Science of the Total Environment 576 (2017) 374–380

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Occurrence of cocaine and benzoylecgonine in drinking and source water

in the São Paulo State region, Brazil
Iolana Campestrini ⁎, Wilson F. Jardim
Environmental Chemistry Laboratory, Department of Analytical Chemistry, Institute of Chemistry, State University of Campinas, Campinas, São Paulo, Brazil

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Cocaine was investigated in natural wa-

ters from important watersheds in
Brazil.
• High use of cocaine and poor sanitation
affect the water quality in Brazil.
• COC and BE levels found are one of the
highest found in urban surface waters.
• Cocaine levels detected in rivers in the
São Paulo State may pose some risk to
aquatic species.
• The presence in tap water indicates hu-
man indirect exposure to multiple com-
pounds.

a r t i c l e i n f o a b s t r a c t

Article history: The occurrence of illicit drugs in natural waters (surface, source and drinking water) is of interest due to the poor
Received 30 June 2016 sanitation coverage and the high consumption of drugs of abuse in Brazil. In addition, little is known about the
Received in revised form 13 October 2016 effects of these compounds on aquatic organisms and human health. This work investigates the occurrence of co-
Accepted 13 October 2016
caine (COC) and its major metabolite, benzoylecgonine (BE), in surface and drinking water collected in rivers
Available online xxxx
from a populated geographic area in Brazil. Surface water samples were collected in 22 locations from 16 differ-
Thomas Kevin V ent rivers and one dam from São Paulo State, whereas drinking water samples were collected in 5 locations. Sam-
ples were collected during the dry and wet season. Among the 34 surface water samples analyzed, BE was
Keywords: detected above the LOD in 94%, while COC in 85%. BE concentrations ranged from 10 ng L−1 to 1019 ng L−1
Cocaine and COC concentrations from 6 ng L−1 to 62 ng L−1. In the drinking water samples analyzed, BE and COC were
Natural waters found in 100% of the samples analyzed. For BE, concentrations were found in the range from 10 ng L−1 to 652
Emerging contaminant ng L−1, and COC was quantified in concentrations between 6 and 22 ng L−1. These concentrations are one of
Biomarker the highest found in urban surface waters and may pose some risk to aquatic species. However, no human health
Risk assessment
risk was identified using the Hazard Quotient. BE is proposed as a reliable indicator of sewage contamination in
both source and drinking water.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction

Illicit drug abuse causes effects on health, social life and world econ-
⁎ Corresponding author. omy. In the last years, the increase in consumption and consequently
E-mail address: iolanacamp@gmail.com (I. Campestrini). the detection in many matrices has motivated the investigation of

