Utility of coloured hair for the detection of drugs and alcohol
Abstract
The aim of this paper is to assess the utility of coloured hair for the detection of drugs and alcohol in a large statistically significant population. The positivity rate, the 1st, 5th, 50th, 95th, and 99th percentiles of five amphetamines, cannabinoids, cocaine, four opiates, methadone, buprenorphine, seven benzodiazepines, and ethyl glucuronide (EtG) in 9488 non-treated and 1026 cosmetically treated (dyed or bleached) authentic hair samples was compared. Analytical methods used were accredited for forensic purposes at the cut-offs defined by the German driving licence re-granting medical and psychological assessment (MPA) guidelines.
Considering only the drug classes for which at least 10 positive samples were detected, the positivity rate in non-treated hair was highest for alcohol (4.50%; measured using EtG at concentrations ≥ 7 pg/mg hair), followed by THC (2.00%), cocaine (1.75%), and amphetamine (0.59%). While the 1st to 99th percentile range was significantly lower for drugs in treated, compared to non-treated hair, no significant change was observed for EtG. Additionally, no significant difference in the positivity rate was observed between treated hair and non-treated hair for both drugs and EtG. This study is the first attempt to evaluate the influence of cosmetic treatment, mainly dying, on the positivity rate for both drugs and EtG in hair samples submitted routinely for abstinence testing and the first to indicate that dyed and eventually bleached hair is not necessarily useless in detecting drugs and/or alcohol consumption, thus making coloured hair analysis still useful, often being the only possibility to prove such misuse. Copyright © 2014 John Wiley & Sons, Ltd.
Introduction
Any physical, chemical, and mechanical alteration of the natural hair such as perming, straightening, dyeing, bleaching, excessive washing, intensive illumination with ultraviolet radiation, and unusual exposure to the sunlight could damage the hair cuticula. Repeated shampooing was shown to have no effect on drug concentrations in hair.1 On the other hand, washing with special shampoos such as ‘Ultra Clean’ (Zydot Unlimited, Tulsa, OK, USA) resulted in a slight decrease in drug contents in hair as follows: cocaine -5%, 6-monoacetylmorphine (6-MAM) -9%, morphine -26%, tetrahydrocannabinol (THC) -36 %.2 Other investigations showed that treatments with shampoos, conditioners, and peroxide bleaches reduced cocaine concentrations in fortified hair by 60 to 80% after 30 treatments.3 Bleaching and blonding irreversibly destruct melanin to different extents; additionally, strong bleaching increases hair porosity. Hence cosmetic treatment is a potential threat to the otherwise stably stored drugs hair in that it may alter the original drug concentrations deposited in hair. In fact in vitro applications of commercially available bleaching and perming formulas have been shown to reduce the initially spiked morphine, codeine and dihydrocodeine (DHC) solutions by 70–80% after perming and 82–98% after bleaching.4 As opposed to in vitro bleaching, in vivo bleaching resulted in no significant trends in authentic hair samples.4 Another study documents the decrease in drug concentrations reported in bleached hair strands compared to non-bleached natural brown hair strands of a female drug addict as follows: 38% for cocaine, 33% for benzoylecgonine (BE), 41% for cocaethylene, 25% for codeine, 11% for morphine, and 14% for 6-MAM.5 Similarly, in other studies, in vivo bleaching and dying resulted in mean differences of 40–60% for cocaine, BE, codeine, 6-MAM, THC-COOH,6 amphetamine (AM),7 methamphetamine (MA),7 and higher than 60% for morphine.6 While confirming these decreases in opiate concentrations (57.5%, 88.6%, and 67.4%, for codeine, 6-MAM and morphine, respectively) and to a lesser extent in cocaine concentrations (24.6% and 36.4% for cocaine and BE, respectively), Yegles et al. showed also a decrease in benzodiazepine concentrations (39.7, 67.7, and 61.8% for diazepam, nordiazepam, and 7-aminoflunitrazepam, respectively) after applying a bleaching product (Poly Blonde by Schwarzkopf & Henkel, containing typically 6–12% H202) for 20 min in vitro to post-mortem hair.8 Similarly, a 40-min application of the same commercial product on 14 hair samples resulted in a decrease of 20% to 39% (median values) of amphetamine (AM), methamphetamine (MA), 3,4-methylenedioxymethamphetamine (MDMA), and 3,4-methylenedioxyamphetamine (MDA) in hair, which however were all still detectable after the bleaching treatment; moreover their enantiomers’ ratios were unaltered after the in vitro bleaching.9
Traces of the four fatty acid ethyl esters (FAEEs) namely ethyl myristate, ethyl palmitate, ethyl oleate, and ethyl stearate used by some as alcohol markers were detected in all of 49 frequently applied hair care products. False positive FAEEs results in hair were observed after daily treatment with a hair lotion containing 62.5% ethanol as well as with non-alcohol-containing deodorant and hair spray.10 As opposed to FAEEs, the widely used alcohol abstinence and alcoholism marker ethyl glucuronide (EtG) in hair does not seem to be effected by cosmetic treatment.11, 12 Moreover, a negative FAEE result does not overrule a positive EtG result in hair.13 Except for FAEEs, cosmetic treatment generally decreases the drug concentrations in hair as shown above. This decrease becomes critical when the drug concentrations in hair are reduced to a concentration below the cut-off, thus producing false negative results. This probability increases for xenobiotics whose cut-off is very low and close to the limit of detection (LOD) of the analytical method – as was the case for testosterone (LOD at 0.5 pg/mg) which was missed in several specimens due to bleaching with hydrogen peroxide.14 Normal hair washing was postulated to result in false negative EtG results in hair (LOD at 5 pg/mg; LOQ at 10 pg/mg)15 but disproved later by another study16 with more sensitive methods (LOD at 0.6 pg/mg; limit of quantification (LOQ) at 2.8 pg/mg).
