State Of the Art/ Review—
`M. Therese Southgate, MD, Section Editor
`
`Rational Therapeutic Drug Monitoring
`
`H. Friedman, MD, D. J. Greenblatt, MD
`
`FOR MANY drugs, the measurement
`of concentrations in serum or plasma
`has
`become widely
`available
`and
`accepted as an important component of
`clinical decision making. While these
`drug levels often do
`allow more
`objective monitoring and titration of
`therapy, the information also has the
`potential to be valueless or even mis-
`leading. Laboratories sometimes report
`that a serum concentration is in the
`“toxic” range, when the patient
`is
`doing well and has no evidence of toxic
`effects. Or, conversely, the drug is not
`detectable in serum. Such discrepancies
`between measured serum drug concen—
`trations and observed clinical drug
`effects may occur for numerous rea-
`sons. This article Will
`review some
`principles
`and problems
`associated
`with therapeutic drug monitoring.
`
`RATIONALE FOR MONITORING
`SERUM DRUG LEVELS
`
`For a serum drug concentration to be
`potentially useful for purposes of ther-
`apeutic monitoring, at least two requi-
`sites must be fulfilled.1 First,
`the
`serum drug concentration must reflect
`the concentration at the receptor site;
`second, the intensity and duration of
`the pharmacodynamic effect must be
`temporally correlated with the recep—
`tor
`site drug concentration. When
`these two conditions are not met, as in
`the case of anticancer drugs showing
`effects long. after they are gone from
`the serum, the likelihood of correlating
`serum levels with therapeutic effect is
`considerably reduced.
`During long-term dosage with any
`drug, the two major determinants of
`Its mean steady—state serum concentra-
`tion are the rate at which the drug is
`administered (dosing rate) and the
`drug’s total clearance in that particular
`Patient?”i The mathematical relation-
`ship is
`
`
`
`me the Division at Clinical Pharmacology. Depart—
`mam; oi Psycl'iiatry and Medicine, Tufls University
`School of Medicine and New England Medical Center,
`on.
`
`RBurial tomes-ts io Elvision of Clinical Pharmacology,
`Bar 100?,
`runs—New England Medical Center, 171
`Harrison Ave, Huston, MA 02111 (Dr Greenblati).
`
`Mean Steady-State Concentration=
`Dosing Rate
`Clearance
`
`Clearance is measured in units of vol-
`ume per unit of time, and describes in
`quantitative terms the capacity of a
`given individual
`to biotransform or
`eliminate a given drug. Drug clearance
`is usually accomplished by hepatic bio-
`transformation, renal excretion, or a
`combination of the two. Thus, under
`usual circumstances,
`the steady-state
`concentration of a particular drug in a
`given individual
`is directly propor—
`tional
`to the dosing rate (With the
`exception of a few drugs with saturable
`or nonlinear kinetics, such as salicy~
`late, phenytoin, and alcohol). Among
`different individuals, however, any giv-
`en dosing rate is likely to produce wide
`variations in steady-state concentra-
`tion, attributable to large interindivi-
`dual differences in clearance (Fig 1). A
`number of identifiable factors can alter
`the clearance of drugs. such as age,
`gender, body habitus, disease states,
`cigarette smoking, and drug interac—
`tions.” However,
`substantial unex-
`plained individual variation in drug
`clearance is commonly observed even
`among healthy, drug—free persons of
`the same sex and within a narrow age
`range.10 Therefore, dosage may not be a
`good predictor of steady-state concen—
`tration.
`
`“Therapeutic range" and “therapeu-
`tic index” are two concepts used to
`quantitate the relationships of serum
`concentration to efficacy and safety,
`respectively. Some drugs have a well-
`defined therapeutic range. When the
`steady—state concentration falls within
`this range, the likelihood of clinically
`effective and nontoxic therapy is maxi-
`mized. Direct measurement of
`the
`serum concentration allows
`appro-
`priate upward or downward titration
`of dosage in the individual patient, to
`attain the desired level. Therapeutic
`ranges, however, are not absolute (Fig
`2). Levels at the “low” therapeutic end
`have a significant likelihood of being
`clinically ineffective, whereas levels at
`the high therapeutic end have a signifi—
`cant likelihood of causing toxic effects.
`
`JAMA, Oct 24/31, 1986—Vol 256, No. 16
`
`Drug Monitoring—Friedman & Greenblatt
`
`In experimental pharmacology, “thera-
`peutic index” is defined as the ratio of
`the median lethal dose to the median
`effective dose.