http://dx.doi.org/10.1016/j.scitotenv.2016.10.089
0048-9697/© 2016 Elsevier B.V. All rights reserved.
 I. Campestrini, W.F. Jardim / Science of the Total Environment 576 (2017) 374–380
these “emerging contaminant” in the environment (Zuccato et al.,
2008).
Many illicit drugs are continuously released into the aquatic envi-
ronment due to their incomplete removal in the wastewater treatment
plants (WWTP). Once consumed, these substances are partially metab-
olized in the human organism and reach the wastewater network as a
complex mixture of parental compounds and metabolites, where they
can be detected from ng L− 1 to μg L− 1 levels (Berset et al., 2010;
Maldaner et al., 2012; Thomas et al., 2012). However, as the majority
of WWTP have been designed to remove organic matter, many specific
compounds like hormones, pesticides and illicit drugs, which are not in-
cluded in the current local legislation, could persist and reach the receiv-
ing surface waters (Zuccato et al., 2008; Postigo et al., 2010). In
countries with poor sanitation coverage, such as Brazil, high loads of
raw and poorly treated sewage are discarded directly into water bodies,
resulting in a severe contamination scenario in these watersheds (Sodré
et al., 2010; Montagner and Jardim, 2011; Martins et al., 2014; Carreira
et al., 2015; Frena et al., 2016).
The frequency and the levels that illicit drugs are detected in the en-
vironment depend on the local prevalence, seasonality, efficiency and
coverage of sanitation infrastructure. The literature shows illicit drugs
concentrations at ng L−1 in surface waters, but even at such low levels,
these substances can affect the ecosystem and reach the drinking water
supply (Berset et al., 2010; Binelli et al., 2013; van Der Aa et al., 2013;
Mendoza et al., 2014). The removal efficiency of illicit drugs depends
on the water treatment process used, which means that it is necessary
to monitor the presence of these substances also in drinking waters to
protect human health. However, considering the impossibility to moni-
tor all emerging contaminants or every type of illicit drugs and metabo-
lites that may be found in surface and drinking waters, selected target
substances that might work as an indicator of contamination from sew-
age is necessary and desired for a rapid and less costly screening of the
water quality.
According to the World Drug Report of the United Nations Office on
Drugs and Crime (UNODC, 2016), about 246 million people used some
illicit drug in 2013, representing an increase of 3 million compared to
the previous year (UNODC, 2015). Cocaine (COC) is one of the most
used illicit drugs in the world. While the use of opioids is a serious prob-
lem in Europe, the prevalence of COC use in South America is the highest
in the world, particularly in Brazil, which has been facing serious prob-
lems with COC and crack, representing the second highest market in the
world.
In Brazil, around 59% of the sewage produced is not treated and
dumped “in nature”. Consequently, the contamination of source waters
occurs, compromising the quality of drinking water concerning the
presence of illegal substances and its major metabolites, which are
only partially removed in drinking water treatment plants (DWTP)
(SNSA, 2016; Sodré et al., 2010; Montagner and Jardim, 2011). This con-
tamination may pose a risk to humans, especially considering that 17%
of Brazilian cities are not fully served by the water distribution network
(SNSA, 2016).
Few studies investigated the presence of these substances in the
drinking water distribution network around the world. In Madrid re-
gion, COC was quantified (1.6 ng L−1) in only one out of the six sam-
pling points (Mendoza et al., 2014). COC and other illicit drugs were
investigated in cities around Europe (Austria, France, Germany,
Switzerland, United Kingdom), Latin America (Panama, Brazil,
Colombia, Argentina, Uruguay) and in Japan by Boleda et al. (2011).
COC was detected in 66% (0.2 to 2.3 ng L−1) and its major metabolite,
benzoylecgonine (BE), in 34% (0.2 to 3.1 ng L−1) of the samples ana-
lyzed in Spain. In drinking water samples from the other countries,
these compounds were detected at sub ng L− 1 levels. The highest
mean values of COC and BE were detected in tap water samples coming
from South America. For BE, the mean values found were: 0.2 ng L−1 in
Europe, bLOQ in Japan, 0.4 ng L− 1 in Spain and 4.5 ng L− 1 in Latin
America.
375
There are few studies evaluating the presence of drugs of abuse in
both Brazilian surface and drinking water, considering their probable
adverse effects for aquatic life and human health (Thomas et al.,
2014; Boleda et al., 2011; Locatelli, 2011). The present study investi-
gated COC and BE occurrence in surface and drinking water in differ-
ent areas of the São Paulo State. Also, COC and BE levels were
evaluated according to regional and seasonal variability, exploring
the potential use of BE as an indicator of wastewater contamination
in natural waters. To the best of our knowledge, this is the first mon-
itoring study of both substances in surface and drinking waters, cov-
ering the most populated geographic area in some major rivers in the
São Paulo State, Brazil.
2. Material and methods
2.1. Chemicals
COC and BE analytical reference standards were purchased from
Cerilliant Corporation (Round Rock, TX, USA), in liquid form with purity
higher than 98%, at concentrations of 0.1 mg mL−1 or 1.0 mg mL−1 in
methanol (MeOH) or acetonitrile (ACN) (stock solution). All stock solu-
tions were stored in the dark at −4 °C. Work standard solutions con-
taining both target compounds at 10 and 1 μg L−1 were prepared in
ACN/water (50:50 v v−1). Cerilliant Corporation BE-d3 was used as in-
ternal standard (IS), and it was prepared separately at 10 μg L−1 follow-
ing the same procedure. Both solvents (99% purity MeOH and ACN)
were purchased from Mallinckrodt Baker, Inc. (Phillipsburg, NJ - USA).
Formic acid (98% purity) was acquired from Sigma-Aldrich Co., LLC (St
Louis, MO - USA). Ultrapure water (Resistivity of 18 MΩ cm) was pre-
pared using a Millipore Milli-Q Plus water purification system
(Millipore Corporation - Burlington, MA, USA).
2.2. Sampling
The investigated area included some of the main rivers in the state of
São Paulo, and an important dam, located in São Paulo city. These
waterbodies together supply over 10 million people. Surface water sam-
ples were collected in 16 rivers from São Paulo State from important
watersheds, which supply approximately 6.3 million people (Fig. 1). Be-
sides, three sampling points (GD1, GD2, and GD3) were selected in the
Guarapiranga dam (São Paulo city), an important source of drinking
water for nearly 5.6 million people in the neighborhood (SABESP,
2014). In some cases, surface water samples were collected in more
than one point along the river flow, reaching 22 sampling points in
total. Some sampling was performed during dry season (between Feb-
ruary and March 2014) and at the beginning of the rainy season (be-
tween September and October 2014), and 34 surface water samples
were analyzed in total. Drinking water samples were collected during
the same period in five cities supplied by some of the selected rivers,
reaching eight drinking water samples.
2.3. Sample treatment and analyte extraction
Surface and drinking water samples were collected as grab sample in
amber flasks (1 L) previously rinsed with ultrapure water, followed by
the sample. Once collected, and during shipment, the samples were
kept refrigerated. In the laboratory, the samples were vacuum filtered
using 1.2 μm pore-size glass microfiber filters (Schleicher&Schuell -
Dassel, LS, Germany) followed by 0.45 μm pore size cellulose acetate
membranes (Sartorius - Göttingen, LS, Germany). Solid phase extraction
(SPE) was performed with Oasis HLB Cartridges.
The SPE method was modified from the previous study to investi-
gate COC and BE in wastewater (Maldaner et al., 2012). Commercial
HLB Oasis® cartridges (6 mL, 500 mg) were obtained from Waters Cor-
poration (Milford, MA, USA).
376 I. Campestrini, W.F. Jardim / Science of the Total Environment 576 (2017) 374–380
Fig. 1. Map of the study area and location of the sampling points (blue circles). The São Paulo State borders with Paraná (PR), Mato Grosso do Sul (MS), Minas Gerais (MG) and Rio de
Janeiro (RJ). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
The HLB cartridges were equilibrated with three mL of MeOH, four
mL of ACN and five mL of ultrapure water. After percolation of the fil-
tered samples (1 L), the cartridges were vacuum dried (10 mm Hg)
for 10 min, and eluted with six mL of an ACN: MeOH, 60:40 (v v−1) so-
lution, and evaporated to dryness under a nitrogen stream. The dry ex-
tract was dissolved in 1.0 mL of 0.1% aqueous formic acid: ACN, 90:10 (v
v−1) solution and transferred to vials for LC-MS/MS analysis.
2.4. LC-MS/MS analysis
The LC-MS/MS analysis was performed using a liquid chromato-
graph Agilent 1200 series, with a binary pump, connected to a 6410 tri-
ple quadrupole mass spectrometer equipped with electrospray
ionization source (Agilent Technologies, Palo Alto, CA, USA). Separation
used a Zorbax SB-C18 (30 mm × 2.1 mm, 3.5 μm) reversed-phase HPLC
column (Agilent Technologies, Palo Alto, CA - USA). Formic acid solu-
tions (0.1%, v v−1) (eluent A) and ACN (Eluent B) were used as mobile
phase solvents. The following gradient program was performed: initial
condition at 10% of eluent B kept for 0.5 min, followed by a linear gradi-
ent reaching 80% of eluent B in 5 min. This condition was kept for 2 min,
and then a 4 min linear gradient back to 10% of eluent B. To equilibrate
the column for the next run, the initial condition was held for 5 min. The
flow and the column temperature rate were held at 0.3 mL min−1 and
35 °C, and 10 μL of each sample were injected into the LC-MS/MS
system.
The electrospray source was operated in the positive polarity. High
purity nitrogen (N 98%) was used as desolvation (350 °C; 10 L min−1),
nebulizer and collision gas. Nebulizer pressure was kept at 30 psi and
the capillary voltage set at 2500 V. The Fragmentor was optimized at
120 V for all analytes, which were analyzed in multiple reaction moni-
toring (MRM) mode, recording the transitions between the precursor
ion and the three most abundant product ions for each target analyte.
The most intense transition was used as a quantifier ion (MRM1) and
the others as a qualifier ion (MRM2). Collision energy (CE) optimization
was done separately for each compound. The precursor and product
ions for each target analyte, including IS, with their respective CE, are
provided in the supplementary information (Table SI1). The identifica-
tion of the analytes was confirmed by comparing the retention time
and the MRM1/MRM2 ratio with the corresponding values of analytical
standards measurements.
2.5. Validation of the method and quantification
The analytical method was validated for surface water and drinking
water samples fortified with COC-d3 and BE-d3 at two different concen-
tration levels, since there were no negative samples for the presence of
COC and BE. The efficiency of the SPE, the matrix effect (ME) and accu-
racy were evaluated, based on Niessen et al. (2006).
For the efficiency of the SPE, matrix effect (ME) and recovery tests,
two independent samples of each water sample (surface and drinking
water) were spiked before the SPE step. Two others independent sam-
ples were spiked after the SPE step and two solutions of COC and BE
were prepared in the solvent alone, at the same concentrations. Com-
paring the analytical responses (areas) obtained, the SPE efficiency
and the ME were calculated using the following equations. The results
obtained were expressed as a mean value.
spike area before SPE
SPE efficiency ð%Þ ¼  100
spike area after SPE
ðspike area after SPE−spike area in the solventÞ
ME ð%Þ ¼  100
spike area in the solvent
The recovery was estimated comparing the areas obtained in the
spiked water sample with the area obtained in the solvent for both
COC and BE.
spike area before SPE
Recovery ð%Þ ¼  100
spike area in the solvent
Intra-day precision was evaluated by analysis of samples prepared
by six times in the same day, and inter-day precision by analyzing of
the same spiked samples prepared on three different days (spikes
were performed before the extraction step). The precision was evaluat-
ed as combined relative standard deviation (RSD) of both precision
tests.
Quantitation of target compounds based on peak areas was per-
formed using the internal calibration method. BE-d3 (100 μg L−1) was
used as IS to quantify both target compounds by a six-point analytic
curve (n = 3) constructed from each target compound. The limit of de-
tection (LOD) and the limit of quantification (LOQ) was determined
 I. Campestrini, W.F. Jardim / Science of the Total Environment 576 (2017) 374–380 377