The challenge presented by hair cosmetic treatment illustrated above can encourage hair sceptics to mistrust hair testing or even completely ruling it out for certain applications such as driving licence re-granting. Since all the studies concerning the influence of hair colouring on drugs and EtG in hair published so far are mainly in vitro studies and involve few subjects (generally less than 30), the aim of this paper is to statistically compare the positivity rate of hair analyzed for alcohol (EtG) and drugs in dyed or bleached hair and non-treated hair in a large population and hence assess the utility of the former for forensic applications such as driving licence re-granting and child custody cases. This study focuses on the driving licence re-granting cohort, in which abstinence from drugs and alcohol is requested for one year using hair analysis.
Methods
Hair samples
One hundred and twenty-six samples out of a total of 1026 hair samples declared to be cosmetically treated were bleached; the rest, and hence the majority, were dyed in a variety of ways: as a whole or in streaks and to different colours: black, brown, red, and blonde, although the majority were dyed to blonde. These bleached samples were not purposely selected, but represent the real percentage of bleached authentic hair samples in a large population submitted routinely for abstinence testing for alcohol and drugs. Of these, eight samples were positive for drugs or EtG as shown in Table 1, thus making a positivity rate of 6.4%. Due to the small numbers of positive samples in the bleached hair samples, statistical analysis was not possible. Moreover even if hair is stated to be only dyed, bleaching cannot be fully excluded since many dying formulations include bleaching as part of the dying process; hence separation of dying and bleaching processes can be done in controlled studies but is not realistic when analysing hair samples routinely. In contrast to controlled in vitro studies involving a few samples, in such statistical studies concerning around 10 000 samples it is very difficult to obtain the correct information concerning the type of dyeing, namely if it was temporary, semi-temporary or permanent and if the dyeing products contained bleaching components or not. For these reasons, dyeing and bleaching are referred to in this paper collectively as ‘coloured’ or ‘treated’ hair.
| Hair length analysed/cm | Original length/cm | Hair colour | [THC]/ng/mg | [Cocaine]/ng/mg | [BE]/ng/mg | [Cocathylene]/ng/mg | [Morphine]/ng/mg | [6-MAM]/ng/mg | [EtG]/pg/mg |
|---|---|---|---|---|---|---|---|---|---|
| 3.0 | 3.0 | blonde | 0.14 | 0.52 | 0.18 | 0 | |||
| 3.0 | 14.6 | middle blonde | 45.3 | ||||||
| 6.0 | 10.9 | dark blonde | 0.24 | 0.14 | 0.01 | ||||
| 3.0 | 10.2 | dark blonde | 13.3 | ||||||
| 12.0 | 23.2 | blonde | 11.8 | ||||||
| 6.0 | 14.5 | blonde | 0.14 | 1.11 | |||||
| 3.0 | 28.0 | middle blonde | 32.1 | ||||||
| 6.0 | 34.0 | dark blonde | 2.05 | 0.92 | 0.53 |
The different hair lengths analyzed for the two cohorts were plotted in Figure 1 as a function of hair lengths analyzed as follows: <3cm, 3cm, >3<6cm, >6<9cm, >9<=12cm and >12cm. As can be seen no significant difference can be observed between the distributions of hair samples analyzed for the two cohorts, thus eliminating any bias.
