Pharmaceuticals, Personal Care Products, Illicit Drugs and Their Metabolites in Screened Municipal Wastewaters

(1)

Pharmaceuticals, Personal Care Products, Illicit Drugs and Their Metabolites in Screened Municipal Wastewaters

by

Christopher James Lowe

B.Sc., University of British Columbia, 1996

A Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of

MASTER OF SCIENCE in the Department of Biology

 Christopher James Lowe, 2011 University of Victoria

(2)

Supervisory Committee

Pharmaceuticals, Personal Care Products, Illicit Drugs and Their Metabolites in Screened Municipal Wastewaters

by

Christopher James Lowe

B.Sc., University of British Columbia, 1996

Supervisory Committee

Dr. Asit Mazumder, Supervisor (Department of Biology)

Dr. Caren C. Helbing, Outside Member (Department of Biochemistry and Microbiology) Dr. Michael G. Ikonomou, Outside Member (Department of Chemistry)

Céline Davis, Additional Member (British Columbia Ministry of Environment)

(3)

Abstract

Supervisory Committee

Dr. Asit Mazumder, Supervisor (Department of Biology)

Dr. Caren C. Helbing, Outside Member (Department of Biochemistry and Microbiology) Dr. Michael G. Ikonomou, Outside Member (Department of Chemistry)

Céline Davis, Additional Member (British Columbia Ministry of Environment)

Two characterization studies were undertaken to assess the concentrations and environmental loadings of 125 pharmaceuticals, personal care products, illicit drugs and their metabolites (PPCPs) in screened municipal wastewaters being discharged into Juan de Fuca Strait from two marine outfalls in the Capital Regional District, British

Columbia, Canada. Two up-stream pump stations were also sampled. The PPCP concentration profiles were generally similar between the four sampling locations. Qualitative seasonal patterns in PPCP concentrations were also observed, primarily due to rainfall events that diluted wastewater contaminants during the winter. Increases in wastewater flow volumes following a rain event appeared to result in consistent shifts in PPCP concentration profiles for at least three of the four sites. Results indicated that the concentrations of PPCPs were similar to those observed in influents from other

jurisdictions. Predicted environmental concentrations were predominantly well below literature concentration thresholds known to induce acute or chronic effects in organisms in the environment. However, there was slight potential for adverse chronic effects as a result of the predicted environmental concentrations of ibuprofen around the outfalls

(4)

based on comparison to literature environmental effects thresholds. In general, sub-lethal and chronic effects endpoints were relatively limited in availability in the literature, as were literature thresholds derived from exposures to PPCP mixtures. Additional adverse chronic effects of these substances may be discovered in the future. Comparisons were made to regional prescription rates and population demographics to determine whether these factors could be good predictors of PPCP concentrations or loadings. Although wastewater concentrations and loadings were proportional to both prescription rates and population size, the regression relationships were statistically weak or insignificant. As such, prescription rates and population size could not be used to accurately predict pharmaceutical wastewater concentrations and loadings on their own. No qualitative relationships were observed between wastewater PPCP concentrations and either population age or gender breakdown. Overall, wastewater flow volumes, derived population equivalents and analytical method variability were also important factors to consider. Minor proportional deviations were observed following a preliminary loading comparison based on the relative population equivalent sizes of each of the four

wastewater system catchment areas. These deviations could have been a result of

disproportional hospital loading inputs and/or wastewater system inflow and infiltration. Comparisons were also made between the concentrations of PPCPs and the

concentrations of conventional wastewater parameters typically used to characterize bulk wastewater loadings (i.e., carbonaceous biochemical oxygen demand, biological oxygen demand, total suspended solids, volatile suspended solids). Only 18 of the 125 PPCPs were positively correlated with all four conventional parameters. This suggests that designing and optimizing treatment plants to efficiently reduce conventional parameter

(5)

loadings may not lead to as efficient or consistent reductions in the concentrations of all of the assessed PPCPs. However, the PPCP results were based on analyses of the filtered aqueous fraction of the wastewater samples, whereas the conventional parameter results were based on whole unfiltered effluent samples. As such, there was no direct link between the two sets of results.

(6)

List of Tables

Table 2-1 – Summary of the three sample preparation and analytical methodologies used in this thesis for the analysis of acidic pharmaceuticals. ... 24 Table 2-2 – Summary of method detection limits (ng/L) reported by each laboratory for the methods presented in Table 2-1. Note that (---) denotes this analyte was not measured by the laboratory. ... 24 Table 2-3 – Summary of spike recoveries (% mean or range and RSD) in tap or distilled water samples as reported by each laboratory for the methods presented in ... 25 Table 2-4 – Macaulay Point pharmaceutical concentrations (ng/L) as determined by the three different analytical methodologies employed by the author, as well as Environment Canada (GBAP) and AXYS Analytical Services. Note that for the University of Victoria samples, the mean (±standard deviation) of the March 2006 grab sample results is

presented (for comparison to the single concurrent GBAP grab sample), along with the range of November 2004 to September 2006 monthly composite sample results (for comparison to the range of daily AXYS composite sample results). Note that (---) denotes this analyte was not measured by the laboratory. ... 25 Table 3-1 – Method detection limits (ng/L), chemical abstract service numbers,

pharmacological use, dosage range, pharmacokinetics, availability and other information for the pharmaceuticals analyzed in Clover Point and Macaulay Point wastewaters. Pharmacokinetics information is from Rubin et al. (1972) for FEN and Canadian Pharmacists Association (2010) for all other drugs. Analytical standard information is also included. ... 70 Table 3-2 – Summary of Clover Point and Macaulay Point monthly wastewater flows (m3/day), pharmaceutical concentrations (ng/L), loadings (kg/month), and prescription rates (kg/month) from 2004 to 2006. Analytical standard recoveries (%) are also

provided. Individual sample results can be found in Appendix A (CLO) and Appendix B (MAC). ... 71 Table 3-3 – Summary of field and laboratory replicate results for pharmaceuticals (ng/L) in Clover Point and Macaulay Point wastewaters. Individual replicate results can be found in Appendix D (CLO) and Appendix E (MAC). ... 72 Table 3-4 – ANOVA results for pharmaceutical wastewater concentrations by year, month and outfall. Shaded results indicate statistically significant differences. ... 73 Table 3-5 - ANOVA results for log10 transformed wastewater flows by year, month and

outfall. Shaded results indicate statistically significant differences. ... 73 Table 3-6 - ANOVA results for pharmaceutical loadings by year, month and outfall. Shaded results indicate statistically significant differences. ... 74

(10)

