| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
a Author for correspondence. Fax 215-322-4226.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Key Words: indexing terms: drug assays • toxicology
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our goals with the technique presented here included having as many unique sets of reactions as possible (high specificity) within 1–2 h of one chemist's bench time and using routine silica-gel TLC plates and other materials readily available in most toxicology laboratories. We also wanted to meet the criterion that general unknown screening should logically determine what is not present as well as what is present through the use of reliable exclusion and accurate positive identifications of xenobiotics. The training in TLC required to make accurate identifications has been an important factor in developing the technique we present. A computer program, SPOT CHEK, was developed to minimize the time spent interpreting data by tediously matching observed TLC migration and reaction characteristics with those previously obtained for drug standards. Detection reagents were chosen to detect 100 ng or less per applied spot of sample for a wide variety of drug groups. Results for 9 TLC reactions and the migration position (the 10th defining variable) relative to nicotine for each drug or drug metabolite have been placed in the database to facilitate a search keyed on any one (or more) of these 10 TLC identifying characteristics. After the search is completed, the program presents a list of the matches between test specimen substances and database entries.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Detection reagents.
The 10 TLC detection/visualization
reagents used (SPOT ID reagents) were prepared with minor modification
of the formulations given elsewhere (11). Composition of
the reagents was as follows:
S1, fluorescamine (Sigma Chemical Co., St. Louis, MO), 0.3 g/L; stable 1 month stored in a brown bottle
S2, ferric chloride/perchloric acid/nitric acid (FPN) reagent: 90 mL of 70% perchloric acid, 135 mL of 70% nitric acid, and 15 mL of 50 g/L ferric chloride added to 100 mL of water; stable 6 months in a brown bottle
S3, Dragendorff reagent, prepared as in Clarke (11)
S4, Marquis reagent: concentrated sulfuric acid
S5, Mandelin reagent: ammonium metavanadate (A-8050; Sigma Chemical Co.), 800 mg/L in concentrated sulfuric acid; stable at room temperature for at least a year
S6, iodinated Dragendorff reagent: 5 g of potassium iodide, 2 g of iodine, and 0.2 g of bismuth subnitrate (Sigma B9009) mixed in water, followed by 0.5 mL each of glacial acetic acid and concentrated HCl, and brought to 250 mL with deionized water
S7, 254 nm (UV) light projected onto a developed TLC plate (must be fluorescent indicator plate)
S8, chlorine vapor, produced as needed by adding concentrated HCl to commercial bleach (5% hypochlorite)
S9, concentrated HCl fumes, produced as needed in a fume hood
S10, the mobile phase (described below).
Names, application technique, and shelf-life of the reagents are given
in Table 1
. Before dipping plates 2B and 2C into S4 and S5 reagents, we
exposed them for ~5 min to vapors from formaldehyde/methanol solution
(37/63 by vol.), then dipped them into the respective acid solutions.
Mercuric sulfate (not one of the 10 SPOT ID reagents) was prepared by
dissolving 5 g of mercuric sulfate in water plus 20 mL of
concentrated sulfuric acid and bringing to 250 mL with water (stable at
least a year).
|
For all chromatographic runs we used a single mobile phase: ethyl
acetate/dichloromethane/methanol/concd. ammonium hydroxide (80/90/15/5
by vol.). Migration of drug substances was defined by a "zone," or
position relative to a set of known substances present in the
normalizing (N) mixture. The compounds included in the N mixture were
(Fig. 1
, reading spots from bottom to top): benzoylecgonine, morphine,
quinine, nicotine, methadone, and cocaine. The area from the origin to
the center of the morphine spot was designated Zone 1, from morphine to
quinine Zone 2, from quinine to nicotine Zone 3, from nicotine to
cocaine Zone 4, and from cocaine to the solvent front Zone 5.
Substances that eluted even with a boundary line were referenced to the
higher-numbered zone. The TLC chamber must be two-sided or have a small
container in which to place mobile phase for preassay vapor-loading of
the plate. The calibrating mixtures in M and P (Fig. 1
) are neutral
drugs (M mixture) and barbiturates (P mixture); S1,
S2, and S3 are benzodiazepines and other
neutral/acidic drugs; A, B, and C mixtures are basic amine-type drugs.