using the linear regression of the analytical curve (Miller and Miller, Table 2
2005). COC and BE concentrations detected in surface water samples in some rivers in the São
Paulo State.

3. Results and discussion City River Concentration (ng L–1)

Sampling 1 Sampling 2
3.1. Performance of the method BE COC BE COC

Ribeirão do Pinhal watershed

For the development of method used, C18 rapid resolution cartridge, Limeira RPW1 (agriculture) 10 bLOQ n. c. n. c.
with 30 mm and 3.5 μm particle size was used, which provides cost- RPW2 (agriculture) bLOD bLOQ n. c. n. c.
effective and high-speed liquid chromatography analysis, associated RPW3 (agriculture) bLOQ bLOQ n. c. n. c.
with quality separation. For the purpose of this study, the Zorbax SB- RPW4 (agriculture) bLOQ bLOQ n. c. n. c.

C18 column has provided acceptable results, once the separation of PCJ watershed
the analytes was satisfactory. A gradient elution starting with polar mo- Campinas CpR (urban) 522 bLOQ 127 10
bile phase dragged out hydrophilic compounds that could interfere with AtR1 (urban) 292 8 70 8
AnR (urban) 655 12 305 13
the analysis. The method validation parameters (Table 1) showed to be Jaguariúna AtR2 (urban) n. c. n. c. bLOQ bLOQ
acceptable for both target compounds in both types of matrix. The SPE CdR (urban) 143 7 64 bLOQ
procedure using HLB cartridges was efficient, being consistent with pre- JaR2 (urban) n. c. n. c. 37 bLOQ
vious works (Maldaner et al., 2012; van Nuijs et al., 2011). However, the Cosmópolis JaR1 (urban) 95 8 n. c. n. c.
PiR (urban) n. c. n. c. bLOQ bLOQ
matrix effect observed as ion suppression led to a recovery lower than
Rio Claro CbR1 (urban) 284 40 n. c. n. c.
100% for both analytes, ranging from 70 to 96% for COC and BE, consid- Piracicaba CbR2 (urban) 117 6 57 10
ering all samples studied and the two concentration levels tested. BE-d3 PrR (urban) 1019 21 55 bLOD
was used as stable-isotope to quantify the analytes in the samples, Indaiatuba JuR (urban) n. c. n. c. 90 10
which was added to the samples before extraction step, and all subse- Mogi-Guaçu watershed
quent quantifications in water samples were performed using internal Mogi-Guaçu M-GR(urban) 47 bLOD 27 bLOD
calibration prepared in ultrapure water. The method precision was Mogi-Mirim M-MR(urban) 329 62 307 48
higher than 89%, considering all conditions tested. Espírito Santo do Pinhal CchR(rural) 10 bLOD bLOD bLOD