Table 3-7 - ANOVA results for pharmaceutical prescription rates by year, month and outfall. Shaded results indicate statistically significant differences. ... 75 Table 3-8 – Non-parametric Kruskal-Wallis results for ASA prescription rates by year, month and outfall. Shaded results indicate statistically significant differences. ... 75 Table 3-9 - Non-parametric Mann-Whitney test results for ASA prescription rates by year. Shaded results indicate statistically significant differences. ... 75 Table 3-10 – Results of the paired sample t-tests assessing differences between log10

transformed pharmaceutical prescription rates and loadings. Shaded results indicate statistically significant differences. ... 76 Table 3-11 – Results of the regression analyses of pharmaceutical wastewater

concentrations and loadings versus prescription rates. Shaded results indicate statistically significant regressions. ... 76 Table 3-12 – Population size by age group and gender for the Clover Point and Macaulay Point outfall catchment areas in 2006. Population sizes were extrapolated from 2001 census data. ... 77 Table 3-13 – CLO & MAC pharmaceutical concentrations in relation to those observed in municipal wastewater influents and effluents from other locations. General descriptions of the levels of treatment are also provided (1° - Primary; 2° - Secondary; 3° - Tertiary). ... 78 Table 3-14 – Environmental concentrations observed in various receiving environments for the pharmaceuticals assessed in this study. CLO & MAC environmental

concentrations were predicted by applying oceanographic model determined minimum initial environmental dilution factors (CLO – 175:1; MAC – 245:1). ... 79 Table 3-15 – Environmental toxicological effects observed for the pharmaceuticals assessed in this study. CLO & MAC environmental concentrations were predicted by applying oceanographic model determined minimum initial environmental dilution factors (CLO – 175:1; MAC – 245:1). Shaded results indicate effects test results that were within the range of CLO & MAC wastewater and predicted environmental

concentrations. Freshwater (FW) or marine (MAR) test species are noted accordingly. 81 Table 3-16 – A comparison of CLO & MAC total and per capita pharmaceutical loadings (based on the mean loadings over the 2004 to 2006 sampling period) to those observed in other locations. For a number of the other sources, some values were not available, while others were calculated using information provided. ... 84 Table 3-17 - Summary of the statistically significant differences and regressions for Clover Point and Macaulay Point pharmaceutical concentrations, loadings and

(11)

Table 4-1 – Summary of PPCP and conventional parameter concentrations (ng/L unless otherwise stated) in municipal wastewaters collected from the CLO, PEN, MAC and CRG pump stations from November 2nd to 6th, 2009. Results are presented as the means ± standard deviations of up to five daily samples (sample size = N). The number of non-detect (ND) samples is also indicated. Results for the individual daily samples can be found in Appendix O through Appendix R... 124 Table 4-2 – Stress and fit measures for the NMDS ordination based on conventional parameter and PPCP concentrations in CLO, PEN, MAC and CRG municipal wastewater samples collected in November 2009. ... 129 Table 4-3 - Summary of the statistically significant Spearman rank correlations between the two NMDS dimensions and conventional wastewater parameter and PPCP

concentrations in CLO, PEN, MAC and CRG municipal wastewater samples from November 2009. All Spearman rank correlation results, including those that were

insignificant can be found in Appendix T. ... 129 Table 4-4 - Summary of the statistically significant Spearman rank correlations between conventional wastewater parameter and PPCP concentrations in CLO, PEN, MAC and CRG municipal wastewater samples from November 2009. Correlations were positive unless noted by “neg.”, in which case they were negative. All Spearman rank correlation results, including those that were insignificant can be found in Appendix S. Octanol-water partition coefficient (log Kow) and Octanol-water solubility estimates are from the United States Environmental Protection Agency (2009). ... 130 Table 4-5 – Summary of 2009 population equivalents at CLO, PEN, MAC and CRG pump stations. These numbers were estimated using 2005 population equivalents and an annual population growth rate of 1%. The proportions of upstream pump station (i.e., PEN and CRG) catchment area population sizes to their respective total catchment area (i.e., CLO and MAC) population sizes are included... 133 Table 4-6 – Summary of gemfibrozil (GEM), ibuprofen (IBU) and naproxen (NAP) concentrations (ng/L) and loadings (g/day) at CLO, PEN, MAC and CRG pump stations. Daily flow volumes (m3/day), on which the loadings were based, are also included. The percentage contributions of the upstream pump stations (i.e., PEN and CRG) to their respective outfall (i.e., CLO and MAC) total loadings are presented for each analyte. . 134 Table 5-1 - Municipal wastewater treatment removal efficiencies observed at the CRD’s Saanich Peninsula Treatment Plant (SPTP) and by other jurisdictions for the

pharmaceuticals assessed in this study. SPTP influent and effluent results, upon which the treatment removal efficiencies were calculated, can be found in Appendix U. Levels of treatment are identified as 1° - Primary; 2° - Secondary; 3° - Tertiary. ... 150

(12)

List of Figures

Figure 3-1 – Location of the Clover Point and Macaulay Point municipal wastewater outfalls in the Capital Regional District, British Columbia. Wastewater collection

catchment areas for each outfall are also shown circa 2005. ... 87 Figure 3-2 – Monthly wastewater flow volumes (m3/day) at Clover Point and Macaulay Point from 2004 to 2006. ... 88 Figure 3-3 – Monthly salicylic acid (acetylsalicylic acid metabolite) concentrations (ng/L), environmental loadings (kg/month) and prescription rates (kg/month) in Clover Point and Macaulay Point wastewaters from 2004 to 2006. ... 89 Figure 3-4 - Monthly diclofenac concentrations (ng/L), environmental loadings

(kg/month) and prescription rates (kg/month) in Clover Point and Macaulay Point

wastewaters from 2004 to 2006. ... 90 Figure 3-5 - Monthly fenoprofen concentrations (ng/L), environmental loadings

wastewaters from 2004 to 2006. ... 91 Figure 3-6 - Monthly gemfibrozil concentrations (ng/L), environmental loadings

wastewaters from 2004 to 2006. ... 92 Figure 3-7 - Monthly ibuprofen concentrations (ng/L), environmental loadings

wastewaters from 2004 to 2006. ... 93 Figure 3-8 - Monthly ketoprofen concentrations (ng/L), environmental loadings

wastewaters from 2004 to 2006. ... 94 Figure 3-9 - Monthly naproxen concentrations (ng/L), environmental loadings (kg/month) and prescription rates (kg/month) in Clover Point and Macaulay Point wastewaters from 2004 to 2006. ... 95 Figure 3-10 - Population size by age group and gender for the Clover Point and Macaulay Point outfall catchment areas in 2006. Population sizes were extrapolated from 2001 census data. ... 96 Figure 3-11 – Regressions of salicylic acid (acetylsalicylic acid metabolite) wastewater concentrations (ng/L) and loadings (mg/month) versus prescription rates (mg/month). Data from both CLO and MAC were pooled prior to running the regression. ... 97

(13)

Figure 3-12 - Regressions of diclofenac wastewater concentrations (ng/L) and loadings (mg/month) versus prescription rates (mg/month). Data from both CLO and MAC were pooled prior to running the regression. ... 98 Figure 3-13 - Regressions of gemfibrozil wastewater concentrations concentrations (ng/L) and loadings (mg/month) versus prescription rates (mg/month). Data from both CLO and MAC were pooled prior to running the regression. ... 99 Figure 3-14 - Regressions of ibuprofen wastewater concentrations concentrations (ng/L) and loadings (mg/month) versus prescription rates (mg/month). Data from both CLO and MAC were pooled prior to running the regression. ... 100 Figure 3-15 - Regressions of ketoprofen wastewater concentrations (ng/L) and loadings (mg/month) versus prescription rates (mg/month). Data from both CLO and MAC were pooled prior to running the regression. ... 101 Figure 3-16 - Regressions of naproxen wastewater concentrations (ng/L) and loadings (mg/month) versus prescription rates (mg/month). Data from both CLO and MAC were pooled prior to running the regression. ... 102 Figure 4-1 - Location of the Clover Point and Macaulay Point municipal wastewater outfalls, and the Penhryn and Craigflower Pump Stations in the Capital Regional District, British Columbia. Wastewater collection catchment areas for each outfall and pump station are also shown circa 2005. ... 136 Figure 4-2 – Results of the NMDS ordination based on conventional parameter and PPCP concentrations in CLO, PEN, MAC and CRG municipal wastewater samples collected in November 2009, with colour coding by (a) station and (b) sampling date. The numbers following station names denote the sampling day (e.g., CLO1 corresponds to the sample from November 2nd, CLO2 corresponds to November 3rd, etc.). ... 137 Figure 4-3 – Summary of mean (± standard deviation) gemfibrozil (GEM), ibuprofen (IBU) and naproxen (NAP) loadings (g/day) from samples collected at CLO, PEN, MAC and CRG in November 2009... 138