The total number of individual drugs present in the set of mixtures is
48, each at 1 g/L in methanol. Mixture compositions are given below
under Assay Performance (Separations) and in Fig. 1
.
|
computer and computer program
Data for >240 drug substances have been placed into a PC program,
compressed, compiled, and installed onto an MS DOS hard drive, version
6.2. The original program, SPOT CHEK, was written in dBase IV and
requires an IBM or IBM-compatible 286 or higher PC, DOS 6.0 or later
100%-compatible version, at least 640 K of memory plus 2 Mb of
extended RAM, at least 3 mB of hard disk space (4.5 mB at
installation), and an CGA, EGA, or VGA monochrome or color monitor. A
peripheral printer is optional but is useful for printing lists of
matches after data entry.
Data cards.
TLC reactions were detailed on 250 separate
identification cards that give all of the information database for each
drug. Fig. 2
shows as an example the data card for carbamazepine. All
substances have a relative Rf, nicotine being assigned
Rf = 1.0. The data cards are indexed alphabetically and
sorted in order of increasing relative Rf. As shown in Fig. 2
, the data cards contain, in addition to the 10 reactions in SPOT
CHEK, information that will greatly assist in making accurate drug
identifications. Data cards illustrate portions and reactions of
metabolites of parent drugs, list drugs that react similarly (can be
mistaken for suspect drug), give the most sensitive reagents for
visualization, and offer important details of the TLC behavior of the
various drugs.
|
sample preparation/drug recovery from urine and serum
An aliquot of serum (0.5–1 mL) not pH-adjusted was extracted with
6 volumes of a solution of dichloromethane/isopropanol (19/1 by vol.).
Whole blood was diluted with 1 volume of water before extraction with
the 6 volumes of solvent. Two 5-mL aliquots of urine, one adjusted to
pH 9 and the other to pH 12, were extracted with 10 mL of the
dichloromethane/isopropanol solvent and combined into one extract
fraction. The serum and urine extracts were passed through coarse
filter paper and evaporated to dryness in disposable plastic 16-mL
conical beakers. Before evaporation, two drops of 1 mL/L HCl in
methanol were added to urine extracts to prevent loss of volatile
amines. This isolation technique provides a weak acid/neutral fraction
and an amphoteric/amine base fraction. The serum extract, which might
contain neutral drugs, weakly acidic drugs (barbiturates,
acetaminophen), and benzodiazepines, was spotted onto Plates 1A and 1B
(Fig. 1
), and the basic/amphoteric fraction from urine was spotted onto
Plates 2A, 2B, and 2C.
We examined some drugs—from very polar (benzoylecgonine) to nonpolar (phencyclidine)—for approximate recoveries in the liquid–liquid extraction technique described above. Because on the silica-gel TLC plate the drugs migrate according to their polarity, recovery correlates well with migration position; i.e., recovery increases as Rf increases. Benzoylecgonine, found low in Zone 1, has ~45% recovery and phencyclidine ~95% recovery. Morphine recovery is ~70%, and drugs migrating farther than morphine are 75–95% recovered. Benzodiazepines, barbiturates, acetaminophen, caffeine, and other neutral compounds and weak acids are recovered from serum in high yield (>80%) with no pH adjustment. Recovery was determined by measuring the fraction remaining unextracted after the initial extraction.
assay and techniques
Application of samples.
The plates used throughout were
factory-scored hard-layer silica-gel plates, 20 cm wide x 10 cm
high (running distance), with an organic binder and fluorescent
indicator. Biological fluid extracts, powders, and other solid dosage
forms (e.g., tablets, pills, residues) were applied directly to the
plates in methanol or dichloromethane. Residues from both the neutral
and the pH 9 plus 12 fractions were taken up in a few drops of
dichloromethane/isopropanol solvent and applied in 10-µL capillaries
or by an automated applicator. The neutral extract was spotted about
equally on Plates 1A and 1B and the pH 9 plus 12 fraction from urine
was spotted about equally on Plates 2A, 2B, and 2C, ~20% of this
residue being reserved for optional tests. The plates being spotted
were placed over a 2.54-cm-wide heating tape so that application spots
dried quickly and fit within a standard 3-mm-diameter size for mixtures
and extraction fractions (see Fig. 1
for spotting pattern). The scored
plates were sectioned after development, except that Plate
2C (Fig. 1
) was separated from the other sections before development.
This permitted the placing of all 5 sections of plate into a single
small development chamber (2C was placed facing the other plate
sections) and developed in the mobile phase described above. Serum and
urine extracts from one or two subjects were accommodated on a
12.5 x 10 cm (w x l) plate. Larger plate
sections of 1A, 1B, 2A (etc.) and tanks holding 20 x 10 cm plates
were used when we were analyzing specimen extracts from more than two
subjects.
Plate development.