Analytical curves showed linearity between 0.5 and 150 μg L−1, with
a coefficient of linear correlation (r2) of 0.999 for both. LOD and LOQ of
the method were obtained taken into account the concentration factor
in the SPE step. LOD of 2 ng L− 1 was obtained for both compounds.
LOQ ranged from 6 ng L−1 and 5 ng L−1 for COC and BE, respectively.
When the drug concentrations exceeded the linear working range, the
concentration was estimated using a second calibration curve in the
concentration range of 150–1000 μg L−1 injected into LC-MS/MS (r2 of
0.998 for both BE and COC).
3.2. Occurrence in surface water
Table 2 shows the concentrations obtained for COC and BE in all sur-
face water samples in both sampling periods. Each sample was identi-
fied by the city, river code (Fig. 1) and watershed name. BE was
detected in concentrations above LOQ in 94% of the samples analyzed,
while COC in 85% of them. The concentration levels of BE (from 10 to
1019 ng L−1) were also higher than the concentrations obtained for
COC (6–62 ng L−1). These results are in accordance considering the ex-
cretion metabolism of COC and the stability of both compounds (COC
and BE) in the environment. After consumed, COC is excreted mainly
in the urine, and between 35 and 55% of the dose consumed is excreted
as BE, while only between 1 and 9% is excreted as COC (EMCDDA, 2008).
Stability studies show that COC is less stable than BE in different envi-
ronmental conditions (Postigo et al., 2011; van Nuijs et al., 2012;
Bijlsma et al., 2013; McCall et al., 2016). In aerobic and anaerobic
biofilms, and under natural conditions of pH and temperature, the
transformation of COC reaches 100%, while for BE this value is 80% at
the most. Hydrolysis of COC seems to be pH-dependent, and in treated
Guarapiranga Dam
São Paulo GD1 17 bLOQ 10 bLOQ
São Paulo GD2 179 12 127 bLOQ
São Paulo GD3 18 bLOQ bLOQ bLOQ
Sampling 1: February–March, 2014; Sampling 2: September October, 2014.
bLOD: Below of limit of detection; bLOQ: Below of limit of quantification; n. c.: not
collected.
sewage as well as in surface water samples, with typical pH around 7–
8, the remaining portion of COC is susceptible to hydrolysis. In these
conditions and at room temperature, BE is considered highly stable
(transformation b20%), while the stability of COC is considered to be
lower (60–100%) (McCall et al., 2016).
Highest concentrations of BE (522–1019 ng L−1) were observed in
samples from Capivari (CpR), Anhumas (AnR) and Piracicaba River
(PrR), in areas receiving high loads of sewage. In these surface water
samples, BE concentrations were higher than those typically found in
urban surface waters elsewhere. In rivers of Spain, Postigo et al.
(2010) found BE at concentrations ranges from 1.4–346 ng L−1, in
Italy BE levels ranged from 2 to 34 ng L−1 (Zuccato et al., 2008), and
in Belgium from 2 to 530 ng L−1 (van Nuijs et al., 2009). Higher COC
and BE levels were found by Mendoza et al. (2014), who detected up
to 823 ng L− 1 of both compounds in Manzanares river located in
Madrid (Spain). In Brazil, Thomas et al. (2014) detected COC and BE at
levels higher than concentration detected in this study, i.e., between
366 and 5896 ng L−1 in two Rio Negro tributaries (Igarapé do 40 and
Igarapé do Mindú) located in Manaus, Amazon, a shallow creeks that re-
ceive large amounts of untreated sewage; In the Rio Negro, COC and BE
were below LOQ. Also in Brazil, Locatelli (2011) found BE concentration
ranging from 5 to 170 ng L−1 in rivers forming the Atibaia watershed,

Table 1
Method performance parameters (SPE efficiency, matrix effect, % recoveries, and precision) for COC-d3 and BE-d3 at two concentrations levels in surface and drinking water.

Compound SPE efficiency (%) Matrix effect (%) % Recovery (%RSD)

Surface water Drinking water Surface water Drinking water Surface water Drinking water
−1 −1 −1
ng L ng L ng L

0.5 1000 0.5 1000 0.5 1000 0.5 1000 0.5 1000 0.5 1000

COC-d3 100 91 100 99 −25 −15 −22 −14 69 (8) 79 (2) 77(9) 80 (1)
BE-d3 90 91 99 98 −11 −8 −15 −10 97 (3) 85 (2) 85 (3) 96 (1)
378 I. Campestrini, W.F. Jardim / Science of the Total Environment 576 (2017) 374–380

also located in the São Paulo State, similar to those values detected in Table 3
samples collected in September–October 2014 (rainy season). COC and BE concentrations detected in the drinking water samples from the São Paulo
State.
Considering the 22 sampling points, 12 were sampled in both sea-
sons. Evaluating the concentrations of COC and BE detected in the sam- City River Concentration (ng L−1)
ples from these 12 locations, a seasonal variability could be observed. Sampling 1 Sampling 2
The highest concentrations of both compounds were found during Feb-
BE COC BE COC
ruary–March 2014, when a severe drought reached the southeast re-
PCJ watershed
gion of Brazil, including São Paulo State region. During February and
Limeira JaR1 62 bLOQ n. c. n. c.
March 2014, the flow rate of the sampled rivers was far below normal Campinas CpR 652 22 105 bLOQ
levels, and the Environmental State Agency of São Paulo reported high Campinas AbR 394 14 89 6
loads of organic matter and low concentrations of dissolved oxygen in Santa Bárbara do Oeste PiR 10 bLOQ n. c. n. c.
the studied catchment area (CETESB, 2015). The BE concentration in Piracicaba CbR2 n. c. n. c. 60 bLOQ