(14)

List of Appendices

Appendix A – Clover Point monthly wastewater flows (m3/day), pharmaceutical

concentrations (ng/L), loadings (kg/month), and prescription rates (kg/month) from 2004 to 2006. Analytical standard recoveries (%) are also provided. Replicate results are presented as mean±SD; individual replicate results can be found in Appendix D. ... 170 Appendix B – Macaulay Point monthly wastewater flows (m3/day), pharmaceutical concentrations (ng/L), loadings (kg/month), and prescription rates (kg/month) from 2004 to 2006. Analytical standard recoveries (%) are also provided. Replicate results are presented as mean±SD; individual replicate results can be found in Appendix E. ... 173 Appendix C – Results of deionized water blank sample analyses. All wastewater samples were corrected by the mean of the blank samples for each analyte. ... 176 Appendix D – Clover Point laboratory (LR) and field replicate (FR) results.

Concentrations are in ng/L and analytical standard recoveries are in %. ... 177 Appendix E – Macaulay Point laboratory (LR) and field replicate (FR) results.

Concentrations are in ng/L and analytical standard recoveries are in %. ... 179 Appendix F – Results of the monthly freezing assessment of the March 2006 Macaulay Point grab samples. Pharmaceutical concentrations are in µg/L and the error bars represent mean ± standard deviations of n=2 field replicates in March and n=3 lab replicates in August. Concurrent Environment Canada Georgia Basin Action Plan (GBAP) results are also included. ... 181 Appendix G – Clover Point and Macaulay Point annual mean (and standard deviation) wastewater flows (m3/day), pharmaceutical concentrations (ng/L), loadings (kg/month), and prescription rates (kg/month) for 2004, 2005 and 2006. Annual mean analytical standard recoveries (%) are also provided. Individual monthly results can be found in Appendix A (CLO) and Appendix B (MAC). ... 185 Appendix H – Results of the Kolmogorov-Smirnov tests of normality for wastewater flows, pharmaceutical concentrations, loadings and prescription rates. Shaded results indicate tests that did not meet the assumption of a normal distribution; in these instances data were log10 transformed to meet the assumption. ... 186

Appendix I – Results of the Levene’s tests of homogeneity of variances for wastewater flows, pharmaceutical concentrations, loadings and prescription rates. Shaded results indicate tests that had heterogeneous variances; in these instances data were log10

transformed to meet the assumption. ... 187 Appendix J – Results of the Tukey post-hoc tests of SA (ASA metabolite) concentrations by month. Shaded results indicate statistically significant differences... 188

(15)

Appendix K - Results of the Tukey post-hoc tests of wastewater flows by

month. Shaded results indicate statistically significant differences... 192 Appendix L - Results of the Tukey post-hoc tests of SA (ASA metabolite), IBU and NAP loadings by month. Shaded results indicate statistically significant differences. ... 195 Appendix M - Results of the Tukey post-hoc tests of DCF and KET (log10 transformed)

prescription rates by year. Shaded results indicate statistically significant differences. 204 Appendix N – Summary of LC/ESI-MS/MS ionization modes and method detection limits for the pharmaceuticals, personal care products, illicit drugs and their metabolites that were analyzed at AXYS Analytical Services Ltd. in November 2009. Log Kow (octanol-water partition coefficient) and water solubility estimates are from United States Environmental Protection Agency (2009). ... 205 Appendix O – Concentrations (ng/L unless otherwise stated) of pharmaceutical and personal care products in wastewaters collected at the Clover Point municipal wastewater outfall in November 2009. Note that (---) denotes no sample collected for that

analyte/day. ... 208 Appendix P - Concentrations (ng/L unless otherwise stated) of pharmaceutical and personal care products in wastewaters collected at the Macaulay Point municipal

wastewater outfall November 2009. Means (± standard deviations) of the field triplicate results are included. Note that (---) denotes no sample collected for that analyte/day. . 211 Appendix Q - Concentrations (ng/L unless otherwise stated) of pharmaceutical and personal care products in wastewaters collected at the Penhryn Pump Station in

November 2009. Note that (---) denotes no sample collected for that analyte/day. ... 214 Appendix R - Concentrations (ng/L unless otherwise stated) of pharmaceutical and personal care products in wastewaters collected at the Craigflower Pump Station in November 2009. Means (± standard deviations) of the field triplicate results are

included. Note that (---) denotes no sample collected for that analyte/day. ... 217 Appendix S – Spearman rank correlations between wastewater conventional parameters and frequently detected PPCPs. (**) denotes significance at p<0.01. (*) denotes

significance at p<0.05. ... 220 Appendix T - Spearman rank correlations between NMDS DIM 1 and 2, and wastewater conventional parameters and frequently detected PPCPs. (**) denotes significance at p<0.01. (*) denotes significance at p<0.05. ... 222 Appendix U - Pharmaceutical concentrations (ng/L) in Saanich Peninsula Treatment Plant screened influent and secondary effluent. Analytical standard recoveries (%) are also provided. The lab replicate results are presented as mean±SD. ... 224

(16)

Appendix V – Summary of the statistically significant differences and

regressions for CLO and MAC pharmaceutical concentrations, loadings and prescription rates using only results for wastewater samples that had IS and SS recoveries within 50-150%. P-values are also provided. Shading indicates where differences in statistical significance occurred when compared to the statistical results for the full dataset; results for the full dataset, including samples with standard recoveries outside the 50-150% range, can be found in Chapter 3. ... 225 Appendix W - Monthly pharmaceutical concentrations (ng/L), environmental loadings (kg/month) and prescription rates (kg/month) in Clover Point and Macaulay Point

wastewaters from 2004 to 2006 using only results for wastewater samples that had IS and SS recoveries within 50-150%. Results for the full dataset, including samples with standard recoveries outside the 50-150% range, can be found in Chapter 3. ... 226 Appendix X – Box plots of DCF and GEM concentrations at CLO and MAC for good recovery data (i.e., only samples with standard recoveries of 50-150%) and all data (i.e., all samples regardless of standard recovery). ... 233 Appendix Y - Box plots of SA (ASA metabolite), DCF, IBU, and NAP loadings by month for good recovery data (i.e., only samples with standard recoveries of 50-150%) and all data (i.e., all samples regardless of standard recovery). ... 234 Appendix Z – Regressions of DCF, GEM, IBU and NAP wastewater concentrations (ng/L) versus prescription rates (mg/month) using only results for wastewater samples that had IS and SS recoveries within 50-150%. Data from both CLO and MAC were pooled prior to running the regression. Results for the full dataset, including samples with standard recoveries outside the 50-150% range, can be found in Chapter 3. ... 236 Appendix AA – Daily flow data (m3/day) from January and July 2005 at Clover Point and Macaulay Point. The composite sampling dates were January 17-18 and July 19-20 and are indicated by a dashed vertical line on the figures. ... 237