Analytical procedures
Two analytically different methods were used as required by international forensic guidelines namely ELISA screening followed by confirmation using gas chromatography-mass spectrometry (GC-MS), gas chromatography-tandem mass spectrometry (GC-MS/MS), or liquid chromatography-tandem mass spectrometry (LC-MS/MS) for the detection of drugs and EtG. Both analytical methods are presented in previous publications.17, 18 The lower limits of quantification (LLOQs) were all below the recommended MPA cut-offs shown in Table 2; As previously published17 the LLOQ of EtG in hair was 2.8 pg/mg and the lower limit of detection (LLOD) detection was 0.6 pg/mg using linear regression at 99% significance level, hence well below the ‘abstinence’ cut-off at 7 pg/mg hair. Analytical methods used for the detection of drugs in hair were accredited according to DIN EN ISO/IEC 17025 for forensic purposes at the cut-offs shown in Table 2 as defined by the German driving licence re-granting medical and psychological assessment (MPA) guidelines.19
| Urine | Hair | |||
|---|---|---|---|---|
| old cut-offs | new* cut-offs | old cut-offs | new* cut-offs | |
| in ng/mL | in ng/mL | in ng/mg | in ng/mg | |
| Cannabinoids | 50 | 10 | 0.1 | 0.02 |
| Cocaine | 300 | 30 | 0.5 | 0.1 |
| Amphetamines | 500 | 50 | 0.2 | 0.1 |
| Opiates | 300 | 25 | 0.2 | 0.1 |
| Benzodiazepines | 200 | 50 | 0.05 | |
| Methadone (& EDDP) | 300 | 50 | 0.1 | |
| Ethyl glucuronide | 100 | 0.007 | ||
| Opioids (Tilidine, Tramadol, Burpenorphine) | 50‡ | 0.01‡ |
Briefly, hair was washed with dichloromethane for 10 min followed by methanol and dried at 50°C in a drier. Decontamination is recommended by international guidelines20-23 in order to avoid/reduce external contamination of drugs. The hair was then grinded using a ball mill, and a 10 mg aliquot was weighed into glass test tubes with hermetic screw caps. Phosphate buffer (pH 6.0, 1.0 mL) was added and the tubes were capped and incubated for five hours at room temperature. The supernatant was used for ELISA screening for amphetamines, cocaine, opiates, methadone and benzodiazepines. For cannabinoids, an extra aliquot was weighed (10 mg), which, after digestion and extraction was subsequently screened using LUCIO-direct ELISA THC kit. The assays were subsequently run according to the manufacturer's instructions. Minor modifications were introduced over the years since our first accreditation to ISO/IEC 17025 and to ISO 15189 in 2005 in order to increase the performance of the screening tests. In 2009 seven ELISA screening tests were re-accredited to DIN EN ISO/IEC 17025 for forensic purposes as required by the new MPA guidelines19 using the validation guidelines of the German Society of Toxicological and Forensic Chemistry (GTFCh)24, 25 for the detection of cannabinoids at 0.02 ng/mg hair, amphetamines, cocaine, opiates, and methadone at 0.1 ng/mg hair and benzodiazepines at 0.05 ng/mg hair using authentic hair samples.
Quality assurance
Apart from running successfully the daily internal quality control, we participate successfully in all external proficiency testing programs organized by the Society of Toxicological and Forensic Chemistry (GTFCh) and the Society of hair testing (SoHT) for all the six drug classes and EtG.
Data analysis
The number of positive hair samples for amphetamines, cannabinoids, cocaine, opiates, methadone, opioids, benzodiazepines, and ethyl glucuronide (EtG) in 9488 non-treated and 1026 treated authentic hair samples (dyed or bleached) routinely analyzed at our laboratory for MPA was calculated and expressed as per cent in order to enable easier comparison between the two groups. The differences observed were tested for significance at the 95 and 99% significance level using the Chi squared test. A summary of the results is presented in Table 3. Additionally, percentiles were calculated for the drug classes for which at least 10 positive samples were detected in treated and non-treated hair and compared. In contrast to the positivity rate, the percentiles highlight any statistical changes in the drug concentrations detected in a population, hence identifying trends and expected concentrations due to dyeing or bleaching of the hair (Table 4). In order to examine more closely any difference in the drug concentrations measured, especially around the cut-off concentrations and hence better assess the influence of dyeing or bleaching, frequency histograms were constructed separately for EtG and drugs for which at least ten samples were positive in both cosmetically treated and non-treated hair samples, (a) as a function of drug concentrations (generally using drug concentration ranges of (1) cut-off concentration, (2) twice, (3) five-fold, (4) ten-fold, and (5) greater or equal to a twenty-fold the cut-off concentration; see Figures 3, 4, 6, 8) and (b) as a function of the hair length analyzed (using the same ranges of hair lengths as in Figure 1, namely: <3cm, 3cm, >3<6cm, >6<9cm, >9<=12cm and >12cm; see Figures 2, 5, 7, and 9).
