SEARCH

SEARCH BY CITATION

Keywords:

  • Sediment;
  • Environmental risk assessment;
  • Volume of distribution;
  • Aquatic;
  • Fate

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. SUPPORTING INFORMATION
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Pharmaceuticals released into aquatic systems are expected to sorb to sediments to varying degrees. Their sorption is likely to influence their fate and, ultimately, the risk they pose to aquatic organisms. This has led to the European Medicines Agency requiring an assessment of affinity to solids, using batch sorption methods, for the environmental risk assessment (ERA) of new human medicines. However, a large body of data is generated before pharmaceuticals are released onto the market, including their extent of distribution throughout the human body, measured by the volume of distribution (VD). In the present study, batch sorption experiments were undertaken using 12 different soils and sediments to determine whether VD was a good indicator of experimental Kd values for 21 pharmaceuticals. The r2 values obtained from the regressions ranged from 0.39 to 0.76 (with a median value of 0.5) and all regressions were found to be significant. The use of this more comprehensive set of soils and sediments was consistent with previous studies comparing VD and Kd, despite the Kd values of the selected pharmaceuticals varying greatly between soils. The relationship between Kd and VD was greatly improved when zwitterionic antibiotics and carbamazepine were not included, possibly due to complex sorption or pharmacokinetic behavior. There are likely to be a number of factors affecting the sorption of pharmaceuticals that cannot be explained by VD. However, further work may elucidate how these factors can be accounted for, enabling VD to be effectively used to facilitate the ERA of human pharmaceuticals with already available information.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. SUPPORTING INFORMATION
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Considerable effort has been undertaken to establish the potential ecological risk that pharmaceuticals may pose to aquatic environments. Environmental risk assessments (ERAs) are generally based on the level of exposure to aquatic organisms determined by both the concentrations and environmental fate of pharmaceuticals in aquatic ecosystems. Various studies have demonstrated that sorption of pharmaceuticals is one of the major factors in determining their overall environmental fate in both sediments [1–5] and soils [5–10]. Quantifying the sorption of pharmaceuticals has been incorporated into the European Medicines Agency ERA guideline for human medicinal products [11]. This ERA must be undertaken prior to the registration of medicinal products for human use and the extent of sorption needs to be defined experimentally using standard batch equilibrium systems [12]. While batch sorption systems are commonly used to assess sorption with solid phases varying in their physicochemical properties, the variability encountered within natural systems would far exceed those of the standardized systems [13,14]. The sorption of pharmaceuticals has been shown to be highly dependent on the properties of receiving environments, and distribution coefficient (Kd) values can vary widely for a particular pharmaceutical [5,6,15,16]. Given the large number of pharmaceuticals currently marketed (and potentially being developed for market) and the range of environments pharmaceuticals are discharged into, it is highly desirable to at least have approaches that could reduce the reliance on experimentally generated sorption data.

It has been previously suggested that the use of pharmacological data may be one area worth exploiting for obtaining an estimate of the potential effects of pharmaceuticals in aquatic systems for ERA guidance [17–19]. Also, pharmacological data have been used to estimate the fate of pharmaceuticals in aquatic systems, particularly in relation to the potential extent of sorption [20,21]. In this case, the volume of distribution (VD), used to measure the extent a dosed pharmaceutical will distribute from the blood into the tissues, was suggested as being a useful indicator of the relative Kd value of pharmaceuticals. Determining VD is based on the ratio between the total amount of pharmaceutical dosed in humans and the amount remaining within the bloodstream, effectively measuring the fraction of a dosed pharmaceutical that distributes into tissue, fats, and the central nervous system [20]. The extent of VD is related to the interaction between a number of parameters specific to the human body and those of a dosed pharmaceutical. Parameters of the human body that can influence VD include lipid content of the body, protein binding affinity for drugs, pH gradients across membranes, and membrane composition [22]. Physicochemical properties of a drug that can influence VD include acid-dissociation coefficient (pKa) values of present functional groups, pH-dependent octanol-water coefficient (KOW) values, and molecular weight [22]. Similarly, the extent of distribution of an organic contaminant to the solid phase is dependent on the interaction between a number of environmental parameters and the physicochemical properties of the contaminant. Environmental parameters important in determining the extent of distribution include ion exchange capacity, organic carbon content and quality of solids, and solution properties including pH and presence and type of ionic and colloidal materials [13,14].

Table Table 1.. Pharmaceuticals selected for batch sorption experiments, including their log octanol-water partition coefficient (log KOW), acid dissociation constant (pKa) and volume of distribution (VD) values
PharmaceuticalTherapeutic categoryLog KOWapKabVDc (L)
  1. a Values obtained from the Syracuse Research Corporation ([43]; www.syrres.com/esc/physdemo.htm).

  2. b Compounds have either acidic (a) or basic (b) functional groups or were neutral (n); values obtained from the Syracuse Research Corporation [43].