(17)

List of Abbreviations

2,3-D – 2,3-dichlorophenoxyacetic acid ANOVA – Analysis of variance

ASA – Acetylsalicylic acid BC – British Columbia

CAS – Chemical Abstracts Service

CBOD – Carbonaceous biochemical oxygen demand CLO – Clover Point

COD – Chemical oxygen demand CON - Concentration

CRG – Craigflower Pump Station CRD – Capital Regional District DCF – Diclofenac

DIM - Dimension DL – Detection limit

DQO – Data quality objective EC50 – Effect concentration 50% FEN – Fenoprofen

FR – Field replicate FS – Full scan FW - Freshwater

GBAP – Georgia Basin Action Plan GC – Gas chromatography

GC/IT-MS/MS – Gas chromatography/ion trap tandem mass spectrometry GEM – Gemfibrozil

GIS – Geographic information system

HPLC – High performance liquid chromatography IBU – Ibuprofen

IC50 – Inhibition effect concentration 50% IS – Internal standard

(18)

IT – Ion trap KET – Ketoprofen

Kow – Octanol-water partition coefficient

LC/ESI-MS/MS – Liquid chromatography/electrospray ionisation tandem mass spectrometry

LC50 – Lethal concentration 50% LD – Loading

LOEC – Lowest observable effect concentration log – logarithmic (base 10)

LR – Lab replicate

LRMS – Low resolution mass spectroscopy LWMP – Liquid Waste Management Plan MAC – Macaulay Point

MAR - Marine

MDL – Method detection limit MoE – Ministry of Environment MS/MS – tandem mass spectroscopy MSR – Municipal Sewage Regulation MXR – Multixenobiotic resistance N/A – Not available

NAP – Naproxen neg. – negative

NMDS – Non-metric multidimensional scaling NOEC – No observable effect concentration NS – No sample

OTC – Over the counter PEN – Penhryn Pump Station

PPCP – Pharmaceuticals & personal care products PR – Prescription rate

QA/QC – Quality assurance/quality control RPD – Relative percent difference

(19)

RSD – Relative standard deviation SA – Salicylic acid (ASA metabolite)

SCADA – Supervisory control and data acquisition SD – Standard deviation

SIM – Selective ion monitoring SPE – Solid phase extraction

SPTP – Saanich Peninsula Treatment Plant SS – Surrogate standard

STP – Sewage treatment plant TSS – Total suspended solids UK – United Kingdom

USA – United States of America VSS – Volatile suspended solids WQG – Water quality guideline

(20)

Acknowledgments

First off, I would like to thank my wife and daughter, Kamala and Nadia, for being patient with me. They have put up with a lot of grumpiness and frustration related to this work!

Funding for this study was provided from numerous sources. I was supported by a Natural Sciences and Engineering Research Council (NSERC) of Canada Industrial Postgraduate Scholarship, a Michael Smith Foundation for Health Research/University of Victoria Health Research Fellowship, Dr. Asit Mazumder (via an NSERC Senior

Industrial Research Chair) and the University of Victoria Department of Biology. The Capital Regional District (CRD) provided both in-kind support for sample collection and direct financial support of my NSERC Industrial Postgraduate Scholarship. Analytical method development was supported by Dr. Asit Mazumder via an NSERC Senior Industrial Research Chair. Dr. Sergei Verenitch was instrumental in developing the analytical methodologies for use in Chapter 3; without him, this work would not have been possible. Ingrid Sorensen, an undergraduate co-op student, was also of great assistance with sample preparation and analyses. Pharmacological assistance was provided by Dr. William Dyson (Director) and various nursing staff at the University of Victoria Health Services clinic. A number of pharmacists at various pharmacies in the region were also very helpful and willing to answer my numerous questions; those that were most helpful have been identified in the thesis. Various CRD Marine Programs staff and technicians coordinated sampling and collected wastewater samples on my behalf including Shirley Lyons, Avrael Perreault, Dalia Hull-Thor, Rob Shoemaker, Heather Sinnot, Joe Pennimpede, James McAloon, and Simon Grant. Shane Ruljancich at the CRD provided all of the GIS maps and postal code information for the outfall catchment areas in the region.

(21)

Chapter 1 General Introduction

Pharmaceuticals and personal care products (PPCPs) are a large class of compounds that have the potential for unintended effects on organisms in the

environment. This thesis investigates the presence of PPCPs in municipal wastewaters from the Capital Regional District (CRD) that are destined to be discharged into the marine environment of Juan de Fuca Strait (British Columbia, Canada).

The remainder of this chapter will provide a brief overview of the environmental concerns about PPCPs, the CRD and the impetus for this thesis, and a summary of the overall thesis objectives.

1.1 PPCPS AS SUBSTANCES OF ENVIRONMENTAL CONCERN

There are approximately 11,910 different prescribed and over-the-counter (OTC) drug products (representing 1,776 unique active ingredients) approved for use by humans in Canada today (Health Canada 2011), and new ones are under development on a regular basis. There are also 1,356 different veterinary drugs (representing 695 unique active ingredients) approved for use in Canada (Health Canada 2011), many of which also have human use. These substances are used specifically to have biological effects in either humans or animals and, once administered to a target organism, are metabolized (to some extent) and excreted. There are also numerous illicit drugs used by humans, also because they have a biological effect. Personal care products (with many of the active ingredients included in the drug product numbers above) are a broad class of chemicals that include ingredients in products such as perfumes, soaps and toothpastes. Similar to pharmaceuticals, many personal care products can have a direct biological effect (e.g., antimicrobial preservatives). Other personal care products can have direct

(22)

physical modes of action (e.g. surfactants in soaps and shampoos), while others have only an aesthetic intent (e.g. fragrance compounds in perfumes).

Pharmaceuticals and personal care products can potentially enter the aquatic and marine environments via discharge from municipal wastewater systems, runoff from agricultural fields and feed lots, direct deposit from aquaculture, or washing off via bathing. Because PPCPs are designed to have biological or physical effects, there have been significant concerns that they could impact non-target organisms in the

environment.

1.1.1 History of PPCP Detection in Environment

The earliest detections of pharmaceuticals in wastewaters were in the 1970’s by Garrison et al. (1976) and by Hignite and Azarnoff (1977). These authors both detected clofibric acid (a blood lipid regulator) and salicylic acid (a metabolite of acetylsalicylic acid, an analgesic) in wastewaters from Ohio and Kansas, respectively. The first detection of PPCPs in Canadian wastewaters was by Rogers et al. (1986). Synthetic estrogens used in birth control were also some of the earliest PPCPs assessed in

wastewater treatment systems (Tabak and Brunch 1970; Tabak et al. 1981). These early studies recognized the potential for the continuous discharge of substantial quantities of PPCPs into the environment via wastewaters and that, in the receiving waters, non-target organisms could be exposed to these drugs. Since these early studies, and particularly since the early-1990s, the number of investigations into PPCPs in wastewaters has increased substantially (Daughton and Ternes 1999; Kummerer 2009; Wong and MacLeod 2009).