| Drug type | Pos. treated | Total treated | % Pos. treated | Pos. non-treated | Total non-treated | % Pos. non-treated | Chi2 | Chi2 at p = 0.05 | Significant difference | Chi2 at p = 0.01 | Significant difference |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Cannabinoids | |||||||||||
| THC | 15 | 1026 | 1.46% | 190 | 9488 | 2.00% | 1.41 | 3.84 | No | 10.83 | No |
| Cocainics | |||||||||||
| Cocaine | 14 | 1026 | 1.36% | 166 | 9488 | 1.75% | 0.82 | 3.84 | No | 10.83 | No |
| Benzoylecgonine | 13 | 1026 | 1.27% | 125 | 9488 | 1.32% | 0.02 | 3.84 | No | 10.83 | No |
| Amphetamines | |||||||||||
| Amphetamine | 11 | 1026 | 1.07% | 56 | 9488 | 0.59% | 3.40 | 3.84 | No | 10.83 | No |
| MDMA | 4 | 1026 | 0.39% | 19 | 9488 | 0.20% | 1.53 | 3.84 | No | 10.83 | No |
| Metamphetamine | 4 | 1026 | 0.39% | 9 | 9488 | 0.09% | 6.52 | 3.84 | Yes | 10.83 | No |
| MDA | 1 | 1026 | 0.10% | 4 | 9488 | 0.04% | 0.60 | 3.84 | No | 10.83 | No |
| MDE | 0 | 1026 | 0 | 9488 | |||||||
| Opiates | |||||||||||
| Dihydrocodeine | 0 | 1026 | 0.00% | 5 | 9488 | 0.05% | |||||
| Codeine | 0 | 1026 | 0.00% | 26 | 9488 | 0.27% | |||||
| Morphine | 1 | 1026 | 0.10% | 10 | 9488 | 0.11% | 0.01 | 3.84 | No | 10.83 | No |
| 6-MAM | 2 | 1026 | 0.19% | 8 | 9488 | 0.08% | 1.19 | 3.84 | No | 10.83 | No |
| Benzodiazepines | |||||||||||
| Diazepam | 1 | 1026 | 0.10% | 1 | 9488 | 0.01% | 3.68 | 3.84 | No | 10.83 | No |
| Alprazolam | 0 | 1026 | 0 | 9488 | |||||||
| Bromazepam | 2 | 1026 | 0.19% | 2 | 9488 | 0.02% | 7.36 | 3.84 | Yes | 10.83 | No |
| Nordiazepam | 1 | 1026 | 0.10% | 3 | 9488 | 0.03% | 1.06 | 3.84 | No | 10.83 | No |
| Oxazepam | 1 | 1026 | 0.10% | 0 | 9488 | 0.00% | 3.84 | 10.83 | |||
| Lorazepam | 0 | 1026 | 0 | 9488 | |||||||
| Flunitrazepam | 0 | 1026 | 0 | 9488 | |||||||
| Methadone | 5 | 1026 | 0.49% | 17 | 9488 | 0.18% | 4.21 | 3.84 | Yes | 10.83 | No |
| EDDP | 5 | 1026 | 0.49% | 17 | 9488 | 0.18% | 4.21 | 3.84 | Yes | 10.83 | No |
| Opioids | |||||||||||
| Buprenorphine | 0 | 1026 | 0.00% | 2 | 9488 | 0.02% | |||||
| Tilidine | 0 | 1026 | 0 | 1026 | |||||||
| Tramadol | 0 | 1026 | 0 | 1026 | |||||||
| EtG | 42 | 1026 | 4.09% | 427 | 9488 | 4.50% | 0.36 | 3.84 | No | 10.83 | No |
| N | Percentile 1 | Percentile 5 | Median | Percentile 95 | Percentile 99 | Max | ||
|---|---|---|---|---|---|---|---|---|
| EtG | non-treated | 427 | 8.23 | 9.40 | 23.10 | 179.97 | 395.54 | 683.10 |
| treated | 42 | 9.34 | 11.21 | 32.60 | 216.35 | 311.36 | 354.00 | |
| Cannabinoids | non-treated | 190 | 0.02 | 0.02 | 0.14 | 1.16 | 2.32 | 3.00 |
| treated | 15 | 0.02 | 0.03 | 0.12 | 0.55 | 0.72 | 0.76 | |
| Cocainics | non-treated | 166 | 0.13 | 0.15 | 0.60 | 6.49 | 20.80 | 28.50 |
| treated | 14 | 0.25 | 0.28 | 0.78 | 5.25 | 10.01 | 11.20 | |
| BE | non-treated | 125 | 0.05 | 0.06 | 0.18 | 2.24 | 5.97 | 6.43 |
| treated | 13 | 0.07 | 0.09 | 0.28 | 1.54 | 2.29 | 2.48 | |
| DHC | non-treated | 5 | 0.14 | 0.14 | 0.17 | 24.16 | 28.83 | 30.00 |
| treated | 0 | — | — | — | — | — | — | |
| COD | non-treated | 26 | 0.10 | 0.11 | 0.23 | 0.67 | 1.27 | 1.46 |
| treated | 0 | — | — | — | — | — | — | |
| MOR | non-treated | 10 | 0.11 | 0.11 | 0.30 | 1.54 | 2.04 | 2.16 |
| treated | 1 | 0.14 | 0.14 | 0.14 | 0.14 | 0.14 | 0.14 | |
| 6MAM | non-treated | 8 | 0.16 | 0.16 | 0.33 | 8.04 | 8.75 | 8.93 |
| treated | 2 | 0.60 | 0.62 | 0.85 | 1.08 | 1.10 | 1.11 | |
| Amphetamine | non-treated | 56 | 0.15 | 0.16 | 0.91 | 22.45 | 34.53 | 36.40 |
| treated | 11 | 0.28 | 0.31 | 1.22 | 6.90 | 8.82 | 9.30 | |
| MDMA | non-treated | 19 | 0.13 | 0.16 | 0.70 | 3.26 | 3.82 | 3.96 |
| treated | 4 | 0.14 | 0.17 | 0.45 | 6.19 | 7.00 | 7.20 | |
| MDA | non-treated | 4 | 0.14 | 0.14 | 0.17 | 0.23 | 0.24 | 0.24 |
| treated | 1 | 0.35 | 0.35 | 0.35 | 0.35 | 0.35 | 0.35 | |
| Metamphetamine | non-treated | 9 | 0.49 | 0.55 | 2.33 | 9.76 | 12.83 | 13.60 |
| treated | 4 | 0.23 | 0.37 | 1.76 | 2.49 | 2.54 | 2.55 | |
| MDE | non-treated | 0 | — | — | — | — | — | — |
| treated | 0 | — | — | — | — | — | — | |
| Methadone | non-treated | 17 | 2.98 | 3.32 | 10.50 | 22.78 | 29.44 | 31.10 |
| treated | 5 | 2.73 | 3.24 | 11.70 | 30.74 | 33.35 | 34.00 | |
| EDDP | non-treated | 17 | 0.12 | 0.18 | 0.60 | 1.84 | 2.61 | 2.80 |
| treated | 5 | 0.24 | 0.40 | 1.30 | 6.48 | 6.98 | 7.10 | |