  3. c Values obtained from Thummel and Shen [38].

Amoxicillin (AMX)Antibiotic0.332.4, 9.6 (a/b)15
Atenolol (ATL)Antihypertensive–0.119.6 (b)66.5 ± 10.5
Caffeine (CAF)Stimulant–0.07— (n)42.7 ± 1.4
Carbamazepine (CBZ)Anti-epileptic2.5— (n)98 ± 35
Cimetidine (CIM)Anti-ulcerative0.216.8 (b)70 ± 14
Chlorpheniramine (CHP)Antihistamine3.389.13 (b)224
Diclofenac (DCF)Anti-inflammatory4.44.5 (a)11.9 ± 7.7
17α-Ethynylestradiol (EE2)Contraceptive3.67— (n)265 ± 56
Ibuprofen (IBU)Anti-inflammatory3.54.4 (a)10.5 ± 1.4
Imipramine (IMI)Antidepressant4.89.5 (b)1,260 ± 140
Ketoprofen (KET)Anti-inflammatory3.124.45 (a)11
Metoprolol (MET)Antihypertensive1.889.7 (b)294
Naproxen (NAP)Anti-inflammatory3.184.2 (a)11
Norfloxacin (NFX)Antibiotic–1.036.22, 8.51 (a/b)215
Paracetamol (PAC)Analgesic0.519.7 (a)66.5 ± 8.4
Promethazine (PRM)Antihistamine4.819.1 (b)970
Propranolol (PRL)Antihypertensive3.569.45 (b)301 ± 4.2
Sulfamethoxazole (SFM)Antibiotic0.891.69, 5.57 (a)14.7
Trimethoprim (TRM)Antibiotic0.917.12 (b)112
Verapamil (VER)Antihypertensive3.798.92 (b)350

Therefore, VD and Kd are both parameters that integrate a number of processes that influence distribution in their respective systems, including how the properties of the system can interact with the physicochemical properties of the pharmaceutical and vice versa. These previous studies suggested VD could explain a reasonable degree of experimentally obtained Kd values although, being a proof of concept, they were undertaken in a limited number of systems. In the present study we have investigated the application of the approach on a much larger dataset in order to further establish the potential of this approach for indirect assessment of experimentally measured Kd using the surrogate parameter VD. The specific objectives of the present study were to assess the extent that published values of VD for a range of classes of pharmaceuticals (21 pharmaceuticals in total) can be utilized to estimate relative Kd in 12 soils and sediments with varying soil properties and to assess what types of pharmaceutical compounds are particularly amenable to such an approach.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. SUPPORTING INFORMATION
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Chemicals

The 21 pharmaceuticals used in the present study were all analytical grade reagents and were obtained from SigmaAldrich. The mass of the pharmaceuticals used to prepare stock solutions took into account the purity of the received pharmaceuticals, as well as the mass of salts. A summary of the pharmaceuticals selected for the present study is provided in Table 1.

Calcium chloride (CaCl2; Biolab), sodium formate (Aldrich), formic acid (Aldrich), and high performance liquid chromatography (HPLC)-grade acetonitrile (Mallinckrodt Baker) were also used for experimental work.

Soils

A total of eight soils collected from various agricultural regions around Australia and four sediments collected from a catchment area for a local reservoir were selected for the batch sorption experiments. All soils were air-dried at 30°C and sieved through a <0.425 mm mesh-size. The soils were selected based on the diversity of a number of their physicochemical properties, such as pH, organic carbon content clay content, and cation exchange capacity, which are summarized in Table 2.

Table Table 2.. Selected physicochemical properties of soils and sediments selected for batch sorption experiments. Soils are named from their geographical locations while sediments are named alpha-numerically. All soils and sediments were collected from regions in Australiaa
Soil/sedimentpHEC (mS/cm)%OCCEC (meq/100 g)%Clay
  1. a EC = electrical conductivity (1:5 soil:solution); %OC = % organic carbon content of soil and sediment; CEC = cation exchange capacity.

Berrigan6.150.050.42.57
Booleroo7.310.11.08.523
Cooke Plains6.350.040.746
Emerald Black7.490.10.937.159
Minnipa7.190.030.231
Mintaro6.630.132.620.341
Mt Shank5.510.37.018.822
Tepko6.260.091.04.48
A547.400.150.44.65
A577.171.258.636.255
A516.220.020.0802.5
A026.320.10.611.322.5

Analytical methods

Analysis of pharmaceuticals was undertaken using an Agilent 1100 HPLC coupled with photodiode array detection. An Alltima C18 (250 × 4.6 mm; 5 μm) with an Alltima C18 guard column (7.5 × 4.6 mm; 5 μm) (Grace Davison) was used for separation of the pharmaceuticals. A formate buffer (10 mM sodium formate, 1% formic acid [w/v], pH 3) and acetonitrile were used as the mobile phase and had the following gradient program for the formate buffer: 90% for 4 min, 65% buffer for 6 min, 45% for 12 min, 30% for 16 min, 90% for 22 min, and 8 min re-equilibration giving a 30 min run-time per sample. The flow rate of the mobile phase was 1.2 ml/min into which 100 μl of sample was injected.