`In clinical medicine,
`however, therapeutic index is usually
`estimated as the ratio of the highest
`potentially therapeutic concentration
`divided by the lowest potentially thera-
`peutic concentration (Fig 2). Some
`drugs (such as gentamicin, digoxin, and
`lithium) have a narrow therapeutic
`range and therefore a low therapeutic
`index. For such drugs, one can antici—
`pate considerable overlap among inef-
`fective, effective, and possibly toxic
`concentrations, thereby increasing the
`importance of serum level monitoring.
`Serum drug concentrations may still
`be of considerable value even when a
`therapeutic range has not been defi—
`nitely established. Consider a patient
`with no apparent clinical response to
`drug therapy despite seemingly ade-
`quate desage. A measured steady—state
`concentration that appropriately re-
`flects the dosage rate suggests that the
`patient may actually be a “nonrespond—
`er.” If, however, the measured level is
`very low or undetectable, this suggests
`that the patient either is not taking the
`medication (noncompliance) or has un-
`usually high metabolic clearance. An-
`other example is the patient with a
`sign or symptom (such as loss of
`appetite during digitalis therapy) that
`could be attributable either
`to an
`adverse drug reaction or
`to the
`underlying disease itself.11 In this case,
`a high serum drug level suggests that
`the medication might be responsible
`for the-adverse effect; a low serum
`level, on the other hand, could indicate
`that the underlying disease, or some
`other factor, explains the reaction.
`Drug concentrations frequently are
`measured for medicolegal reasons. In
`cases of deliberate or accidental drug
`overdosage, verification of the particu-
`lar substances ingested, and their con—
`centrations in serum, may have impor—
`tant
`therapeutic and forensic value.
`“Screening” of current and prospective
`employees for the presence of “illicit
`drugs” is becoming increasingly com-
`mon, although these tests are usually
`done on samples of urine.12
`
`TEVA1 059
`2227
`
`1
`
`TEVA1059
`
`

`

`etc), or even by intravenous injection,
`the entire administered dose does not
`have immediate and complete access to
`its receptor site mediating pharmaco-
`logic activity (Fig 3). After intravenous
`injection, the entire dose reaches the
`systemic circulation and by definition
`has 100% bioavailability. However, the
`drug is distributed not only to the
`tissue where it is active, but also to a
`number of other sites (Fig 4). Further—
`more, once the drug has reached the
`systemic circulation, it also encounters
`serum or plasma proteins. Drugs are
`bound to proteins
`to varying de-
`grees.”15 The principal binding pro-
`teins are albumin and al-acid glycopro-
`tein. The affinity of a drug for serum
`protein limits its freedom to diffuse
`across cell membranes, hence further
`limiting its accessibility to the receptor
`site.
`When a drug is administered by an
`extravascular route, it reaches the sys-
`temic
`circulation
`indirectly,
`often
`yielding less than 100% bioavailabili-
`ty.” Oral bioavailability of drugs in
`tablet and capsule form can be influ-
`enced by incomplete absorption due to
`incomplete dissolution, which in turn
`depends on packaging and drug parti—
`cle size. Oral solutions overcome the
`dissolution problem. Other factors that
`can influence oral bioavailability in-
`clude changes in gastrointestinal mo—
`tility, gastric and intestinal pH, mal-
`absorption
`syndromes,
`and
`the
`coadministration of foods and drugs
`(especially
`antacids,
`antidiarrheal
`agents, and chelating agents).
`After absorption of the drug from
`the gastrointestinal tract, systemic bio-
`availability may be reduced because of
`metabolic transformation in the gut
`wall, or by extraction from the portal
`circulation during the “first pass”
`through the liver. This is the case for
`certain drugs characterized by high
`hepatic clearance, including proprano-
`lol, lidocaine, tricyclic antidepressants,
`opiate analgesics, neuroleptics, hydral-
`azine, nitroglycerin, verapamil, and
`prednisone.17
`Incomplete bioavailability after in-
`tramuscular injection is also possible.
