Isotopic Dilution Analysis

May 15th, 2020 in Pond's Pondering

"Pat, if my method says that I add 10 ng internal standard ("IS") into my samples, but the internal standards have degraded by 50%, how could my quantitation be correct? If I mistakenly call my IS "10 ng" but really there's only 5 ng there, how could my results still be accurate?"

And Pat began:

"Isotopic dilution involves spiking known amounts of isotopically-labelled analytes (aka "internal standards") into a known amount of sample prior to being processed through extraction and clean-up. In theory, any loss of the IS during handling and extraction should be mirrored by the same loss of the corresponding native (i.e. non-labelled) analytes, yielding a corrected value of ~100% recovery of native.

Figure 1: Native PFBS (left) and the labelled internal standard [13C]-PFBS (right).
These two have essentially identical chemistries.

So, if you have a sample that is known to contain 4 ng of PFOA, and your method demonstrates 75% recovery of the [13C]-PFOA you added to the sample, then you can assume 75% recovery of the native PFOA in the sample. Then, it will look like you have 3 ng PFOA but that's only 75% of what's in the sample! The 3 ng will be divided by 75% to yield a corrected value of 4 ng, which is what's in the sample. It doesn't matter if you recovered 5, 6, 7, 8, 9, or 10 ng of [13C]-PFOA because the % recovery of IS is used to correct the concentration of native PFOA."

"But," I answered, "If my standards vial contains 5 ng internal standard from the start and I call it 10 ng, then won't that quantify a customer sample incorrectly?"

Pat answered, "If your standards vial has analyte loss from 10 ng to 5 ng, then you will see that reflected in your (laboratory) control spike ("LCS"). It will be out of control (assuming it is composed from the same standards vial). If the LCS is out of control, then you cannot accept the sample data until the issue is fixed. You will see the IS out of control because the IS are quantified off the recovery standards ("RS") which is another set of labelled analytes added to every sample just before LC/MS." 

"That's what the LCS is for," he continued. "The LCS is the calibration standard used to check how the last calibration is performing that day. Remember, the LCS is not the matrix spike. The matrix could have something in it to throw off the amounts."

"Okay," I retorted, "What if only 1 or 2 IS are out of control, and the other 10 are okay?"

"Then your method has chosen the wrong RS, or the RS is degrading for some reason and you need to check that," Pat replied. "The RS doesn't need to be chemically similar to all the other compounds, but it does need to be extremely stable across time. In contrast, the IS set should consist of reasonably stable compounds of similar chemistry to the native analytes you are analyzing because they are used to correct for analyte losses in extraction, clean-up and dilutions. This is why we use 13C and 2H isotopes of analytes in an isotope dilution method: The chemistry is assumed to be identical to the native analytes.


"For example, if you are looking for DDT and DDE (see picture above), you'll want IS and native to both do the same things when the sample is being processed. The best way to do that is to have DDT that contains high numbers of 13C. Once you've done that, then you work to put in controls and make it part of your method."


"Back to the original question," I began, "What if I actually have 5 ng of IS, but I call it "10 ng" in the method and calibrate as if it was 10 ng?"

"Well," Pat replied, "in reality, you don't know that you have 5 ng, you're assuming. If you have 5 ng but call it 10 ng, then something will be out of control according to the previous calibration. Correct it and then your quantitation will be corrected. Still, there could be dozens of acceptable reasons why you are measuring half of what you expect to see and you can't assume that it's because the analyte has degraded, especially with POPs. There could still very well be 10 ng in there."

He continued, "Spikes, blanks and control standards show when a system is out of control. This is defined in your SOP."

"Even if you're a rebel, you have to trust your system."

"If people want to ask, we can only show 'this is what happened' according to the method. Then again, sh-t happens. See Rule 3.a.ii.A.1. So, we absolutely never speak in absolutes."

"That completely makes sense," I agreed.

~ Filtered by Dave Hope ~