Mining through complex data and non-target analysis is made easier with GCxGC-TOFMS

I am currently working on a project looking for pesticides in herbal tea.  At this point, I am using the contents of one tea bag (~1.5g) and 10mL of freshly boiled water with a QuEChERS extraction.  I am still working through the details on the impact of the hot water and what type of cleanup will be necessary for pesticide determination, so that will be a blog for another day. I first decided to do some split injections of the raw extract to see what type of natural products are in some of the herbal teas I picked up at the grocery store.

Using GCxGC-TOFMS with a 30m x 0.25mm x 0.25µm Rxi-5Sil MS in the first dimension and a 1m x 0.25mm x 0.25µm Rtx-200 in the second, I processed the data by doing a “peak find” and library search for anything with a S/N of at least 100.  I then get my peak list and sort according to the NIST library similarity number, so that the highest matches are at the top of the list.  I start scanning through the list and see what names sound either like a pesticide or something else interesting.  For example, I came across the name niacinamide with a similarity of 912, and a spectra that seemed fairly unique.  I did a quick google search on the name and (according to Wikipedia) it is a main ingredient in an acne medication and has shown to have anti-anxiety properties among other potential medicinal uses.  This type of peak find is made possible by the spectral deconvolution of the time-of-flight mass spectrometer.

I also mine through the data by taking advantage of the ordered chromatograms produced by GCxGC.  This means that compounds with a similar structure, or a homologous series, will elute in a band across the chromatogram.  I found another compound, lupeol, which is a triterpenoid that has been noted to have several medicinal properties.  By looking at peaks that are eluting near lupeol, I found sitosterol which is a phytosterol that may reduce cholesterol.

While I don’t have standards to positively confirm peak identity, a high NIST match similarity with a spectra that contains abundant high m/z ions increases my confidence in the peak assignments.  I really enjoy digging through the data and seeing if anything interesting pops up.  Using GCxGC-TOFMS can give you A LOT of data to mine through, but with spectral deconvolution and ordered chromatograms, it makes it much easier.

The NIST library spectra, the peak true (spectral deconvoluted) and caliper spectra all show a high match similarity for Niacinamide

Sitosterol and Lupeol have a similar structure and elute in a band in the GCxGC-TOFMS chromatogram