`This has been attributed to poor drug
`solubility at physiologic pH and precip—
`itation at
`the injection site after
`administration of chlordiazepoxide, di—
`goxin, phenylbutazone, phenytoin, and
`quinidine.18
`A number of recent studies have
`evaluated drug absorption after sublin-
`gual or buccal administration.19'2l In
`principle, this route of administration
`delivers the drug directly into the
`systemic circulation, bypassing both
`the gastrointestinal tract, where some
`
`drugs are degraded or metabolized, and
`the portal circulation and consequent
`first-pass hepatic extraction. For most
`drugs evaluated to date, bioavailability
`after sublingual dosage is equivalent to
`or greater than that after oral admin-
`istration. A similar principle holds for
`rectal drug administration, since ap—
`proximately 50% of the hemorrhoidal
`circulation empties into the systemic
`rather than the portal venous system.2
`Finally,
`the transdermal23 or pulmo-
`nary route can be used to administer
`some drugs.
`For all these reasons, drug concen-
`trations in blood, serum, or plasma
`often
`reflect pharmacologic
`action
`more closely than administered dosage
`alone.
`
`FACTORS INFLUENCING ‘
`INTERPRETATION OF
`SERUM DRUG CONCENTRATIONS
`Total vs Free Serum
`Concentrations
`
`'Although only the unbound or free
`drug can passively cross cell mem-
`branes and interact with receptors,
`free drug levels nonetheless are still
`not routinely monitored. This is partly
`because their measurement is techni-
`cally more difficult
`to perform than
`that of total
`levels. Fortunately, for
`most drugs, the ratio of free to total
`serum concentration (free fraction)
`usually remains
`relatively constant
`during a given patient’s course of thcr:
`apy, with salicylate and ibuprofeni'f‘
`being
`among
`notable
`exceptions.
`Therefore, a doubling of the total con-
`centration will also lead to a doubling
`of the free serum drug concentration at
`steady state. In most clinical circum-
`stances, variability between patients in
`free fraction may also be relatively
`small.26 When within- and between-
`individual differences in serum protein
`binding are small, monitoring of total
`serum concentration should prove to be
`as useful therapeutically as monitoring
`of free or unbound concentrationm'fl'fi
`In some conditions, however, drug
`binding to serum protein may be sub—
`stantially altered. For example, protein
`binding of a given drug may be reduced
`(increased free fraction) when another
`drug displaces
`it
`from its binding
`sitesmm Such interactions in them-
`selves are unlikely to be of direct
`clinical
`hnportance,1527'”'30
`since the
`increased "free” concentrations will be
`only transient due to rapid equilibra-
`tion with tissues. However, the total
`drug concentration will consequently
`fall, and may lead to a lowering of thfl
`therapeutic and toxic ranges for ti}?
`total serum drug level (Fig 5].“ Uremifi-
`and hypoalbunlinemia are other clifll'
`
`Finally, the availability of methods
`for measurement of drug concentra-
`tions provides the impetus for clini—
`cians to increase their expertise and
`understanding of pharmacologic and
`pharmacokinetic principles of drug
`therapy. Enhanced awareness of dose-
`concentration relationships, and fac-
`tors influencing these relationships,
`may lead to an overall improvement in
`the quality of drug treatment.”
`
`DRUG DISTRIBUTION AND
`ACCESS TO ITS RECEPTOR
`
`When a drug is given by an extravas-
`cular route of administration (orally,
`intramuscularly,
`rectally,
`subcuta-
`neously,
`sublingually,
`transdermally,
`
`2228
`
`JAMA, Oct 24/31, 1986—Vol 256, No. 16
`
`2
`
`Drug Monitoring—Friedman Bx Greenblafi
`
`4;
`
`SerumDigoxin,ng/mL
`
` .--l _._._t_;'-
`
`:4.-...'
`
`‘
`
`t_.‘_-
`.
`.
`12 3 4 5 6 7 8 9101112131415
`Digoxin, ug/kg/d
`
`_ F
`
`ig 1.——Relation of steady-state serum digoxin
`concentration to daily dose per kilogram for 100
`patients receiving long-term digoxin therapy.
`Correlation is poor (r=.069),
`indicating sub-
`stantial variability in steady-state concentration
`that is not explained by dosage (Hermann R.
`Ochs, MD, unpublished data,~1979).
`
`
`Toxic
`Possibly
`Toxic
`
`
`
`
`
`
`
`
`
`
`
`Therapeutic
`
`'5
`
`E‘EmoC
`
`8E
`
`E
`8
`
`Toxn:
`.
