If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Pollution from the production of pharmaceuticals has usually been considered to be negligible in comparison to emissions due to the consumption of medicines. In some contrast to this perception, recently published articles have highlighted drug manufacturing as a major contributor to local pharmaceutical pollution 1–7. The Patancheru area around Hyderabad, India, is well known for its large-scale production of bulk drugs, serving the global pharmaceutical market 5. A central treatment plant in Patancheru receives process water from approximately 90 different manufacturers. The treated effluent contains a complex mixture of many different active pharmaceutical ingredients (APIs) up to milligram-per-liter levels 1, 4. The industrial activities have led to unprecedented pharmaceutical contamination of surface, ground, and drinking water in the region 4. Some of the APIs found at high concentrations in the effluent belong to the following classes: fluoroquinolone antibiotics, angiotensin II receptor antagonists, angiotensin-converting enzyme inhibitors, β1-adrenoreceptor antagonists, H1- and H2-receptor antagonists, selective serotonin reuptake inhibitors, folic acid synthesis inhibitors, and antimycotics. The effluent has previously been screened for approximately 200 other organic chemicals and metals, but the pharmaceuticals clearly stand out as the most abundant contaminants investigated 8. Detailed information about other effluent characteristics has previously been published 1.
Standard toxicity tests on species commonly used in environmental risk assessments have shown that the effluent is toxic to bioluminescent bacteria (Aliivibrio fischeri), lettuce (Lactuca sativa), and water fleas (Daphnia magna) at concentrations ranging from 1.6 to 35% effluent 1. In addition, for zebrafish embryos (Danio rerio) 144 h postfertilization, the median lethal concentration values varied between 2.7 and 8.1% effluent 8. A considerably lower effluent concentration (0.2%) resulted in a 40% growth reduction in developing frog tadpoles (Xenopus tropicalis) 8. Exposure of rainbow trout (Oncorhynchus mykiss) to the same low concentration of effluent also had marked effects on global hepatic gene transcription, elevation of blood plasma phosphate and cholesterol levels, and increased Cyp1a activity in the liver 9. High dilutions of the same effluent have also been shown to induce different cyp1 mRNA expressions in various tissues in the three-spined stickleback (Gasterosteus aculeatus) 10. Another recent study revealed high levels of resistance genes for several different classes of antibiotics in river sediments sampled up- and downstream from the Indian treatment plant. Furthermore, a high prevalence of genetic elements associated with mobility and exchange of resistance genes between bacteria was also identified in the antibiotic-contaminated sediment samples 11.
To the best of our knowledge, peer-reviewed publications investigating the health of people living in the polluted area are lacking, but a report by Greenpeace suggests an increased prevalence of a series of disorders (including diseases within the respiratory, nervous, and circulatory systems; congenital malformations; and cancer) among local villagers as well as unexplained deaths of livestock 12. Indeed, cattle drink river water contaminated by effluent. Also, people have complained about rashes after bathing in effluent-contaminated river water 13. Evidently, the risks of acute effects on humans and livestock associated with exposure to effluent-contaminated water need to be investigated. The present study addresses the potential effects from short-term oral exposure, using the rat as a model species for humans and livestock. Despite the limited amounts of individual pharmaceutical compounds ingested via water, mixture effects from the complex effluent are conceivable. We have therefore combined traditional clinical blood chemistry assays with analysis of global hepatic mRNA expression using microarrays, a highly sensitive and exploratory research approach, which may give clues about the mechanism(s) behind the potential toxicity of the complex effluent. The complexity of the mixture makes it difficult to theoretically estimate the bioavailability and uptake of drug residues from the effluent; in a parallel exposure experiment, the actual concentrations of 97 different pharmaceutical compounds were measured in rat blood serum at two different time points (1 and 24 h, respectively) after the final exposure.
MATERIALS AND METHODS
The chemical characteristics of the effluent and descriptions of the treatment plant and water-sampling procedures were previously published 1, 8. The effluent used in the present study was sampled on November 7, 2006. Samples from the same sampling occasion have previously been used in the studies of Larsson et al. 1, Carlsson et al. 8, Gunnarsson et al. 9 and Beijer et al. 10. Effluent samples were frozen on dry ice in India and kept frozen until pharmaceutical concentration analysis and exposure experiments.