(23)

Some of the earliest PPCPs to be detected in the environment were the

synthetic musk fragrances used in cosmetics and perfumes. In the first comprehensive monitoring program targeting PPCPs, Yamagishi et al. (1981; 1983) discovered synthetic musks in municipal wastewaters, freshwater receiving environments upstream and downstream of wastewater treatment plants, and in freshwater fish and marine shellfish. Additional studies have since identified the presence of other PPCPs in surface waters (Aherne et al. 1990; Samuelsen et al. 1991; Hirsch et al. 1999; Kolpin et al. 2002b; Metcalfe et al. 2003b), ground waters (Eckel et al. 1993; Holm et al. 1995; Godfrey et al. 2007) and the terrestrial environment (Addison 1984; Nessel et al. 1989; Yang and

Metcalfe 2006; Edwards et al. 2009). Recently, PPCPs were detected in surface waters in a national park designated as a United Nations Educational, Scientific and Cultural

Organization (UNESCO) Biosphere Research and Human Heritage site (Camacho-Munoz et al. 2010a; Camacho-(Camacho-Munoz et al. 2010b). More recent studies have identified additional PPCPs in plant and animal tissues, including antidepressants in fish (Brooks et al. 2005) fragrances and antibiotics in earthworms (Kinney et al. 2008), and

anticonvulsants and antimicrobials in soybeans (Wu et al. 2010a).

1.1.2 Implications of PPCPs in the Environment

Although the vast majority of PPCPs are not persistent in the environment, they are “pseudo-persistent” in that they are continuously discharged via wastewater outfalls and other sources. Non-target organisms in the environment, therefore, are potentially continuously exposed to these compounds albeit at generally low levels. Acute effects of PPCPs have been widely demonstrated, but typically only at exposure concentrations far above those found in the environment (Ferrari et al. 2004; Quinn et al. 2008; Nassef et al.

(24)

2009). However, there is potential for chronic sub-lethal effects in the

environment, and it is these effects that are of ongoing concern (Kostich and Lazorchak 2008).

Antibiotic resistance is one of the most significant concerns resulting from the discharge of antimicrobial PPCPs to the environment. Antibiotic resistance has been identified in hospital settings since the 1950s (Finland et al. 1959), and more recently in wastewater treatment systems (Reinthaler et al. 2003; Zwenger and Gillock 2009) and the environment (Attarassi et al. 1993; Kerry et al. 1996; Goni-Urriza et al. 2000; Sengeløv et al. 2003; Chee-Sanford et al. 2009). Antibiotic resistance occurs as a result of the exposure of bacteria to concentrations of antibiotics that are not high or prolonged enough to kill the entire bacterial population. Ultimately, the resistent bacteria survive and replicate resulting in a bacterial community that is much more resistant to the original antibiotic. Evidence is beginning to confirm that the prevalence of antibiotic resistant bacterial strains in the environment is on the rise (Zhang et al. 2009; Knapp et al. 2010). Algae have also been found to be particularly sensitive to the effects of antimicrobials at environmentally relevant concentrations (Halling-Sorensen 2000; Wilson et al. 2003; Porsbring et al. 2009).

The presence of synthetic estrogens in the environment also raises significant concerns. A number of the synthetic estrogens used in human birth control pills are significantly more potent than natural estrogens, meaning they can potentially have estrogenic effects at much lower concentrations. Synthetic hormones have been shown to have a number of impacts on non-target organisms, including affecting vitellogenin induction in male fish (Brian et al. 2005). It is this effect that implicates these

(25)

anthropogenic chemicals as one of the causes of male feminization in various

species, an effect which is increasingly being observed around the world (Folmar et al. 1996; Jobling et al. 1998; Pettersson and Berg 2007) and has potentially serious implications on population reproductive success (Kidd et al. 2007). Other observed impacts of these substances at environmentally relevant exposure concentrations include reproductive delay and decreased size in Daphnia magna (Dietrich et al. 2010b). It should be noted, however, that endocrine disruption is not limited to environmental exposure to synthetic hormones; numerous other anthropogenic chemicals can have similar effects, including organochlorine pesticides, polychlorinated biphenyls, dioxins and furans, bisphenol-A, alkylphenolic chemicals, fungicides, tributyl tin, and some phthalates (Tyler et al. 1998; Tyler and Jobling 2008).

Nonsteroidal anti-inflammatory drugs also have the potential to cause adverse effects in non-target organisms in the environment. In the terrestrial environment, they have been implicated in the deaths of Gyps vultures as a result of kidney failure following consumption of domesticated ungulates that have been treated with the drugs (Naidoo and Swan 2009; Naidoo et al. 2010). At environmentally relevant concentrations, it has also been demonstrated that these drugs can delay the age at which Daphnia magna first reproduce (Dietrich et al. 2010b) and can affect riverine biofilm communities (Winkler et al. 2001; Lawrence et al. 2005).

Multixenobiotic resistance (MXR) is a cellular defence mechanism employed by a number of aquatic organisms. The MXR mechanism works as a result of

transmembrane proteins that are capable of carrying a wide variety of natural or

(26)

to outside of the cell. A number of PPCPs have been shown to inhibit the

MXR mechanism in some organisms including musks (Luckenbach and Epel 2005) and some antihypertensives (Kurelec 1992). Exposure to these PPCPs in the environment has the potential to make some organisms more sensitive to other environmental

contaminants.

Antidepressants are another group of PPCPs of environmental concern. A number of these drugs have been shown to be potent inducers of spawning in bivalves (Fong 1998; Fong et al. 1998) and have other reproductive effects on aquatic organisms (Kulkarni et al. 1992). As such, exposure to these compounds in the environment could have significant implications on the reproductive success of some species.

Lipid regulators have also been shown to have effects at environmentally

relevant concentrations. Isidori et al. (2009) demonstrated the potential for these drugs to induce estrogenic activity and they have also been shown to induce oxidative stress in fish (Mimeault et al. 2006). Research involving these compounds and their potential for chronic environmental effects is on the rise due to their prevalence in municipal

wastewaters.

The above summary of PPCPs effects demonstrates the potential for these compounds to elicit effects in the environment and why they have become an

environmental concern. It is anticipated that more and varied effects will be identified over the next few years as the number of studies looking for chronic sub-lethal effects of PPCPs is on the rise (Richardson 2009). Ecotoxicologists are undertaking more varied exposure experiments, including some that are more cognizant of organisms having differing sensitivities to anthropogenic substances depending on where the organism is in

(27)

its life cycle (e.g., embryonic versus adult exposure), and the fact that

organisms are potentially exposed to mixtures of PPCPs and anthropogenic chemicals, rather than individual compounds (Dietrich et al. 2010b). Although our ability to detect these substances in the environment is increasing rapidly (Gilpin and Gilpin 2009), the biological and environmental consequences of exposure to low level mixtures of PPCPs, and other anthropogenic chemicals, are very challenging to detect and assess. In

addition, direct cause and effect relationships that can be attributed to individual compounds are still relatively lacking in the literature (Tyler et al. 1998).

