Monitoring compliance through analysis of drug and metabolite levels

Monitoring Compliance Through Analysis of Drug and Metabolite Levels Joseph A. Mollica Ciba-Geigy Corporation, Summit, New Jersey

ABSTRACT: Cold analytical methodology is usually available for drug and metabolite monitoring during clinical trials, since a procedure is required for the bioavailability, pharmacokinetic, and dose proportionality studies that must be conducted by the sponsor. Such methods can and have been applied to monitoring patient compliance. Examples from several classes of drugs with different pharmacokinetic profiles illustrate the type of data that can be obtained, along with their applicability and inherent limitation in assessing compliance. The effects of concomitant medication, drug half-life, volume of distribution, and sampling time on observed levels are also discussed. Several other approaches involving trace metals, microtaggants, and an electronic monitor are also presented. Various m e t h o d s have been employed to assess patient adherence to a dosage regimen during clinical trials. As one progresses from initial phase I studies in volunteers to large, multicenter, long-term intervention trials, the available options become more limited. The techniques employed d e p e n d on m a n y factors, including the nature of the illness or disease, w h e t h e r the treatment is short-term or chronic, the dosage form employed, and w h e t h e r the drug is administered to volunteers or to patients. It makes a difference w h e t h e r the subjects are institutionalized or are ambulatory or bedridden. The clinical phase (I, II, III, or IV) of the drug trial, where the s t u d y is c o n d u c t e d - - a n in-house unit or in multiple centers--also dictates which methods can be employed. Finally, the accuracy and precision of the m e t h o d o l o g y and, therefore, the data derived from such assessments are ultimately subject to wide variation. The techniques to monitor compliance can be grouped u n d e r five headings: administration, observation, consultation, reconciliation, and biochemical measurement. These are well d o c u m e n t e d and have been the subject of other studies and reports [1-7]. Before discussing the area of biochemical measurement, I would like to make a few observations on a widely used technique, the pill count. In a clinical setting, it is difficult to control the n u m b e r of dosage units, since the Address reprint requests to: Joseph A. Mollica, Ph. D., Senior Vice President, Development, Ciba-Geigy Corporation, 556 Morris Avenue, Summit, New Jersey 07901. Controlled Clinical Trials 5:505-514 (1984) © Elsevier Science Publishing Co., Inc. 1984 52 Vanderbilt Avenue., New York, New York 10017

5O5 0197-2456/84/$03.00

506

Joseph A. Mollica time at which a prescription is filled, the accuracy of the prescription count, and the availability of medication already on hand or of multiple prescriptions all contribute to the uncertainty of the count. In a controlled clinical trial, however, most of these sources of error are eliminated. The investigator is provided with supplies identified by patient number, visit number, package number, and bottle count. The investigator records on the patient record the amount dispensed and returned. All unused materials, including empty and partially filled bottles, are returned to the sponsor for reconciliation and are inventoried and identified by patient number and visit. Table I provides an example of a completed initial inventory form in which the returns are designated empty, partial, full, or missing. In this protocol, only 6 out of 720 bottles were missing. This can be reconciled with the data entered on the patient record by the investigator. From such information, the precentage of medication taken for a specific time period can be calculated. Obviously, this does not assure whether the medication was taken according to the prescibed regimen, whether the patient shared the drug with family or friends, or whether the missing tablets were thrown out. The reconciliation data can, and should, be used by the investigator during consultation with the patient and serve as a baseline comparison for biochemical methods. Biochemical analysis of the drug, a metabolite, an endogenous blood or urine constituent known to be responsive to the drug, or of an added marker can be combined with a pill count or used as an independent ~neasure to determine drug compliance. For example, in the study of sulfinpyrazone as a preventive of cardiac death after myocardial infarction (Anturane Reinfarction Trial), drug compliance was monitored by tablet counts and measurement of the patients' serum uric acid level [8]. Uric acid was reduced to 470% of baseline in 89% of the patients. Measurement by pill count of the entire sample demonstrated that 87% had an overall compliance of 980%. In a study on the use of propranolol in patients with acute myocardial infarction, serum propranolol determinations were made at 1.5 and 6 months after randomization and every 6 months thereafter [9]. As in the Anturane trials, the laboratory results were not made available to the clinical center personnel. During the conduct of phase I trials, cold methodology for determining the drug or metabolite in plasma or urine is usually developed and validated to establish pharmacokinetic parameters. Hence, methodology is usually available to monitor compliance via these techniques. Although analytical methods (GC, HPLC, GCflVIS) for direct monitoring are complex and time consuming, they are specific for the drug or metabolite, and one can easily detect levels below 1 ng/ml with such chromatographic techniques. A disadvantage is that unless a drug is in widespread use, which is usually not the case during clinical trials, only one or a few laboratories will have the capability to analyze for the agent. Figure 1 presents steady-state blood level data on samples taken from individual volunteers prior to the first dose of the day. The bars represent average concentrations over several days (number of visits) and the lines represent the standard deviation. Thus one sees quite large intersubject variation larger than intrasubject variability as represented by the standard deviation. Therefore, to use this technique to assess continued adherence to a regimen, one would need a knowledge of the steady-state level expected for an individual. Figure 2 provides steady-state concentrations for the same drug