The detection wavelengths were generally 280 nm, although a number of pharmaceuticals were detected at other wavelengths, including 230 nm (cimetidine [CIM] and ibuprofen [IBU]), 254 nm (amoxicillin [AMX], paracetamol [PAC], and trimethoprim [TRM]) and 290 nm (carbamazepine [CBZ], imipramine [IMI], and propranolol [PRL]).

Calibration curves were generated from seven points with a range from 10 to 1,000 μg/L, with two calibration curves generated per run. Calibration standards were prepared from a 1 g/L stock solution, while 20, 75, and 750 μg/L quality control samples were prepared separately in duplicate. Calibration was accepted if variation was within 10% of intra- and interday analyses and regression values were >0.99, while quality control samples were accepted if variation was within 20% of intra- and interday analyses. Limits of quantification were determined from six repeat injections at concentrations no greater than five times the apparent detection limit of the respective pharmaceuticals; a limit of quantification was defined as 10 standard deviations of the mean response.

Batch sorption

In general, the batch sorption experiments followed Organization for Economic Co-operation and Development guidelines [12]. Experiments were undertaken in 15 ml glass tubes with polytetrafluoroethylene-lined lids. The solution used was 10 mM CaCl2, with a total volume of 10 ml. Prior to the batch sorption experiment, soil samples were pre-equilibrated with 5 ml of 10 mM CaCl2 for 24 h, with the final 5 ml containing the pharmaceuticals added after this period. Methanol was necessary to solubilize a number of pharmaceuticals, although the final methanol concentration in the total volume of solution used for the batch sorption experiments was no more than 0.1%. Pharmaceuticals were spiked at seven individual concentrations, with initial solution concentrations ranging from 500 to 5,000 μg/L, in order to create sorption isotherms.

The amount of soil used was determined from preliminary experiments to ensure the final proportion of pharmaceutical in solution was between 20 and 80% of that originally spiked. The soil:solution ratios were then selected as 1:2, 1:100, and 1:1,000, due to the marked variability in sorption. This also ensured that there was no co-elution of pharmaceuticals during HPLC analysis. Preliminary experiments also determined that there was no effect on the respective sorption values whether the pharmaceuticals were spiked individually or as a mixture in the batch sorption system (data not shown). An apparent equilibrium of solution concentration was found to occur after 16 h for all pharmaceuticals. Degradation was unlikely to have occurred over this period since no change in solution concentration occurred for up to 24 h during the batch sorption experiments for all pharmaceuticals.

Glass tubes were shaken on a rotating shaker for 16 h, followed by centrifugation at 650 g for 30 min. A 1 ml aliquot was taken from the supernatant for analysis of the pharmaceuticals by HPLC. A 1 ml aliquot of pharmaceutical spiking solution was also analyzed by HPLC to determine the initial amount added to the batch sorption system.

Based on the concentration of pharmaceuticals measured in solution and the amount initially spiked into the batch sorption system, a Freundlich sorption isotherm could be used to determine the extent of sorption based on the following equation:

  • equation image(1)

which can also be expressed in the logarithmic form, following rearrangement:

  • equation image(2)

where Caq is the measured concentration (μg/L) of the pharmaceutical in solution, Cs is the indirectly measured concentration (μg/kg) of the pharmaceutical in the solid phase, KF is the Freundlich distribution coefficient, and n is the linearity coefficient [12]. A plot of Cs against Caq can then be used to determine the value of n (slope) and KF (x-intercept) using Equation 2. When n = 1, this indicates there is no concentration dependence evident and Equation 1 can be simplified to

  • equation image(3)

where Kd is the distribution coefficient. To enable Kd values to be used for comparison with VD values, KF was only used to estimate the value of Kd for comparison with VD where n (Eqn. 2) was found to exist within the range of 0.8 to 1.1. This ensured Kd would be consistent over a range of concentrations.

RESULTS AND DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. SUPPORTING INFORMATION
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Soil-dependent variability of sorption

Sorption was found to vary markedly for individual pharmaceuticals, depending on the soil or sediment used. There was also a large variation in the sorption values between respective pharmaceuticals, highlighted in sorption isotherms obtained in Minnipa soil (Fig. 1) but were generally consistent with literature. The greatest Kd values were found with cationic pharmaceuticals, especially for chlorpheniramine (CHP) (Kd range from 11 to 370 L/kg), IMI (44 to 7,333 L/kg), promethazine (PRM) (206 to 1,575 L/kg), verapamil (VER) (1,341 to 5,876 L/kg), and PRL (0.6 to 129 L/kg). At the pH range of the batch sorption experiments, norfloxacin (NFX) and AMX would have existed as zwitterions, where they had both cationic and anionic functional groups. Norfloxacin had the highest Kd values of all the pharmaceuticals, with values ranging from 2,200 to 273,000 L/kg. Other studies assessing the sorption of fluoroquinolone antibiotics, including NFX, have determined Kd values of up to more than 28,000 L/kg [9,23,24]. The Kd values of AMX ranged from 15 to 599 L/kg in the eight soils and four sediments tested.