1.2 OVERALL THESIS OBJECTIVES

In March 2003, the CRD’s Core Area Liquid Waste Management Plan (LWMP) (Capital Regional District 2000) was approved by the British Columbia Ministry of Environment (BC MoE) (formerly the Ministry of Water, Land and Air Protection). The intent of this plan was to guide the CRD in its management of liquid wastes for the next 25 years and it included commitments to ensure that human health and the environment were protected from impacts of the region’s liquid waste discharges. As part of the LWMP approval, BC MoE required the CRD to undertake collaborative studies on PPCPs and other emerging scientific issues. This thesis is the first comprehensive assessment intended to satisfy BC MoE’s requirement.

The objectives of this research were to 1) characterize PPCP concentrations in CRD wastewaters and loadings to the environment; 2) assess the potential for

environmental impacts of the PPCPs being discharged; 3) determine the suitability of utilizing prescription rates and population demographics to predict wastewater PPCP concentrations and loadings; 4) determine the relationships between the concentrations of

(28)

PPCPs and a number of conventional wastewater parameters that are typically

used to characterize effluent quality and treatment efficiency at wastewater processing facilities; 5) assess PPCP concentration patterns in relation to chemical characteristics used to predict partitioning behaviour during wastewater treatment; and 6) determine the relationship between whole wastewater system PPCP loadings and those from smaller catchment areas within the system. Additional rationale for these objectives will be provided in Chapters 3 and 4.

Chapter 2 will provide a brief comparison and critique of the three different analytical methodologies used throughout this thesis. In Chapter 3, the results of approximately two years of monthly wastewater sampling for a select group of pharmaceuticals will be assessed in relation to prescription rates and population

demographics and for their potential to impact non-target organisms in the environment. Chapter 4 will provide a summary of a five day wastewater characterization study that was undertaken to assess the concentrations of a larger suite of PPCPs and how they related to the concentrations of conventional wastewater parameters and partitioning behaviour indicators. Chapter 4 will also include a preliminary concentration and loading assessment with respect to the relative inputs of various municipalities into the CRD’s wastewater system. Finally, Chapter 5 will provide a discussion of the overall

implications of the PPCPs in CRD wastewaters, the implications of the region’s plans for more advanced wastewater treatment and suggestions for future research.

(29)

Chapter 2 Analytical Methodology and Results Comparison

The intent of this chapter is to compare and contrast the three different analytical methodologies discussed in the following chapters of this thesis. A brief comparison of the results of the three sets of analyses, as applied to Macaulay Point samples, will also be provided to give some indication of the relative effectiveness of each method.

2.1 METHODS

2.1.1 Sample Collection

2.1.1.1 Samples from 2004 to 2006

University of Victoria samples were collected on a monthly basis at the MAC pump station from November 2004 to September 2006. Field replicates were collected for three of the months to assess field variability. All samples were collected from the facility wet well (post-screening) as 24 hr composites using Teledyne ISCO, Inc. (Lincoln, Nebraska, USA) automatic water samplers concurrently with the CRD’s routine wastewater sampling program (Capital Regional District 2005; 2006; 2007). The samples consisted of 400 mL of wastewater collected at 30 min intervals over the 24 hr period and composited into a pre-cleaned 20 L carboy. At the end of the 24 hr period, the carboy was placed on ice and transported to CRD sample processing facilities. The carboys were continuously stirred while 1 L subsamples were drawn for the

pharmaceutical analyses. All 1 L subsamples were immediately frozen at -20 °C. In March 2006, additional grab samples were collected at MAC (on a different day than the March 2006 24 hr composite sample) using a pre-cleaned plastic bucket lowered in the the wet well (post-screening). The grab samples were collected

concurrently with sampling that was being done on behalf of Environment Canada and the Georgia Basin Action Plan (GBAP). Eight 1 L samples were collected with a

(30)

pre-cleaned bucket and immediately frozen at -20 °C for later analysis at the

University of Victoria. The GBAP grab samples were collected using the same pre-cleaned bucket, transferred to large carboys, and transported on ice to Environmental Canada facilities in North Vancouver.

All CRD samples were kept frozen for up to eight months prior to extraction and analyses, except for the additional March 2006 MAC grab samples. These eight samples were extracted, but not necessarily analyzed, on a monthly basis for six

consecutive months to assess the implications of the extended freezer holding times. The GBAP samples were kept at 4 °C and analyzed within 14 days of sample collection (Osachoff 2008; Environment Canada 2009).

2.1.1.2 Samples from 2009

Wastewater samples were collected on a daily basis at the MAC pump station from November 2nd to 6th, 2009 with one field triplicate collected to assess field

variability on November 2nd. These samples were collected from the same location as the 2004 to 2006 samples (i.e., the facility wet well, post-screening). All samples were collected as 24 hr composites using Teledyne ISCO, Inc. (Lincoln, Nebraska, USA) automatic water samplers. The samples consisted of 400 mL of wastewater collected at 30 min intervals over the 24 hr period and composited into one pre-cleaned carboy. At the end of the 24 hr period, the carboy was placed on ice and transported to CRD sample processing facilities. The carboys were continuously stirred while three or four 1 L subsamples were drawn for the PPCP analyses. All subsamples were immediately put on ice and dispensed to the analytical laboratory.

(31)

2.1.2 Analyses

Table 2-1 contains a summary of the major steps taken by the three laboratories to filter, extract and analyze the samples collected in 2004 to 2006, and in 2009. Only a brief summary of each method will be provided below, with the focus of the chapter being on significant differences between the methods that could lead to differences in how effective each method is at determining the concentrations of the same PPCPs. Additional methodological details can be found in the following three documents:

 University of Victoria - (Verenitch et al. 2006);

 Environment Canada/GBAP - (Environment Canada 2009); and

 AXYS Analytical Services - (AXYS Analytical Services Ltd. 2010). For any samples that were replicated, either in the field or the laboratory, relative percent differences (RPD) were calculated by dividing the absolute difference between the duplicates by their mean and multiplying by 100%, while relative standard deviations (RSD) were calculated by dividing the standard deviation of the replicates by their mean and multiplying by 100%.

All mean and standard deviation calculations used a value of half the method detection limit (Table 2-2) for non-detect (ND) values.

2.1.2.1 Samples from 2004 to 2006

The CRD/University of Victoria samples were thawed overnight at room temperature prior to filtration, extraction and analysis, following up to eight months of freezer storage. One sample was split prior to extraction to assess laboratory analytical replication. In addition, one blank sample (i.e., distilled water) was analyzed per analytical batch. Extraction, derivitization, and analyses followed the methods of

(32)

Verenitch et al. (2006). In short, all samples were filtered, spiked with 2,3-dichlorophenoxyacetic acid (2,3-D) as a surrogate standard to assess extraction efficiency, and acidified with HCl to a pH of 2.0. The acidified samples were passed through solid phase extraction (SPE) cartridges and then eluted. The resulting extracts were concentrated, spiked with meclofenamic acid (MCF), to assess derivitization

efficiency, and derivtized using methanol and BF3. The methylated extracts were cooled,

concentrated, dehydrated, and then transferred to a gas chromatograph (GC) vial by pipette. The extracts were then analyzed for the methylated derivatives of acetylsalicylic acid (ASA), diclofenac (DCF), fenoprofen (FEN), gemfibrozil (GEM), ibuprofen (IBU), ketoprofen (KET), and naproxen (NAP) using a gas chromatography – ion trap tandem mass spectroscopy (GC/IT-MS/MS). Salicylic acid (SA) was analyzed as a surrogate for ASA because ASA is known to breakdown to SA during the methylation process

(Metcalfe et al. 2003a). Method detection limits are presented in Table 2-2. All University of Victoria pharmaceutical concentrations were quantified using MS

chromatogram peak areas. The results were blank corrected, but not surrogate (2,3-D) or internal (MCF) standard recovery corrected.