I

b,.o

r~

fD

o

t'o

~°

>

508

Joseph A. Mollica

500

ng/ml

400

300

200

100

0

!:i:!:,::

A

B

C

D

E

F

G

No. of Visits 11

12

11

11

12

20

16

Subject

Figure 1 Steady-state blood levels of an antidepressant before daily dose of 150 rag/day.

600

nglml

5OO

400

300

:,:.

11111

o Visit

A

B

C

D

E

G

H

I

J

K

L

M

N

O

P

Q

R

$

Figure 2 Steady-state blood levels of an antidepressant of one patient receiving 150 mg daily over several weeks.

509

Analysis of Drug and Metabolite Levels

collected just prior to the first dose of the day over a period of 42 days. Levels vary from 200 to 600 ng/ml. This variation is seen for a drug with a large volume of distribution (Vd = 23 l/kg), which is indicative of extensive tissue distribution and quite a long half-life (43 hours). Visit A is 8 days after start of therapy when levels should be at least 90% of steady state. The range of steady-state plasma levels found in individual patients following doses of 300 or 500 rag/day of a nonsteroidal antiinflammatory drug (NSAID) are plotted in Figure 3. This drug has a small volume of distribution (Va = 0.18 1/kg) and a very long half-life (2-5 days). Samples were collected prior to the first dose of the day and then 6 hours later. Considerable intrasubject variation is depicted by the standard deviation of the zero hour data points and also variation between subjects. The steady-state plasma levels for another nonsteroidal antiinflammatory drug are depicted graphically in Figure 4. This drug has a moderate half-life ( - 7 hours) and a small volume of distribution (Va = 0.171/kg). Considerable

300

~g/ml

2OO

1

i

~:~: ::'4~ ~::~.~:;.: ~:::2.:.-:~. :i:'.
Time*

0,6

o,s

0,6

6

0,4

Number of samite days

5,2

5,1

2,3

2

2

1

S

D

E

F

G

H

300

30O

3OO

3OO

50O

5OO

Subject Do~lday

0,4

*Houri post rooming dole

Figure 3 Steady-state plasma levels of an NSAID averaged over several visits at two

time points.

