Detecting diclofenac in livestock carcasses in India with an ELISA: A tool to prevent widespread vulture poisoning

Environmental Pollution 160 (2012) 11e16

Contents lists available at SciVerse ScienceDirect

Environmental Pollution journal homepage: www.elsevier.com/locate/envpol

Detecting diclofenac in livestock carcasses in India with an ELISA: A tool to prevent widespread vulture poisoning Mohini Sainia, Mark A. Taggartb, c, Dietmar Knoppd, Suchitra Upretia, Devendra Swarupa, e, Asit Dasa, Praveen K. Guptaa, Reinhard Niessnerd, Vibhu Prakashf, Rafael Mateob, Richard J. Cuthbertg, * a

Indian Veterinary Research Institute, Izatnagar, 243 122 Uttar Pradesh, India Instituto de Investigación en Recursos Cinegéticos, IREC (CSIC-UCLM-JCCM), Ronda de Toledo, 13005, Ciudad Real, Spain Environmental Research Institute, University of the Highlands and Islands, Castle Street, Thurso, Caithness, Scotland KW14 7JD, UK d Institute of Hydrochemistry, Technical University Munich, Marchioninistrasse 17, D-81377 Munich, Germany e Central Institute for Research on Goats, Makhdoom, 281 122 Uttar Pradesh, India f Bombay Natural History Society, Hornbill House, S. B. Singh Road, Mumbai 400 001, India g Royal Society for the Protection of Birds, The Lodge, Sandy, Bedfordshire SG19 2DL, UK b c

a r t i c l e i n f o

a b s t r a c t

Article history: Received 9 May 2011 Received in revised form 31 August 2011 Accepted 3 September 2011

Diclofenac, a non-steroidal anti-inflammatory drug (NSAID), has caused catastrophic vulture declines across the Indian sub-continent. Here, an indirect ELISA is used to detect and quantify diclofenac in 1251 liver samples from livestock carcasses collected across India between August 2007 and June 2008, one to two years after a ban on diclofenac manufacture and distribution for veterinary use was implemented. The ELISAs applicability was authenticated with independent data obtained using LCeESI/MS. Of 1251 samples, 1150 (91.9%) were negative for diclofenac using both methods, and 60 (4.8%) were positive at 10e4348 and 10e4441 mg kg
1 when analysed by ELISA and LCeESI/MS, respectively. The residue level relationship in the 60 positive samples was highly significant (p < 0.001, r2 ¼ 0.644). Data suggest that this immunological assay could be used not only for cost effective sample screening, but also for residue level semi-quantification. Ó 2011 Published by Elsevier Ltd.

Keywords: Non-steroidal anti-inflammatory drugs Vulture conservation Diclofenac poisoning Indirect competitive ELISA Diclofenac analysis

1. Introduction Vultures are at the apex of the terrestrial food web, and like other avian species are considered to be sentinel biomonitors of prevailing environmental conditions and quality (Sekercioglu et al., 2004). Recently, three species of Gyps vulture endemic to South Asia (Gyps bengalensis, G. indicus and G. tenuirostris) have undergone catastrophic declines across the Indian sub-continent, and are now listed as Critically Endangered (Prakash et al., 2007; IUCN, 2010). These declines have been clearly linked with the extensive use of the non-steroidal anti-inflammatory drug (NSAID) diclofenac (Oaks et al., 2004; Shultz et al., 2004; Cuthbert et al., 2009). Widely used (until recent restrictions were put in place) to treat ailing livestock across the Indian sub-continent, residues of this drug are commonly present in the carcasses of domesticated ungulates (buffalo, cow, goat, camel, horse) that provide a principal food source for these scavengers (Taggart et al., 2007, 2009). Diclofenac

* Corresponding author. E-mail address: [email protected] (R.J. Cuthbert). 0269-7491/$ e see front matter Ó 2011 Published by Elsevier Ltd. doi:10.1016/j.envpol.2011.09.011

is now known to be extremely toxic towards all Gyps species tested so far (Oaks et al., 2004; Swan et al., 2006a; Naidoo et al., 2009; Das et al., 2011), causing fatal extensive visceral gout in exposed individuals. Another NSAID, ketoprofen, is now also known to be toxic to vultures, causing the same toxicological symptoms as diclofenac (Naidoo et al., 2010a, 2010b). In 2006, following extensive safety testing on meloxicam, another NSAID now known to be “vulture safe” (Swan et al., 2006b; Swarup et al., 2007), an India wide ban on the manufacture and distribution of veterinary diclofenac was imposed by the Drug Controller General in India (Kumar, 2006). Similar restrictions were also put in place in the same year in Pakistan and Nepal. In India, regulations were tightened further in 2008, and it is now an imprisonable offence to sell diclofenac for veterinary use or to use it on livestock (Singh, 2008). In order to assess how prevalent diclofenac residues are in vulture food sources, and to monitor the eventual affect these usage restrictions may have within India, two surveys have been published which have documented diclofenac levels and prevalence in livestock carcasses across the region (Taggart et al., 2007, 2009). Liquid chromatography e electrospray ionisation mass spectrometry (LCeESI/MS) analytical methods