A paper pad was placed on one side of
a two-sided TLC chamber for 10 x 10 cm plates, and 6 mL of
developing solvent was run (pipetted) down the pad, followed by 3 drops
of concentrated NH4OH (pad side). The plate was placed on
the dry side of the chamber for 10 min before developing to full plate
height by the addition of 6 to 8 mL of solvent. Chromatography per se
took 10.5 min. The plates were then dried and observed in the viewing
chamber at 254 and 366 nm, where we marked the UV-quenching spots
lightly with a pencil and made note of any fluorescent spots. Relative
Rf values were not routinely measured, but zone positions
were accurately noted by reading the positions of detected substances
vs those of the N calibrator drugs, either under the UV 254 light
(reaction S7) or after a visualization reaction.
Plate reactions.
Reagents were applied (after segmenting
the plates as in Fig. 1
) by classical techniques, either dipping or
spraying as indicated in Table 1
. Barbiturates, meprobamate,
glutethimide, and other nonbarbiturate sedatives were detected or ruled
out by use of mercuric sulfate dip (neutral/weak acid extract) on plate
1A. After positioning plate 1B over a 600-mL beaker on a warm hot plate
(in a fume hood), we added several milliliters of commercial bleach
followed immediately by the same volume of concentrated HCl. This
produced sufficient Cl2 gas to react with drugs on the
developed plate. After exposure to Cl2 vapor for 10–15 s,
some drugs give a visible color (acetaminophen), whereas others
fluoresce strongly (some benzodiazepines, carbamazepine). We next
exposed plate 1B to HCl vapors (in a fume hood). After each reaction
(Cl2 and HCl) we examined plate 1B for fluorescence with UV
254 and 366 nm light. Detection is vastly improved by placing the plate
over a hand-held UV lamp (face up) inside the dark chamber rather than
having the light source above the plate.
Plates 2A, 2B, and 2C were processed according to the scheme in Fig. 1
and all reactions were noted. The sequence of reactions for plate 2 was
as follows: 2A, NH4OH vapor, fluorescamine, FPN, D1;
2B, formaldehyde for 5–10 min, then concentrated
H2SO4 dip; 2C, formaldehyde for 5–10 min, then
Mandelin reagent dip. After the acid dips for 20–50 s, plates 2B and
2C were dipped first into water and then into reagent D2 for 2–5 min,
which fixes the location of most drug substances. The two acid dips
produce colors in the spots, so a color number is assigned to all spots
on the plates that give colors with the acid dips. The color code was:
1, pink; 2, red; 3, copper/rust/red-brown; 4,
orange/peach/yellow-orange; 5, tan/salmon/gold; 6, yellow; 7, green; 8,
olive green/gray-green; 9, blue-green/aqua/lime; 10, blue (light and
dark); 11, violet/purple/lavender; and 12, gray/blue-gray. Reaction
results and the zone positions of unknown spots on the plate were
recorded. These nine reactions plus the zone position, collectively
called SPOT ID reactions, are numbered S1 through S10 for convenient
tabulating and entry into the SPOT CHEK program.
assay performance
Separations.
The silica-gel plates and mobile phase used
resolved individual drugs within classes sufficiently so that 48 drugs
seen very frequently could be applied as calibrators and fully
separated. Mixtures of these calibrators [1 g/L (1 µg/µL) each]
were applied to TLC plates as long as 2 weeks before plate development,
~2 µg per spot. No decomposition of the calibrators occurred if the
prepared plate was covered with a glass plate and kept in a closed box.
The components of each calibration mixture are listed in the legend to
Fig. 1
. This large number of calibrators allows for visual
determination that an unknown drug is between two calibrators rather
than measuring the Rf or relative Rf of the
unknown. For example, for a drug that migrates between methadone and
cocaine (in the N mixture), this area of the TLC plate presents few
matches. Migration relative to a known calibrator is more consistent
than relative Rf, which varies as much as 15% from plate
to plate. Some drugs are more susceptible to slight solvent changes
than others. For example, phenobarbital is moderately dependent on the
ammonia concentration within the chamber, and benzoylecgonine migration
depends on the moisture content of the TLC plate. Whereas relative
Rf numbers can change from plate to plate, migration
positions relative to the N mixture set remain constant if the
developing conditions are according to the protocol given above.
Detection limits.
Detection limits, in terms of quantity
(ng)/spot, are dependent on spotting technique and the reaction used to
visualize the drugs. Table 2
indicates the approximate detection limits expected for several
drugs and visualization reagents, assuming that the procedure is
performed by someone with reasonable analytical skill and experience.