510

Joseph A. MoUica 32 ~ /ag/ml I

7

I

T

24

"//A N Visit

A

B

POpU(ation

1

n

6

Time*

AB

AB

COE

AB

A

AB

CDE

AB

AB

AB

2

3

4

I

2

3

4

I

2

3

5

7

4

6

5

7

4

6

5

0

4

-

-

6

7 -

-

*HOurs post dose

Figure 4

Steady-state plasma levels of an NSAID.

interpopulation variability is seen for both visit A and B and is somewhat greater for the samples collected prior to the morning dose. Concomitant medication can also significantly affect the observed plasma levels of a drug. In Figure 5, average drug levels are seen in 20 patients receiving a daily dose of 800 mg of the NSAID depicted in Figure 4 and 2.4 g of aspirin. The decrease in the presence of aspirin is quite apparent. In a study of another NSAID, we monitored, in a sense, negative compliance. The protocol requires that aspirin not be taken. To detect what are colloquially called "aspirin cheaters," urine samples are analyzed for presence of salicylate. We have attempted to show through the preceding examples that both intrasubject and intersubject variation are encountered even when dosage, .Q/ml

~/rnl 881i¢ f |lllil 120

Drug

~ruu Asl~rin

Day 14

DQy 42

Oa¥ 56

6O

0

4

6

0

2

4

6

0

$

Hou~ Po~t A,M. [3oee

Figure 5

Average NSAID ( , ) and salicylate ({~) plasma concentrations in patients (20) receiving daily drug (800 rag) and aspirin (2400 mg) doses alone and concomitantly.

Analysis of Drug and Metabolite Levels

511

administration, sampling time, storage, and shipment of the samples are well controlled and the analyses are conducted in a single laboratory. In a large srudy of an antidepressant, the control drug was imipramine. The parent compound and its desmethyl metabolite were monitored to assess compliance. The data are presented in Table 2. The blood samples reportedly obtained at visit 11, the final visit, did not give measurable levels of the tricyclics for the majority of the subjects. The blood data through visit 9 (13 weeks) and in some cases visit 10 (19 weeks) appear reasonable. The lack of reliable blood levels to support compliance after 19 weeks also corresponded with some confusing entries in the patient record form. The levels of ~10 ng/ml obtained at the final visit for a number of patients were thought to be related to the lag time of 1 month from the collection of blood to the actual laboratory analysis. The drug was known to be stable for this time period, but a possible factor that was thought to contribute to these low levels was a change in plastic tubes used to collect the sample. However, studies conducted in our laboratories to determine if the drugs could be adsorbed onto or absorbed into the plastic after 1 month's exposure did not support this hypothesis. The conclusion of poor compliance for this portion of the study was quite evident. Hence monitoring a drug, metabolite, or another parameter can show compliance of a type. The degree of assurance will depend on the pharmacokinetic parameters, dose proportionality, whether levels change due to enzyme induction, sampling time, or effect of concomitant medication and accuracy of the method. Even so, one may not know if a patient has skipped intervening doses (or has taken one dose prior to the sample), or has taken his medication only some of the time, or has only taken half the prescribed dose, and so forth. In addition to these biological or technical limitations, there are the cost and logistics of sampling and analysis, lack of patient acceptance or outright resistance or refusal, and the potential for breaking the blind. There are times when it is desirable to have the study triple blind so that the monitor evaluating the data does not know which patient received which treatment. Hence, information on adherence via biochemical monitoring as opposed to other approaches must be thoroughly weighed. One can infer that no single substance could be ideally used for all situations. However, metals and trace elements as a class deserve some consideration for use as markers. Various elements have been added to products for identification. Some tablets or capsules have trace mixtures added to identify a product where the potential for counterfeiting exists. Recent advances in spectroscopy make the detection of low levels quite feasible. In the last 5 years, a variety of revitalized, modified, and new techniques have emerged for the analysis of trace elements. Analytical methods typically used for biological samples are given in Table 3. A combination of metals could be employed depending on their pharmacokinetic properties and those of the drug under study. Another technique we may wish to address is the use of microtaggants. For example, microscopic plastic particles of approximately 40-400 ~m composed of seven distinct colored layers have been employed to identify explosives, oil spills, and medicated premixes for animal and poultry feed [11,12]. The colors and sequences of the layers can be interchanged, thereby making available several million possible codes. These would not be absorbed, but factors such as transit time, propensity to become lodged, and the like must