the AtR1, sampled within the Campinas city limits (292 ng L−1), was
around four times higher than the level found in samples collected dur-
ing September 2014 (70 ng L−1), when a better quality scenario was ob-
served. In the PrR, a sample collected in March 2014 showed the highest
concentration of BE (N1000 ng L−1) among all samples analyzed. In the
same period, CETESB recorded one of the lowest values of dissolved ox-
ygen in this water body.
Spatial variability in the concentration of COC and BE were also ob-
served. The highest concentrations were found in samples collected
from cities with high population density, such as Campinas or
Piracicaba. Only one of the samples collected in the basin of Ribeirão
do Pinhal, which flows through small towns, BE could be quantified.
However, the concentration detected was approximately hundred
times lower compared to concentrations detect in samples collected
near urban areas. In Mogi-Guaçu watershed, along with GD1 and GD3
sampling points collected in São Paulo city, BE concentrations were
lower than 100 ng L−1, except M-MR sample, which showed 329 ng
L−1 of BE. COC and BE concentrations above 100 ng L−1 were measured
at GD2 (Guarapiranga dam), as a result of the highest degree of human
impact at this sampling site, associated with inputs of domestic sewage
in this area. This variability in concentrations observed according to the
region may be associated not only to differences in the prevalence of
drug use, but also shows the problem of dumping raw or partially treat-
ed sewage in these water bodies. The environmental contamination re-
vealed in this study by the presence of BE tends to aggravate during long
periods of drought, since water scarcity decreases the river flow, a de-
crease that is not proportionally followed by the sewer discharge.
Due to some physicochemical properties of COC and BE and the pH of
the sample analyzed (pH ranged between 6.4 and 7.9 considering all sam-
ples), these compounds are expected to be freely dissolved in surface wa-
ters. Sorption onto the organic matter is pH dependent and in the
conditions found in natural waters, COC (pKa 8.6) and BE (pKa 10.1/3.6)
are predominately in the highly polar form. Studies evaluating the occur-
rence of COC and BE in the aqueous phase and sediment of sewage sam-
ples showed the adsorption of COC in the sediment reaching only 1.4 to
2.0%, and b 0.01% for BE (Domènech, 2009), justifying the high concentra-
tions detected in the aqueous phase in surface waters.
A Hazard Quotients (HQ) approach can be used to estimate the envi-
ronmental risk of a compound present in natural waters. This approach is
defined as the ratio between the measured environmental concentration
(MEC) of a target substance and its chronic toxicity, based on the no ob-
served effect concentration (NOEC) or the predicted non-effect concen-
tration (PNEC). In the lack of chronic data for COC and BE, Mendoza
et al. (2014) derived the PNEC for COC (2.28 10−3 mg L−1) and BE
(6.81 mg L−1) via ECOSAR, applying an assessment factor (AF) of 1000.
Using these derived PNEC values against the highest MEC found in
this study, the results obtained for COC (b0.1) and BE (≪0.1) indicate
no risk to aquatic organism. However, this statement should be taken
with caution, since that in a real scenario one cannot exclude long-
term exposure and a possible synergistic association with other psycho-
active substances present in the ecosystems studied.
Few other papers addressed the toxicological effects of illicit drugs
occurring at environmental concentrations to aquatic organism. Binelli
et al. (2012) observed that COC exposition to zebra mussels (Dreissena
Mogi Guaçu watershed
Espírito Santo do Pinhal CchR bLOQ bLOQ n. c. n. c.
Sampling 1: February–March, 2014; Sampling 2: September–October, 2014.
bLOQ: Below of limit of quantification; n. c.: not collected.
polymorpha), at 40, 220 ng L−1 and 10 μg L−1 caused sub-lethal effects
after three days of exposition, even at the lowest concentration tested.
DNA damage, inactivation of defensive compounds release and oxida-
tive stress were some of the effects observed. Capaldo et al. (2012) de-
tected BE in eel tissue (a vertebrate aquatic organism), and Gay et al.
(2013) showed that when eels were exposed to COC at 20 ng L−1 by
30 days, impairs in their dopamine brain catecholamines and pituitary
activity were observed. Once COC is hydrolyzed to BE in pH found in
natural waters, it becomes evident that the effects could also be associ-
ated to BE exposition, even as an inactive metabolite of COC, as pointed
out by Binelli et al. (2013). Parolini et al. (2016) study the zebra mussel
specimens exposed for 14 days to an illicit drug mixture at environmen-
tal levels, including COC (50 ng L−1) and BE (300 ng L−1), and observed
a significant increase of genotoxicity.
These experimental toxicity studies carried out using freshwater
non-target organism exposed to COC and BE at environmental levels
suggest that is plausible to consider a potential negative impact on the
aquatic ecosystem at concentrations detected in the samples investigat-
ed in this study.
3.3. Occurrence in drinking water
Concentrations of COC and BE detected in drinking water samples are
listed in Table 3. BE concentrations ranged from 10 to 652 ng L−1, and
COC levels from 6 to 22 ng L−1. In general, the results obtained for drink-
ing water samples were consistent with those observed in surface waters:
COC concentrations were always lower than BE concentrations. The
highest concentrations of both compounds were observed in the samples
collected between February and March 2014, and in drinking water pro-
duced from rivers with higher concentrations of COC and BE.
Also, levels detected in drinking water were, in general, similar to
those detected in surface water samples (Fig. 2), showing that the
Fig. 2. COC and BE (ng L−1) in the surface and drinking water from Capivari, Atibaia,