Detection limits in terms of concentrations in urine and serum (ng/mL
specimen) are dependent on recovery, spotting technique, and reagent
for visualization. Tables
3 and
4 present results for several drugs added to 0.5-mL serum and
5-mL urine samples and are typical of results for all 240 drugs
studied. These results should be considered approximate because
experience, plate quality, and TLC skills are factors in the overall
detection. Nevertheless, the technique presented here offers lower
detection limits for all classes of drugs relative to previously
published detection limits
(2)(3)(9)(10)(12),
either per spot or per volume of specimen.
|
Validation.
Data on many drugs and drug metabolites have
been accumulated over several years in the process of broad-based drug
screening of >15 000 individual urine and serum/blood sets.
Validation of the TLC techniques was accomplished by using at least one
of several alternative techniques—RIA, gas chromatography, HPLC, UV
spectrophotometry, fluorescence polarization immunoassay (FPIA), and
GC/MS—to confirm the TLC identifications. All positive results for
cocaine/benzoylecgonine, opiate, barbiturate, phencyclidine,
amphetamine, and benzodiazepine were tested by RIA; tricyclics,
barbiturates, and various neutral drugs positive by TLC were tested by
FPIA; drugs with unique UV spectra such as naproxen and methaqualone
were scanned by UV spectrophotometry; gas chromatography, GC/MS, or
HPLC was used in numerous instances where TLC findings were not
clearcut.
The Toxi-Lab compendium, consisting of color photographs of spots after
the Mandelin reaction, helped us confirm how drugs and metabolites
react with this reagent. Urines from persons taking a single known
medication were processed on several occasions to obtain metabolite
patterns (e.g., for tramadol, ibuprofen, and codeine). Patient
histories in which drug intake was reasonably documented provided a
check on the overall reliability of the SPOT CHEK technique.
Information for the data cards (Fig. 2
) was based on positive,
confirmed identifications with authentic drugs.
Specificity.
Identification problems most often involved
choices between closely related drugs or drugs with few or no SPOT ID
reactions other than migration position. For example, orphenadrine and
diphenhydramine have very similar reagent reactions but can be
differentiated by slight migration differences. Brompheniramine,
pheniramine, and chlorpheniramine are difficult to distinguish by these
TLC techniques and when present in a mixture could not be resolved with
the solvent system presented here.
Persons who are color-blind can use SPOT CHEK, but they need to use the Marquis and Mandelin color codes guardedly or have others' eyes note the colors.
data entry and effective use of spot chek computer program
To operate the program, the user notes systematically the
reactions and zone for each spot. Weak reactions are noted as such and
colors are described in words, then assigned the corresponding number
from the color key. The program, installed in MS DOS onto a PC, is then
opened. The result of each SPOT ID reaction (the 9 reactions) is asked
for with a yes/no (Y/N) statement or a prompted color number
(reactions 4 and 5), as follows.
S1, fluorescamine reaction—Quench, Fluoresce, None? (Q/F/N)
S2, FPN reaction? (Y/N)
S3, D1 Dragendorff reaction ? (Y/N)
S4, Marquis reaction? Enter number corresponding to color, or NN for no color
S5, Mandelin reaction? Enter number corresponding to color, or NN for no color
S6, D2—iodinated Dragendorff reaction? (Y/N)
S7, UV 254 nm reaction—quench fluorescence? (Y/N)
S8, chlorine vapor reaction? (Y/N)
S9, HCl vapor reaction? (Y/N)
S10, zone portion on plate? Enter 1, 2, 3, 4, or 5
Would you like a printout? (Y/N)
Are all entries correct ? (Y/N)
If N (no), go back to start (S1).
Initially, only those reactions that were carried out and gave very clearcut spots should be entered. Leaving the response blank means that the reaction will not be considered. After the last entry, the program produces on the screen a list of drugs matching the designated properties. Only if the list contains more than four entries is one advised to enter more reactions. Specificities and probability matches vary with the compound and to some extent with the skill and experience of the analyst. Once a small list has been generated, the data cards are to be examined carefully for best matches and for those substances similar to the suspect drug.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
gives the results from seven individual drug screens processed
according to the procedure given and to the TLC reactions entered into
the SPOT CHEK computer program. Not all reactions were used for the
individual cases; in some cases, only a single plate was used to
acquire a listing or identification. After obtaining a list, whether
one drug or more, we compared each drug with a data card of the listed
drugs such as the one given in Fig. 2
. The exact position
(Rf relative to nicotine) and the metabolite pattern
assisted in distinguishing close matches. Case 1 was from a pH 7 serum
extract spotted onto plates 1 and 2 (Fig. 1
). If only reactions S6, S7,
S8, S9, and Zone are entered into SPOT CHEK, the search list is too
long for quick identification. By including plates 2A and 2B, the
program lists four possible drugs, which are readily differentiated by
close inspection of the TLC plate for relative Rf values of
the three benzodiazepines and carbamazepine, all having been present in
the calibrators applied. In case 2, SPOT CHEK selected dextromethorphan
as the only choice after plates 2A, 2B, and 2C were developed, reacted,
and read. The acid-neutral side of the reaction system (plate 1A and
1B) was not needed to select dextromethorphan: Basic drugs can usually
be identified with plates 2A, 2B, and 2C, whereas benzodiazepines and
various neutral compounds can be detected and identified on Plate 1.