12

M. Saini et al. / Environmental Pollution 160 (2012) 11e16

have been applied to samples taken during these surveys (undertaken in 2004e2005 and 2006) which in combination determined diclofenac concentrations and prevalence in >3000 individual ungulate carcass liver extracts. These surveys revealed that the prevalence of detectable diclofenac (>10 mg kg
1) in these vulture food sources was between 10 and 11% during both periods (Taggart et al., 2007, 2009). Detailed modeling using this and other existing data (Green et al., 2004, 2006, 2007), has further demonstrated that only 0.13e0.76% of carcasses need contain a lethal dose of diclofenac in order to cause the extreme rates of vulture decline in India that have been recorded over the last one to two decades. The most common Gyps vulture in the region (G. bengalensis), for example, is now thought to have declined by >99.9% since 1992 (Prakash et al., 2007), having previously been considered to be probably the most abundant large raptor in the world (Houston, 1985). Whilst the LCeESI/MS methodology used to undertake diclofenac monitoring across India thus far is precise, accurate and sensitive (Taggart et al., 2007, 2009), it is also expensive, time consuming, and of limited availability across India, Pakistan and Nepal. As an alternative analytical option, an enzyme-linked immuno-sorbant assay (or ELISA) was recently proposed and prevalidated for such an application (Knopp et al., 2007). ELISAs are powerful immunological techniques that have been shown to have potential in monitoring drug residues such as antibiotics, in a range of matrices. ELISAs have been used (for example) to monitor amoxicillin in pigeon serum (Yeh et al., 2008), ciprofloxacin in meat destined for human consumption (Duan and Yuan, 2001), furazolidone in swine liver and muscle tissue (Chang et al., 2008), and marbofloxacin in beef and pork (Sheng et al., 2009). Such assays are often very sensitive, compound specific, reasonably easy to use, and are designed such that a large number of samples can be screened in a relatively short time period (using 96 well plate formats). A diclofenac specific ELISA originally designed for use in water/waste water applications (Deng et al., 2003), has (in trials) been shown to have potential as an assay which could be applied to diclofenac monitoring in tissue samples in India (Knopp et al., 2007), however, extensive validation has not so far been undertaken. The aim of the present study was to investigate whether such an ELISA could be reliably used as a screening tool for large scale diclofenac monitoring in the field. Thus, we present diclofenac residue data produced for 1251 extracts of liver tissue. Samples were collected during monitoring surveys which took place across India between August 2007 and June 2008. The results are compared to independently collated data generated using an existing validated LCeESI/MS methodology. These data give an important first indication as to the effectiveness of restrictions on veterinary diclofenac use that are now meant to be in place across India.

Fig. 1. Map showing the states sampled within India, the number of samples taken per state, and the approximate sample site locations. Liver samples were taken from dead livestock arriving at carcass dumps at these locations between the 18th of August 2007 and the 20th of June 2008.

2.2. Diclofenac extraction A sub-sample from each liver sample was removed using a stainless steel scalpel, and weighed to an accuracy of 0.0001 g, to a recorded wet weight between 0.45 and 0.55 g. Each liver tissue sample was weighed into a new glass test tube, and homogenized with 2 ml of HPLC grade acetonitrile using an Ultra Turrax IKA T8 homogenizer. The homogenate was then centrifuged at 1000g for 10 min, and the supernatant filtered through a 0.45 mm nylon disposable syringe filter unit. Filtrates were passed directly into 2 ml glass LCeMS vials, and stored as two aliquots (one for ELISA, one for LCeESI/MS) at
20  C until analysis. Each aliquot was analysed independently for the presence of diclofenac using an indirect competitive ELISA at the Indian Veterinary Research Institute (IVRI) in India and by LCeESI/MS at the Instituto de Investigación en Recursos Cinegéticos (IREC) in Spain.

2. Methods 2.3. Analysis by ELISA 2.1. Field sampling of livers from domesticated livestock carcasses Samples of liver tissue from 1251 individual carcasses were collected over a ten-month period between the 18th of August 2007 and the 20th of June 2008 (15e25 months after the 2006 veterinary diclofenac ban was announced, and 12e22 months after it was theoretically implemented (Kumar, 2006)) using procedures described previously (Senacha et al., 2008). The majority of samples were taken from carcasses of cow (Bos indicus, Bos taurus and hybrids; n ¼ 752) and buffalo (Bubalus bubalis; n ¼ 437). The remainder were from sheep (Ovis aris; n ¼ 30), goat (Capra hircus; n ¼ 27), horse (Equus caballus; n ¼ 2) and camel (Camelus dromedarius; n ¼ 3); n ¼ 62 in total. Samples were frozen upon collection using a portable field freezer (Engel, Japan), and then transported whilst frozen to laboratory based freezers (
20  C) where they were stored until extraction. Samples were collected from carcass dumps in six Indian states, i.e., in Rajasthan (n ¼ 303), Gujarat (n ¼ 159), Maharashtra (n ¼ 262), Andhra Pradesh (n ¼ 143), Madhya Pradesh (n ¼ 257) and Uttar Pradesh (n ¼ 127). Fig. 1 shows a map of the approximate locations of the sample sites used across India.