Conversely, pharmaceuticals with only an anionic functional group for the experimental pH range, including diclofenac (DCF; Kd range from 1 to 18 L/kg), ketoprofen (KET) (0.1 to 10 L/kg), IBU (0.1 to 11 L/kg), naproxen (NAP) (0.2 to 17 L/kg), and sulfamethoxazole (SFM) (0.05 to 15 L/kg), were found to exhibit the lowest affinity to the solid phase. Also, CBZ and paracetamol were found to have a consistently low extent of sorption, with Kd values ranging from 0.1 to 20 L/kg for CBZ and 1 to 54 L/kg for paracetamol. Previous studies on the sorption of nonsteroidal anti-inflammatory drugs (NSAIDs) show their affinity to solids is generally low, with Kd values of CBZ and SFM also comparably low [15,16,25–29]. However, sorption of acidic DCF has also been found to be substantial, relative to other solids, when the pH of the system is reduced [15,16]. In this case, the pH of the environment could have led to hydrophobic interactions becoming more important for an acidic compound such as DCF. This case would only occur to a significant extent when the environmental pH was within 2 pH units of the pKa of the compound.

thumbnail image

Figure Fig. 1.. Sorption isotherms of test pharmaceuticals in Minnipa soil in which log distribution coefficient (Kd) values were derived from concentrations in the aqueous phase (log Caq) compared with the solid phase (log Cs), demonstrating the large variability in extent of sorption between compounds.

Download figure to PowerPoint

It was expected that sorption would be highly variable for respective pharmaceuticals due to the broad range of soils selected. This is especially the case for a number of the test pharmaceuticals, considering their ionic nature and how factors such as pH and ionic strength of solutions can affect sorption [6,29,30]. For example, the sorption of an acidic antibiotic, a basic antibiotic, and a multiply ionized antibiotic were found to be highly dependent on the pH and ionic strength of soil solutions [31]. In this case, pH affected the extent of ionization of the respective compounds, which is likely to have had an influence on the sorption mechanisms. Also, this influence of ionization was further compounded by variation in ionic strength of solution where, for example, competitive sorption between ions could occur. Variation in soil and sediment physicochemical properties, such as quality of organic carbon and surface charge characteristics, has been demonstrated as having an important influence on sorption of organic compounds [4,13,14,24,32]. For example, a review by Delia Site [14] demonstrated that sorption coefficients of a range of pesticides could be correlated with a number of soil properties, including pH, organic carbon content, cation exchange capacity, and clay content.

The influence of measured soil properties (Table 2) on the Kd values attained in the present study (see Supporting Information; http://dx.doi.org/10.1897/08–587.S1) were assessed using multiple regression analysis ([33]; www.R-project.org). Based on this analysis, only significant soil effects on the obtained Kd values could be demonstrated for electrical conductivity, although this was of minor importance relative to VD. However, based on both previous studies and the present study, it is apparent that both soil and compound effects work in conjunction to influence the extent of sorption.

Concentration dependence of sorption

The values of n (Eqn. 2) for all sorption isotherms of NFX were below 0.8 (range 0.23 to 0.77), showing the sorption of NFX in all of the batch sorption systems had a strong degree of concentration dependence. Therefore, NFX was not included in the regression analysis with VD, since Kd would be highly dependent on concentration, making it difficult to make a valid comparison with other compounds. A number of other drugs apart from NFX were also excluded due to their high degree of concentration-dependent sorption. The compounds showing nonlinearity, where n < 0.8, included cations such as verapamil (Cooke Plains, Berrigan, Minnipa, Tepko, Mount Shank, and Booleroo soils and A54, A51, and A02 sediments), PRM (Cooke Plains and Mount Shank soils), PRL (Mount Shank and Emerald Black soils and A57 and A02 sediments), CIM (Minnipa soil), atenolol (ATL) (A54 sediment), and IMI (Minnipa soil). Another strongly sorbing group of cationic pharmaceuticals, the selective serotonin reuptake inhibitors, also exhibit a considerable degree of nonlinearity [5], which is consistent with VER, PRM, and PRL in the present study. Other pharmaceuticals that exhibited concentration-dependent sorption included 17α-ethynylestradiol (EE2) (Mount Shank soil), TRM (Mount Shank soil), IBU (Cooke Plains and Emerald Black soils and A02 sediment), NAP (Emerald Black soil), and DCF (Cooke Plains and Emerald Black soils and A02 sediment). Previous studies have also demonstrated a similar degree of concentration-dependent sorption of pharmaceuticals, including CBZ, SFM, PRL, EE2, DCF, and NAP [4,10,16,34]. For all Kd values derived in the current study from Equation 2, the r2 value for the linear regression was always ≥0.9 but generally >0.97.

thumbnail image

Figure Fig. 2.. A comparison between distribution coefficient (Kd) values derived from batch sorption experiments compared with volume of distribution (VD) values for test pharmaceuticals in eight soils. The legend indicates the number of pharmaceuticals within respective batch sorption systems and r2 and p values associated with each regression. The location of the amoxicillin (AMX) data is also shown.

