Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

The suitability of hair as a reliable quantitative indicator of systemic exposure to drugs of abuse remains controversial because of a number of unresolved questions. These include the undetermined chemical mechanism of systemic drug incorporation into hair and its differentiation from drug accumulation and retention in hair after external exposure. The ability to distinguish ingested drug from externally applied drug has been debated (Blank and Kidwell, 1995; Kidwell and Blank, 1996; Cone et al., 1991; Baumgartner and Hill, 1992). Ultimately, a solid understanding of the chemical mechanisms of drug deposition into the hair matrix and the dynamics of drug partitioning within the hair matrix is necessary to establish the validity of using hair sampling for drug detection.

Despite the potential problems associated with hair analysis, hair sampling for drug detection remains a powerful alternative for detecting the use of abused drugs for long periods of time. This reduces the necessity for testing to occur within a short time after exposure, as is required for blood and urine tests. Many drugs disappear rapidly from the urine and blood; therefore, these tests, unlike hair tests, are of limited value in determining a history of drug use.

Fentanyl is one such drug. Because of its narcotic and addictive properties, it has a high likelihood of being abused (Poklis, 1995; Schwartz et al., 1994). Fentanyl is primarily used as a surgical analgesic and anesthetic agent; therefore, the primary abusers of fentanyl are reported to be medical personnel, because of their increased access (Poklis, 1995). Schwartz et al. (1994) were able to detect, by GC/MS, fentanyl in urine only up to 48-72 hr after administration of very large doses. Thus, a test that can detect fentanyl use for a longer period of time after dosing is necessary for effective screening of possible abuse.

Although much work on drug detection in hair has been performed, relatively little work has been done regarding the deposition of fentanyl into hair. Selavka et al. (1995) determined (by a GC/MS method) fentanyl concentrations in the hair of one suspect to be 0.02 ng/mg of hair. Wang et al. (1993) determined fentanyl concentrations (using an immunoassay method of detection) in the hair of patients after the administration of fentanyl for surgery. They determined the range of fentanyl concentrations to be 0.013-0.048 ng/mg of hair, with no significant dose-response relationship. To our knowledge, no systematic studies have examined fentanyl deposition in hair after chronic dosing.

The first objective of this study was to determine the detectability of fentanyl in hair, using a controlled animal model involving chronic fentanyl dosing. Animal models of drug deposition into hair are useful, because they allow for greater experimental control than available with humans while still allowing reasonable comparisons with in vivo conditions for human subjects (Gygi et al., 1995). The mouse model used allowed for rigorous control of the quantity and timing of doses. This model also allowed for the control of pigmentation and, by using albino mice, drug deposition resulting from protein interactions could be distinguished from interactions with hair pigments. Additionally, other factors that affect the structure of the hair, such as diet, interindividual genetic variation, and chemical and cosmetic treatments of the hair, are more easily controlled in an animal model than in humans. Lastly, at 23 days of age, mice are entering their second period of synchronized hair growth for 24 days (Hamilton et al., 1974), so the hair was uniformly growing and incorporating drug during the dosing period used in this study. Therefore, an animal model allowed for rigorous scientific control.

We also examined external loading of fentanyl into drug-free hair. Because the use of blank hair samples soaked in solutions of drug is a common practice for producing calibration standards, the suitability of this method for producing standard materials should be meticulously examined. Therefore, the second objective of this study was to examine aspects of the partitioning of externally applied drug into hair.

Incorporation of drugs, such as fentanyl, from the circulation or from external application might involve ionizable polar functional groups (such as amino and carboxylic acid groups) within the hair that could function as binding sites. The same might be true with nonionizable polar functional groups (such as hydroxyl groups). Kidwell and Blank (1996) have proposed a model for drug incorporation in which hair functions as an ion exchange membrane. They suggested that cationic species have greater affinity for hair and that this interaction is generally with carboxyl groups within the hair. To examine this, they esterified carboxylic acid groups within the hair with methanolic HCl and then examined the binding of cocaine to the hair. They indicated that the binding of cocaine would be reduced because of the treatment with acid even without esterification, and they attempted to control for this effect by soaking the hair in an aqueous buffer solution for 5 days before the binding experiments. They found that less cocaine could bind to the treated hair, suggesting that carboxylic acid groups participated in the binding of cocaine to hair.