Prior to ELISA analysis, extracts were thawed, shaken, and then stored overnight at 4  C. Extracts were then screened for the presence of diclofenac using 1:50 extract dilutions, and then (depending on the concentration of the drug detected at screening), quantified using 1:50, 1:100 and 1:200 dilutions in water (HPLC grade). For calibration curve construction on each ELISA plate, a stock solution containing 50 mg l
1 diclofenac (2-[(2,6-dichlorophenyl)amino] benzeneacetic acid sodium salt; Sigma Aldrich, D6899) was first prepared in 1:1 water:acetonitrile. Working calibration standards ranging from 0.01 to 10 mg l
1 were then prepared in an acetonitrile:water ratio which reflected the sample extract dilution being used on any particular plate (i.e., 1:50, 1:100 or 1:200). Coating antigen synthesis, the production of the polyclonal serum against diclofenac, and the development of the indirect competitive ELISA used here has been described in detail previously (Deng et al., 2003; Knopp et al., 2007). In brief, lyophilized powder containing diclofenacethyroglobulin or diclofenaceovalbumin hapten conjugate was prepared using the mixed anhydride method, and was reconstituted in 0.05 M sodium carbonate buffer (pH 9.6) and then used as the plate coating antigen. Diclofenacebovine serum albumin conjugate was used as an

M. Saini et al. / Environmental Pollution 160 (2012) 11e16 immunogen to raise antiserum in rabbits, and the polyclonal serum served as the primary antibody in the ELISA. Firstly, a 96 well microtiter plate (Nunc Maxisorp) was coated with 200 ml per well of diclofenacethyroglobulin conjugate (20 ng ml
1) or diclofenaceovalbumin conjugate (300e400 ng ml
1) diluted in coating buffer (0.05 M sodium carbonate buffer; pH 9.6). Plates were sealed with sealing film and kept overnight in a refrigerator at 4  C. The plate was then washed three times using PBS-Tween (0.01 M PBS, pH 7.46, containing 0.15 M NaCl and 0.1% Tween 20) in an automated plate washer (Awareness Technology, USA). Unoccupied binding sites on the plate were then blocked by addition of 300 ml per well of 1% casein solution in PBS (pH 7.6), under incubation for 1 h at 37  C on an orbital flatbed shaker (plates were sealed with sealing film at this incubation point and at all subsequent points mentioned below). Plates were then washed again. Diluted liver extracts and standards (including blanks) were then added in quadruplicate wells (at 100 ml/well). Then, 100 ml per well of primary antibody was immediately added (at 1:18 000 in PBS when using the thyroglobulin conjugate and at 1:15 000 in PBS when using the ovalbumin conjugate), and the plates were incubated again for 1 h at 37  C on an orbital flatbed shaker. After washing again, goat anti-rabbit peroxidase conjugate (Pierce, 31 460; 1:8000 in PBS; 200 ml/well) was added. Plates were incubated for 1 h at 37  C on the orbital shaker, and then washed for a final time. Lastly, 200 ml of “ready to use” TMB substrate (Invitrogen) was added to each well and the plates were incubated at 37  C on the orbital shaker in darkness until a clear colour (blue) gradation was apparent amongst standards. The colour reaction was stopped using 100 ml/well of stop solution (5% sulphuric acid), and light absorbance was recorded at 450 nm using an automatic plate reader (Awareness Technology, USA). Two alternative substrates (to TMB) were also tested, i.e., ABTS (2,20 azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt; Sigma A9941) and OPD (o-Phenylenediamine; Sigma P9029). ABTS did not generate a significant reaction, whereas OPD was found to be a suitable TMB alternative.

13

Fig. 2. A typical plot (for illustration purposes only) showing light absorbance (at 450 nm) versus diclofenac concentration in standards and real extracts (to the right) for the ELISA technique. The dashed lines indicate the adjusted blank absorbance 3SDs. At the right side, examples of sample readings are given, showing where a sample would be deemed negative or positive for diclofenac.

2.4. Procedure used to differentiate between positive and negative ELISA samples

3. Results During validation, it became apparent that the use of pure acetonitrile:water based blanks was not ideal in terms of allowing the test to reliably distinguish between diclofenac positive and negative liver extracts. Likewise, since the sample extracts analysed were of liver tissue of varying age, degree of decomposition, and variable species, a reliable, simple, “matrix matched” blank (or set of standards) could not be realistically generated. Therefore, when using pure acetonitrile:water based blanks, the mean blank reading given by the ELISA was invariably a little higher than the mean reading in real liver extracts known to be negative (as determined by LCeESI/MS analysis). Whilst this remained the case, the ELISA would be prone to generating a high proportion of false positive data. To overcome this issue, we determined that (based on previous data; Taggart et al., 2007, 2009) it was highly unlikely that all samples tested on any one plate (between 12 and 16, depending on the plate design and number of standards used) would be positive for diclofenac. As such, when reading plates, we assumed the highest mean sample extract absorbance reading (OD) would actually provide a baseline “matrix matched” blank reading, against which all other samples could be assessed. While assuming that our pure (acetonitrile:water based) blank should have the same reading (as the highest real extract blank), we then applied an adjustment factor (af) to all blanks and standards on a plate, where the value of af was equal to that required to bring the pure blank value down to a reading equal to that of the highest real extract OD. For example, if our highest mean real extract OD reading was 0.9, but our mean pure blank reading was 0.95, our adjustment factor af was simply 0.9/0.95 ¼ 0.947. All mean pure blank and pure standard ODs were then multiplied by this af value, and our real samples were then assessed against the calibration curve gained. This generated a plate specific matrix matching system. When analysing the plate data for our 1251 real sample extracts, our positive vs negative cut-off point was set at the mean adjusted blank reading minus three times the standard deviation (SD) of the 8 blank readings taken on each plate. At the same time, using our real extract readings (n ¼ 4), we added three SDs to the mean sample reading obtained. The two resultant values were then compared, and where the sample mean plus 3SDs was greater than the blank mean minus 3SDs, the extract was deemed to be negative for diclofenac. Where the value for the sample mean plus 3SDs was lower than the blank mean minus 3SDs, the extract was deemed to be positive (see Fig. 2 for a diagramatic representation of the technique used). The linear analytical range for the test commonly fell between approximately 0.025 mg l
1 and 2.5 mg l
1. Accordingly, all extracts deemed positive were re-analysed at 1:50, 1:100 and 1:200 dilutions, and quantification was achieved using results which fell nearest to the center of this linear range. 2.5. Liquid chromatography e mass spectrometry (LCeESI/MS) In addition to analysing all our extracts by ELISA, diclofenac levels were also determined using liquid chromatography e mass spectrometry (with electrospray ionisation; LCeESI/MS). An Agilent 1100 series LC instrument with a 6110 series single quadrupole MS was used, and full details regarding the methods utilised have been described previously (Taggart et al., 2007, 2009). The limit of quantification (LOQ; back calculated to within wet tissue concentrations) and the limit of detection (LOD) for this technique were 10 mg kg
1 and 4 mg kg
1, respectively.