`Therapeutlc
`Partly
`artlv
`Effective
`Effective
`
`Not
`Not
`
`Effective
`Effective
`Narrow
`Wide
`Therapeutic
`Therapeutic
`Range
`Range
`
`Fig 2.—Schematic relation of serum or plasma
`drug concentration to clinical efficacy or toxicity
`tor hypothetical drugs having wide or narrow
`thtzerapeutic ranges (from Greenblatt and Shad-
`er ).
`
`2
`
`

`

`
`
`Intramuscular,
`Subcutaneous,
`or Transdermal
`
`Dose
`
`Oral Dose
`
`‘6
`9!
`E1
`.5
`2C
`'§
`‘Q
`
`i
`
`Site of
`
`
`
`Absorption
`
`\- "k
`
`
`
`Intravenous Dose
`
`\ 908/
`\
`
`\ .\
`
`.
`_> Liver
`
`4—h-
`
`
`.
`
`Site of
`A .
`.
`
`
`\ 55.
`Systemic
`/ CtMty
`Circulation
`
`.
`
`Protein-
`Bound Drug
`
`0 \
`
`Rectal
`Dose
`
`Unabsorbed
`Drug and
`Metabolites
`
`
`\
`d /
`Metabolites
`
`Tissues
`
`Kidney
`
`*
`Excretion
`
`
`Fig 3.—Schematic diagram of pathways of drug absorption, distribution, elimination, and clearance.
`
`Fig 4.—Estimated distribution pattern of benzodiazepine derivative nordazepam (dgsmethyldiaze—
`cal situations in which serum protein
`pam} In normal healthy woman [30% body fat). based on human autopsy BlUthS. Nordazeparn
`binding of drugs is reduced, causing
`(dssmethyldiazeparn) is major metabolite of diazepam (Vatium) and halazepam (Paxiparn). and Is
`lowered therapeutic and toxic ranges
`principal' active substance present
`in blood durlng treatment with clorazepate dipotassium
`for total drug concentration For exam-
`(Tranxene) and prazepam (central
`ple, phenytoin free fraction, which usu—
`ally falls between 10% and 20%, may§
`become as high as 30% in uremics.31
`Alternatively, oil-acid glycoprotein, an
`acute phase reactant, may be tran-
`siently elevated in acute myocardial
`infarction, shock, severe burns,
`inju—
`ries, or infectious processes,” causing
`increased binding of some basic drugs,
`and result
`in increased total serum
`drug levels without an enhancement of
`clinical effect. Examples of such drugs
`include lidocaine, propranolol, imipra-
`mine, phenytom, quinidine, and disopy-
`ramide. For drugs not extensively
`bound to serum proteins, such as cime-
`tidine, digoxin, and gentamicin, lithium,
`procainamide and acecainamide (N-
`acetylprocainamide), changes in pro—
`tein binding are of far
`less conse—
`quence.
`
`Adrenal (036%)
`Kidney (1.5%)
`
`Heart (1.9%)
`
`Optimal Sample Timing
`Proper choice of sampling time is
`Crucial for the interpretation of serumfl
`
`JAMA, Oct 24/31. 1986—Vol 256, No 16
`
`Drug Monitoring—Friedman & Greenblatt
`
`2229
`
`Muscle (37%)
`
`Fat (34%)
`
`
`
`
`
`3
`
`

`

`
`
`PlasmaConcentration.mg/L
`
`Bc
`
`15.0
`
`10,0
`
`F"o
`
`Fig 5.—Influence of change in protein binding on total and unbound serum or plasma
`concentrations of hypothetical drug at steady state. It is assumed that drug is being administered at
`constant dosing rate (8 mg/kg/d), and that drug's total clearance is also constant. When free
`fraction (FF) is 0.05 (left), total plasma drug concentration is 20 mg/L, and free concentration is
`1.0 mg/L (dotted lines). If for some reason extent of protein binding is reduced, and FF is increased
`to 0.10 (right), steady—state free drug concentration remains at 1.0 mg/ L because there has been
`no change in either dosing rate or clearance. However, total concentration falls to 10 mg/ L. Thus,
`change in protein binding (free fraction) by itself causes no alteration in free drug concentration, but
`will cause reciprocal change in total drug concentration (from Greenblatt and Shaderz).
`
`Css (After Reduction)
`
`Reduction
`in Dosage
`
`,
`
`
`
`
`
` MeanSerumConcentration
`
`it
`In general,
`drug concentrations.