Jaguari and Cachoeira rivers. Samples were collected in the dry season, between
February and March of 2014 (sampling 1).
 I. Campestrini, W.F. Jardim / Science of the Total Environment 576 (2017) 374–380
poor quality of source water can persist into the public supply system
due to a partial removal in DWTP. A more realistic study of the removal
efficiency of these compounds by conventional DWTP is necessary to
subsidize strategies to improve DWTP performance; although surface
and drinking water samples (Fig. 2) were collected in the de same
river, the hydraulic retention time in the DWTP and the distribution net-
work has not been considered. At this moment, it is reasonable to affirm
there is no potential risk to human health from this indirect exposition
of the illicit drug; the levels found are much smaller than a minimal dose
consumed by a drug user (100 mg), even considering a daily water in-
take of 2 L.
In order to assess a more consistent risk evaluation of these sub-
stances found in drinking water, a human risk characterization was un-
dertaken following Schwab et al. (2005) and Mendoza et al. (2014). HQ
was calculated as EDI/ADI, where EDI (estimated daily intake) was cal-
culated taking the concentration found in the drinking water multiplied
by the ingestion rate, divided by the body weight. ADI (acceptable daily
intake) was calculated dividing the average daily dose by an AF that in-
cludes uncertainty factors to assure that no effect will occur for human
health. For COC, it is a concern to use 100 mg as the average dose. Once a
lower dose can impair human health, even though it is not enough to
provide psychoactive action, the average daily dose of 100 mg day−1
was used, divided by an AF of 10,000. This AF of 1000 considers an un-
certain value of 10 related to intra-species variability in humans, of 10
for different sensitive subpopulations (i.e., pregnant women, elderly,
and children), of 10 considering an extrapolation from no-observable-
adverse-effect-level (NOEC), as proposed by Mendoza et al. (2014). It
is recommended use an uncertainty of 10 for each factor when no rele-
vant data is available (Schwab et al., 2005). This calculation also consid-
ered 70 kg as the average weight and a daily water intake of 2 L for a
standard European adult (EFSA, 2012).
Considering the highest COC concentration detected in the drinking
water samples analyzed (22 ng L−1), the calculated HQ was 4.4 × 10−4,
which is smaller than 0.1, suggesting no toxic effect at the levels found
(Mendoza et al., 2014). Even considering BE concentrations detected
in the drinking water samples around 30 times higher than COC, the
HQ would be smaller than 0.1.
Considering that toxicity data of COC and BE intake via drinking
water are scarce, this exercise has to be considered as a very preliminary
and rough estimate of the risk to humans. Moreover, considering that
this contamination tends to persist or even worsen in the case of sanita-
tion deteriorates chronic exposure to illicit drugs and effects of mixtures
of different active substances may pose a real risk to the health of indi-
viduals. Studies using aquatic vertebrates have shown that the effects of
COC and BE are more pronounced at longer exposure time (Binelli et al.,
2012; Gay et al., 2013). In humans, Sofuoglu et al. (2001) showed that
single intravenous administration of 0.46 mg kg− 1 of COC increased
plasma norepinephrine and epinephrine concentrations in male COC
users. This dose was around three times smaller than the minimal
daily dose of COC.
3.4. BE as an indicator of wastewater contamination in natural waters
The improvement of sensitive of analytical techniques made possi-
ble to detect emerging organic compounds at low concentrations in en-
vironmental samples. Caffeine has already been used as a reliable
indicator of sewage contamination and able to predict estrogenic activ-
ity in natural waters (Montagner et al., 2014). Other contaminants of
emerging concern, such as estradiol, triazine, and carbamazepine,
have been studied as possible indicators of pollution and environmental
risk for better management and complementary strategies in watershed
management (Naidu et al., 2016; Daneshvar et al., 2012). This study
suggests that BE is a candidate to be used as a reliable biomarker of psy-
choactive substances such as illicit drugs in both surface and drinking
water. Many of the emerging contaminants present high hydrophilicity,
thus posing a challenge to attain high levels of removal during the
379
conventional processes of sewage and drinking water treatment. The ef-
ficient removal of COC and BE depends on the wastewater treatment
process applied, but in general with poor efficiency, especially for the
metabolite (Postigo et al., 2010; Kasprzyk-Hordern et al., 2009;
Campestrini, 2015).
Once in the environment, COC can be rapidly and totally degraded,
mainly by hydrolysis, whereas BE can persist in the environment even
after 15 days from the discharge (McCall et al., 2016). Due to this envi-
ronmental stability, BE has been detected in surface waters even in
countries with high levels of sanitation, as well in drinking water sam-
ples (Berset et al., 2010; Mendoza et al., 2014). Furthermore, high con-
centrations and frequency of detection of BE in raw sewage have been
used to calculate COC consumption by a population in many sewage-
based epidemiology applications (van Nuijs et al., 2011; van Wel et al.,
2016).
4. Conclusion
This a pioneer work focused in the occurrence of illicit drugs in both
surface and drinking waters in the region of the São Paulo State, Brazil,
embracing a water system supplying approximately 10 million people
(considering the rivers and dam investigated). The results obtained
clearly reflect the degrading situation concerning basic sanitation as
well as the social problem related to illicit drug use. COC and BE levels
detected in surface water samples were between 10 and 100 times
