A semi-target analytical method for quantification of OH-PCBs in environmental samples

Abstract

Hydroxylated polychlorinated biphenyls (OH-PCBs) are oxidative metabolites of PCBs and residuals found in original Aroclors. OH-PCBs are known to play a role as genotoxicants, carcinogens, and hormone disruptors, and therefore it is important to quantify their presence in human tissues, organisms, and environmental matrices. Of 837 possible mono-OH-PCBs congeners, there are only ~ 70 methoxylated PCB (MeO-PCB) standards commercially available. Hence, a semi-target analytical method is needed for unknown OH-PCBs. The mass concentrations of these unknowns are sometimes determined by assuming the peak responses of other available compounds. This can bias the results due to the choices and availabilities of standards. To overcome this issue, we investigated the peak responses of all commercially available MeO-PCB standards with gas chromatography (GC) coupling with triple quadrupole (QqQ) mass spectrometry (MS) system, with positive electron impact (EI) ionization at 20-70 eV in selected ion monitoring (SIM) mode. We found correlations between the relative peak responses (RRFs) and the number of chlorine (#Cl) in the molecules of MeO-PCBs. Among the studied models, the quadratic regression of #Cl is the most suitable model in the RRF prediction (RRF = β₁ × #Cl^² + β₀) when the peak responses are captured at 30 eV. We evaluated the performance of the model by analyzing 12 synthesized MeO-PCB standards and a PCB-contaminated sediment collected from a wastewater lagoon. We further demonstrate the utility of the model using a different chromatography column and GC-EI-MS system. We found the method and associated model to be sufficiently simple, accurate, and versatile for use in quantifying OH-PCBs in complex environmental samples.

Keywords: Hydroxylated polychlorinated biphenyls; OH-PCBs; Quantification.

Figure 1.

The relative response factors (RRF) of the molecular ions of 70 certified MeO-PCBs in hexane solution at 30 eV (left plot). The RRF increases with the number of chlorines in the MeO-PCB molecules. The average RRF for all compounds in each homolog group (#Cl) exhibits the greatest range of response at the election impact (EI) ionization energy of 30 eV (right plot).

Figure 2.

The R-squared (RSQ), Adjusted RSQ (ADJ.RSQ), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC) of the predictive model of unknown MeO-PCBs’ RRFs from homolog group (#Cl) at 30 eV (left plot). The quadratic regression shows the dramatical improvement when compared with the others. The linear, quadratic-1, and quadratic-2 predictive models (right plot). Among them, quadratic-2 regression is the most suitable predictive model.

Figure 3.

The predictive quadratic model of unknown MeO-PCBs’ relative response factors (RRFs) from homolog group (#Cl) at 30 eV whose outliers have been removed with Cook’s distance (CD) at the threshold of 4 / (N – k – 1), where N is the sample size and k is the number of independent variables. The root mean square errors (RMSEs) are then computed with 10 times 10-fold cross-validation (10×10.CV) to obtain the 95% prediction intervals (PIs). *All data points of MeO-nonaCB (#Cl=9) were discarded as outliers, so the PI and RMSE could not be calculated.

Figure 4.

The plot of the actual concentrations at 25 ng/mL of the synthetic MeO-PCB standards (open circles), the predicted concentrations calculated from the RRFs of the model (solid circles), and the 95% prediction intervals (PIs; error bars).

Figure 5.

The concentrations of known and unknown OH-PCBs in a sediment sample from Altavista, Virginia. The unknowns are named by homolog group (#Cl) and retention time (RT) in Agilent DB-1701 capillary column. The compounds measured with authentic standards (Knowns, black) are 13.5% and the compounds measured using the method developed in this study (Unknowns, white) are 86.5% of the of the total quantified OH-PCBs. The error bars of the Unknowns indicate 95% prediction intervals (PIs).

Figure 6.

The procedure to generate of the predictive quadratic model of unknown MeO-PCBs’ relative response factors (RRFs) from homolog groups (#Cl). Root mean square errors (RMSEs); prediction intervals (PIs); and gas chromatography coupling with electron impact mass spectrometry (GC-EI-MS). *Coeluting congeners are not included in the model.