Download figure to PowerPoint

Comparison between Kd and VD

The number of pharmaceuticals used for the comparison in each batch sorption experiment ranged from n = 13 to n = 18, depending on the number of pharmaceuticals within each soil displaying strong concentration dependence. Using the Kd values for respective pharmaceuticals in each test system, the r2 value of the regression between VD and Kd ranged from 0.39 (p = 0.001) in the Cooke Plains soil to 0.76 (p < 0.001) in the Mintaro soil (Figs. 2 and 3), with a median r2 = 0.5. All of the regressions were found to be significant, with p values ranging from <0.001 to 0.01 (Figs. 2 and 3).

The r2 values obtained in the present study are comparable with data from our previous, less comprehensive study, where r2 values ranged from 0.62 to 0.72 in three different systems using 12 or 13 pharmaceuticals [20]. Another study comparing Kd and VD values in a biosolid, using 19 pharmaceuticals distinct from our work, also found an r2 value of 0.44 (p = 0.005) [21]. Therefore, despite a number of different batch sorption systems being used for a wide range of pharmaceuticals, a reasonably consistent degree of relative Kd values can be estimated from the value of VD. The use of a number of pharmaceuticals in a range of soil and sediment types has also allowed comparisons between pharmaceuticals of similar therapeutic and physicochemical properties. For example, the NSAIDs used in the present study (DCF, IBU, KET, and NAP) have a range of log KOW values while their VD values are much more comparable (Table 1). The VD values of these NSAIDs are relatively low, compared with other pharmaceuticals, which would also suggest that their Kd values should also be relatively low. Indeed, Kd values of the NSAIDs were comparatively much lower than the other pharmaceuticals tested and generally had similar Kd values in all of the test soils and sediments. Sulfamethoxazole was also found to have a comparably low extent of sorption in all systems, while also having a comparably low VD value (Table 1).

Test pharmaceuticals not amenable to VD estimation of Kd

The Kd value of CBZ was also consistently comparable with those of the NSAIDs, although it has a VD that is nearly 10 times greater than those of the NSAIDs (Table 1). Conversely the antibiotics NFX and AMX both had relative Kd values considerably higher than what was generally predicted from the Kd/VD plots. Although NFX was not included in any of these Kd/VD plots due to its concentration dependent sorption, its log Kd value calculated from Equation 2 ranged from 3.34 to 5.44 L/kg, despite having a VD of 215 L. Also, the log Kd values calculated for AMX ranged from 1.18 to 2.72 L/kg, while its VD is only 15 L. A previous assessment of AMX sorption in sterilized activated sludge also demonstrated it has a relatively high affinity to the solid phase [35]. Sorption of zwitterions can be complicated by a number of processes controlling sorption, including ionic interactions of the cationic and anionic species and sorption of the electrically neutral zwitterion [8,36,37]. In this case, an increasingly complex sorption profile may decrease the effectiveness of VD as a surrogate parameter. Exclusion of AMX from the dataset comparing Kd with VD led to an increase in the r2 value and significance of the regressions for all soils and sediments. For example, r2 increased to 0.58 (p < 0.001) in Cooke Plains soil, for Berrigan r2 = 0.71 (p < 0.001), for Minnipa r2 = 0.59 (p = 0.001), for Booleroo r2 = 0.73 (p < 0.001), and for sediment A51 r2 = 0.75 (p < 0.001) when AMX was not included in the regression analysis.

thumbnail image

Figure Fig. 3.. A comparison between Kd values derived from batch sorption experiments compared with VD values for test pharmaceuticals in four sediments. The legend indicates the number of pharmaceuticals within respective batch sorption systems and r2 and p values associated with each regression. The location of the amoxicillin (AMX) data is also shown.