Joseph et al. (1997) suggest that drugs are incorporated into hair not as an ion exchange process but as ligand binding to specific sites within the protein matrix of the hair. They contend that hair has sites that can act as cocaine receptors, because they observed cocaine binding to be reversible, stereoselective, and saturable. Additionally, they suggest that this binding site for cocaine is melanin. However, melanin might not be the principle binding site for all drugs; Ishyama et al. (1983) found that methamphetamine bound similarly to both light and dark hair, suggesting that drug binding might be drug specific and for some drugs might involve protein binding sites.

The third purpose of this study was to determine to what extent chemical modification of specific, polar, ionizable functionalities (carboxylic acid and amino groups) within the hair protein matrix affected the deposition and recovery of fentanyl from hair subjected to external loading. These functionalities were chemically modified to less polar, nonionizable groups, and the dynamics of external drug loading were observed. These modifications used chemical methods that did not require exposure to extreme pH conditions, as required by esterification with methanolic HCl (Kidwell and Blank, 1996), which could adversely affect the protein structure through methanolysis of amide linkages and introduce confounding effects. Modifications were independently verified by FT¹-IR and by pH titration.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Hair Samples. For all procedures, hair samples were obtained from untreated or treated BALB/c mice by shaving with an electric shaver to within 1/100 inch of the skin. For the in vitro procedures and modified hair procedures, hair obtained from untreated mice was washed thoroughly in methanol to remove external oils or other contaminants that might affect subsequent experiments.

GC/MS Analysis. All samples were analyzed using a Hewlett-Packard 5890 gas chromatograph coupled to a 5972 mass-selective detector. GC separation used an HP-5 column (25 m × 0.25 mm × 0.25 µm). Head pressure was maintained at 21 kPa. An injection aliquot of 5 µl was used. The temperature was maintained at 80°C for 1 min, ramped at 30°C/min to 150°C, and then ramped at 20°C/min to a final temperature of 300°C; the purge valve was turned on at 1 min. The retention time for fentanyl under these conditions was approximately 13.0 min (fig. 1).

View larger version (28K): [in this window] [in a new window]   Fig. 1.   Sample ion chromatograph showing ions monitored in the single-ion monitoring mode. This sample was from a solution control. The concentration was determined to be 9.5 ng/ml by using fentanyl-d5 (m/z 250, 151, and 194) as the internal standard and taking the 245/250 ratio.

Calibration. A calibration curve was established for a 1-ml sample volume. Standards were diluted in phosphate buffer or methanol, from certified drug standards (Radian), to 0.100, 0.500, 1.0, 5.0, 10.0, 50.0, and 100.0 ng/ml. Two to four replications were used for each of the concentrations. A simple linear regression was performed using Statgraphics 6.0 software (Manugistics). The model line obtained had the equation of ratio = 0.196·concentration, with r² = 99.7% and r = 0.998. Lack-of-fit testing demonstrated no significant model lack of fit. The limit of quantification was approximately 0.2 ng/ml (or 0.02 ng/mg of hair), as determined by the response that was significantly distinguishable from 0 at a 95% confidence level.

In Vivo Procedures. Twenty, 23-day-old, male BALB/c mice were obtained from The Jackson Laboratory (Bar Harbor, ME) and housed in four groups of five mice each. Because almost all excreted fentanyl appears in the urine as norfentanyl (Poklis, 1995) and we analyzed fentanyl levels, urine contamination of the hair was not a significant problem with group housing. Mice were housed at the University of Colorado Health Sciences Center Animal Resource Center, under approved protocols. Fentanyl was obtained as fentanyl citrate (50 µg/ml USP; Abbott Labs, Chicago, IL). Mice were weighed and then given ip injections on Monday, Wednesday, and Friday for 3 weeks. Solutions used to achieve doses of 0.02, 0.05, and 0.1 mg/kg were prepared in 0.9% saline solution. These doses were selected to encompass the ED₅₀ for rats of 52 µg/kg, as determined by Shingu et al. (1983). Control mice were given injections of saline. Mice were weighed once each week, to ensure that the mice gained weight during the study period.