The sample collected concurrently as part of the GBAP program at MAC in March 2006 was analyzed by Environment Canada (North Vancouver, British Columbia) (Environment Canada 2009) for the same seven pharmaceuticals (including SA as a surrogate for ASA). No field or laboratory replicates were collected for this sample. In short, the GBAP method involved extracting and analyzing the fresh (i.e., without any freezing) sample within 14 days of collection. The sample was acidified with sulphuric acid, but not filtered prior to the liquid/liquid dichloromethane extraction. The extract

(33)

was analyzed using high performance liquid chromatography – low resolution mass spectroscopy (HPLC/LRMS) in SIM mode. All Environment Canada

pharmaceutical concentrations were quantified using MS chromatogram peak areas, with the concentrations of ibuprofen corrected post-analyses based on the sample specific recovery of the d3-IBU standard. Detection limits for each PPCP can be found in Table

2-2.

2.1.2.2 Samples from 2009

The 2009 sample analyses were performed by AXYS Analytical Services Ltd. (Sidney, British Columbia, Canada) using two or three of the 1 L subsamples from each composite sample. In short, all PPCP samples were assessed fresh (i.e., without any freezing). The samples were filtered, acidified to a pH of 2.0, and the aqueous portions were cleaned up by solid phase extraction (SPE). The extracts were analyzed by high performance liquid chromatography/electrospray ionisation tandem mass spectrometry (LC/ESI-MS/MS) (triple quadrupole) in negative ionization mode using AXYS method MLA-075 Rev 01 (AXYS Analytical Services Ltd. 2010). Detection limits specific to each analyte are summarized in Table 2-2. Numerous laboratory blanks and spiked matrix samples were run concurrently with the field sample analyses; samples were also spiked with surrogate standards. Not all PPCPs were measured in every sample due to financial constraints. All AXYS pharmaceutical concentrations were quantified using MS chromatogram peak areas, with the concentrations corrected post-analyses based on the sample specific recoveries of the d6-GEM, 13C3-IBU, and 13C-d3-NAP standards.

(34)

2.2 RESULTS

2.2.1 Samples from 2004 to 2006 2.2.1.1 University of Victoria Results

Standard recoveries ranged from ND to 179% for the surrogate standard 2,3-D, and from 44 to 219% for the internal standard MCF (Appendix B). Although these ranges exceed analytical method expectations (Verenitch et al. 2006) and the CRD data quality objectives (DQO) for standard recoveries of 50-150% (Golder Associates Ltd. 2009b), most recoveries were within 50-150%. The implications of surrogate recoveries outside of the DQO ranges will be discussed in detail in Chapter 3.

Blank results are presented in Appendix C. Blank results for all analytes except FEN met the CRD DQO of blank results being less than 10x sample concentrations (Golder Associates Ltd. 2009b). All wastewater sample results were blank corrected based on the mean blank results for each analyte.

Laboratory and field replicate sample results are presented in Appendix E. As expected, lab replicates were less variable than field replicates. Lab replicate relative standard deviation (RSD) and relative percent difference (RPD) values ranged from 1 to 16% and were well within the expectations of the analytical method (Verenitch et al. 2006) and the CRD’s DQO requiring RSD and RPD values to be below 50% (Golder Associates Ltd. 2009b). Field replicate RSD and RPD values ranged from 3 to 188%, with the majority below 50%. Fenoprofen field replicate results were the most variable due to the number of ND values for this analyte. The observed field variability was slightly higher than expected based on analytical method development results (Verenitch et al. 2006). However, the field variability was similar to that occasionally seen for other organic analytes in CRD wastewaters (Capital Regional District 2010a).

(35)

Results of the MAC freezing assessment can be found in Appendix

F. The implications of the six month freezer holding time on the concentrations of the seven pharmaceuticals were not assessed statistically due to the small dataset. However, qualitatively it appears that concentrations of DCF, IBU, KET, and NAP declined slightly by approximately 5 to 15% over the six months. Gemfibrozil appeared to have the

largest decrease in concentrations of approximately 45% over the six months. Slight increases in concentrations, of approximately 20% over the six month holding time, were observed for SA (ASA metabolite). Due to the number of ND values for FEN, the implications of freezer holding time could not even be qualitatively assessed for this analyte. No corrections were made to any of the other sample results as a result of these freezer holding time observations.

Macaulay Point monthly composite pharmaceutical concentrations are presented in Appendix B with the March 2006 grab sample freezing assessment results in Appendix F. Summaries of both sets of results can be found in Table 2-4.

2.2.1.2 Georgia Basin Action Plan Results

The Environment Canada/GBAP results for the single March 2006 grab sample collected concurrently with the University of Victoria freezing assessment grab samples can be found in Table 2-4 and Appendix F. Concentrations of SA (ASA metabolite), FEN, or KET were non-detect at MDLs ranging from 3 to 29 ng/L. Concentrations of DCF, GEM, IBU and NAP were detected by Environment Canada in the GBAP samples, but at concentrations less than half of the levels found concurrently using the University of Victoria methods. There were no batch specific QA/QC sample results (i.e., spiked samples, blanks, etc.) available from GBAP for use in this methods comparison.

(36)

2.2.2 Samples from 2009 (AXYS Analytical Results)

Only three of the PPCPs measured by the University of Victoria and

Environment Canada/GBAP from 2004 to 2006 were also measured by AXYS in 2009: GEM, NAP, and IBU. The concentrations of GEM and NAP from MAC in 2009 were on the low end of, or slightly below, the 2004 to 2006 ranges as determined by the

University of Victoria and were much higher than the concurrent GBAP results (Table 2-4; Appendix P).

Relative standard deviations (RSD) of the field triplicate results met the CRD data quality objectives (DQO) of RSD values less than 50% (Golder Associates Ltd. 2009b) (Appendix P). At least one laboratory blank sample was run per analytical batch and all blank results for GEM, IBU and NAP were ND.

2.3 DISCUSSION

All three methods employed sample acidification prior to filtration and

extraction (Table 2-1). Therefore, any implications of pH on target analyte solubility or ionization state would have been similar for all three methods.

2.3.1 Implications of Analytical Method Differences

Municipal wastewater is a complex matrix and can pose challenges for analyses of organic compounds. There were a number of differences in the three analytical methodologies employed that could have led to the rather large differences observed when GBAP results were compared to the University of Victoria and AXYS results, and the more subtle differences observed when University of Victoria and AXYS results were compared.

First off, the tandem MS analytical acquisition methods used both by the University of Victoria and AXYS analytical services by can be particularly powerful

(37)

relative to single MS methods, as tandem MS methods allow for enhanced

target compound specificity (Monaghan et al. 1992) and results precision (Choi and Antoniewicz 2011). Indeed, Helenkar et al. (2010) found that MS/MS methods were much more sensitive than selective ion monitoring (SIM) or full scan (FS) GC-single MS methods for IBU, NAP, KET and DCF in Danube River samples.