`takes four times the drug’s half-life at
`a constant dosing rate for the steady-
`state condition to be more than 90%
`attained. Similarly, an increase or
`decrease in dosage will
`require the
`same time interval to reach the new
`steady-state level (Fig 6). After initiat-
`ing therapy with long half-life drugs, a
`considerable length of time may be
`required for steady state to be attained
`(Fig 7). Therefore, sampling before the
`attainment of the actual steady-state
`condition may lead to premature dos-
`age adjustments. This is of particular
`importance for drugs such as theophyl-
`line that are administered to infants
`and children.
`Occasionally, the need may arise to
`hasten the attainment of therapeutic
`concentrations. This can be achieved by
`giving an initial loading dose, the size
`of which has been appropriately chosen
`based on the desired therapeutic con-
`centration and the pharmacokinetic
`characteristics of the drug.”34 How-
`ever, even the ideally selected loading
`dose ha
`tenti l
`i
`n
`. The
`S Do
`a d sadva tages
`Fig 7.—Time course of attainment of steady-state condition. assuming drug is given once daily.
`rapld tttamment 0f therapeuuc con'
`Case A indicates drug with short half-life; case B, drug with long half-life (from Greenblatt and
`centratlons precludes gradual adapta-
`Shaderz).
`— tion to therapeutic or adverse drug
`effects, such as sedative, hypotensive,
`bradycardic, or anticholinergic proper-
`ties.
`,
`Once the steady-state condition has
`been achieved, the mean steady-state
`serum drug level should remain con-
`stant as long as the dosing rate and
`clearance are constant (as indicated in
`the equation in the first section of this
`article). However, the interdose fluc-
`tuation depends on the dosage interval-
`A proportional increase or decrease in
`both the size of each dose and the
`interval between doses, such that the
`overall dosing rate remains constant.
`does not change the mean steady-state
`concentration, but will alter the inter-
`— dose fluctuation (Fig 8). More frequent
`
`Css (Initial Dosage)
`
`012 34 56 780123456
`
`MumP'eS 0‘ HalHife
`
`Fig 6.'—Tlme course (in multiples of half-life) of mean serum concentration during attainment of
`steady-state condition after starting therapy and after reducing dosage. Css indicates mean serum
`concentration at steady state (from Greenblatt and Shaderz).
`
`Css (Case B)
`
`Concentration
`Serum
`
`_ _ 035(Case A)
`
`0123456789101112131415
`
`Days
`
`2230
`
`JAMA, Oct 24/31, 1986—Vol 256, No. 16
`
`Drug Monitoring—Friedman & Greenblafi
`
`4
`
`

`

`drugs with narrow therapeutic indexes,
`such as aminoglycosidcs or lithium.
`Unfortunately, however,
`the time of
`peak concentration can seldom be pre-
`dicted with certainty.
`If the dosage interval is not regular,
`or if the drug is taken intermittently,
`then the best time to sample is not
`necessarily so obvious, since there is no
`single “trough” concentration (Fig 10).
`
`Artefacts due to Collection Tubes
`The Vacutainer brand of blood col-
`lection tubes’is reported to contain
`TRIS (2-butoxyethyl) phosphate,
`a
`plasticizing
`agent. Blood
`samples
`drawn into these tubes can give spu-
`riously low serum drug levels when the
`serum is analyzed for
`imipramine,
`alprenolol, propranolol,
`lidocaine, and
`quinidinef“ The mechanism for
`the
`lowering of serum levels appears to
`involve displacement of drugs from
`oil-acid glycoprotein (but not
`from
`albumin)
`by TRIS (2-butoxyethyl)
`phosphate. This in vitro phenomenon
`results in an increase in unbound drug,
`which quickly diffuses into and equili-
`brates with the red blood cells present
`in the tube. Thus, when the serum is
`aspirated after centrifugation, the re-
`sultant serum drug level is spuriously
`low. However, the whole blood level is
`unchanged. Any drug that is extensive-
`ly bound to all-acid glycoprotein is
`likely to be influenced by this collection
`artefact.
`
`Analytic Methodology
`Knowledge of the methodology used
`by a laboratory in analyzing serum for
`drug levels may be of critical impor-
`tance for the clinician in interpreting
`the results. Ideally, an assay procedure
`for a particular drug should (1} resolve
`compounds of similar structure, such
`as the parent drug and its metabolic
`products or other substances present in
`the serum (specificity); (2) consistently
`conform to accepted standards for
`accuracy and replicability for the range
`of concentrations encountered clini—
`cally, and (3) be sensitive enough to
`quantitate levels. well below the thera-
`peutic range. In addition, the need for
`cost containment must always be con-
`sidered. Procedurally straightforward
`analytic methods that can be auto-
`mated are generally less expensive and
`therefore preferred. However,
`such
`procedures, although less costly, may
`not provide adequate specificity, accu-
`racy,
`replicability,
`and sensitivity.