External Loading of Nonmodified Hair. pH Titration of Hair. Ten milligrams of drug-free albino hair were suspended in 10 ml of distilled deionized water. The pH values were monitored with an Accumet glass pH probe and meter (Fisher Scientific, Pittsburgh, PA) as equivalents of NaOH or HCl were added. The titration curve was constructed by expressing all acid and base additions in micromole-equivalents of NaOH per milligram of hair. Two titrations of separate aliquots of hair were performed, to determine the reproducibility of the titrations.

Loading of Hair at Different pH Values. To examine the effect of hair matrix ionization on the external loading of fentanyl into hair, hair was loaded after addition of varying concentrations of NaOH to the methanolic solution. Ten milligrams of hair were suspended in 10 ml of methanol, and 10 ng/ml fentanyl was added as fentanyl citrate. Sodium hydroxide was added to achieve concentrations of 0, 0.001, 0.005, and 0.008 µM. Samples were then continuously stirred for 24 hr. Solution aliquots, methanol washes, and hair extracts were prepared and analyzed as described above.

External Loading Experiments. Drug-free BALB/c mouse hair was obtained, and 10-mg samples were incubated in 10 ml of methanol made basic by the addition of 0.001 µM NaOH. Fentanyl was added to achieve concentrations of 10, 50, and 100 ng/ml. Control samples, without hair, were concurrently analyzed to quantitate fentanyl loss resulting from the basic conditions of the solution. Samples were continuously stirred for the entire time period. At each time point (4, 20, 44, and 68 hr), 1-ml aliquots of the methanolic solutions were removed. These were evaporated to dryness and reconstituted in 50 µl of ethyl acetate.

Modified Hair Experiments. Chemical Modification of Hair. Esterification of solution-accessible carboxylic acid groups within the hair to methyl esters was accomplished by reaction with diazomethane in ether. Diazomethane was produced by reacting 1 g of Diazald (Aldrich, Milwaukee, WI) with 1 g of KOH in a diazomethane generator (Aldrich). Hair (200 mg) was presoaked in ether to reduce the potential formation of explosive diazomethane crystals. This hair was then combined with the diazomethane/ether mixture and allowed to react overnight. The excess diazomethane and ether were allowed to evaporate.

Characterization of Modified Hair. To characterize changes in the chemistry of the hair samples after each chemical modification, titration curves were constructed by suspending 10 mg of hair in 10 ml of distilled deionized water. A glass pH probe (Accumet Basic; Fisher Scientific) was used to monitor the pH as aliquots of NaOH and HCl were added and allowed to equilibrate. Two separate samples of each type of modified hair were titrated, to ensure the reproducibility of the results.

1

External Loading Experiments with Modified Hair. Aliquots (10 mg) of drug-free BALB/c mouse hair were incubated in 10 ml of methanol with 0.001 µM NaOH, so that fentanyl would be present as the free base. Identical samples were incubated in methylene chloride with free-base fentanyl extracted into the loading solution. Fentanyl was added to achieve a concentration of 10 ng/ml. Samples were continuously stirred for 24 hr, after which 1-ml aliquots of solution were removed. These were evaporated and reconstituted in 50 µl of ethyl acetate. This allowed determination of fentanyl removed from the incubation solution. Controls without hair were used to determine drug stability in solution.

Statistical Analysis. All data from these procedures were analyzed by analysis of variance, followed by least-significant difference, multiple-range testing, using Statgraphics 6.0 software. Significance was assumed for all interactions at  = 0.01.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

In Vivo Results. The aqueous extractable fentanyl levels (mean ± SE, N = 5) are presented in fig. 2. The extractable fentanyl concentrations of the dosed groups were significantly different from those of the control animals, although no significant differences were observed among dosed groups. The extractable fentanyl levels appeared to plateau with increasing doses. Additionally, no fentanyl was detectable in the methanol washes of the mouse hair, suggesting that group housing of the mice did not result in significant contamination of the hair.

