation. The data is quite information rich in that there's lots of peaks to consider, even so there's nothing that really stands out as a peak that's masking the THCA signal at the tR for THCA.

5. We are assuming the screening method is the problem since the GCMS is generally rock solid in it's performance and is very well established, and tested with PTs etc.

6. When we repeat either the LCMS or the GCMS analysis of the affected samples we get the same results. 

My approach going forward is to 

1. Use the QQQ to determine if there's adducts of THCA being formed that are more prevalent than the M+1/M-1 species we already look at (we look at both)

2. We have a 6540 QTOF available to us, I was considering using this to perform a Find By Molecular Feature injection of the affected samples and see what masses it detects with the tR range of THCA

3. We have an alternative SPE cartridge we can use for the blood extraction, it might yield different results for screening.

4. We  have an APCI probe for the 6470 instruments, so I could re-inject the affected blood samples using this and compare with the GCMS

Has anyone got further suggestions as to how to troubleshoot this ? 

Thank you!

  • Do you see Delta 9 THC on your LCMS screen? If so Delta 9 THC and THCA would come out as a single peak on GCMS as THCA breaks down to Delta 9 THC unless you derivatize it

  • We do see a little THC on the screen, but generally the peak is tiny and doesn't give a good peak shape, so we don't pay much attention to it on screening. We keep it in the method with the hope of being able to improve it later by making some adjustments to the MRMs it uses. The GCMS picks up both THC and THCA very nicely and we do perform derivatization on both the THC and THCA using different derivatization agents. In both cases we get plenty of signal, great peaks and multiple MRMs all behaving nicely.

Was this helpful?