23 20 13 17 16 12 22 15 34 12 20 111 15 11 117 66 28 17 10 6 29 25

19" 82~ 5

116

150mg

50 m g

50 mg

50 mg

50 mg

50 mg

50 mg

50 m g

50 mg

50 m g 70 50 m g

39 18 205 225

45 45 74 54 205 254 92 571 140 48 442 165

24

--

56 136

150 mg

150 m g

150 mg

150 mg

150 m g

150 m g

922 242 157 106 49 66 <20

73 688 <20

56 34 --

200 m g

300mg

300mg

Dosage

10B

<10

<10

<10

<10

<10

<10

<10

<10

300 mg

300 m g

50 m g

200 mg

150 m g

300 m g

200 m g

300 m g

Visit 11 26 weeks Level Dosage

D r o p p e d Out . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

300 m g

150mg

90 46

169

Level

Visit

300 m g -<10 150 mg (Dosage Reduced Due To H i g h Level) 150 mg -31 200 m g 13 150 m g -86 50 mg (Dosage Reduced Due To H i g h Level) 300 m g 93 300 m g 127 300 m g 20 71 300 m g -52 300 m g 74 300 mg Dropped Out ............

33 300 mg 59 150 m g . . . . . . . . . . . . . . . . . . . .

150 m g

<20

143 49 <10

--

300mg

150 mg

Dropped Out

150mg

300mg

Dropped Out

60 181 <20

Visit 10 19 weeks Level Dosage

150 m g . . . . . . . . . . . .

31 150mg 32 50mg ...........................

50mg 255 50mg

°Imipramine level (in ng/ml). bDesipramine level (in ng/ml).

025

024

023

022

021

016

014

013

010

006

OO4

O03

O01

Patient

Visit 9B 13 weeks Level Dosage

Visit 2 1 week Level Dosage

Visit 5 4 weeks Level Dosage

Imipramine and Desipramine Plasma Levels

Table 2

oi...al I,o

3

Normal serum (ppm)

.

0.02 0.33 0.07 0.5 2 1 0.9 0.009

.

0.07 0.06 0.04 -0.6

0.003 0.01 0.003 0.02 --

.

.

0.00006 0.00004 0.00004 -0.000008

0.000004 0.0001 0.000007 0.0004 --

ETA-AAS sensitivity (ppm)

0.1 1 0.1 100 1

O.5 3 O.1 0.1 --

"Data from reference 10.

0.0001 0.00008 0.0004 -0.0005

0.0002 -0.000003 0.000001 0.02

ICP-ES detection limits (ppm)

0.00001 0.1 0.00001 0.0002 1 0.0002 0.0001 0.2 0.0002 0.0001 100 0.03 0.007 0.5 0.01 0.0004 0.5 0.003 0.0005 0.5 0.00006 0.000002 5 0.0001 Analytical Techniques Surveyed

AAS sensitivity (ppm)

Anodic-stripping voltametry (ASV) Atomic absorption spectrometry (AAS) Electrothermal atomization-atomic absorption spectrometry (ETA-AAS) Emission spectroscopy (ES) Inductively coupled plasma-emission spectrometry (ICP-ES) Neutron activation analysis (NAA) Spark-source mass spectrometry (SSMS) X-ray fluorescence spectrometry (XRF) Proton-induced x-ray excitation spectrometry (PIXE)

-0.002 -0.00004

--

----

.

0.000005 ---

--

--

ASV detection limits (ppm)

DC arc ES detection limits (ppm)

0.001 0.1 0.7 0.01 -0.03 0.002 0.1

0.002

0.01 0.3 0.002 -2

0.004 0.2 0.5 4 --

NAA sensitivity (ppm)

0.0004 0.003 0.002 0.002 0.0002 0.004 0.0004 0.002

0.001

0.0005 0.0005 0.002 0.0002 O.0005

0.0002 0.0003 0.0002 0.0003 0.0003

SSMS detection limits (ppm)

A Summary of Several Elements and the Detection Limit of Methods Commonly Used in Biological Analysis a