View larger version (14K): [in this window] [in a new window]   Fig. 2.   Fentanyl extractable, by soaking in phosphate buffer for 24 hr at 45°C, from hair of mice dosed at various doses (mean ± SE, N = 5). Dosed mice had significantly more extractable fentanyl in their hair than did nondosed mice. *, below the limit of quantification.

pH Titration Results. Titration results are presented in fig. 3. The initial pH of the hair suspension was approximately 4.45. Upon titration with NaOH, a reproducible buffering effect was seen at approximately pH 7.4, as demonstrated by the plateauing in the curve at pH 7.4. It is apparent from this plot that enough NaOH needed to be added to the solution to overcome the buffering capacity of the hair and make the solution basic enough that fentanyl was present as its free base. This was achieved at an NaOH concentration of 0.001 µM NaOH (represented in fig. 3 as 5 µmol-equivalents of NaOH/mg of hair). This concentration was used in all of the loading experiments. This concentration of NaOH produced a solution pH of approximately 9.5. 

View larger version (13K): [in this window] [in a new window]   Fig. 3.   pH titration curves for normal and modified hair. Two replicates of titration of hair are shown. Hair was titrated from its initial pH when suspended in water with NaOH or HCl. The initial pH of the solution after equilibration of the hair with the water was 4.5. A reproducible buffering effect was evident at approximately pH 7. The 0.001 µM concentration of NaOH used in subsequent loading experiments is represented on this graph as 5 µmol-equivalents of NaOH/mg of hair; this concentration of NaOH resulted in a pH of 9.5. Acetylated hair appears more acidic, as marked by the shift in the titration curve to the right of the curve for control hair. This is consistent with a loss of amino groups that could act as bases. Esterified hair appears more basic, as marked by the shift in its curve to the left of the curve for control hair. Again, the results are consistent with a loss of ionizable carboxylic acid groups. Both modifications result in the loss of the buffering capacity evident in the control hair at pH 7.6.

Results of Loading at Different pH Values. Fig. 4 shows the loss of fentanyl from the loading solution as a difference from the solution controls (loading solution with no hair present). A significant pH effect was observed in the amount of fentanyl lost from solution, relative to controls, with significantly more fentanyl being lost from solution as the concentration of NaOH was increased.

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Fentanyl remaining in solutions used to load hair at various concentrations of NaOH in methanol (mean ± SE, N = 3). The plot represents the differences between solution controls (with identical solution conditions but with no hair) and sample loading solutions after 24 hr. The pH effect was significant at the  = 0.01 level.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Fentanyl recovered in methanol washes of hair exposed to fentanyl in methanolic solutions containing various concentrations of NaOH (mean ± SE, N = 3). The plot represents the amount of fentanyl detected in the 30-sec methanol washes of hair after loading in methanolic solutions with varying concentrations of NaOH. Only the 0.008 µM samples were significantly greater than 0 at the  = 0.01 level.

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Aqueous extractable fentanyl from hair exposed to fentanyl in methanolic solutions containing various concentrations of NaOH (mean ± SE, N = 3). The plot represents the amount of fentanyl detected in the phosphate buffer extractions of hair after loading in methanolic solutions with varying concentrations of NaOH. The pH effect was significant at the  = 0.01 level.

External Loading Experiment Results for Nonmodified Hair. Fig. 7 illustrates the concentration of fentanyl remaining in the methanolic solution used for the external loading experiments at each of the time points. The decreases in the concentration in solution, at each of the concentrations, over time were significant. In all concentration series, the concentration reached a plateau after approximately 20 hr. No significant loss of fentanyl from solution was observed at any of the concentrations for the analogous aqueous phosphate buffer experiments.

View larger version (9K): [in this window] [in a new window]   Fig. 7.   Concentrations of fentanyl remaining in methanolic solutions used to load hair vs. time (mean ± SE, N = 3). A, Results for loading solutions with the initial fentanyl concentration of 100 ng/ml; B, results for loading solutions starting with 50 ng/ml; C, results for loading solutions starting with 10 ng/ml. Upper lines, concentrations of fentanyl in solution; lower lines, concentration of fentanyl lost at each time point in matched controls with no hair. Trends with time were significant at the  = 0.01 level.