Experimental animals and exposure experiments
The present study includes results from two separate exposure experiments. Biological effects from acute oral exposure to undiluted treated effluent were studied, partly through analysis of clinical blood parameters that serve as markers for functional disturbances of several organs of the rat but also by measuring the potentially more rapid but subtle responses affecting liver mRNA expression. In the other exposure experiment, pharmaceutical concentrations in rat blood serum at the end of the exposure were determined, with the aim of comparing them with human therapeutic blood serum levels.
For both exposure experiments, male Sprague-Dawley rats, aged five to six weeks, were purchased from Taconic. Animals were left to acclimatize for at least 7 d prior to exposure. During this period, as well as during the whole experiments, rats had free access to food and water and were kept in a controlled environment with a 12:12-h photoperiod. The experiments were approved by the Gothenburg Animal Research Ethical Committee.
Biological effect experiment
Twenty rats were caged in pairs and assigned to either exposed or control treatment. Once daily, rats were individually weighed and tube-fed 1.5 ml of effluent or tap water according to our animal ethics permit. As the effluent has an unpleasant smell and is probably distasteful for the rats, providing effluent as the only water source was not an option. By tube-feeding the rats, all animals were ensured to receive an equal volume of effluent (or tap water). The rats' average body weight at the beginning of the study was 210.7 g (± standard deviation 22.5 g). In accordance with the previous trout study, the exposure experiment went on for 5 d. Indeed, genes are often rapidly regulated following chemical exposure, thus allowing responses to be identified before more serious effects occur. The reasons for specifically choosing a 5-d exposure period were threefold: (1) it was due to a limited access to effluent; (2) it would allow some comparisons to be made with the 5-d fish exposure experiment, where clear responses were found 9; and (3) it allowed comparison with a database on short-term gene responses in rats exposed to hundreds of different pharmaceuticals 14. However, due to the lack of clearly regulated genes after effluent exposure, as well as several limitations with the database, the latter comparison eventually turned out not to be useful.
On the sixth day, the animals were killed using carbon dioxide, followed by cutting the heart. A piece of the median liver lobe was taken, snap-frozen in liquid nitrogen, and later used for microarray and quantitative polymerase chain reaction (qPCR) analyses. Blood was obtained from the vena cava, and serum samples were stored at −20°C until further clinical chemistry analysis.
Clinical chemistry analysis of rat blood serum
Blood serum from 10 exposed and eight control rats in the biological effect study were analyzed at the Institution for Clinical Chemistry at the Sahlgrenska University hospital (sera from two control rats were excluded from the analysis due to hemolysis). The concentrations of 10 metabolites (albumin, chloride, cholesterol, creatinine, phosphate, potassium, sodium, urea, uric acid, and total protein) and the activities of five enzymes (alanine aminotransferase, alkaline phosphatase, aspartate aminotransferase, creatinine kinase, total lactate dehydrogenase) were measured with standard protocols for blood sample screenings (Roche Diagnostics). Significant differences in metabolite concentrations and/or enzyme activities between the treatment groups were determined using two-sided Student's t tests.
Liver samples were removed from storage in liquid nitrogen and homogenized using the TissueLyser (Qiagen). Total RNA was extracted with the RNeasyPlus Mini Kit (Qiagen). The RNA quality and quantity were assessed by spectrophotometric measurements (Nanodrop 1000; NanoDrop Technologies), and potential RNA degradation was measured with the RNA StdSens Analysis Kit and Experion (Experion Electrophoresis Station; Bio-Rad Laboratories).
Global hepatic gene transcription was analyzed in five exposed and five control rats by microarray experiments performed at the Swegene Centre for Integrative Biology at Lund University, Lund, Sweden. In short, 5 µg total RNA was processed following the GeneChip Expression 3′-Amplification Reagents One-cycle cDNA synthesis kit instructions (Affymetrix) to produce double-stranded cDNA. This was used as a template to generate biotin-targeted cRNA following the manufacturer's specifications. Fifteen micrograms of the biotin-labeled cRNA were fragmented to strands between 35 and 200 bases in length, 10 µg of which were hybridized onto the GeneChip Rat Genome 230 2.0 Array overnight in the GeneChip Hybridisation 6400 oven using standard procedures. Arrays were washed and then stained in a GeneChip Fluidics Station 450. Scanning was carried out with the GeneChip Scanner 3000, and images were analyzed using GeneChip Operating Software. The microarray data were analyzed in the statistical language R 2.12.1 (http://www.r-project.org) using the Bioconductor package 15. Preprocessing was performed using robust multiarray average from the rma package 16. Differences in mRNA expression were then identified by the moderated t statistic implemented in the LIMMA package 17. The Benjamini-Hochbergs false discovery rate was used to estimate the proportion of false positives 18. All microarray data comply with the Minimum Information about a Microarray Experiment guidelines, and the raw data have been submitted to the Gene Expression Omnibus, accession number GSE23983.