| Diazepam | non-treated | 1 | 0.11 | 0.11 | 0.11 | 0.11 | 0.11 | 0.11 |
| treated | 1 | 0.16 | 0.16 | 0.16 | 0.16 | 0.16 | 0.16 | |
| Alprazolam | non-treated | 0 | — | — | — | — | — | — |
| treated | 0 | — | — | — | — | — | — | |
| Bromazepam | non-treated | 2 | 0.19 | 0.21 | 0.37 | 0.53 | 0.55 | 0.55 |
| treated | 2 | 0.43 | 0.44 | 0.58 | 0.71 | 0.72 | 0.72 | |
| Nordiazepam | non-treated | 3 | 0.10 | 0.10 | 0.10 | 0.14 | 0.14 | 0.14 |
| treated | 1 | 0.46 | 0.46 | 0.46 | 0.46 | 0.46 | 0.46 | |
| Oxazepam | non-treated | 0 | — | — | — | — | — | — |
| treated | 1 | 0.16 | 0.16 | 0.16 | 0.16 | 0.16 | 0.16 | |
| Lorazepam | non-treated | 0 | — | — | — | — | — | — |
| treated | 0 | — | — | — | — | — | — | |
| Flunitrazepam | non-treated | 0 | — | — | — | — | — | — |
| treated | 0 | — | — | — | — | — | — | |
| Buprenorphine | non-treated | 2 | 0.07 | 0.09 | 0.39 | 0.69 | 0.71 | 0.72 |
| treated | 0 | — | — | — | — | — | — |
Results and discussion
Comparing the positivity rate between treated and non-treated authentic hair samples
Surprisingly, as can be seen in Table 3, the positivity rate for none of the drugs and EtG routinely tested for the MPA in Germany in 1026 cosmetically treated authentic samples was significantly lower than the positivity rate for the same drugs in 9488 non-treated hair samples:
4.09 % (dyed or bleached hair) vs 4.50 % (non-treated hair) for EtG, 1.46 % (dyed or bleached hair) vs 2.00 % (non-treated hair) for THC, 1.27 % (dyed or bleached hair) vs 1.32 % (non-treated hair) for the cocaine metabolite benzoylecgonine and 1.07 % (dyed or bleached hair) vs 0.59 % (non-treated hair) for amphetamine, considering only the drugs for which at least ten samples were positive in both groups. The positivity rates were higher for four, two and five dyed or bleached hair samples in which metamphetamine, bromazepam, and methadone were found respectively (Table 3); these were not considered to be relevant in spite of significance at the 95% confidence levels due to the small sample size. Similar results were previously observed26 and concluded that ‘it follows that this (referring to bleaching perming or dyeing) does not play an important role in practical hair analysis’ and that ‘according to these statistics, there is no essential difference in the percentage positive results between natural, dyed and bleached hair.’
Comparing percentiles and frequency histograms between treated and non-treated authentic hair samples
For the abstinence control for drugs and alcohol as is required for the MPA in Germany, the concentration of drugs and EtG is not relevant as long as these are above the defined cut-offs shown in Table 2 and hence classified as positive. Nonetheless, for the purpose of this study the concentrations of the drugs and EtG expressed by the 1st, 5th, 50th, 95th, and 99th percentiles were also compared for treated and non-treated hair for the drug classes for which at least 10 positive samples were detected. The observed results are summarized in Table 4.
Cannabinoids
For cannabinoids, while the range percentile 5 to percentile 95, representing 90% of the positive THC concentrations measured decreased from 0.02–1.16 ng/mg in 190 out of 9488 non-treated hair samples to 0.03–0.55 ng/mg in 15 out of 1026 treated hair samples, no significant change in the median was observed, between treated and non-treated hair (Table 4). In order to investigate any bias in the results due to a different distribution profile, the different hair lengths analyzed for THC in the two cohorts were plotted in Figure 2 as a function of hair lengths analyzed as follows: <3cm, 3cm, >3<6cm, >6<9cm, >9<=12cm and >12cm. As can be seen no significant difference can be observed between the distributions of hair samples analyzed for the two cohorts for THC, thus eliminating any hair length bias.
Previous larger population studies involving 6194 samples analyzed for a wider range of contexts resulted in the same median value for THC.27 This can infer that our population is statistically relevant and hence represents quite well the real population consuming THC. A closer look at the distribution of THC positive samples as function of THC concentrations measured shown in Figure 3, reveals clearly a non-significant difference between treated and non-treated hair.