higher than the ones found in other countries such as Italy and Spain,
pointing to the fact that this environmental problem might worsen if
the low efficiency and coverage of the sanitary system persist over the
years, especially over long periods of drought. A matter of concern is
that this study area is inserted in the Brazilian region holding the
highest rate of collection and sewage treatment. Although the concen-
trations of both compounds found in surface water do not pose an ap-
parent risk to the aquatic organism based on HQ, experimental
toxicological studies have shown negative effects on the maintenance
of the aquatic life at low ng L−1 concentrations of COC and BE. It
opens space for a more thorough discussion on the risks related to
long-term exposure and possible synergy of psychoactive substances
mixtures exposure even at very low levels. The same analysis is valid
for human health; the occurrence of BE in the levels found in drinking
water warns about the indirect chronic exposure to multiple psychoac-
tive compounds, even at levels much lower than the effective dose. Fur-
ther work on toxicology is necessary to understand the effects of the
exposure to COC and metabolites to predict the associated risk better
using appropriate values of uncertainty factors. Finally, BE seems to be
a powerful candidate to be used as an indicator of sewage contamina-
tion in the screening process of other emergent contaminants originat-
ed from human use. Future toxicity studies may contribute to the
establishment of BE limits concentrations in natural waters in an at-
tempt to ensure the safety of all forms of life.
Supplementary data to this article can be found online at doi:10.
1016/j.scitotenv.2016.10.089.
Acknowledgement
The authors would like to thank the INCTAA (FAPESP, proc. 2008/
57808-1 and CNPq, proc. 573894/2008-6 for financial support to this
study. I.C. is grateful to IQ-UNICAMP, and CAPES for the Ph.D. fellowship.
References
Berset, J.-D., Brenneisen, R., Mathieu, C., 2010. Analysis of licit and illicit drugs in waste,
surface and lake water samples using large volume direct injection high-
performance liquid chromatography – electrospray tandem mass spectrometry
(HPLC–MS/MS). Chemosphere 81, 859–866.
Bijlsma, L., Boix, C., Niessen, W.M.A., Ibáñez, M., Sancho, J.V., 2013. Investigation of degra-
dation products of cocaine and benzoylecgonine in the aquatic environment. Sci.
Total Environ. 443, 200–2008.
380 I. Campestrini, W.F. Jardim / Science of the Total Environment 576 (2017) 374–380
Binelli, A., Pedriali, A., Riva, C., Parolini, M., 2012. Illicit drugs as new environmental pol-
lutants: cyto-genotoxic effects of cocaine on the biological model Dreissena
polymorpha. Chemosphere 86, 906–911.
Binelli, A., Marisa, I., Fedorova, M., Hoffman, R., Riva, C., 2013. First evidence of protein
profile alteration due to the main cocaine metabolite (benzoylecgonine) in a fresh-
water biological model. Aquat. Toxicol. 268, 140–141.
Boleda, M.R., Huerta-Fontela, M., Ventura, F., Galceran, T.M., 2011. Evaluation of the pres-
ence of drugs of abuse in tap waters. Chemosphere 84, 1601–1607.
Brasil. Ministério das Cidades. Secretaria Nacional de Saneamento Ambiental – SNSA,
2016a. Sistema Nacional de Informações Sobre Saneamento: Diagnóstico dos Serviços
de Água e Esgotos – 2014. Brasília: SNSA/MCIDADES.
Campestrini, I., 2015. Determination of Cocaine and Metabolites in Users' Urine Samples
and Environmental Samples. (Ph.D. Thesis). University of Campinas:p. 147 (http://
www.bibliotecadigital.unicamp.br/document/?code=000952501&opt=4).
Capaldo, A., Gay, F., Maddaloni, M., Valiante, S., Falco, M., Lenzi, M., Laforgia, V., 2012. Pres-
ence of cocaine in the tissues of the European Eel, Anguilla anguilla, exposed to envi-
ronmental cocaine concentrations. Water Air Soil Pollut. 223, 2137–2143.
Carreira, R.S., Albergaria-Barbosa, A.C.R., Arguelho, M.L.P.M., Garcia, C.A.B., 2015. Evidence
of sewage input to inner shelf sediments in the NE coast of Brazil obtained by molec-
ular markers distribution. Mar. Pollut. Bull. 90, 312–316.
CETESB (São Paulo), 2015. Qualidade das águas superficiais no estado de São Paulo, 2014.
CETESB, São Paulo: CETESB. http://aguasinteriores.cetesb.sp.gov.br/wp-content/
uploads/sites/32/2013/11/agua-doce-parte1-corrigido.pdf.
Daneshvar, A., Aboulfadl, K., Viglino, L., Broséus, R., Sauvé, S., Madoux-Humery, A.S.,
Weyhenmeyer, G.A., Prévost, M., 2012. Evaluating pharmaceuticals and caffeine as in-
dicators of fecal contamination in drinking water sources of the Greater Montreal re-
gion. Chemosphere 88, 131–139.
Domènech, X., Peral, J., Muñoz, I., 2009. Predicted environmental concentrations of co-
caine and benzoylecgonine in a model environmental system. Water Res. 43,
5236–5242.
EFSA, 2012. Guidance on selected default values to be used by the European Food Safety
Authority (EFSA) Scientific Committee, Scientific Panels and Units in the absence of
actual measured data. EFSA J. 3.
European Monitoring Center for Drugs and Drug Addiction – EMCDDA Insights Series No
9, 2008E. Assessing Illicit Drugs in Sewage: Potential and Limitations of a New Mon-
itoring Approach. Project Leaders: Frost, N.; Griffiths, Paul. Luxembourg.
Frena, M., Bataglion, G.A., Tonietto, A.E., Eberlin, M.N., Alexandre, M.R., Madureira, L.A.S.,
2016. Assessment of anthropogenic contamination with sterol markers in surface
sediments of a tropical estuary (Itajaí-Açu, Brazil). Sci. Total Environ. 544, 432–438.
Gay, F., Maddaloni, M., Valiante, S., Laforgia, V., Capaldo, A., 2013. Endocrine disruption in
the European Eel, Anguilla anguilla, exposed to an environmental cocaine concentra-
tion. Water Air Soil Pollut. 224, 1579–1591.
Kasprzyk-Hordern, B., Dinsdale, R., Guwy, A., 2009. The removal of pharmaceuticals, per-
sonal care products, endocrine disruptors and illicit drugs during wastewater treat-
ment and its impact on the quality of receiving waters. Water Res. 43, 363–380.