Download figure to PowerPoint

However, the complexity of ionic sorption processes cannot be used to explain the anomalous behavior of nonionic CBZ. Since its Kd was consistently found to be low in the present study (as in other studies), its exclusion from the Kd/VD dataset also considerably improved the ability of VD to estimate sorption, often to a similar extent as when AMX was removed. While removing these pharmaceuticals from the regression analysis may improve the relationship between Kd and VD, it does not address the reasons why the sorption of zwitterionic pharmaceuticals and CBZ cannot be well explained by their VD values. It is therefore not clear whether the anomaly in the comparison between the two parameters is due to physiology-specific factors in humans or sorption-specific factors in soils or sediments. In the case of the multivalent pharmaceuticals, the respective mechanisms for their distribution in either the body or the environment may be too complex and, therefore, divergent to allow VD to effectively estimate the Kd value. The zwitterionic character of AMX and NFX could be an important factor in their relatively high degree of sorption, as can be found with other drugs with zwitterionic functional groups such as other fluoroquinolones and tetracyclines [8,23,31,37]. At the same time, tetracyclines and other fluoroquinolone antibiotics have VD values that are generally less than 100 L [38]. All of the above ionic drugs with complex chemistry are not particularly suitable to estimates of sorption from a surrogate parameter such as VD. This is analogous with other widely used approaches to estimate Kd values, including those of ionizable or polar organic contaminants using KOC or KOW [29,30].

Alternatively, the pharmacokinetic profile of CBZ is complex due its enhancement of metabolizing enzyme systems within the body [39,40]. Therefore, pharmaceuticals such as CBZ that exhibit complicated pharmacokinetic profiles may also mean their VD has a diminished ability to estimate their potential extent of sorption in aquatic systems.

In general, it has been demonstrated for a broad range of pharmaceuticals under a wide range of conditions that the use of VD as an estimate of relative Kd is reasonably consistent. The ultimate use of VD would be to use the ratio of VD and a known Kd value to estimate the Kd of another pharmaceutical in the same environment or to at least be used as to estimate the relative Kd if there was no previous information on Kd. However, both within the present study and previous studies, the variability of Kd explained by VD has ranged from 0.4 to nearly 0.8, although r2 values of 0.5 to 0.6 are more common [20,21]. Therefore, as expected, some of the variability of Kd cannot be explained by VD alone. A number of factors that are likely to have a strong influence on sorption need to be identified to help account for the factors that cannot be explained by VD alone. Apart from the percentage of organic carbon content of soil and sediment, a number of factors can affect the extent of sorption of pharmaceuticals, particularly those with ionizable functional groups [41,42]. For example, the presence of competing ions in solution and solution pH has an important influence on the sorption of a number of antibiotics [31]. In this case, VD could still be used as a first step in estimating the extent of sorption, followed by further analysis for pharmaceuticals in which ionic interactions may need to be taken into account. It may even be possible to flag pharmaceuticals for which these ionic interactions may be important by using available, drug-specific plasma protein binding data, which are also dependent on ionic interactions [22].

The effect of solution pH on sorption is also worth considering as being a reason for VD not being able to fully explain Kd, since pharmacological pH is constant while environmental pH values vary considerably, as they have in our batch sorption systems. One further step that could be taken to assess the utility of VD would be to assess a large number of structurally similar pharmaceuticals with a range of VD values and physicochemical properties, such as the benzodiazepines, and assess their Kd values under a variety of conditions. Testing the ability of VD to estimate the relative extent of partitioning between structurally related pharmaceuticals would give further insight into whether the already determined strength of the relationship is fortuitous or there is indeed scope for its use within environmental risk assessment.

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. SUPPORTING INFORMATION
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

The relationship between Kd and VD was tested for 21 pharmaceuticals in 12 soils and sediments of diverse physico-chemical properties. Under these conditions, the r2 value ranged from 0.39 to 0.76 (with a median r2 of 0.5), indicating VD could give a reasonable estimate of the relative extent of sorption of each drug within the varying systems. These r2 values are consistent with previous work, despite the high degree of variation in the Kd values of each pharmaceutical in respective systems. The present study therefore establishes the potential utility of an approach for estimating the distribution coefficient of pharmaceuticals in a range of solid phases.

However, while VD may provide an estimate of relative sorption, it cannot fully explain the likely extent of sorption of pharmaceuticals, particularly when the antibiotics AMX and NFX and the anti-epileptic CBZ are used in the analysis. There is therefore a need to seek additional environmental parameters that could allow normalisation to soil or sediment properties to ensure its effective use for ERAs.

SUPPORTING INFORMATION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. SUPPORTING INFORMATION
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Table S1. HPLC parameters for pharmaceuticals selected for batch sorption experiments. LQC, MQC, and HQC are low, mid, and high quality control samples, used throughout each analysis and LLOQ is the lower limit of quantification.

