toring-Randomized controlled trial-Clinical trial-Anti- .... Characteristics of the patients included in the trial ..... pellaro (Verona, 27); P. Benedetti, M. Brinciotti, F.
E p i k p i u , 41(2):222-230, 2000 Lippincott Williams & Wilkins, Inc., Philadelphia 0 International Leaguc Against Epilepsy
A Multicenter Randomized Controlled Trial on the Clinical Impact of Therapeutic Drug Monitoring in Patients with Newly Diagnosed Epilepsy G. Jannuzzi, *P. Cian, C. Fattore, G. Gatti, A. Bartoli, *F. Monaco, and E. Perucca, on behalf of The Italian TDM Study Group in Epilepsy Clinical Pharmacology Unit, University of Pavia, Pavia; and *Neurology Clinic, University “Amedeo Avogudro, ” Novara, Italy
Summary: Purpose: To assess the clinical impact of monitoring serum concentrations of antiepileptic drugs (AEDs) in patients with newly diagnosed epilepsy. Methods: One-hundred eighty patients with partial or idiopathic generalized nonabsence epilepsy, aged 6 to 65 years, requiring initiation of treatment with carbamazepine (CBZ), valproate (VPA), phenytoin (PHT), phenobarbital (PB), or primidone (PRM) were randomly allocated to two groups according to an open, prospective parallel-group design. In one group, dosage was adjusted to achieve serum AED concentration within a target range (10-20 p,g/ml for PHT, 1 5 4 0 pg/ml for PB, 4-11 pg/ml for CBZ, and 40-100 pg/ml for VPA), whereas in the other group, dosage was adjusted on clinical grounds. Patients were followed up for 24 months or until a change in therapeutic strategy was clinically indicated. Results: Baseline characteristics did not differ between the two groups. Most patients with partial epilepsy were treated with CBZ, whereas generalized epilepsies were most commonly managed with PB or VPA. PHT was used only in a small minority of patients. A total of I16 patients completed 2-year follow-up, and there were no differences in exit rate from any cause between the monitored group and the control group. The
proportion of assessable patients with mean serum drug levels outside the target range (mostly below range) during the first 6 months of the study was 8% in the monitored group compared with 25% in the control group (p < 0.01). There were no significant differences between the monitored group and the control group with respect to patients achieving 12-month remission (60% vs. 61 %), patients remaining seizure free since initiation of treatment (38% vs. 41%), and time to first seizure or 12-month remission. Frequency of adverse effects was almost identical in the two groups. Conclusions: Only a small minority of patients were treated with PHT, the drug for which serum concentration measurements are most likely to be useful. With the AEDs most commonly used in this study, early implementation of serum AED level monitoring did not improve overall therapeutic outcome, and the majority of patients could be satisfactorily treated by adjusting dose on clinical grounds. Monitoring the serum levels of these drugs in selected patients and in special situations is likely to he more rewarding than routine measurements in a large clinic population. Key Words: Therapeutic drug monitoring-Randomized controlled trial-Clinical trial-Antiepileptic drugs-Epilepsy.
In the early 1960s, the introduction of techniques for measuring serum levels of antiepileptic drugs (AEDs) led to a better understanding of the many factors that affect the pharmacokinetics of these drugs, and to the use of this knowledge for the design of rational dosing schemes (1,2). Likewise, identification of concentration ranges at which patients achieve optimal seizure control resulted in the widespread application of serum drug level measurements (therapeutic drug monitoring, TDM) as an aid to dosage adjustment (3). In a overview summarizing the contribution of serum
drug concentration monitoring to therapeutics, Spector et al. (4) identified three historic phases in TDM. The first phase came with the birth of the discipline and was characterized by overenthusiastic faith in its usefulness, without a clear definition of the limits and criteria that should guide the use of the service. The second phase, which evolved through the 1970s and the 1980s, resulted in the definition of these criteria, mostly on the basis of theoretic considerations. The third phase, which is still in its infancy, consists of the practical verification of these principles to provide scientific evidence for the costeffectiveness of TDM in specific target populations (5). In epilepsy, the wave of overenthusiasm for TDM subsided many years ago (6), partly because of the observation that there is considerable variation in the serum level at which individual patients achieve optimal re-
Accepted July 21, 1999. Address correspondence and reprint requests to Dr. E. Perucca at Clinical Pharmacology Unit, Department of Internal Medicine and Therapeutics, University of Pavia, Piazza Botta 10, 27100 Pavia, Italy.