Locatelli, M.A.F., 2011. Assessment of antibiotics and illicit drugs in the Atibaia River wa-
tershed. (Ph.D. Thesis). University of Campinas:p. 164 (http://www.bibliotecadigital.
unicamp.br/document/?code=000840148&opt=4).
Maldaner, A.O., Schmidt, L.L., Locatelli, M.A.F., Jardim, W.F., Sodré, F.F., Almeida, F.V.,
Pereira, C.E.B., Silva, C.M., 2012. Estimating cocaine consumption in the Brazilian Fed-
eral District (FD) by sewage analysis. J. Braz. Chem. Soc. 23, 861–867.
Martins, C.C., Cabral, A.C., Barbosa-Cintra, S.C.T., Dauner, A.L.L., Souza, F., 2014. An inte-
grated evaluation of molecular marker indices and linear alkylbenzenes (LABs) to
measure sewage input in a subtropical estuary (Babitonga Bay, Brazil). Environ.
Pollut. 188, 71–80.
McCall, A.K., Bade, R., Kinyua, J., Lai, F.Y., Thai, P.K., Covaci, A., Bijlsma, L., van Nuijs, A.L.N.,
Ort, C., 2016. Critical review on the stability of illicit drugs in sewers and wastewater
samples. Water Res. 88, 933–947.
Mendoza, A., Rodríguez-Gil, J.L., González-Alonso, S., Mastroianni, N., López de Alda, M.,
Barceló, D., Valcárcel, Y., 2014. Drugs of abuse and benzodiazepines in the Madrid Re-
gion (Central Spain): seasonal variation in river waters, occurrence in tap water and
potential environmental and human risk. Environ. Int. 70, 76–87.
Miller, J.C., Miller, J.N., 2005. Statistics and Chemometrics for Analytical Chemistry. 5th ed.
Pearson, London (ch. 5).
Montagner, C.C., Jardim, W.F., 2011. Spatial and seasonal variations of pharmaceuticals
and endocrine disruptors in the Atibaia River, São Paulo State (Brazil). J. Braz.
Chem. Soc. 22, 1452.
Montagner, C.C., Umbuzeiro, G.A., Pasquini, C., Jardim, W.F., 2014. Caffeine as an indicator
of estrogenic activity in source water. Environ. Sci. Process. Impacts 16, 1866–1869.
Naidu, R., Espana, V.A.A., Liu, Y., Jit, J., 2016. Emerging contaminants in the environment:
risk-based analysis for better management. Chemosphere 154, 350–357.
Niessen, W.M.A., Manini, P., Andreoli, R., 2006. Matrix effects in quantitative pesticide
analysis using liquid chromatography-mass spectrometry. Mass Spectrom. Rev. 25
(6), 881–899.
Parolini, M., Magni, S., Castiglioni, S., Binelli, A., 2016. Genotoxic effects induced by the ex-
posure to an environmental mixture of illicit drugs to the zebra mussel. Ecotoxicol.
Environ. Saf. 132, 26–30.
Postigo, C., López de Alda, M.J., Barceló, D., 2010. Drugs of abuse and their metabolites in
the Ebro River basin: occurrence in sewage and surface water, sewage treatment
plants removal efficiency, and collective drug usage estimation. Environ. Int. 36,
75–84.
Postigo, C., Sartori, C., Oller, I., Malato, S., Maldonado, M.I., Alda, M.L., Barceló, D., 2011.
Solar transformation and photocatalytic treatment of cocaine in water: kinetics, char-
acterization of major intermediate products and toxicity evaluation. Appl. Catal. B En-
viron. 104, 37–48.
SABESP (São Paulo) - Companhia de Saneamento Básico do Estado de São Paulo, 2014a.
Relatório de Sustentabilidade. http://site.sabesp.com.br/uploads/file/sociedade_
meioamb/rs_2014.pdf.
Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M., Cragin, D., Meyerhoff,
R.D., D'Aco, V.J., Anderson, P.D., 2005. Human pharmaceuticals in US surface waters:
a human health risk assessment. Regul. Toxicol. Pharmacol. 42, 296–312.
Sodré, F.F., Locatelli, M.A.F., Jardim, W.F., 2010. Occurrence of emerging contaminants in
Brazilian drinking waters: a sewage-to-tap issue. Water Air Soil Pollut. 206, 57–67.
Sofuoglu, M., Nelson, D., Babb, D.A., Hatsukami, D.K., 2001. Intravenous cocaine increases
plasma epinephrine and norepinephrine in humans. Pharmacol. Biochem. Behav. 68,
455–459.
Thomas, K.V., Bijlsma, L., Castiglioni, S., Covaci, A., Emke, E., Grabic, R., Hernández, F.,
Karolak, S., Kasprzyk-Hordern, B., Lindberg, R.H., López de Alda, M.J., Meierjohann,
A., Ort, C., Pico, Y., Quintana, J.B., Reid, M., Rieckermann, L., Terzic, S., van Nuijs,
A.L.N., de Voogt, P., 2012. Comparing illicit drug use in 19 European cities through
sewage analysis. J. Sci. Total Environ. 432, 432–439.
Thomas, K.V., Silva, F.M.A., Langford, K.K., Souza, A.D.L., Nizzeto, L., Waichman, A.V., 2014.
Screening for selected human pharmaceuticals and cocaine in the urban streams of
Manaus, Amazonas, Brazil. J. Am. Water Resour. Assoc. 50 (2), 302–308.
UNODC - United Nations Office on Drugs and Crime, 2015. Word Drugs Report. http://
www.unodc.org/wdr2015.
UNODC - United Nations Office on Drugs and Crime, 2016. Word Drugs Report. http://
www.unodc.org/wdr201.
van Der Aa, M., Bijlsma, L., Emke, E., Dijikman, E., Van Nuijs, A.L.N., Van de Vem, B.,
Hernández, F., Versteegh, A., De Voogt, P., 2013. Risk assessment for drugs of abuse
in the Dutch water cycle. Water Res. 47, 1848–1857.
van Nuijs, A.L.N., Pecceu, B., Theunis, L., Dubois, N., Charlier, C., Jorens, P.G., Bervoets, L.,
Blust, R., Neels, H., Covaci, A., 2009. Cocaine and metabolites in waste and surface
water across Belgium. Environ. Pollut. 157, 123–129.
van Nuijs, A.L.N., Castiglioni, S., Tarcomnicu, I., Postigo, C., Lopez de Alda, M.J., Neels, H.,
Zuccato, E., Barcelo, D., Covaci, A., 2011. Illicit drug consumption estimations derived
from wastewater analysis: a critical review. Sci. Total Environ. 409, 3564–3577.
van Nuijs, A.L.N., Abdellati, K., Bervoets, L., Blust, R., Jorens, P.G., Neels, H., Covaci, A., 2012.
The stability of illicit drugs and metabolites in wastewater, an important issue for
sewage epidemiology? J. Hazard. Mater. 239, 19–23.
van Wel, J.H.P., Kinyua, J., van Nuijs, A.L.N., Salvatore, S., Bramness, J.G., Covaci, A., Van Hal,
G.A., 2016. Comparison between wastewater-based drug data and an illicit drug use
survey in a selected community. Int. J. Drug Policy 34, 20–26.
Zuccato, E., Castiglioni, S., Bagnati, R., Chiabrando, C., Grassi, P., Fanelli, R., 2008. Illicit
drugs, a novel group of environmental contaminants. Water Res. 42, 961–968.