Total RNA samples (1 µg) from all 20 rats from the effect study were reverse-transcribed with a mixture of random hexamers and oligo (dT) primers, using the iScript cDNA Synthesis Kit (Bio-Rad). The cDNA synthesis was performed according to the manufacturer's instructions. Template cDNA corresponding to 20 ng total RNA, TaqMan Gene Expression Mastermix (Applied Biosystems), and preformulated primers and probes purchased as TaqMan Gene expression assays (Applied Biosystems) were mixed and run in accordance with the manufacturer's specifications. Amplification reactions were carried out in triplicate with a 2-min initial hold at 50°C and a 10-min denaturation step at 95°C, followed by 40 cycles of 15 s at 95°C and 1 min at 60°C. All amplification reactions were carried out in 384-well plates using the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems) and analyzed with SDS software 2.1 (Applied Biosystems). Two reference genes, β-actin and β2-microglobulin, were selected for the qPCR assay, and both had stable mRNA expression levels between the treatment groups (p = 0.32 and p = 0.62, respectively, two-sided Student's t test). Primer sequences and accession numbers are presented in Supplemental Data, Table S1.
The qPCR assay results were analyzed by normalizing the median Ct value of the three technical replicates with the average of the median Ct values from the two reference genes. The resulting ΔCt values were used for the statistical analyses where tests for differential expression between control and exposed rats were performed using a one-sided Student's t test.
Calculations of rat API dosage
The administered dose of nine of the residual APIs found in the highest concentrations in the effluent was calculated using the average rat weight over the whole biological effect study (226 g) and chemical characterization analyses of the effluent 1 (Table 1). The dosage was then compared with the lowest human defined daily dose (DDD) data found at the World Health Organization's Collaborating Centre for Drug Statistics Methodology (http://www.whocc.no/atcddd/) divided by 70 kg. Two of the fluoroquinolone substances, lomefloxacin and enrofloxacin, were excluded from the calculations due to lack of available DDD data; enrofloxacin was excluded also because it is a veterinary medicine. The results of the estimation showed that rats were exposed to less than 6.1% of the human DDD (adjusted for body weight) for any of the nine APIs. For four of the compounds (cetirizine, losartan, citalopram, and ciprofloxacin) the estimated rat dose exceeded 1% of the adjusted DDD.
Table 1. Theoretical doses of nine active pharmaceutical ingredients for rats orally exposed to industrial effluent compared to human defined daily doses (DDDs) per kilogram body weighta
Concentration in effluent (ng/ml)
Rat daily dose (mg/kg)
Rat dose related to DDD/kg (%)
Calculations based on pharmaceutical concentrations in effluent presented in Larsson et al. 1.
Angiotensin II receptor antagonist
Serotonin reuptake inhibitor
Study of pharmaceutical concentrations in rat blood serum
Fifty-five rats were randomly divided into groups of five and assigned to one of five different treatment groups (two cages per treatment) in addition to one control group. Once a day, rats were weighed and then tube-fed 2 ml of either (1) one of four premade single-substance solutions or (2) effluent from the Indian treatment plant. Rats in the control group were tube-fed an equal volume of tap water. The substances in the prepared single-substance solutions were cetirizine, losartan, citalopram, and ciprofloxacin, which were the four pharmaceuticals detected in the effluent that corresponded to the highest rat dosage compared with the human DDD per kilogram body weight. To ensure that the concentrations of pharmaceutical compounds had not decreased significantly since the effluent sampling and to enable comparisons with the previous biological effects study, the effluent was reanalyzed with respect to the concentrations of 97 different pharmaceuticals and the four single-substance solutions were prepared to match the concentration of the corresponding compound in the effluent as closely as possible (Table 2). The rats' average body weight at the start of the study was 247.0 g (± standard deviation 11.6 g).