Cocaine
For the cocaine group, the data for the metabolite benzoylecgonine was taken into account. In all these samples the benzoylecgonine concentration was > 0.05 ng/mg and cocaine concentration was > 0.1 ng/mg, thus providing evidence of cocaine consumption.20, 22 Here, 90% of the positive BE concentrations, represented between the range percentile 5 to percentile 95 decreased from 0.06–2.24 ng/mg (in 125 out of 9488 non-treated hair samples) to 0.09–1.54 ng/mg (in 13 out of 1026 treated hair samples; Table 4). Again, as Figure 2 demonstrates, minimal changes were observed in the distribution of positive BE samples in dyed or bleached hair samples compared to those in non-treated hair (Figure 4). While for the BE concentration range 0.05 to < 0.1 ng/mg hair, a three-fold decrease in the percentage BE positive samples for dyed or bleached hair were observed, compared to non-treated hair, the opposite was observed for higher concentrations, namely a doubling of percentage positive BE samples in dyed or bleached hair samples (31%) compared to those in non-treated hair (16%) for the BE concentration range of 0.5 to < 2.0 ng/mg hair (Figure 4). Hence, both the percentile and frequency histogram data provide no evidence that dyeing or bleaching significantly reduces the positivity rate for cocaine in hair. Again, as observed for THC, for the cocaine determination in hair the frequency distribution of hair lengths analyzed was similar in the two cohorts (Figure 5).
Amphetamines
The range percentile 5 to percentile 95, representing 90% of the amphetamine concentrations measured decreased from 0.16–22.45 ng/mg in 56 out of 9488 non-treated hair samples to 0.31–6.90 ng/mg in 11 out of 1026 treated hair samples (Table 4). Previous larger population studies involving 3279 samples analyzed for amphetamines resulted in a higher range median value possibly due to a wider range of contexts than in this study and due to the fact that the cut-off used were a three-fold higher than in this study, namely 0.3 ng/mg.27 Similarly to cocaine, the decrease in drug concentration as reported in various studies1-8, 14, 28 is seen only at the lower concentration ranges, below the SoHT cut-off,20, 22 namely from 0.1 to less than 0.2 ng/mg hair (Figure 6). While five non-treated hair samples (9%) were present in this concentration range, no treated samples were present. Even though it seems that dyeing or bleaching can affect the positivity rate at concentrations lower than 0.2 ng/mg hair, first the sample size of 5 is not statistically significant and second, overall the positivity rate for amphetamines in dyed or bleached hair (1.07 %) was not lower than that for non-treated hair (0.59 %; Table 3). Again, for amphetamines, no significant differences were observed in the distribution of the hair lengths analyzed for both cohorts; only for hair lengths between 9 and 12 cm were the percentage number of treated samples double the non-treated hair samples (Figure 7).
The difference in percentiles when compared to an earlier publication27 can be attributed to the different context and the different hair lengths analyzed. In this work,27 only 3% of the samples analyzed were longer than 6 cm compared to 33% in the present study, half of which (15% of the total samples analyzed) were between 0–9 cm and 0–12 cm long. As expected, analysis of such long hair strands in subjects in whom a period of abstinence from drugs follows after a history of drug consumption, result in lower ‘averaged out’ drug concentrations than the analysis of the separate distal segment. Moreover, 54% of the samples analyzed in the previous study27 were for clinical purposes compared to abstinence context in this study.
EtG
For EtG the 5th to 95th percentile range varied from 9.40–179.97 pg/mg hair in dyed or bleached hair to 11.21–216.35 pg/mg hair in non-treated hair (Table 4); similarly no significant change was observed between the 1st to 99th percentile range (8.23–395.54 to 9.34–311.36 pg/mg hair) for the two cohorts respectively (Table 4). The frequency histogram for EtG showed also no change between dyed or bleached hair and non-treated hair (Figure 8). Hence, it seems that dyeing or bleaching does not increase or decrease the concentration of EtG in hair, in accordance to previous publications.11, 12 A recent in vitro study has also shown that temporary colouring has no significant effect on the EtG concentration in hair, whereas bleaching and perming does.29 As opposed FAEEs, EtG was found not to be influenced by hair spray, gel, wax, oil, or grease.30 The same study reports a decrease of 20–40% of the mean EtG concentration by bleaching or dyeing.30 Only one study documents the total cancellation of the EtG chromatographic peak due to bleaching;31 this may however be attributed to ion suppression effects seen in LC-MS/MS but not in GC-MS/MS methods as in our case.17 Also only one publication reports a single false positive result for EtG in hair carried out on one subject using an alcohol containing lotion (35g/L).32 As for THC, cocaine and amphetamine, the percentage positive samples for EtG in treated and non-treated authentic hair samples were grouped and plotted as a function of hair lengths analyzed as follows: <3cm, 3cm, >3<6cm, >6<9cm, >9<=12cm and >12cm. No significant different distributions were observed for the two cohorts, thus eliminating any bias due to different hair lengths analyzed in the two cohorts (Figure 9).