Of 1251 carcass liver extracts screened by the ELISA and the LCeESI/MS, a total of 1150 (91.9%) were found to be negative using both techniques, 101 (8.1%) were positive by one or both of the two methods used, and 60 (4.8%) were positive using both methods. The positive concentrations ranged between 10e4348 and 10e4441 mg kg
1 (in liver tissue by wet weight), when analysed by ELISA and LCeESI/MS, respectively. Hence, in total, 96.7% of all data produced by these 2 quite different techniques was in good agreement. Overall, the ELISA suggested that 7.3% (n ¼ 91) of samples were positive for diclofenac, while the LCeESI/MS indicated 5.6% (n ¼ 70) were positive. The ELISA highlighted 31 samples (2.5%) as positive for diclofenac which were negative by LCeESI/MS, whilst the LCeESI/MS detected diclofenac in 10 extracts (0.8%) that were suggested to be negative by ELISA. Where there were discrepancies between the two techniques, these often tended to occur at quite low concentrations. For example, although the ELISA did not identify 10 positives that were confirmed by LCeESI/MS, only one of these positives was at a high level (i.e., at 369 mg kg
1). The other 9 samples were all at levels below 32 mg kg
1 in tissue. Likewise, of the 31 samples determined to be positive by ELISA, which were not confirmed as such by LCeESI/MS, 24 were at <77 mg kg
1, while 20 were <45 mg kg
1. Seven quite high values (in 0.6% of samples) were, however, presented at levels between 135 and 902 mg kg
1. One of our objectives was to determine how well these two datasets related. Fig. 3 shows the regression for all the samples found to be positive for diclofenac by both techniques (n ¼ 60). The r2 value for this regression was 0.644 (p < 0.001), and the slope indicated that the overall distribution of the data was very similar for the ELISA and LCeESI/MS, even though the precise values often differed to some degree. 4. Discussion Across the Indian sub-continent, various vulture conservation programmes are underway, and it is essential that practical and reliable diclofenac monitoring techniques are available for these. Techniques can be applied at a country or state wide level to monitor the effectiveness of restrictions in place on the use of

14

M. Saini et al. / Environmental Pollution 160 (2012) 11e16

(a) ELISAs are highly manual, multi-step “wet” techniques. They are therefore (by nature), more prone to human error. (b) ELISAs can be non-specific because the antiserum can crossreact with non-target compounds with very similar structures. In this case, the antiserum is known to cross-react directly with certain tested diclofenac metabolites, with some other tested NSAIDs, and it may have the potential to crossreact with other as yet untested/unidentified (structurally similar) analytes (Deng et al., 2003). (c) Various elements and/or compounds within the highly complex and variable sample matrix (liver extracts) may also interfere non-specifically with the accurate use of the ELISA at various points during the extended multi-step process (Deng et al., 2003; Knopp et al., 2007). (d) Obtaining a reliable positive vs negative “cut-off” point, for use during the ELISA, can be challenging. This may increase the chances of obtaining false positive or negative data, especially at the lowest concentration levels (see Methods, Results and Fig. 2).

