384 / 471
P a g e
|
3
Discussion of Data
An amended SMPR Evaluation Form has been submitted to the ERP, containing all the SLV data for
carnitine and choline. Tab 1, Method Evaluation Form, is filled out at the top with the SMPR
requirements (from the website) and the CARN-05 performance summary. In LOD/LOQ section, note
that we have chosen to spike Adult Milk Powder, 11750017V3, at the required LOQ limits in the SMPR,
rather than supply the lowest limit we can actually quantitate. This is because we had some difficulty in
pinning down the LOQ. Measuring the blanks, which were essentially pure water, with some purified
ammonium nitrate after the digestion, provided very erratic baseline signals, especially in comparing the
two instruments used. Therefore, multiplying this noise by 10 to calculate the LOQ seemed speculative.
The more informative approach was to spike a product matrix at the LOQ limit expressed in terms of
solution concentration:
For choline, the SMPR requirement is LOQ=2.0 mg/100g RTF. That’s 20 ppm or 20,000 ppb. With a
dilution factor of roughly 1000x for a 3g sample (see method outline above), the quantitation of a 20
µg/L addition of choline in solution is needed.
For carnitine, the SMPR says LOQ = 0.16 mg/100g RTF. That’s 1.6 ppm, or 1600 ppb. Applying DF for a 3g
sample, we must be able to quantitate 1.6 ppb in solution (we chose 1.0 ppb to be conservative).
Table 1, Tab 2, shows the spike recovery data on each matrix. This was performed in duplicate on each
of three days (duplicate spikes as well as duplicate baseline values which were averaged). The spiking
was carried out at two levels: an additional 40% and 80% of the native level. This caused us to have to
extend the calibration curve a bit to accommodate the spiked samples that would normally be outside
the calibration range. Choline and carnitine were spiked together in one set of spiking experiments, then
acetylcarnitine was spiked separately in another 3 days of experiments. For a couple of products, the
native level for carnitine and acetylcarnitine is shown in red (zero level) because we spiked into these
Adult Powder products at a more routine level to demonstrate normal recoveries – the level was not
40% or 80% of the native level. Note that we later reduced the spike level into 11750017V3 to
demonstrate the LOQ for carnitine – this matrix proved to be an adequate placebo for carnitine. This
powder was also low in choline. We had to weigh an abnormally large sample size of 7.5g to try to
adequately measure the choline and carnitine in the unspiked sample, which may explain the poor
recovery (shown in red) for 11750017V3 at the 40% level. When the sample size was brought down to a
more reasonable level of 3.0g, the choline recovery was very good (see Table 5b and LOQ data).
Tables 2a and 2b, Tabs 3&4, show the means and SD of nine independent days of SRM 1849a analyses.
The total choline mean was 94.2% of the certified mean and the total carnitine mean was 110.3% of the
certified mean. The NIST values tab, Tab 5, shows that Abbott values were almost identical to those
obtained by the NIST lab themselves, with their very similar LC-MS method (publication referenced
earlier). The NIST certificate labels the result “free” carnitine, but it is in fact determined after a
microwave digestion and should be compared to Abbott’s “total” carnitine result. The true free carnitine
is about 88% of the total in SRM 1849a (see Table 7d). Perhaps there was some confusion in the
Candidates for 2016 Method of the Year
322