The major cosmetic treatment effects observed in this population consisting of around 90% dyed hair are differences for the drugs at the higher concentration ranges, namely above the 95th percentile for THC and amphetamine and above the 99th percentile for cocaine. In fact the THC concentration at the 95th percentile in dyed or bleached hair is roughly half that for non-treated; the THC concentration at the 99th percentile was one third that in non-treated hair. Similarly for cocaine, the BE concentration at the 99th percentile in dyed or bleached hair is roughly half that for non-treated hair; the amphetamineconcentration at the 95th percentile and the 99th percentile is roughly one-third that in non-treated hair, thus a reduction of 70%. It seems that here we find again the cosmetic treatment effects previously published in both in vitro3, 4, 8, 9 and in vivo studies.5-7 So possibly, cosmetic treatment reduces the drug concentration, but only at high drug concentrations, represented in the population studied in this paper by the 95th percentile and higher. At least a 50% decrease in drug concentration was observed when comparing percentiles at THC concentration greater than 1.16 ng/mg hair, at BE concentration greater than 5.97 ng/mg hair and at amphetamine concentration greater than 22.45 ng/mg hair, corresponding to the 95th, 99th, and 95th percentile for non-treated hair, respectively.
To date the reduction in drug concentration in hair due to dyeing or bleaching was explained either by chemical degradation or physical removal from the damaged keratinic matrix.3-9, 31 The decrease in drug concentration in hair due to dyeing or bleaching was reported to be proportional to the degree of hair damage6 caused by the treatment; hence explaining why in vitro bleaching produced higher drug decreases than dyeing did. Other studies have shown that cocaine, opiates, benzodiazepines, and methamphetamine were chemically unstable in presence of hydrogen peroxide, a component of bleaching and dyeing products.4, 8 On the other hand the much smaller decrease of drug concentrations observed in this study indicates that possibly the experimental conditions chosen for in vitro dyeing or bleaching studies are generally more aggressive than what is usually applied in practice and hence tend to be unrealistic. Others have attributed the apparent cosmetic treatment effect to external contamination.33 Special washing procedures are needed to eliminate external contamination, which can otherwise be misinterpreted as cosmetic treatment effect. This overestimated reduction of drug concentration falsely attributed to cosmetic treatment was not observed in this study. Moreover, ten years earlier it was shown that external in vitro contamination of treated hair by artificial sweat or sebum containing cocaine, benzoylecgonine, 6-acetylmorphine, morphine and codeine at high concentrations (500 ng/g) did not manage to increase the drug concentration in hair significantly. This indicated that the drug uptake into the hair fiber from the endogenous-exogenous shunt via sebum or sweat is of minor importance.28 At lower drug concentrations, as in this study, this can however explain the apparent increase in the drug concentration seen for the first time at very low drug concentration: at THC concentration around 0.02 ng/mg hair, at BE concentration around 0.05 ng/mg hair and at amphetamine concentration around 0.15 ng/mg hair, corresponding generally to the 5th percentile or lower for non-treated hair. In order to examine carefully the dyeing or bleaching effect, the percentiles for THC, BE, amphetamine and EtG were calculated for both treated and non-treated positive hair samples and expressed as percentage change. In order to compare the general trends, the data was fitted to a quadratic function, as shown in Figure 11. Considering changes greater than 40% or less than -40% to be significant and hence possibly due to cosmetic treatment and not due to other noise or uncertainty of measurement, it seems that amphetamine is mostly affected by dyeing or bleaching, both at low and high drug concentration followed by benzoylecgonine. THC seems to be much less effected by dyeing or bleaching; EtG not at all.
The observed gradients for amphetamine, BE, and THC seem not to be due to external contamination and the consequent lack of appropriate pre-analytical washing of the hair as previously suggested,33 or the possible external incorporation from sebum or sweat.34 Incorporation of the drugs from blood capillaries into growing hair cells by passive diffusion and thus along the pH gradient seems to be the predominant pathway, since the intracellular pH of keratinocytes and melanocytes is lower (more acidic) than in plasma.35, 36 It has been reported that the cell membrane complex consisting of proteins and a protein–lipid complex, originating from previous cell membranes, is most vulnerable to chemical and mechanical attack and is the primary diffusion point for incorporation and elimination of drugs.34 Hence, as previously suspected,28 dyeing or bleaching increases the porosity of the hair thus allowing loss of drugs but also increased incorporation at very low drug concentration in hair compared to plasma. Moreover, the fact that a higher gradient is observed for amphetamine and BE in Figure 10, both being more polar and basic than THC, indicates a predominant role of basic pH over the lipophilicity of the drug. On the other hand, EtG, being acidic and extremely hydrophilic seem not to be effected by the increased porosity of the melanocytes’ and keratinocytes’ membrane; here the incorporation pathway into the hair from sweat plays also a role.16, 34 In fact in another publication16 it was shown that this pathway, rather than the washout effect explains the significantly higher EtG concentrations observed in hair locks smaller than 3 cm.