Table 2. Results from the reanalysis of seven pharmaceutical concentrations in effluent (October 2011): Human therapeutic plasma concentrations (HTPC) and drug levels detected in rat blood serum at two different time points after the final tube-feeding with contaminated water are compared
Drug concentration in effluent (ng/ml)
Drug concentration in rat serum (ng/ml)
Maximum drug concentration in rat serum/HTPC (%)
ND = not detected; NA = not applicable.
8 × 10–4
Approximately 1 h after the final tube-feeding on the fifth and last day of the experiment, one-half of the animals within each treatment group were killed using carbon dioxide, followed by cutting the heart. The rest of the animals were killed in the same way on day 6, approximately 24 h after the final dose of effluent. From all 55 rats, blood was obtained and the serum stored at –20°C and later analyzed with respect to pharmaceutical concentrations.
Chemical analysis of drug concentrations in rat blood serum
All 97 pharmaceutical reference standards were classified as analytical grade (>98%). The 2H6-amitriptyline, 2H10-carbamazepine, 13CN-ciprofloxacin, 2H5-fluoxetine, 13C6-sulfamethoxazole, 13C2H3-tramadol, and 13C3-trimethoprim were bought from Cambridge Isotope Laboratories. The 2H5-oxazepam, 2H7-promethazine, 2H4-risperidone, and 13CN-tamoxifen were bought from Sigma-Aldrich. Methanol and acetonitrile were purchased in liquid chromatography–mass spectrometry (LC–MS)–grade quality (Lichrosolv, hypergrade; Merck). Purified water was prepared by a Milli-Q Gradient ultrapure water system (Millipore), equipped with an ultraviolet radiation source. Single-substance water solutions of cetirizine, losartan, citalopram, and ciprofloxacin were prepared with Milli-Q water and pharmaceutical reference standards purchased from Sigma-Aldrich with a purity of >98%. Buffering of the mobile phases was performed by addition of 1 ml of formic acid (Sigma-Aldrich) to 1 L of solvent.
Rat serum (100 µl) was diluted with 100 µl methanol plus 0.1% of formic acid, and 50 ng of each internal and surrogate standard were added. After freezing the samples (1 h at –20°C), 200 µl of water was added; and the samples were centrifuged at 14,000 revolutions per minute for 10 min. Supernatants were then transferred to insert vials and directly analyzed on a triple-stage quadrupole tandem mass spectrometry TSQ Quantum Ultra EMR (Thermo Fisher Scientific) coupled with an Accela LC pump (Thermo Fisher Scientific) and a PAL HTC autosampler (CTC Analytics). Twenty microliters of the sample was loaded onto a Hypersil GOLD column (50 × 2.1 mm i.d., 5 µm particles; Thermo Fisher Scientific) preceded by a guard column (2 × 2.1 mm i.d., 5-µm particles) of the same packing material and from the same manufacturer. A gradient of flow and methanol and acetonitrile in water (all solvents buffered by 0.1% formic acid) was used for elution of analytes. The elution conditions were programmed as follows: 200 µl min−1 10% methanol in water for 1 min isocratically, then composition changed to 30/10/60 water/acetonitrile/methanol and flow of 250 µl min−1 at 8 min. Then, the column was washed by a mixture of ACN/methanol 60/40 and flow of 300 µl min−1 for 9 min. These parameters were kept for 1 min and then switched to starting conditions and equilibrated for 4 min before the next run. Heated electrospray in positive and negative mode was used for ionization of target compounds. Both the first and third quadrupoles were operated at resolution 0.7 FWHM (full width half maximum), and two or three selected reaction monitoring (SRM) transitions were monitored for each analyte. Several calibration standards, both in spiked blank serum and in mobile phase, covering all concentration ranges were measured before, in the middle of, and at the end of sample sequences. The maximum difference between results at quantification and qualification mass transition was set to 30% as a criterion for positive identification. The setting of key parameters, SRM transitions, absolute recoveries, and so on, is described in Grabic et al. 19.