`More complex and often more expen-
`sive analytic methods may be needed
`.to provide meaningful serum concen-
`tration data.
`Historically, spectrophotometry and
`
`Drug Monitoring—Friedman & Greenblatt
`
`2231
`
`K 500 mg Every 24 Hours
`
`
`
`——————————————————
`K 250 mg Every 12 Hours
`\
`125 mg Every Six Hours
`
`
`
`035
`
`Therapeutic
`Flange
`
`25
`
`20
`
`15
`
`10
`
`
`
`SerumConcentration
`
`
`
`0
`
`6
`
`12
`
`24
`
`Hours
`
`§ F
`
`ig 8.—lnterdose fluctuation of semm drug concentration as function of dosage schedule,
`assuming that drug is given in overall total dosage of 500 mg/24 h, but with different dosing
`schedules. Note that mean serum concentration at steady state (055) is same for each regimen,
`and that interdose fluctuation is largest for once-daily therapy (from Greanblatt and Shader2).
`
`small fluctuations in serum drug con-
`centrations.36
`At steady state, each discrete drug
`dose is followed by an “absorptive”
`phase, during which serum concentra—
`tions exceed the mean. Transient side
`effects may be associated with the
`absorptive peak. After peak concentra-
`tions are reached, the serum level then
`falls as distribution and clearance pre-
`dominate. Just before the next dose,
`levels are at a minimum during the
`“trough” phase. Sampling shortly after
`a dose, during the absorptive phase, is
`not
`recommended for evaluation of
`therapeutic efficacy since the measured
`level does not necessarily correspond to
`the peak. Furthermore, even if the
`peak level was found to be in the
`therapeutic dosing range, this would
`not ensure therapeutic levels through—
`out
`the entire dosage interval. The
`optimal time to sample for evaluation
`of efficacy is just before the time of
`dosing (Fig 9),
`to ensure that
`the
`minimum drug level falls with the
`therapeutic range. If the trough level is
`found to be subtherapeutic, the clini-
`cian may elect to give smaller doses
`more frequently while maintaining the
`same total dose per 24 hours (Fig 8).
`This change would reduce the interdose
`fluctuation and possibly bring the
`trough level to within the therapeutic
`range. Measurement of peak serum
`concentrations after an individual dose
`may be of value when clinicians wish to
`evaluate potential drug side effects
`coinciding with peak concentrations.
`Knowledge of both peak and trough
`concentrations may be desirable for
`5
`
`— l
`
`Dose
`
`Dose
`
`i
`
`i
`
`Trough
`
`Absorption
`Peak
`_ i
`i .
`i<—->l
`(Best Time
`(Not Optimal
`In' Sample}
`for Sampling)
`I,_|_.|____|_I_I——|—|—I
`0
`2
`4
`10
`12
`6
`B
`Hours After Dose
`
`c
`.9
`
`1
`EEa:or:o
`
`O E
`
`:1
`._
`(2‘
`
`— F
`
`ig 9.—Tlme course of serum drug concentra-
`tion at steady state during oral dosage every 12
`hours, with illustration of optimal sampling time
`(from Greenblatt and Shaderz).
`
`dosing is useful to minimize transient
`effects due to high peak levels that
`some people find objectionable, such as
`Sedation and drowsiness from certain
`psychotropic drugs.“5 0n the other
`hand, dosing schedules that require
`Very frequent dosing are inconvenient,
`and may be associated with reduced
`patient compliance.
`_Certain sustained—release formula-
`tIons of drugs have been designed to
`prolong drug action after each dose,
`1Shcreby allowing less frequent dosing.
`If the rate of drug entry into the
`i‘iiStemic circulation precisely mimics a
`flmdfate infusion,
`then the serum
`free level will not'flnctuate. Although
`his is not an attainable ideal, some
`.sPstained-release preparations do in
`act allow infrequent dosing, with only
`
`JAMA. Oct 24/31, 1986—Vol 256, No. 16
`
`5
`
`

`

`enzymatic label for a radioactive one.