View larger version (16K): [in this window] [in a new window]   Fig. 8.   Fentanyl present in methanol washes of hair loaded in methanol vs. time (mean ± SE, N = 3). The plot shows fentanyl present after various times in the methanol washes of hair loaded in solutions containing the indicated fentanyl concentrations and 0.001 µM NaOH. *, below the limit of quantification.

View larger version (14K): [in this window] [in a new window]   Fig. 9.   Aqueous extractable fentanyl from hair exposed to fentanyl in methanol vs. time (mean ± SE, N = 3). The plot represents results of 24-hr 100 mM phosphate buffer extraction (45°C) of hair loaded in methanolic solutions with 0.001 µM NaOH.

Modified Hair Experiments. pH Titration Results. Results of pH titrations of control, esterified, and acetylated hair are shown in fig. 3. TMS-derivitized hair was not titrated, because the TMS derivatives are labile in aqueous solutions (Nakamuro and Kuwajima, 1993). Results from titrations of two separate aliquots of esterified, acetylated, and control hair are shown in fig. 3, to demonstrate reproducibility.

FT-IR Results. FT-IR results for each of the three modification methods are presented in figs. 10- 12. All figures show two scans from two separate aliquots of control hair and modified hair. In all cases, the two control spectra agree closely. FT-IR spectra demonstrated no significant differences between control and TMS-derivitized hair samples, either in the full spectra (400-4000 cm¹) or in the second derivative of the amide I region (fig. 10). These data are consistent with the lability of the TMS derivatives in the methanolic fentanyl loading solution (Nakamuro and Kuwajima, 1993). Acetylation of hair resulted in a reduction in the intensity of absorbance at 1695 cm¹, compared with controls (fig. 11). Esterified hair samples showed a reduction in the intensity of absorbance at 1594 cm¹ (fig. 12).

View larger version (17K): [in this window] [in a new window]   Fig. 10.   Second-derivative FT-IR spectra of the amide I region for TMS-derivitized hair, compared with control hair. Hair was derivitized with bis(trimethylsilyl)trifluoroacetamide/1% trimethylchlorosilane. No significant differences were evident in the FT-IR analysis. TMS derivatives were possibly labile under the experimental conditions (Nakamuro and Kuwajima, 1993).

View larger version (17K): [in this window] [in a new window]   Fig. 11.   Second-derivative FT-IR spectra of the amide I region for acetylated hair, compared with control hair. Hair was acetylated by heating with acetic anhydride and pyridine, to acetylate amino groups to amides. The loss of absorption at 1695-1700 cm1 indicates a loss of anti-parallel -sheet secondary structure associated with this band (Arrondo et al., 1993). This is consistent with a loss of hydrogen bonding interactions, which typically stabilize -sheet structures, as would be expected with the formation of amides and acetyl esters.

View larger version (19K): [in this window] [in a new window]   Fig. 12.   Second-derivative FT-IR spectra of the amide I region for esterified hair, compared with control hair. Hair was esterified by reaction with diazomethane in ether. The loss of absorption at 1595 cm1 indicates a loss of the COO stretch. This is consistent with the esterification of carboxylic acid groups to methyl esters.

Loading Experiment Results. Fig. 13 illustrates the solution concentrations of fentanyl upon the loading of control hair and chemically modified hair. The results represent the loss of fentanyl from the loading solution and are presented as the differences of the test groups from matched solution controls containing no hair. Results for both methanolic and methylene chloride solutions are shown in fig. 13. Significantly less fentanyl was lost from the methanolic and methylene chloride solutions of the esterified hair than from all other groups. Also, significantly less fentanyl was lost from the methylene chloride solutions of all groups than from the methanolic solutions.