Major Essentials Na 140 K 4.4 Mg 1.85 Ca 5 P 114 Trace Essentials Co 0.0003 Cr 0.0028 Cu 1.14 F 0.0028 Fe 1.15 I 0.07 Mn 0.0014 Mo 0.004 Ni 0.03 Se 0.011 Si 0.08 Sn 0.033 V 0.001 Zn 1.0

Element

T a b l e

0.2 1.5 0.2 -11 -0.15 0.2

0.2

0.2 0.7 0.4 1 0.3

1 2 -0.5 4

XRF detection limits (ppm)

ol e.o

O"

>

514

Joseph A. Mollica be addressed. H o w e v e r , these markers are retrievable (for example, from b o m b site debris) and hence could be isolated from feces, since they are easily detected t h r o u g h fluorescence and can be lifted out with magnets. A n o t h e r interesting technique to monitor compliance was d e v e l o p e d by ALZA [13] to assess a d h e r e n c e to pilocarpine drops in the treatment of ocular h y p e r t e n s i o n or glaucoma. A m e m o r y element, controlled b y a digital clock, was built into an e y e d r o p container. Each h o u r a n e w m e m o r y location was addressed. If during that h o u r the cap was u n s c r e w e d (activating a microswitch) and the bottle was inverted (activating a n o t h e r microswitch), t h e n the m e m o r y location is loaded. If neither e v e n t occurred or only one, the location remained e m p t y . This p r o v i d e d information on which hours a patient unscrewed and inverted the bottle, which was taken as evidence that a d r o p was directed to the eye. There was sufficient capacity for 1000 hours. One can readily envision that such a miniature unit could be built into a tablet bottle or any other d r u g container. In s u m m a r y , our direct experience is with monitoring biological fluids for drugs or metabolites and reconciliation of supplies d i s p e n s e d and returned. We haste had experience with trace metals and microtaggants to identify brands and to establish manufacturing site a n d time. The use of an electronic monitor incorporated into an e y e d r o p container has b e e n used to assess adherence.

REFERENCES 1. Haynes RB, Mattson ME, Engebretson TO, Eds: Patient Compliance to Prescribed Antihypertensive Medication Regimens: A Report to the National Heart, Lung, and Blood Institute. NIH Publication No 81-2102, October 1980 2. Haynes RB: The effect of therapeutic regimen on patient compliance and the possible influence of controlled-release dosage forms. In: Controlled Release Pharmaceuticals, Urquhart J, Ed. Washington, D.C.: American Pharmaceutical Association, 1981, p 121 3. Haynes RB, Taylor DW, Sackett DL, Eds: Compliance in Health Care. Baltimore, Johns Hopkins University Press, 1979 4. Roth HP, Caron HS: Accuracy of doctors' estimates and patients' statements on adherence to a drug regimen. Clin Pharmacol Ther 23:361-370, 1978 5. Roth HP, Caron HS, Hsi BP: Measuring intake of a prescribed medication. A bottle count and a tracer technique compared. Clin Pharmacol Ther 11:228-237, 1970 6. Vaisurb S: Monitoring for mendacity. J Am Med Asoc 241:2194, 1979 7. Stark JE, Ellard GA, Gammon PT, Fox W: The use of isoniazid as a marker to monitor the self-administration of medicaments. Br J Clin Pharmacol 2:235-358, 1975 8. The Anturane Reinfarction Trial Research Group: Sulfinpyrazone in the prevention of cardiac death after myocardial infarction. The Anturane reinfarction trial. N Engl J Med 298:289-295, 1978 9. A randomized trial of propranolol in patients with acute myocardial infarction. I. Mortality results. J Am Med Assoc 247:1707-1714, 1982 10. Morrison GH: Elemental trace analysis of biological materials. CRC Crit Rev Anal Chem 8:287-320, 1979 11. Tagged explosives---Reading the code after the blast. Chem 50:23-24, 1977 12. Product Data Sheet, 3M Brand Microtaggant Particles. 3M, Minneapolis, Minnesota 13. Urquhart J: Personal communication. ALZA Corp., Palo Alto, California