`The inhibition of labeled enzyme activi-
`ty by antibody is the basis for the very
`popular enzyme—multiplied immunoas-
`say
`technique. This
`technique
`is
`claimed to be procedurally straightfor-
`ward and inexpensive, but is considera-
`bly less sensitive than RIA and can
`have variable specificity, particularly
`when applied to screening for drugs of
`abuse in urine.12'4°'“ Most nonisotopic
`immunoassay labels are inactivated by
`antibody; hence,
`they do not require
`the separation of bound from free
`labeled ligand as in RIA.
`Chromatography is a method of sep-
`arating mixtures of substances based
`on their physicochemical characteris-
`tics, so that one or more of
`those
`substances may be
`specifically de-
`tected. The principal methods used in
`drug measurement are gas-liquid chro—
`matography (GLCI’: and high-pressure
`liquid chromatography (HPLC).“3 Se—
`rum, or a concentrated extract thereof,
`is
`injected onto a column through
`which flows a mobile phase. For GLC,
`the mobile phase is a purified gas such
`as helium, nitrogen, or argon. For
`HPLC, the mobile phase contains mix-
`tures of an aqueous buffer and an
`organic solvent such as acetonitrile or
`methanol. Separation of
`the serum
`components by the mobile phase is
`influenced by interactions with the
`column’s stationary phase. Actual sep-
`arations are based on lipophilicity,
`polarity, molecular size,
`temperature,
`and boiling point (for GLC). The HPLC
`separations may be further refined by
`varying the pH and polarity of the
`mobile phase. Often it is possible to
`detect and quantitate the parent drug
`and some or all of its important metab-
`olites simultaneously.
`For some applications, a mass spec-
`trometer is coupled to a chromato-
`graph’s effluent and thereby acts as
`the detector. This combination pro-
`vides the “gold standard” in specificity
`and sensitivity in drug analysis. How-
`ever, mass spectroscopy is expensive
`and requires complex instrumentation
`as well as highly trained personnel.
`When the drug mixture is well sepa-
`rated chromatographically, other less
`complex detection systems usually suf-
`fice. In GLC, the most commonly used
`detectors are flame ionization, nitro—
`gen-phosphorus, and electron capture.
`Flame ionization will respond to all
`organic compounds. Nitrogen-phospho-
`rus has enhanced response to nitro-
`gen- and phosphorus-containing com-
`pounds, and electron capture responds
`to drug-containing electronegative sub-
`stituents such as halogens, nitrates, and
`conjugated carbonyls. The sensitivity
`
`l
`
`of nitrogen-phosphorus and, electron
`capture is in the subparts-per-billion
`range; that of flame ionization is on
`the order of parts per million. High-
`pressure liquid chromatography most
`commonly utilizes spectrophotometry,
`fluorescence, or electrochemistry for
`the detection and quantitation of drugs
`after chromatographic separation. Flu-
`orescence detection is applicable to
`molecules with rigid
`polyaromatic
`structures or extensive
`conjugation
`having the property of absorbing light
`and then emitting it at a lower wave—
`length. Electrochemical detection is
`employed for the recognition of easily
`oxidized or
`reduced groups such as
`phenols,
`indoles, and secondary and
`tertiary nitrogens. In favorable cases,
`fluorescence and electrochemical detec-
`tion can extend sensitivity two orders
`of magnitude beyond the spectrophoto-
`metric range.
`be
`can
`information
`The
`above
`extremely useful to clinicians in inter-
`preting serum drug concentration re-
`ports. Consider, for example, a “nonde—
`tectable” serum concentration report.
`This must be interpreted in light of the
`lower limits of sensitivity of the partic-
`ular assay technique.
`If
`the assay
`technique has a high degree of sensitiv-
`ity, a reported zero level may actually
`mean that there is no drug present in
`serum. Conversely, it may imply that
`clinically important amounts of drug
`may be present in serum, but the assay
`is not Sufficiently sensitive to quanti-
`tate levels in this range. At the other
`extreme, laboratories may report very
`high serum drug concentrations in a
`patient taking “usual doses” and hav-
`ing no manifestations of toxicity. This
`might be attributable to the use of a
`very nonspecific assay technique that
`quantitates not only the drug in ques-
`tion but also its metabolic products, or
`possibly other endogenous unrelated
`substances present in serum. Finally,
`laboratories may report widely discrep-
`ant serum drug concentrations from
`day to day in samples drawn under
`identical conditions in a given patient
`receiving a constant dose of a drug-
`These variations could be due to varia-
`bility over time in the patient’s meta-
`bolic clearance or extent of drug
`absorption. They could also be attribut—
`able to insufficient accuracy in the
`laboratory determination, regardless of
`the method.