View larger version (27K): [in this window] [in a new window]   Fig. 13.   Loading experiment results. Results are presented as the differences between the test group and matched solution controls without hair. Significantly less fentanyl was lost from the loading solution of esterified hair, compared with all other groups. More fentanyl was lost from the methanol loading solutions than from the methylene chloride loading solutions. All results are the mean of three experiments; error bars, 1 SE.

View larger version (24K): [in this window] [in a new window]   Fig. 14.   Fentanyl recovered from externally loaded hair in 30-sec methanol washes (mean ± SE, N = 3). Fentanyl was detectable in the washes of only the esterified and control hairs loaded in methanolic solutions. No fentanyl was detected in any of the methylene chloride-loaded hair. Results indicate that loading solvent has a significant effect and that methylene chloride may have access to more of the hair matrix than does methanol.

View larger version (20K): [in this window] [in a new window]   Fig. 15.   Fentanyl recovered in a 100 mM phosphate buffer, pH 6, extraction of externally loaded hair (mean ± SE, N = 3). Significantly less fentanyl was recovered from esterified hair than from control hair. Significantly more fentanyl was recovered from acetylated hair loaded in methylene chloride, compared with all other groups.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The results of in vivo fentanyl administration indicated that fentanyl was significantly detectable in the hair of animals subjected to chronic fentanyl dosing in a controlled animal model. The range of mean levels of fentanyl extractable from hair (0.025-0.05 ng/mg of hair) is consistent with the concentrations reported by Selavka et al. (1995) for an assumed chronic user of fentanyl and by Wang et al. (1993) for acutely treated surgical patients. This supports the appropriateness of this animal model to examine the deposition of fentanyl into hair. This may also indicate that the three different extraction methods, i.e. methanol (Wang et al., 1993), dilute HCl (Selavka et al., 1995), and phosphate buffer, are equally effective in extracting fentanyl from the hair matrix.

The data from in vivo mouse studies also indicated that fentanyl deposition plateaus as the dose increases; no linear dose-response relationship was observed. This is also consistent with the lack of a linear dose-response relationship observed after acute treatment of surgical patients by Wang et al. (1993). This suggests that the pool of fentanyl extractable from hair by ionization at low pH or with methanol is rapidly saturated at low doses and additional drug might be deposited in less accessible compartments within the hair. Additionally, this suggests that chronic fentanyl intake might not be distinguishable from acute intake by hair sampling.

The results of the experiments using external loading in methanol demonstrate that the solution of fentanyl reaches an equilibrium with the hair after approximately 20 hr, as demonstrated by the plateauing in the solution fentanyl concentrations after this time (fig. 4). The distribution of fentanyl into the hair matrix appeared to be greatly affected by the pH of the loading solution. As the pH was increased by increasing the concentration of NaOH, the amount of fentanyl that partitioned into the hair increased (fig. 6). The significant amounts of fentanyl lost from solution when the hair was soaked in more basic methanolic solutions indicated that the nonionized form of fentanyl has greater access into the hair than does the ionized form of fentanyl in phosphate buffer. Additionally, because the amount of NaOH added to the loading solution far exceeded the stoichiometric amount necessary to form the free base of fentanyl, functional groups within the hair matrix must also have a pH dependence for fentanyl to partition into the hair. This is somewhat contrary to the model, proposed by Kidwell and Blank (1996) and Blank and Kidwell (1995), of hair as an ion exchange membrane that allows cations into the matrix more readily. However, our results may reflect the compound-specific nature of drug incorporation into hair, which may be more dependent on the structure of the drug and factors other than molecular charge (Uhl, 1997). This may reflect a greater affinity of nonionized fentanyl for more lipophilic compartments within the hair (see fig. 17).

Chemical modification of hair by reaction with diazomethane or acetic anhydride and pyridine to esterify or acetylate hair produced reproducible results, as verified by FT-IR. The band at 1695 cm¹ has been associated with anti-parallel -sheets (Arrondo et al., 1993); therefore, these results suggest a loss of anti-parallel -sheet structure of the proteins in acetylated hair. This is consistent with a loss of hydrogen bonding interactions, which typically stabilize -sheet structures in proteins, as would be expected with the formation of amides and acetyl esters.