The third set of plates (case 3) initially used plate 2A, which
resulted in 12 possible identifications. Adding plate 2B shortened the
selection list to 2 substances but did not reduce to 1 "hit."
Guaifenesin and methocarbamol can be differentiated with
p-dimethylaminobenzaldehyde, because methocarbamol is a
carbamate and guaifenesin is not.
|
Some drugs can be selected on the basis of only one TLC plate, e.g.,
case 4 in Table 5
. The absence of UV absorbance and the charring with
FPN reagent are highly characteristic of erythromycin. The last three
searches in Table 5
resulted in single-drug listings: case 5,
azacyclonol; case 6, lidocaine metabolite (monoethylglycinexylidine or
MEG-X), and case 7, amiodarone. Not all reactions were required for the
specific identification of these basic drugs.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The technique described here can be carried out without the assistance of the computer program. Advantages of the TLC technique alone are: (a) in a single developing system, plates provide good separations for drug substances and have a high peak (spot) capacity; (b) plates are convenient to use because ruggedness of the layer allows dips and sprays to be applied without disintegrating the layer; (c) the last reaction of each sequence can be read weeks after it has been carried out; (d) reagents used will detect drugs in quantities <100 ng/spot, in some cases as little as 5 ng/spot; (e) scored plates are efficient and economical; (f) reagents are stable and do not need to be made up daily, (g) new drugs can be readily added to the total screen, and (h) the drug recovery technique is flexible; i.e., laboratories can continue in-house procedures or use the one described here. However, the tedious and somewhat subjective interpretation of TLC results can now be made faster and more objective by utilization of a versatile but simple computer program SPOT CHEK.
In the two papers published on the use of a computer for storage and retrieval of TLC data (8)(13), both studies rely on Rf determinations in several mobile-phase solvents, thereby increasing analysis time. The SPOT CHEK program, however, has the following advantages: (a) from 1 to 10 TLC characteristics (identification variables) can be entered, depending on how many reactions were done (i.e., partial use of the system is feasible); (b) the program works in practical situations, as demonstrated by several years of experience with >15 000 full drug screens from clinical and forensic referrals; (c) the program will often list results by drug class; (d) the identification technique is reaction-based and visual and does not require tedious measurements of migration positions or the use of more than one mobile phase. The program is also very helpful in training analysts in TLC and in understanding drug chemistry. Finally, and most importantly, the computer-assisted interpretation of results obviates much of the tedium involved in broad-spectrum drug screening.
In conclusion, the extent and degree of confirmation of drug substances identified or selected by the TLC technique described here depend on how the results are to be used. The decision of how and when to confirm drug finding will depend on laboratory protocols and on the chemist's perspective, experience, and insight. Some of the drugs in the SPOT CHEK database and card file are uniquely defined by their TLC reactions and metabolite patterns. The approach of Ojanpera and Vuori (4), who presented TLC identifications by scanning individual spots in situ in the UV range, is especially compatible with the SPOT ID reactions and SPOT CHEK program presented here because the UV spectrum can be obtained nondestructively. With a "hit" list from a UV scan, selective SPOT ID reactions can follow. TLC offers an excellent first step in broad-spectrum drug screens and in general unknown protocols; once applied to the plate, the sample is not lost (unless volatile). The techniques to corroborate TLC findings can be intelligently selected because (after TLC) the analyst will know something about the functional groups, aromaticity, polarity, and often the likely identification of the unknown(s) before the confirmatory step. Non-TLC confirmatory techniques appropriate in clinical and forensic toxicology include immunoassay (RIA, FPIA, Emit), UV spectrophotometry, gas and liquid chromatography, and GC/MS.
|
|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 Nonstandard abbreviations: TLC, thin-layer chromatography; FPN, ferric chloride/perchloric acid/nitric acid; FPIA, fluorescence polarization immunoassay. 
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