The survival rate of rats during the course of the experiments was 100%, and although no formal assessments were made, we did not observe any signs of abnormal behavior. As a highly general parameter for discomfort, rats were monitored for alterations in weight gain. The increase in body weight over the 5-d experiments, 12.0 to 15.3%, is within the normal range 14; and no statistically significant differences between treatment groups could be found. The clinical blood chemistry analyses showed no significant differences in serum metabolites or enzyme activities between the control group and the exposed rats in the biological effect study (Table 3). Analysis of global hepatic mRNA expression levels between five exposed rats and five controls revealed that fewer than 1% of the transcripts had a p value below 0.01 (274 transcripts; Fig. 1). Furthermore, the apparent changes in gene transcription levels were generally low, and only five genes differed more than 50% (i.e., absolute fold change above 1.5) between the two treatment groups. In addition, the false discovery rate was estimated to be at least 88%, suggesting a very high prevalence of false positives. For the complete list of all genes on the microarray, see Supplemental Data, Table S2.
Table 3. Average concentrations of 10 different metabolites and the activities of five enzymes in blood serum of control rats and rats orally exposed to effluent from bulk drug industriesa
Control (n = 8)
Exposed (n = 10)
a Standard deviations are shown in parentheses.
ALAT = alanine aminotransferase; ALP = alkaline phosphatase; ASAT = aspartate aminotransferase; CK = creatinine kinase; Total LD = total lactate dehydrogenase.
Albumin (g × L–1)
Chloride (mmol × L–1)
Cholesterol (mmol × L–1)
Creatinine (µmol × L–1)
Phosphate (mmol × L–1)
Potassium (mmol × L–1)
Sodium (mmol × L–1)
Total protein (g × L–1)
Uric acid (µmol × L–1)
Urea (mmol × L–1)
ALAT (U × L–1)
ALP (U × L–1)
ASAT (U × L–1)
CK (U × L–1)
Total LD (U × L–1)
The most significantly regulated gene as measured by the microarray was Mx2, myxovirus (influenza virus) resistance gene 2, with a 1.56-fold change in exposed rats (p = 4.23 × 10−5). The upregulation of Mx2 suggested by the microarray could be confirmed by qPCR when RNA from the same animals was used in the assay (p = 0.004, n = 5 + 5). However, when the sample size was increased to include all rats in the study, the effect disappeared and no differences were found between treatment groups (p = 0.37, n = 10 + 10; Fig. 2). Furthermore, in a previous experiment where rainbow trout were exposed to the same effluent, cyp1a was shown to be the most significantly regulated gene 9. Therefore, Cyp1a1 was also analyzed by qPCR in the present study, despite no indication of regulation by the microarray. Accordingly, qPCR showed no difference in Cyp1a1 mRNA levels between the groups (p = 0.45, n = 10 + 10; Fig. 2). No genes previously associated with exposure to the four APIs in the single-substance solutions were found to be regulated on the array, suggesting that the dose received was not sufficient to cause a target-related pharmacological response neither was any gene related to a more general toxic response or downstream pathways altered 14, 20–22.
Pharmaceutical concentration analysis in rat blood serum
Pharmaceutical screening of rat blood serum after 5 d of oral exposure to contaminated water showed that six different substances were detected among 15 individual rat samples (Table 4). Losartan, the most frequently detected substance, was found in eight different rats, all of them tube-fed with effluent. In contrast, losartan could not be detected in any of the rats fed the losartan solution. Cetirizine could be detected in both animals given the cetirizine single solution and effluent but only in the subgroups where blood was drawn within an hour after the final tube-feeding. Ciprofloxacin was found in only four of the animals treated with the antibiotic single solution. Citalopram could not be detected in any animals. Three additional compounds were detected among the exposed animals, all of them found in effluent-treated rats, and the concentrations of these substances in the effluent were also determined (Table 2). Generally, the pharmaceutical concentrations detected in the blood serum were only slightly above the limit of quantification. The levels of the six drug residues found in rat blood serum were compared with the human therapeutic plasma concentration (HTPC) for each substance, respectively. The concentration of clomipramine corresponded to 3.4% of the HTPC and that of biperiden to 2.5%, whereas the levels of the other four detected drugs (diltiazem, cetirizine, losartan, and ciprofloxacin) were markedly lower. Analysis of the industrial effluent in 2011 showed that the concentrations of the four pharmaceutical compounds selected for single-substance solutions were slightly lower than in the original analyses performed in 2006 by Larsson et al. 1 (Table 2).