The absorbance band at 1594 cm¹ has been associated with the COO stretch (Venyeminov and Kalnin, 1990). The reduction in intensity of this band observed in esterified hair is consistent with the formation of methyl ester groups from carboxylic acid groups and the loss of that stretch component. The increase in absorption at 1585 cm¹ may be attributable to a slower stretch resulting from larger methyl groups substituted onto carboxylic acid groups. The reduced intensity of the band at 1594 cm¹ may be the result of reduction in some -sheet characteristics secondary to the loss of hydrogen bonding by other groups in the protein.

By pH titration, acetylated hair demonstrated a loss of both buffering capacity and basic nature. Additionally, esterified hair lost both its buffering capacity and its acidic nature. Both of these results were consistent with the chemical modifications reducing the amino and carboxylic acid groups within the hair.

Loading experiments demonstrated that significant changes in both loading into hair and recovery from hair of fentanyl were associated with the modification of these functional groups. Esterification of hair resulted in significantly less uptake of drug and significantly lower recovery of drug, relative to controls, suggesting that carboxylic acid groups are necessary for the uptake of drug.

Acetylation of hair resulted in increased removal of fentanyl from methylene chloride loading solutions and increased recovery from these same samples. The same result was not observed in methanol loading solutions. This suggests that the loading solvent has a significant effect and that methylene chloride may have greater access to more lipophilic regions of the hair. Because the amides and acetyl esters formed by acetylation are less polar than the original amino and hydroxyl groups, acetylation potentially resulted in an increase in the lipophilicity of the hair matrix. Thus, the solvent differences observed might have resulted from differences in the polarity of the solvents used in loading.

These results demonstrate that solution-accessible ionizable functionalities of hair play a significant role in the accumulation and retention of fentanyl from external solution. Additionally, these results suggest that functional groups on proteins can play a significant role in the deposition and retention of drugs, specifically nonionized fentanyl, in hair without the presence of melanin.

A compartmental model is suggested in fig. 16, to describe the movement of fentanyl into and within the hair matrix for both ionized and nonionized drug application. K₁ through K₈ represent kinetic coefficients describing the partitioning of fentanyl between compartments. K₁ and K₂ represent the coefficients describing movement into and out of a surface compartment from the loading solution. K₃ and K₄ describe the movement of fentanyl into a more interior lipophilic compartment. K₅ and K₆ describe movement between lipophilic and hydrophilic compartments. K₇ and K₈ describe transfer of fentanyl from the loading solution directly into a hydrophilic compartment. Thus, if fentanyl is present in its free-base lipophilic form, it would partition into the lipophilic compartment via the surface compartment. Aqueous loading of ionized fentanyl would partition directly into a hydrophilic compartment, as described by K₇ and K₈.

View larger version (19K): [in this window] [in a new window]   Fig. 16.   Proposed compartmental model for hair to describe the movement of fentanyl into and within the hair matrix for both ionized and nonionized drug application. K1 and K2 represent the kinetic coefficients describing movement into and out of a surface compartment. K3 and K4 describe the movement of fentanyl into a more interior lipophilic compartment. K5 and K6 describe movement between lipophilic and hydrophilic compartments. K7 and K8 describe transfer of fentanyl from the loading solution directly into a hydrophilic compartment. Thus, if fentanyl is present in the loading solutions as its free-base, lipophilic form, it partitions into the lipophilic compartment via the surface compartment (K1  K8). Alternatively, phosphate buffer loading and extraction accesses the hydrophilic compartment (K8  K1).

The amount of fentanyl extractable from the externally loaded hair also demonstrated no significant relationship with the concentration of the initial loading solution, also indicating possible saturation of the extractable compartment. This process again appeared to reach an equilibrium, because the significant time effect plateaued at about 20 hr. The lack of any relationship between loading concentration and extractable fentanyl is in contrast to the results of Selavka et al. (1995) and Wang et al. (1993), who found linear relations between their spiking concentrations and the amounts of fentanyl extractable from the hair. This may reflect the fact that, over the time that hair was exposed to the loading solutions, equilibria were reached between the solution and the hair and between compartments within the hair, thus saturating the extractable compartment (fig. 16).