Fig. 3. Regression for diclofenac concentrations derived by an ELISA assay and by LCeESI/MS analysis, independently. The inner long dashed line shows the 95% confidence interval, while the outer dashed-dotted line describes the 95% prediction interval. The r2 value is for data found positive by both techniques. The grey squares show data determined as positive by LCeESI/MS but negative by ELISA; the grey triangles show data determined as positive by ELISA but negative by LCeESI/MS (negative values are plotted at the 4 mg kg
1 LOD level for graphical purposes only).

diclofenac (as described here), and in addition, at a more local level. For example, in-situ conservation activities may include establishing “safe wild vulture feeding zones or restaurants” (Gilbert et al., 2007), or, “captive vulture breeding sanctuaries”, and in both cases, only diclofenac-free food must be provided. In the longer term, the diclofenac-free nature of domesticated ungulate carcasses available in the wild across India (and Pakistan and Nepal) must be strictly ensured. After all, a single carcass containing high diclofenac residue levels could completely decimate a wild (captive or re-introduced) long-lived Gyps colony. In terms of Indian Government initiated control measures regarding diclofenac manufacture, distribution, sale and use, which are now in place, the results presented indicate (whether considering the ELISA or LCeESI/MS data) that a decline in residue prevalence in the wild had occurred by 2007/2008. The raw data show that the residue prevalence between August 2007 and June 2008 (15e25 months after the 2006 ban was first initiated) was 5.6 or 7.3% (whether using LCeESI/MS or ELISA data, respectively) in comparison to previously reported values of 10e11% (recorded prior to (Taggart et al., 2007), and immediately after (Taggart et al., 2009) the ban was announced). Whilst this is encouraging, the situation during the period reported here still remains completely incompatible with the recovery of wild Gyps within the region. Again, only a very small proportion of carcasses (1:130e1:760) actually need contain lethal levels of diclofenac in order to cause the rapid population declines recently observed (Green et al., 2004). Whilst these data are important for vulture conservation, the main aim here was to determine whether the ELISA described could be used as an effective diclofenac residue monitoring tool within India (and beyond). In this respect, it is important that the limitations and differences between the two techniques are discussed. These techniques are based on two very different principles, however, we would normally expect LCeESI/MS to be more accurate when determining diclofenac residue levels (and less prone to analytical error) than the ELISA, for several key reasons. These include:

Several relevant reports have previously compared ELISA generated data with data collected using other (instrumental) methods, i.e., GCeMS, LCeESI/MS and triple quadrupole LCeMS (Deng et al., 2003; Himmelsbach and Buchberger, 2005; Knopp et al., 2007; Dolliver et al., 2008; Hu et al., 2008). These indicate that (as a general rule) ELISAs are likely to provide somewhat less compound specific, accurate and reliable field monitoring data when compared to such instrumental techniques because of the differences/limitations noted above. This does not, however, detract from the value of an ELISA as a potentially powerful screening tool in large scale field sampling campaigns for a wide variety of compounds (Shelver et al., 2008; López et al., 2010; Lutter et al., 2011). In terms of point (b) above, this antiserum is known to cross-react directly and specifically with a range of compounds (some of which have been identified; Deng et al., 2003) that are commonly similar in chemical structure to diclofenac. For example, complete (i.e., 100%) cross-reactivity is known to occur against at least one common primary diclofenac metabolite (i.e., the 5-OH derivative), and crossreactivities up to 1.5% (and 0.2%) have been identified for another four diclofenac metabolites (Deng et al., 2003). Essentially, any addition (substituent) to the dichlorophenyl ring present in diclofenac (and the 5-OH derivative) seems to greatly reduce the potential for such cross-reactivity (Knopp et al., 2007). In contrast, the LCeESI/ MS has the advantage of being (by nature) highly compound specific. The LCeESI/MS is also set (in specific ion monitoring mode) to target and detect the parent compound alone (diclofenac and its characteristic fragments), and is not currently used to identify diclofenac metabolites. As such, it seems highly likely that the cross-reactivity inherent within the ELISA will be a primary cause for at least some of the discrepancies we see between the two datasets (both in terms of positive vs negative, and also in terms of the quantification level recorded). In this respect, the ELISA will be more prone to overestimating the concentration of the parent compound, since in reality it is commonly detecting diclofenac plus certain metabolites. This tendency is reflected in the regression slope recorded here (Fig. 3), and a similar trend (towards overestimation) was also noted previously when comparing this ELISA with GCeMS data generated regarding wastewater samples (where GCeMS values were, on average, around 25% lower; Deng et al., 2003; Knopp et al., 2007). Within the samples tested, metabolite levels are also likely to vary widely from sample to sample, since the drug metabolising capacity of any liver will be influenced (for example) by a tested animal’s age, its nutritional status, or whether it is suffering from a particular disease at the time of death (Jorquera et al., 1996; Meyer, 1996; Wynne, 2005; Rendic and Guengerich, 2010). This fact will in turn