Table 4. Results from pharmaceutical screening (liquid chromatography–tandem mass spectrometry) of rat blood serum after 5 d oral exposure to industrial effluent from drug manufacturing or one of four different single-substance solutionsa
Industrial effluent from drug manufacturing
Cetirizine single-substance solution
Ciprofloxacin single-substance solution
Citalopram single-substance solution
Losartan single-substance solution
No. of rats with detected drugs
No. of rats with detected drugs
No. of rats with detected drugs
No. of rats with detected drugs
No. of rats with detected drugs
No. of rats with detected drugs
No. of rats with detected drugs
Blood was drawn after 1 or 24 h following the final tube-feeding with contaminated water. The number of rats with a positive detection within each treatment group is shown together with the average concentration of the substance.
LOQ = limit of quantification; conc. = concentration; ND = not detected.
In the present study, both exploratory and more targeted methods were used to detect potential short-term effects in rats orally exposed to effluent from drug manufacturing. In summary, the complete survival and general state of health of the rats together with analyses of body weight increase, blood serum metabolites and enzyme activities, and global hepatic mRNA expression indicate that the effluent is not acutely toxic to rats. This result does not, however, exclude negative effects associated with a higher or longer exposure or indirectly through, for example, promotion of antibiotic-resistant bacteria.
The estimation of the administered dose of nine of the most abundant APIs detected in the effluent showed that the rats would be exposed to less than 7% of an individual human DDD, adjusted for body weight. Despite subtherapeutic doses of individual drugs, we could not exclude that these would affect rats. Furthermore, it was possible that other, yet unidentified pharmaceuticals or chemicals in the highly complex effluent could act synergistically, antagonistically, or additively.
While some rats were tube-fed undiluted industrial effluent, others were given an equal amount of a premade single-substance solution containing one of the four effluent pharmaceuticals that corresponded to the highest rat dosage compared to the human DDD per kilogram body weight. When blood serum was analyzed with regard to pharmaceutical residues after five days of oral exposure, several substances were found in exposed rats; but the concentrations were generally low. Some of the pharmaceuticals detected in the effluent-exposed animals could not be found in rats tube-fed the solution with only that particular substance and vice versa, stressing that the uptake of APIs from treatment with a single substance may, due to multiple factors, differ from exposure to a complex mixture, such as an effluent. For example, before tube-feeding of the rats, the effluent was left to sediment; but the solution was still rich in particles that some pharmaceuticals may adsorb to in the effluent and therefore not be bioavailable to the animals. Also, other compounds present in the effluent may have the potential to influence absorption, distribution, metabolism, or excretion of the drugs and thereby affect their pharmacological activity.
The lack of clear responses in the present exposure study stands in some contrast to previous experiments where the same effluent has been used with aquatic vertebrates 8, 9. Despite a comprehensive and largely similar selection of end points as in the experiment with fish, no effects from the exposure could be observed in rats. This is likely mainly due to the difference in exposure route—that is, exposure over the gills versus oral exposure, and thus the internal doses received. The volume of water breathed through the gills may greatly exceed volumes ingested by drinking. Indeed, many chemicals, including certain pharmaceuticals, bioconcentrate efficiently from water to fish 23, 24. In the present study, the pharmaceutical substances detected in rat blood in concentrations closest to their respective HTPC were administered in very low doses compared with DDD, again stressing that the uptake of, and hence potential effects from oral exposure to, pharmaceuticals is not straightforward to predict, even though initial water concentrations are known. Moreover, the pharmaceutical compound that was detected in rat blood in the concentration closest to its corresponding HTPC value only reached approximately 3% of the plasma levels in a treated patient, despite the exceptionally high levels of drugs detected in the industrial effluent. The volume of effluent needed to be consumed by humans to reach corresponding DDD values was calculated for the nine most abundant substances in the effluent. For cetirizine, the volume of effluent needed was 7.7 L/d, and for the other drugs even higher volumes would be required. Cattle may drink contaminated surface water directly, but this is not likely the case for humans. Pharmaceutical concentrations in nearby water wells in the Patancheru region are, to the best of our knowledge, the highest found in drinking water anywhere; still, the levels are far below those found in the effluent tested here. Indeed, for the most contaminated well water used for drinking 4, the concentration of cetirizine was 2,500 ng/L. To reach the DDD for cetirizine of 10 mg, one would have to drink 4,000 L/d. For citalopram the equivalent volume would be 14,286 L and for ciprofloxacin, 71,429 L. However, potential risks to human health from drug-contaminated drinking water are often coupled to effects from chronic exposure to subtherapeutic concentrations of pharmaceutical compounds 25–27 rather than exposure to water with pharmaceutical concentrations certainly up to 100,000 times higher than the levels normally found in drinking water but administered only at a few isolated occasions. Nevertheless, a risk-assessment study on pharmaceuticals in U.K. drinking water by Rowney et al. 28 found that the margin of exposure could be as low as 4 to 40 for specific drugs under low-flow conditions, which may raise concerns also for short-term exposures, especially for particularly sensitive population subgroups such as fetuses and nurslings.