The inefficiency of our aqueous extraction suggests that a large pool of tightly bound or inaccessible fentanyl within the hair was created by external loading with nonionized fentanyl. Kidwell and Blank (1996) and Blank and Kidwell (1995) found that significant amounts of phencyclidine may remain in hair exposed to phencyclidine vapor and then subjected to decontaminating washes. Cone et al. (1991) found that significant amounts of externally applied cocaine could be retained after numerous sequential washes of loaded hair. Therefore, our results are consistent with others in indicating that some drugs may be tightly bound or significantly occluded within the hair matrix even when applied exogenously.

A functionally larger, hydrophobic compartment, into which fentanyl is loaded, and a smaller, hydrophilic compartment, from which fentanyl is extracted, could explain the sequestration of a large amount of fentanyl from aqueous extraction. The surface compartment depicted in fig. 16, through which externally applied drug passes, could provide an easily extracted compartment from which the 30-sec methanol washes removed some drug.

Fig. 17 depicts a potential model for the roles of carboxylic acid and amino groups in the deposition of a free-base, nonionized, amine drug into hair. Proton transfers from protein to ligand are commonly invoked in receptor functioning (Giraldo et al., 1992). A generalized amine drug is represented in fig. 17A by R-NH₂, with ionized amine and carboxylic acid groups of the protein matrix. A proton is exchanged to the drug from the charged amine (fig. 17B), thus facilitating the interaction of the drug with the COO group (fig. 17C). This interaction allows a nonionized drug to associate with a more polar, hydrophilic aspect of the hair matrix. Exchange of the proton back to the COO group (fig. 17D) removes the charge from the drug molecule and allows its association with a more lipophilic compartment of the hair matrix. Therefore, modification of COOH groups to nonionizable methyl esters would result in reduced storage of fentanyl.

View larger version (41K): [in this window] [in a new window]   Fig. 17.   Proposed model for the mechanism of ionizable functional group participation in the deposition of nonionized drugs into hair. A, Generalized nonionized drug in proximity to an ionized protein in the hair matrix; B, proton exchange from the protein to the drug, thus adding a charge to the drug; C, facilitation, by this charge, of the association of the drug with the COO group, subsequent exchange of the proton back to the COOH group, and movement of the nonionized drug into a lipophilic compartment within the protein matrix. This model is based on the model of the H2 receptor proposed by Giraldo et al. (1992).

Although this model differs, with respect to the function of carboxylic acid groups, from that proposed by Kidwell and Blank (1996), it reconciles the increased storage of nonionic fentanyl, relative to ionic fentanyl, and the observed participation of ionizable functionalities in drug uptake and retention. This model is consistent with the possibility of binding sites for drugs within the hair, as proposed by Joseph et al. (1997). This type of binding interaction is not dependent on the presence of melanin and could account for drugs that appear to bind independently of hair color, such as methamphetamine (Ishyama et al., 1983).

In the preparation of externally loaded samples as control or calibration samples, it is unclear whether externally applied drug enters and leaves the hair matrix in a manner analogous to that of drugs deposited from the circulation. This is supported by the observation that the externally loaded hair samples, after 44 hr of exposure to either 10, 50, or 100 ng/ml fentanyl, showed extractable fentanyl levels 10 times those of the in vivo samples after administration of 0.02, 0.05, or 0.1 mg/kg for 21 days. This makes the use of externally loaded hair samples as calibration standards problematic. These calibration standards are used in a legitimate effort to eliminate matrix effects from the analysis of unknown samples but, if the extractable drug concentration within the hair is dependent on the time of exposure to the loading solution and the pH of the loading solution, resulting in differential partitioning of the drug within the hair, then these calibrators may not be representative of the concentrations of the drug in unknown samples. The large amount of fentanyl that remains in the hair after extraction also suggests, as do the results reported by Kidwell and Blank (1996) and Blank and Kidwell (1995), that endogenously deposited drug and exogenously deposited drug may be difficult to distinguish from one another.