Balancing the false negative and positive rates in suspect screening with high-resolution Orbitrap mass spectrometry using multivariate statistics.

Modern high-resolution mass spectrometry (HRMS) enables full-spectrum trace level analysis of emerging environmental organic contaminants. This raises the opportunity for post-acquisition suspect screening when no reference standards are a priori available. When setting up a conventional screening identification train based on successively different identification criteria including mass error and isotope fit, the false negative rate typically accumulates upon advancing through the decision tree. The challenge is thus to elaborate a well-balanced screening, in which the different criteria are equally stringent, leading to a controllable number of false negatives. Presented is a novel suspect screening approach using liquid-chromatography Orbitrap HRMS. Based on a multivariate statistical model, the screening takes into account the accurate mass error of the mono isotopic ion and up to three isotopes, isotope ratios, and a peak/noise filter. As such, for the first time, controlling the overall false negative rate of the screening algorithm to a desired level (5% in this study) is achieved. Simultaneously, a well-balanced identification decision is guaranteed taking the different identification criteria as a whole in a holistic statistical approach. Taking into account 1, 2, and 3 isotopes decreases the false positive rate from 22, 2.8 to <0.3%, but the cost of increasing the median limits of identification from 200, 2000 to 2062 ng L(-1), respectively, should also be considered. As proof of concept, 7 biologically treated wastewaters were screened toward 77 suspect pharmaceuticals resulting in the indicative identification of 25 suspects. Subsequently obtained reference standards allowed confirmation for 19 out of these 25 pharmaceutical contaminants.