Screening of metabolites and enzyme activities in blood serum is a well-established tool to detect (adverse) effects from pharmaceuticals both in drug development and in a clinical setting 14. Increased levels of alanine aminotransferase and creatinine or urea can, for example, be signs of liver and kidney damage, respectively. In the present study, the results from the screening revealed no differences between the treatment groups (n = 8 + 10), which suggests that the rats were not adversely affected with regard to overall kidney or liver integrity.
The microarray analysis is an explorative assay in which the expression of over 30,000 transcripts is measured simultaneously. The liver was chosen as the target organ because of its importance for detoxification of chemicals and to facilitate a comparison with fish exposed to the same effluent 9. In the present study, the effluent exposure caused only minor apparent mRNA expression differences in a limited number of hepatic genes as measured by the microarray in accordance with a lack of effect on blood serum constituents. We acknowledge that the relatively small sample size (n = 5 per group) for the microarray analysis provides limited power for the exploratory analyses. However, the predefined hypothesis of an induced Cyp1a mRNA expression together with altered blood serum constituents, both derived from the previous experiment on rainbow trout exposed to the same effluent, are not subjected to the same limitations. Given the large number of endpoints studied in any microarray experiment, some genes may be falsely identified as regulated. Indeed, our estimates show a high false discovery rate for all of the tentatively regulated genes, suggesting that they may, to a large extent, be false positives. In addition, the array results showed good agreement with qPCR, suggesting that the microarray used in the present study is reliable and able to detect true biological differences even if they are small and few.
The fact that the increase of Mx2 mRNA levels in effluent-fed rats as measured by both the microarray and qPCR disappeared when the sample size was doubled highlights the risks of incorrectly assigning effects to a given exposure in studies where many endpoints are studied in parallel. Nevertheless, due to the nature of high-throughput microarray analysis, we cannot exclude that the mRNA expression from a few isolated genes was, indeed, slightly affected by the treatment. However, there is as yet no support that these gene regulations, if present, are of any major physiological relevance.
The lack of any observable effects in the present study, together with an estimated human exposure far below therapeutic levels, thus indicates that the risk for acute effects on humans from direct pharmaceutical exposure through drinking water is small, even in this highly contaminated region. Long-term effects can, however, not be excluded. For cattle, goats, and so on, which may drink surface water, the exposure may be much higher since detected surface levels of some pharmaceuticals, including cetirizine and ciprofloxacin, in nearby lakes and rivers exceeded well-water levels by more than two orders of magnitude 4. For human health, the strongest concern associated with the pollution of pharmaceuticals is, in our view, the potential indirect effects via exposure to antibiotic-resistant bacteria promoted by the presence of antibiotics 29. The previous finding of a high prevalence of both resistance genes and genetic transfer elements in sediment bacteria living near the treatment plant 11 further emphasizes this apprehension.
To identify the primary effects of concern and to facilitate making adequate priorities in risk assessments, legislation, and policies, it is of great importance that negative findings, such as these, are also made publicly available 30.
Tables S1–S2. (16 KB DOCX; 4.2 MB XLS).
We acknowledge help with microarray laboratory work and analysis from the Microarray Resource Centre at Lund University. We thank A. Eliasson and S. Berntsson, Laboratory of Experimental Biomedicine, University of Gothenburg, for help with tube-feeding of the rats and the Genomics Core Facility platform at the Sahlgrenska Academy, University of Gothenburg, for providing qPCR instruments.