`.
`
`Laboratories should readily provide
`to any inquiring physician all details 0f
`analytic quality control procedurefi-
`Nonetheless, clinicians have means at
`their disposal to test and compare the
`performance of laboratories. If a pure
`reference standard of the drug in ques-
`
`Drug Monitoring—Friedman 8. Greenblan
`
`Doses
`I—‘"
`l
`l
`l
`
`l
`
`l
`
`l
`
`,5 40'
`E*E
`30%
`ED0CO
`
`0 20
`
`EE
`
`:95 10-
`
`.1‘
`I
`I
`-
`-
`0
`4
`8
`12
`24
`Hours
`
`g.
`
`Fig 10.—Plasma concentrations of drug at
`steady state with four-times—daily dosing sched-
`ule, with individual doses given at times shown
`by arrows. This complex dosing schedule
`makes monitoring of plasma drug levels more
`difficult (from Greenblatt and Shaderz).
`
`colorimetry were the first methods
`widely used in laboratories for mea-
`surement of drug levels.37 These proce-
`dures may require one or more solvent
`extractions, often coupled with chemi—
`cal reactions, to yield a solution con-
`taining primarily the drug of interest
`The absorption of visible, infrared, or
`ultraviolet
`light at a specific wave—
`length by the drug in solution is used to
`quantitate its presence. The level of
`sensitivity is usually on the order of
`parts per million to parts per thousand.
`These methods are limited by poor
`sensitivity and variable specificity,”
`and high cost.
`Immunoassays for drugs have be-
`come popular within the last 20 years.38
`In principle, they rely on the interac—
`tion between a drug acting as an
`antigen and an antibody to it. Since
`most drugs are nonimmunogenic, they
`first must be conjugated by a bridge or
`linkage group to a substance of high
`molecular weight, such as a protein. In
`this conjugated mode, the drug behaves
`like a hapten and is used to immunize
`an animal. Antibodies may be gener-
`ated against the drug if the conjuga-
`tion bridge or linkage keeps the drug
`sufficiently far from the larger protein
`molecule. The necessity for an antibody
`to seek out a hapten creates the inher-
`ent variability in specificity provided
`by immunoassays. Some antibodies
`may show cross-reactivity with metab-
`olites and congeners of the drug of
`interest,
`thereby rendering the anti-
`bodies relatively nonspecific.39
`In the radioimmunoassay (RIA),
`drug present in a serum sample com-
`petes with
`a
`radioisotope-labeled
`ligand for antibody binding sites. The
`RIA procedures are stated to be very
`sensitive, but
`require
`the use
`of
`radioactive material and are costly.
`Enzyme immunoassays substitute an
`
`2232
`
`JAMA, Oct 24/31, 1986—VOI 256, N0 16
`
`6
`
`

`

`physicians may
`available,
`tion is
`“spike” drug-free control serum with
`various known concentrations of drug,
`and send them “blindly” to the labora-
`tory. This procedure should provide
`information on the sensitivity and
`accuracy of the laboratory’s methodo-
`logy. Another procedure that can be
`used to test the replicability of deter—
`minations either within the same labo-
`ratory or between laboratories is to
`split a given serum sample into two
`aliquots. The aliquots can be then sent
`to different
`laboratories
`and the
`results compared. Alternatively,
`the
`same divided sample under different
`designations can be sent to the same
`laboratory on different days. The labo-
`ratory should quantitate the identical
`or nearly identical results in these split
`samples. When results on identical
`samples differ by less than 10%, the
`laboratory performance under
`these
`circumstances is acceptable.
`References
`
`1. Koch-Weser J: Serum drug concentrations as
`therapeutic guides. N Engl J Med 1972;287:227-
`231.
`2. Greenblatt DJ, Shader RI: Pharmacokinetics in
`Clinical Practice. Philadelphia, WB Saunders Co,
`1985.
`3. Greenblatt DJ, Koch-Weser J: Clinical pharma—
`cokinetics. N Engl J Med 1975;293:702—705, 964-
`970.
`4. Williams RL: Drug administration in hepatic
`disease. N Engl J Med 1983;309:1616-1622.
`5. Hoyumpa AM, Branch RA, Schenker S: The
`disposition and effects of se