M. Saini et al. / Environmental Pollution 160 (2012) 11e16

cause further differences within the data regarding the quantification levels obtained (by ELISA or LCeESI/MS). In addition, other structurally similar NSAIDs such as tolfenamic acid (0.2% crossreactivity) and meclofenamic acid (3.5% cross-reactivity), if present in our extracts, would also generate false positive or enhanced data (Deng et al., 2003). However, at least for these two NSAIDs, at these low cross-reactivity levels, this is unlikely to be a highly significant issue. In fact, in addition to testing for diclofenac, during our LCeESI/ MS analyses we also routinely collect data (unpublished) regarding eight other common NSAIDs (Taggart et al., 2009), some of which are in widespread use within India, and analysis of this data indicates that there is no evidence that any cross-reactivity is occurring toward these particular eight compounds. Specifically, for the 31 samples determined to be positive by ELISA (but negative by LCeESI/MS), only two contained another of the eight NSAIDs tested. In both cases, this was meloxicam (the second most common NSAID detected), but, meloxicam alone was also detected by LCeESI/MS in >50 other samples (at concentrations up to w4000 mg kg
1), yet none of these samples produced any response when analysed by ELISA. Finally, in the longer term, the ability of the ELISA to detect diclofenac plus some of its metabolites may not be entirely negative. The potential toxicity of common diclofenac metabolites towards Gyps vultures is currently unknown, but, being able to assess levels of the parent compound and at least one primary metabolite may be useful within future work. For example, if metabolites (such as the 5-OH derivative) were indeed identified to be highly toxic to Gyps, it is important that field monitoring has the capacity to identify carcasses that contain this metabolite (even when diclofenac itself has been entirely metabolised). In terms of point (c) above, in addition to compound specific (antibody recognition) cross-reactivity, certain matrix constituents within the sample extracts tested may also have the potential to modulate the results generated by, and therefore the accuracy/ sensitivity of, the ELISA. The fact that results generated here for real extracts known to be negative for diclofenac did not commonly tend to be in line with data generated using pure blanks on the same plate (see point (d), Methods, Results and Fig. 2), gives a clear indication that the highly variable liver extract matrix was actively modulating and affecting the way in which the ELISA should ideally function. Again, when this ELISA was developed (Deng et al., 2003), it was noted that the presence of humic acids/dissolved organic carbon (for example) in surface/wastewater samples could affect (in a negative manner) the sensitivity of the test. Although the exact mechanism by which such test interference could occur is largely unknown, solvent extracts of fresh and partially decomposed liver tissue (as tested here) are likely to contain a wide variety of organic compounds with associated functional groups. It seems likely that given the potential diversity of organic compounds that could be present in such extracts, a range of modulating reactions may occur, either between these and the analyte of interest (diclofenac), or, between these and the antibodies and/or coating antigens used. In either case, non-specific binding and reaction site interference/ blocking may well be occurring by virtue of hydrogen bonding, and/ or nonpolar or ionic interactions (Deng et al., 2003). Large, complex organics may also be partially recognised by the ELISA, if a reactive (exposed) dichlorophenyl ring (as a constituent of the larger hydrocarbon) is presented to the antibody. Finally, during initial validation of this ELISA for this purpose (Knopp et al., 2007), it was also noted that the test could not be used with bile extracts (again because of matrix interference), and, that the tendency for the ELISA to overestimate results potentially increased in relation to tissue age, i.e., greater overestimation may occur when using more decomposed tissues (in comparison to fresh tissues). To conclude, the large (n ¼ 1251) blind comparison presented here between ELISA and LCeESI/MS data demonstrates that this

15

ELISA could indeed be applied to diclofenac monitoring on the Indian sub-continent (or beyond). It is likely that this technique may be most useful when applied to large scale field monitoring where many thousands of samples are required to be analysed (Taggart et al., 2007, 2009). The test has been shown to act as a very effective screening tool, which is highly unlikely to generate false negatives, especially when diclofenac levels are above about 30 mg kg
1 in tissue. While the technique may generate a low proportion of false positive data, if it were to be used primarily as a screening tool and all positives were to be confirmed using MS techniques, significant cost, manpower and time savings could potentially be made. In addition to monitoring diclofenac prevalence (i.e., positive vs negative), the relationship between the ELISA and LCeESI/MS data regarding concentrations, also indicates that the technique could be used for semi-quantification. If, as a next step, the ELISA were to be mounted on a rapid portable test system (such as on a “dipstick assay”), human/manual error may also be significantly reduced, and applications beyond large scale field surveys could become much more viable, i.e., tests could be used to confirm suspected diclofenac poisoning within avian rescue centres, or, during post mortem forensic “cause of death” determinations, or, for field analysis at carcass dumps and/or in remote locations lacking laboratory facilities. Unfortunately, as clearly indicated by these data, the eradication of diclofenac in vulture food sources across India (and beyond) is likely to represent a long term conservation challenge. Existing restrictions regarding veterinary use may well take some significant time to take full effect across the Indian sub-continent, and it is important in the meantime that economically viable diclofenac testing options are made available to those working in the field. Even though licensed veterinary diclofenac manufacture, sale and use is now banned in India, injectable human formulations remain very widely available for unrestricted purchase. As long as this is the case, such formulations are likely to be used on animals and thus the threat to Gyps (wild or re-introduced) will remain.

Acknowledgements We thank the Director and Joint Director (Research) at IVRI and the Director of BNHS for providing the necessary facilities to carry out this work. Financial support came from the UK Government’s Darwin Initiative and the RSPB. We acknowledge laboratory assistance provided by Mohan Bhat at IVRI.