Simple mathematical model
A simple model is presented below solely to schematically show the variation in drug concentration in hair in a subject who consumes drugs occasionally (such that 1 A.U. of drug is accumulated every week in the hair) and simultaneously dyes the hair (typically once every six weeks). Furthermore two scenarios are considered, namely a uniform reduction in the accumulated drug by (1) 75% (worst case scenario) and (2) 50 % on dying, namely the typical drug decreases mentioned previously in the literature both in in vitro3, 4, 8, 9 and in vivo studies5-7 and observed in this study only at high drug concentrations in hair. The results of the decrease of drug concentration in hair based on the above assumptions are shown in Figure 12 for (A) 3 cm, (B) 6 cm, (C) 9 cm and (D) 12 cm hair lock.
- 16–59% of the drug concentration in non-treated hair (i.e. decrease in drug concentration by 41– 84%) depending if sampling occurs just before or just after the cosmetic treatment for case (a); 38–73% of the drug concentration in non-treated hair assuming 50% reduction, i.e. for case (b) when a 3 cm hair lock was sampled.
- 8–30% of the drug concentration in non-treated hair depending if sampling occurs just before or just after the cosmetic treatment for case (a); 23–45% of the drug concentration in non-treated hair assuming 50% reduction, i.e. for case (b) when a 6 cm hair lock was sampled.
- 6–20% of the drug concentration in non-treated hair depending if sampling occurs just before or just after the cosmetic treatment for case (a); 16–31 % of the drug concentration in non-treated hair assuming 50% reduction, i.e. for case (b) when a 9 cm hair lock was sampled.
- 4–15 % of the drug concentration in non-treated hair depending if sampling occurs just before or just after the cosmetic treatment for case (a); 12–23 % of the drug concentration in non-treated hair assuming 50% reduction, i.e. for case (b) when a 12 cm hair lock was sampled.
This simple model shows that limiting the hair length to be analyzed from 12 to 6 cm shall reduce the decrease in drug concentration by half and hence result in a doubling of the drug concentration to be found in dyed or bleached hair.
- The type of cosmetic product used.
- How often is the hair dyed or bleached.
- The frequency and dose of drug consumed before and after the cosmetic treatment.
- The drug incorporation rate into the hair, (partly still unknown).
- How uniformly is the drug incorporated into the hair shaft (relevant in incorporation from sweat).
- The mechanisms of drug incorporation into the hair, (partly still unknown).
- The time of hair sampling with respect to the cosmetic treatment.
- The length of sampled and analyzed hair.
Nonetheless, this simple model shows that even under the extreme conditions of a cosmetic treatment effect of 75%, drug consumption only once a week, drugs will possibly still be found in hair, even when sampling a 12 cm hair lock. Higher drug concentrations are expected when analysing shorter locks of hair. Thus, to compensate for a possible reduction in the drug concentration caused by cosmetic treatment, a shorter hair strand can be requested for analysis and/or segmentation of a hair lock if the lock is longer than 6 cm. Even though cosmetic treatment can cause a reduction in the concentration of drugs and EtG in hair, these xenobiotics remain present in dyed or bleached hair, which can be detected by modern, sensitive MS/MS methods, thus excluding abstinence from drugs and alcohol over the period investigated. As hair is often the only evidence available, many months after the intake of drugs or alcohol, it should be made use of, especially in cases where no correlation between drug concentration in hair and dose consumed is required.
Conclusion
Although this study needs further investigation, it is the first attempt, to my knowledge to study the influence of hair colouring on the positivity rate in hair for both drugs and EtG in a large statistically significant population including more than 10 000 authentic hair samples, thus overcoming the limitations of the previously published in vitro and in vivo cosmetic treatment studies. Based on the observed facts that (1) no significant difference in the positivity rate was observed between treated and non-treated hair for both drugs and EtG tested for abstinence at the MPA cut-offs and that (2) significant changes in drug concentrations in hair were observed only at high drug concentrations (corresponding to 20 times the MPA cut-off and 95th percentiles or higher in the population studied), it can be concluded that dyed or bleached hair is not automatically useless for drug and EtG detection; conversely, hair analysis is a powerful tool, often the only means to detect retrospective drugs and/or alcohol consumption even in dyed or bleached hair. Consequently, guidelines considering the exclusion of dyed or bleached hair shall be reviewed, especially because it has been shown that urine is a worse alternative to coloured hair for the detection of drugs of abuse and especially EtG as required for the driving license re-granting programme MPA in Germany.37
- Investigate whether the same trends are observed in other drugs other than THC, cocaine, amphetamine and EtG.
- Clarify if the apparent increase in the concentration of drugs detected in dyed or bleached hair at concentrations well below the median in this study, is significantly higher than in non-treated hair or if this is due to measurement uncertainty.
- Explain the apparently smaller cosmetic treatment effect in hair containing EtG as seen in this study (more in vivo studies).
- Clarify possibly new drug incorporation mechanisms which also explain why cosmetic treatment plays only a role at relatively high drug concentrations as observed in this study.
Acknowledgements
I would like to thank particularly Ms Graute and Mr Peters for their excellent assistance with the statistical calculations, all our customers who provided the samples, and all our team at the Department of Forensic and Clinical Toxicology for their good work and support.