Other likely contributors leading to the substantially lower GBAP results were the specific filtration and extraction methods employed by Environment Canada.

Because the GBAP detection limits were relatively low (Table 2-2), extraction efficiency was likely the greatest contributor to the relatively low GBAP results as opposed to the limitations of single-MS techniques. This can be observed in the recoveries of analytes spiked in distilled or tap water and analyzed by each analytical method (Table 2-3). Even in these clean samples, the GBAP recoveries were generally lower than the other two labs. Environment Canada used liquid-liquid extraction on unfiltered samples

(Environment Canada 2009) whereas both the University of Victoria and AXYS used SPE on filtered samples (Verenitch et al. 2006; AXYS Analytical Services Ltd. 2010). Solid phase extraction generally leads to better target analyte recoveries, and results in cleaner extracts and a reduction in the matrix interference often encountered with complex samples such as wastewaters. As such, SPE has been used far more frequently than liquid-liquid extraction in the analysis of pharmaceuticals in wastewater (Lopez de Alda and Barcelo 2001). One disadvantage of SPE is the fact that samples need to be filtered prior to extraction to prevent clogging of the SPE substrate. As such, the GBAP results using the unfiltered sample were more representative of the whole wastewater sample, whereas the University of Victoria and AXYS results were only representative of

(38)

the aqueous portions of the wastewater samples. Although the target acidic

pharmaceuticals are relatively water soluble (Appendix N) and unlikely to be associated with the particulate fractions of the wastewater samples, the fact that University of Victoria and AXYS samples were filtered as opposed to the unfiltered GBAP sample ultimately means that the results were not directly comparable. The filtered particulate fractions could have been analyzed independently, but this was not done in this study. Additional implications filtered versus unfiltered samples will be discussed in Chapter 4.

The power of MS/MS versus SIM, and SPE versus liquid/liquid extraction can be demonstrated in this study by comparing the University of Victoria and AXYS SPE and MS/MS results to those of the GBAP sample which was analyzed in SIM mode using HPLC/LRMS following liquid/liquid extraction. The Environment Canada analytical method did not detect SA (ASA metabolite), FEN, or KET at MDLs ranging from 3 to 29 ng/L, whereas the concurrent University of Victoria results detected all three substances at substantially higher concentrations and lower detection limits (Table 2-2; Table 2-4). Concentrations of DCF, GEM, IBU and NAP were detected by Environment Canada in the GBAP samples, but at concentrations less than half of the levels found by the University of Victoria and AXYS.

Both Environment Canada and AXYS utilized liquid chromatographic

separation techniques whereas the University of Victoria employed gas chromatography. The LC techniques had an advantage over the GC technique, in that GC based method required sample derivitization prior to analyses. The sample derivitization process can lead to target analyte degradation, increased variability in the results, and ultimately underestimations of concentrations. As an example of potential for analyte degradation,

(39)

SA was measured as an ASA surrogate by the University of Victoria due to

the fact that ASA is known to breakdown to SA during the methylation process (Metcalfe et al. 2003a). The potential for variability as a result of derivitization was also

demonstrated in the University of Victoria results when one looks at the wide recovery range of MCF (the standard added to assess derivitization efficiencies) (Appendix B).

Even with the relative sensitivity of analytical equipment used by the University of Victoria, there was a fair bit of variability observed the QA/QC results from this lab. Surrogate and internal standard recoveries were relatively variable compared to the results observed during method development (Verenitch et al. 2006). The standard recoveries in the four screened wastewater samples assessed by Verenitch et al. (2006) ranged from 79-90% for MCF and 80 to 103% for 2,3-D. In this study, standard

recoveries fell within the CRD DQOs of 50-150% for only 9 of the 21 samples analyzed. The % recoveries fell outside of this DQO range in seven of the samples for 2,3-D and four of the samples for MCF (Appendix B). With respect to these extreme (i.e., <50% or >150%) standard recoveries, there did not appear to be any correlation with

pharmaceutical concentrations; low surrogate recoveries were not always associated with low pharmaceutical concentrations and vice versa. This is particularly evident when looking at the results of field replicate analyses (Appendix E). As such, no reliable corrections could be made to the individual sample results based on their standard recoveries.

The recoveries of surrogates and standards spiked into tap or distilled water samples are presented in Table 2-3 for each of the three methods. No comparisons to these expected recoveries could be made for the GBAP results as they were only from a

(40)

single sample. On the other hand, a number of the University of Victoria

replicate sample results neither came anywhere close to the RSD values observed during method development (Verenitch et al. 2006) nor met the CRD DQOs of RSD and RPD ≤50% (Golder Associates Ltd. 2009b). These replicate DQO failures were most likely a result of the complex nature of wastewater from a matrix interference perspective and the fact that wastewater is naturally heterogeneous. Failures of a similar magnitude have occasionally been observed during the CRD’s routine wastewater monitoring (Capital Regional District 2010a). Relatively few wastewater samples were analyzed by the University of Victoria during method development and I believe that more variability would have been observed had more wastewater samples been analyzed by Verenitch et al. (2006). Although all AXYS results were within the expected RSD ranges, this lab has had to broaden their acceptable surrogate and standard recovery ranges over time as the analyses of ongoing spiked samples have indicated that their original DQOs were not wide enough (i.e., see table of “OPR” and “IPR” in AXYS Analytical Services Ltd. [2010]). In the University of Victoria results, FEN replicate results had the highest level of variation, but this was primarily due to the FEN concentrations being ND or very near the MDL.

The relatively wide variation in University of Victoria results could likely have been reduced substantially had the laboratory employed the use of deuterated or other similarly labelled standards, a practice employed by both Environment Canada and AXYS (Table 2-1). The use of deuterated standards allows one to correct the

concentrations of the target PPCPs based on the recoveries of the very close-in-structure deuterated equivalents. Correcting results in this manner can substantially reduce

(41)

variation by better accounting for analyte loss or other artefacts that arise at all stages throughout the analytical procedure including during extraction and/or

derivitization. Such PPCP standards were not available when the University of Victoria first developed their methodology, and were not added to the method even when they became readily available commercially (e.g. d3-IBU became available in approximately

June 2004 and d6-GEM became available in 2006). This lack of standards is a significant

shortcoming of the University of Victoria method. Corrections based on 2,3-D and MCF recoveries were not appropriate, as these surrogates were only intended to give a general indication of extraction and derivitization efficiency, and are not close enough in

structure to all of the target PPCPs.

Another difference between the three sampling and analytical methodologies was that the University of Victoria samples were frozen and thawed prior to analyses, whereas the GBAP and AXYS samples were analyzed fresh. Keeping samples cool or frozen reduces inherent biological activity and resulting target analyte degradation (Hunt and Wilson 1986). Sample freeze/thaw could have impacted the results in two ways. First, the University of Victoria samples were filtered after thawing, and during the overnight and room-temperature thawing process, biological activity in the samples could have led analyte degradation prior to analyses (Hunt and Wilson 1986). Had filtration occurred prior to freezing, the biological activity during the room-temperature thawing process would likely have been reduced. However, the extent of biological degradation in the thawing samples was not quantified. Secondly, the freeze/thaw process could have had implications on the suspended particle size distribution of the samples (Singleton et al. 1960) and could have affected the proportion of the PPCPs in solution as opposed to