References Chang, C., Peng, D., Wu, J., Wang, Y., Yuan, Z., 2008. Development of an indirect competitive ELISA for the detection of furazolidone marker residue in animal edible tissues. Journal of Agricultural and Food Chemistry 56, 1525e1531. Cuthbert, R., Prakash, V., Bowden, C., Das, D., Green, R.E., Jhala, Y., Pain, D.J., Senacha, K.R., Shah, N., Taggart, M.A., 2009. Role of veterinary diclofenac in decline of vulture populations in South Asia. Indian Journal of Veterinary Medicine 29, 80e85. Das, D., Cuthbert, R.J., Jakati, R.D., Prakash, V., 2011. Diclofenac is toxic to the Himalayan vulture Gyps himalayensis. Bird Conservation International 21, 72e75. Deng, A., Himmelsbach, M., Zhu, Q., Frey, S., Sengl, M., Buchberger, W., Niessner, R., Knopp, D., 2003. Residue analysis of pharmaceutical diclofenac in different water types using ELISA and GC-MS. Environmental Science and Technology 37, 3422e3429. Dolliver, H., Kumar, K., Gupta, S., Singh, A., 2008. Application of enzyme-linked immunosorbent assay analysis for determination of monensin in environmental samples. Journal of Environmental Quality 37, 1220e1226. Duan, J., Yuan, Z., 2001. Development of an indirect competitive ELISA for ciprofloxacin residues in food animal edible tissues. Journal of Agricultural and Food Chemistry 49, 1087e1089. Gilbert, M., Watson, R.T., Ahmed, S., Asim, M., Johnson, J.A., 2007. Vulture restaurants and their role in reducing diclofenac exposure in Asian vultures. Bird Conservation International 17, 63e77.

16

M. Saini et al. / Environmental Pollution 160 (2012) 11e16

Green, R.E., Newton, I., Shultz, S., Cunningham, A.A., Gilbert, M., Pain, D.J., Prakash, V., 2004. Diclofenac poisoning as a cause of vulture population declines across the Indian subcontinent. Journal of Applied Ecology 41, 793e800. Green, R.E., Taggart, M.A., Das, D., Pain, D.J., Sashikumar, C., Cunningham, A.A., Cuthbert, R., 2006. Collapse of Asian vulture populations: risk of mortality from residues of the veterinary drug diclofenac in carcasses of treated cattle. Journal of Applied Ecology 43, 949e956. Green, R.E., Taggart, M.A., Senacha, K.R., Raghavan, B., Pain, D.J., Jhala, Y., Cuthbert, R., 2007. Rate of decline of the oriental white-backed vulture population in India estimated from a survey of diclofenac residues in carcasses of ungulates. PLoS ONE 2 e686. Himmelsbach, M., Buchberger, W., 2005. Residue analysis of oxytetracycline in water and sediment samples by high-performance liquid chromatography and immunochemical techniques. Microchimica Acta 151, 67e72. Houston, D.C., 1985. Indian white-backed vulture (G. bengalensis. Technical Publication No.5. In: Newton, I., Chancellor, R.D. (Eds.), Conservation Studies on Raptors. International Council for Bird Preservation, Cambridge, pp. 465e466. Hu, D., Fulton, B., Henderson, K., Coats, J., 2008. Identification of tylosin photoreaction products and comparison of ELISA and HPLC methods for their detection in water. Environmental Science and Technology 42, 2982e2987. IUCN, 2010. Red list of threatened species, Version 2010.3. http://www.iucnredlist. org (accessed 18.10.10). Jorquera, F., Culebras, J.M., González-Gallego, J., 1996. Influence of nutrition on liver oxidative metabolism. Nutrition 12, 442e447. Knopp, D., Deng, A., Letzel, M., Taggart, M.A., Himmelsbach, M., Zhu, Q., Perobner, I., Kudlak, B., Frey, S., Sengl, M., Buchberger, W., Hutchinson, C., Cunningham, A., Pain, D., Cuthbert, R., Raab, A., Meharg, A., Swan, G., Jhala, Y., Prakash, V., Rahmani, A., Quevedo, M., Niessner, R., 2007. Immunological determination of the pharmaceutical diclofenac in environmental and biological samples. American Chemical Society Symposium Series 966. In: Kennedy, I.R., Soloman, K.R., Gee, S.J., Crossan, A.N., Wang, S., Sánchez-Bayo, F. (Eds.), Rational Environment Management of Agrochemicals: Risk Assessment, Monitoring, and Remedial Action. Oxford University Press, pp. 203e226. Kumar, A., 2006. Subject: diclofenac for veterinary use. Letter to ‘All State Drug Controllers’ from the ‘Drug Controller General (India)’, 11th May 2006, F.No. 1803/2006/DC. Lutter, P., Savoy-Perroud, M.C., Campos-Gimenez, E., Meyer, L., Goldmann, T., Bertholet, M.C., Mottier, P., Desmarchelier, A., Monard, F., Perrin, C., Robert, F., Delatour, T., 2011. Screening and confirmatory methods for the determination of melamine in cow’s milk and milk-based powdered infant formula: validation and proficiency-tests of ELISA, HPLC-UV, GC-MS and LC-MS/MS. Food Control 22, 903e913. López, P., Martello, S., Bermejo, A.M., De Vincenzi, E., Tabernero, M.J., Chiarotti, M., 2010. Validation of ELISA screening and LC-MS/MS confirmation methods for cocaine in hair after simple extraction. Analytical and Bioanalytical Chemistry 397, 1539e1548. Meyer, U.A., 1996. Overview of enzymes of drug metabolism. Journal of Pharmacokinetics and Biopharmaceutics 24, 449e459. Naidoo, V., Wolter, K., Cuthbert, R., Duncan, N., 2009. Veterinary diclofenac threatens Africa’s endangered vulture species. Regulatory Toxicology and Pharmacology 53, 205e208. Naidoo, V., Wolter, K., Cromarty, D., Diekman, M., Duncan, N., Meharg, A.A., Taggart, M.A., Venter, L., Cuthbert, R., 2010a. Toxicity of non-steroidal antiinflammatory drugs to Gyps vultures: a new threat from ketoprofen. Biology Letters 6, 339e341. Naidoo, V., Venter, L., Wolter, K., Taggart, M.A., Cuthbert, R., 2010b. The toxicokinetics of ketoprofen in Gyps coprotheres: toxicity due to zero-order metabolism. Archives of Toxicology 84, 761e766.