Found at DOI: 10.1897/08–587.S1 (29 KB PDF).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. SUPPORTING INFORMATION
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

The authors would like to thank Debra Gonzago (CSIRO) for technical assistance and Ray Correll (Rho Environmetrics) for statistical analysis. The authors would also like to thank the two anonymous reviewers of this manuscript for their useful comments.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. SUPPORTING INFORMATION
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information
  • 1
    Beausse J. 2004. Selected drugs in solid matrices: A review of environmental determination, occurrence and properties of principal substances. Trends Anal Chem 23: 753761.
  • 2
    Loftier D, Rombke J, Meller M, Ternes T. 2005. Environmental fate of pharmaceuticals in water/sediment systems. Environ Sci Technol 39: 52095218.
  • 3
    Schultz MM, Furlong ET, Kolpin DW, Werner SL, Schoenfuss HL, Barber LB. 2008. Occurrence and distribution of antidepressant pharmaceuticals in water, sediment, and fish brain tissue in two effluent-dominated streams in the U.S. Abstracts, 5th SET AC World Congress, Protecting our Global Environment, Sydney, Australia, August 3–7.
  • 4
    Stein K, Ramil M, Fink G, Sander M, Ternes TA. 2008. Analysis and sorption of psychoactive drugs onto sediment. Environ Sci Technol 42: 64156423.
  • 5
    Kwon JW, Armbrust KL. 2008. Aqueous solubility,n-octanol-water partition coefficient, and sorption of five selective serotonin reuptake inhibitors to sediments and soils. Bull Environ Contam Toxicol 81: 128135.
  • 6
    ter Laak TL, Gebbink WA, Tolls J. 2006. Estimation of soil sorption coefficients of veterinary pharmaceuticals from soil properties. Environ Toxicol Chem 25: 933941.
  • 7
    Kinney CA, Furlong ET, Zaugg SD, Burkhardt MR, Werner SL, Cahill JD, Jorgensen GR. 2006. Survey of organic wastewater contaminants in biosolids destined for land application. Environ Sci Technol 40: 72077215.
  • 8
    Carrasquillo AJ, Bruland GL, Mackay AA, Vasudevan D. 2008. Sorption of ciprofloxacin and oxytetracycline zwitterions to soils and soil minerals: influence of compound structure. Environ Sci Technol 42: 76347642.
  • 9
    Uslu MO, Yediler A, Balcioglu IA, Schulte-Hostede S. 2008. Analysis and sorption behavior of fluoroquinolones in solid matrices. Water Air Soil Pollut 190: 5563.
  • 10
    Chefetz B, Mualem T, Ben-Ari J. 2008. Sorption and mobility of pharmaceutical compounds in soil irrigated with reclaimed wastewater. Chemosphere 73: 13351343.
  • 11
    European Medicines Agency-Committee for Medicinal Products. 2006. Guideline on the environmental risk assessment of medicinal products for human use. EMEA/CHMP/SWP/4447/00. Canary Wharf, London, UK.
  • 12
    Organization for Economic Cooperation and Development. 2000. Adsorption-desorption using a batch equilibrium method. Guideline 106. In OECD Guideline for the Testing of Chemicals. Paris, France, pp 145.
  • 13
    Warren N, Allan IJ, Carter JE, House WA, Parker A. 2003. Pesticides and other micro-organic contaminants in freshwater sedimentary environments—A review. Applied Geochemistry 18: 159194.
  • 14
    Delia Site A. 2001. Factors affecting sorption of organic compounds in natural sorbent/water systems and sorption coefficients for selected pollutants. A review. J Phys Chem Ref Data 30: 187390.
  • 15
    Ternes TA, Herrmann N, Bonerz M, Knacker T, Siegrist H, Joss A. 2004. A rapid method to measure the solid-water distribution coefficient (K(d)) for pharmaceuticals and musk fragrances in sewage sludge. Water Res 38: 40754084.
  • 16
    Drillia P, Stamatelatou K, Lyberatos G. 2005. Fate and mobility of pharmaceuticals in solid matrices. Chemosphere 60: 10341044.
  • 17
    Lange R, Dietrich D. 2002. Environmental risk assessment of pharmaceutical drug substances—Conceptual considerations. Toxicol Lett 131: 97104.
  • 18
    Seiler JP. 2002. Pharmacodynamic activity of drugs and ecotoxicology—Can the two be connected?. Toxicol Lett 131: 105115.
  • 19
    Huggett D, Cook JC, Ericson JF, Williams RT. 2003. A theoretical model for utilizing mammalian pharmacology and safety data to prioritize potential impacts of human pharmaceuticals to fish. Human Ecol Risk Assess 9: 17891799.
  • 20
    Williams M, Saison CLA, Williams DB, Kookana RS. 2006. Can aquatic distribution of human pharmaceuticals be related to pharmacological data?. Chemosphere 65: 22532259.
  • 21
    Hannah R. 2007. Can fate of human pharmaceuticals in wastewater treatment be related to pharmacological data? Compounds of Emerging Concern: What's on the Horizon? Water Environment Federation, Providence, RI, USA, July 29–30.