Oaks, J.L., Gilbert, M., Virani, M.Z., Watson, R.T., Meteyer, C.U., Rideout, B.A., Shivaprasad, H.L., Ahmed, S., Chaudhry, M.J., Arshad, M., Mahmood, S., Ali, A., Khan, A.A., 2004. Diclofenac residues as the cause of vulture population decline in Pakistan. Nature 427, 630e633. Prakash, V., Green, R.E., Pain, D.J., Ranade, S.P., Saravanan, S., Prakash, N., Venkitachalam, R., Cuthbert, R., Rahmani, A.R., Cunningham, A.A., 2007. Recent changes in populations of resident Gyps vultures in India. Journal of the Bombay Natural History Society 104, 129e135. Rendic, S., Guengerich, F.P., 2010. Update information on drug metabolism systems e 2009, Part II. Summary of information on the effects of diseases and environmental factors on human cytochrome P450 (CYP) enzymes and transporters. Current Drug Metabolism 11, 4e84. Sekercioglu, C.H., Daily, G.C., Ehrlich, P.R., 2004. Ecosystem consequences of bird declines. Proceedings of the National Academy of Sciences of the USA 101, 18042e18047. Senacha, K.R., Taggart, M.A., Rahmani, A.R., Jhala, Y.V., Cuthbert, R., Pain, D.J., Green, R.E., 2008. Diclofenac levels in livestock carcasses in India before the 2006 “ban”. Journal of the Bombay Natural History Society 105, 148e161. Shelver, W.L., Shappell, N.W., Franek, M., Rubio, F., 2008. ELISA for sulfonamides and its application for screening in water contamination. Journal of Agricultural and Food Chemistry 56, 6609e6615. Sheng, W., Xia, X., Wei, K., Li, J., Li, Q.X., Xu, T., 2009. Determination of marbofloxacin residues in beef and pork with an enzyme-linked immunosorbent assay. Journal of Agricultural and Food Chemistry 57, 5971e5975. Shultz, S., Baral, H.S., Charman, S., Cunningham, A.A., Das, D., Ghalsasi, D.R., Goudar, M.S., Green, R.E., Jones, A., Nighot, P., Pain, D.J., Prakash, V., 2004. Diclofenac poisoning is widespread in declining vulture populations across the Indian subcontinent. Proceedings of the Royal Society of London Series B Biological Sciences 271, S458eS460. Singh, S., 2008. Subject: Prohibition to manufacture, sale and distribution of diclofenac and its formulations for animal use. Letter to ‘Manufacturer’s Associations’ from the ‘Drug Controller General (India)’, 6th August 2008, F.No. 1801/2007-DC. Swan, G.E., Cuthbert, R., Quevedo, M., Green, R.E., Pain, D.J., Bartels, P., Cunningham, A.A., Duncan, N., Meharg, A.A., Oaks, J.L., Parry-Jones, J., Shultz, S., Taggart, M.A., Verdoorn, G., Wolter, K., 2006a. Toxicity of diclofenac to Gyps vultures. Biology Letters 2, 279e282. Swan, G.E., Naidoo, V., Cuthbert, R., Green, R.E., Pain, D.J., Swarup, D., Prakash, V., Taggart, M.A., Bekker, L., Das, D., Diekmann, J., Diekmann, M., Killian, E., Meharg, A.A., Patra, R.C., Saina, M., Wolter, K., 2006b. Removing the threat of diclofenac to critically endangered Asian vultures. PLoS Biology 4, 395e402. Swarup, D., Patra, R.C., Prakash, V., Cuthbert, R., Das, D., Avari, P., Pain, D.J., Green, R.E., Sharma, A.K., Saini, M., Das, D., Taggart, M.A., 2007. Safety of meloxicam to critically endangered Gyps vultures and other scavenging birds in India. Animal Conservation 10, 192e198. Taggart, M.A., Senacha, K.R., Green, R.E., Jhala, Y.V., Raghavan, B., Rahmani, A.R., Cuthbert, R., Pain, D.J., Meharg, A.A., 2007. Diclofenac residues in carcasses of domestic ungulates available to vultures in India. Environment International 33, 759e765. Taggart, M.A., Senacha, K.R., Green, R.E., Cuthbert, R., Jhala, Y.V., Meharg, A.A., Mateo, R., Pain, D.J., 2009. Analysis of nine NSAIDs in ungulate tissues available to critically endangered vultures in India. Environmental Science and Technology 43, 4561e4566. Wynne, H., 2005. Drug metabolism and ageing. Journal of the British Menopause Society 11, 51e56. Yeh, L.C., Lee, W.M., Koh, B.W., Chan, J.P., Liu, C.H., Kao, J.P., Chou, C.C., 2008. Development of amoxicillin enzyme linked immunosorbent assay and measurement of tissue amoxicillin concentrations in a pigeon model. Poultry Science 87, 577e587.