  • 22
    Wilkinson GR. 2001. Pharmacokinetics. The dynamics of drug absorption, distribution and elimination. In HardmanJG, LimbirdLE, eds, Goodman and Gilman's The Pharmacological Basis of Therapeutics, 10th ed.. McGraw-Hill, New York, NY, USA, pp 330.
  • 23
    Golet EM, Xifra I, Siegrist H, Alder AC, Giger W. 2003. Environmental exposure assessment of fluoroquinolone antibacterial agents from sewage to soil. Environ Sci Technol 37: 32433249.
  • 24
    Zhang JQ, Dong YH. 2008. Effect of low-molecular-weight organic acids on the adsorption of norfloxacin in typical variable charge soils of China. J Hazard Mater 151: 833839.
  • 25
    Scheytt T, Mersmann P, Lindstadt R, Heberer T. 2005. Determination of sorption coefficients of pharmaceutically active substances carbamazepine, diclofenac, and ibuprofen, in sandy sediments. Chemosphere 60: 245253.
  • 26
    Scheytt TJ, Mersmann P, Heberer T. 2006. Mobility of pharmaceuticals carbamazepine, diclofenac, ibuprofen, and propyphenazone in miscible-displacement experiments. J Contam Hydrol 83: 5369.
  • 27
    Jones OAH, Voulvoulis N, Lester JN. 2006. Partitioning behavior of five pharmaceutical compounds to activated sludge and river sediment. Arch Environ Contam Toxicol 50: 297305.
  • 28
    Kimura K, Hara H, Watanabe Y. 2007. Elimination of selected acidic pharmaceuticals from municipal wastewater by an activated sludge system and membrane bioreactors. Environ Sci Technol 41: 37083714.
  • 29
    Carballa M, Fink G, Omil F, Lema JM, Ternes T. 2008. Determination of the solid-water distribution coefficient (K-d) for pharmaceuticals, estrogens and musk fragrances in digested sludge. Water Res 42: 287295.
  • 30
    Kah M, Brown CD. 2008. LogD: Lipophilicity for ionisable compounds. Chemosphere 72: 14011408.
  • 31
    ter Laak TL, Gebbink WA, Tolls J. 2006. The effect of pH and ionic strength on the sorption of sulfachloropyridazine, tylosin, and oxytetracycline to soil. Environ Toxicol Chem 25: 904911.
  • 32
    Ahangar AG, Smernik RJ, Kookana RS, Chittleborough DJ. 2008. Clear effects of soil organic matter chemistry, as determined by NMR spectroscopy, on the sorption of diuron. Chemosphere 70: 11531160.
  • 33
    R Development Core Team. 2009. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
  • 34
    Ying GG, Kookana R, Dillon P. 2003. Sorption and degradation of five selected endocrine disrupting compounds in aquifer material. Water Res 37: 37853791.
  • 35
    Andreozzi R, Caprio V, Ciniglia C, de Champdore M, Lo Giudice R, Marotta R, Zuccato E. 2004. Antibiotics in the environment: Occurrence in Italian STPs, fate, and preliminary assessment on algal toxicity of amoxicillin. Environ Sci Technol 38: 68326838.
  • 36
    Sassman SA, Lee LS. 2005. Sorption of three tetracyclines by several soils: Assessing the role of pH and cation exchange. Environ Sci Technol 39: 74527459.
  • 37
    Nowara A, Burhenne J, Spiteller M. 1997. Binding of fluoroquinolone carboxylic acid derivatives to clay minerals. J Agric Food Chem 45: 14591463.
  • 38
    Thummel KE, Shen DD. 2001. Appendix II. Design and optimization of dosage regimens: Pharmacokinetic data. In HardmanJG, LimbirdLE, eds, Goodman and Oilman's The Pharmacological Basis of Therapeutics. McGraw-Hill, New York, NY, USA, pp 19172024.
  • 39
    Bernus I, Dickinson RG, Hooper WD, Eadie MJ. 1996. Dose-dependent metabolism of carbamazepine in humans. Epilepsy Res 24: 163172.
  • 40
    McNamara JO. 2001. Drugs effective in the therapy of the epilepsies. In HardmanJG, LimbirdLE, eds, Goodman and Oilman's: The Pharmacological Basis of Therapeutics, 10th ed.. McGraw-Hill, New York, NY, USA, pp 521547.
  • 41
    Loke ML, Tjornelund J, Halling-Sorensen B. 2002. Determination of the distribution coefficient (logKd) of oxytetracycline, tylosin A, olaquindox and metronidazole in manure. Chemosphere 48: 351361.
  • 42
    Cunningham VL. 2004. Special characteristics of pharmaceuticals related to environmental fate. In KummererK, ed, Pharmaceuticals in the Environment, 2nd ed.. Springer-Verlag, Berlin, Germany, pp 1324.
  • 43
    Syracuse Research Corporation. 2004. Interactive PhysProp Database. Interactive PhysProp Database. New York, NY, USA.

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. CONCLUSIONS
  7. SUPPORTING INFORMATION
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information
FilenameFormatSizeDescription
10.1897_08-587.S1.pdf35KSupplementary Materials

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.