222
EVALUATION OF THERAPEUTIC DRUG MONITORING sponses (7-1 1),and the realization that inadequate use of the service may result in unacceptable waste of resources (6,12,13) and even harm to patients’ health (14-16). Even though precise criteria for the optimal use of TDM have been formulated (8,9,11,17-20) and the usefulness of a correctly applied service has been suggested on the basis of retrospective surveys (21,22), there has been a paucity of studies in which the impact of serum AED determinations has been evaluated prospectively (11,23, 24). Moreover, most of the studies were limited to assessing whether TDM affected therapeutic decisions or the proportion of patients with serum drug levels within the optimal range, without investigating the actual impact on seizure control or the incidence of adverse effects (25-27). Unquestionably, the randomized controlled trial (RCT) represents the most valuable tool for an unbiased assessment of the impact of a therapeutic intervention ( 5 ) . Although this design has been successfully applied to demonstrate the cost-effectiveness of TDM in other therapeutic areas (28), it is surprising that no RCT has ever demonstrated a positive impact of TDM on clinical outcome in patients with epilepsy (4,s). This consideration provided the background for the present multicenter randomized study in which outcome was compared in two groups of patients with newly diagnosed epilepsy requiring institution of AED therapy. In one group, dosage was adjusted to obtain serum drug levels within the optimal range, whereas in the other group TDM data were not made available to the clinician, and treatment was individualized based solely on clinical response.
METHODS Patients A total of 180 patients was enrolled in the study. Inclusion criteria were (a) age between 6 and 65 years; (b) a diagnosis of untreated partial or idiopathic generalized epilepsy, based on clinical, electrophysiologic, and imaging investigations; (c) a history of at least two seizures in the previous 4 months (seizure clusters being counted as a single seizure); (d) clinical indication to initiate treatment with one of the following AEDs: carbamazepine (CBZ), phenytoin (PHT), valproate (VPA), phenobarbital (PB), or primidone (PRM); (e) ability to comply with the study procedures; and (f) informed consent from the patient and, when applicable, parents or tutors. Exclusion criteria were (a) a diagnosis of benign rolandic epilepsy, absence epilepsy, or epileptic encephalopathy; (b) presence of any known progressive disease, including dysmetabolic disorder or neoplasm; (c) pregnancy; (d) severe hepatic or renal insufficiency; (e) history of drug or alcohol abuse; or (f)treatment with any AED. Assuming a 50% seizure freedom rate in the control group, a sample size of 146 patients is adequate to detect a 25%
223
improvement in response rate (corresponding to a 75% seizure freedom rate) with a probability of 80% and a p value of 0.05 (29).
Experimental protocol The study, whose protocol was approved by Ethics Committees at participating centers (Appendix I), was carried out according to a multicenter, randomized, prospective parallel-group open (nonblind) design. All patients who met the eligibility criteria were randomized through a central office into two groups by using a stratification procedure aimed at ensuring a balanced distribution of idiopathic and symptomatic (or cryptogenic) epilepsies between groups. Physicians were free to choose the AED they considered indicated for that patient. In one group (TDM group), the dosage of the selected AED was adjusted based on serum drug level monitoring to achieve within a period 5 3 months steady-state concentrations within the target range. The target ranges used were 10-20 pg/ml (40-80 pM) for PHT, 15-40 pg/ml(64-172 pM) for PB, 4-1 1 ~ g / m l ( l 7 - 4 6p M ) for CBZ, and 40-100 pg/ml (280-700 phf) for VPA. For PRM-treated patients, only metabolically derived PB was used for TDM purposes. All blood samples were collected at steady state. For patients treated with PHT, PB, PRM, and controlled-release CBZ, samples had to be collected before the morning dose, 5 12 h (PHT, CBZ) or 5 1 5 h (PB, PRM) after the last administration. For VPA and immediate-release CBZ, two samples had to be obtained, one before the morning dose ( 5 1 2 h after the last administration) and one 3 h later, and the physicians were instructed to aim at serum drug concentrations within the target range in both samples. For patients taking VPA, food intake was delayed until the second sample was taken. If seizures persisted despite serum levels in the lower part of the target range, the protocol required that physicians adjust the dosage further to produce AED levels in the upper part of the range. Levels below the target range were allowed only if the patient was unable to tolerate higher concentrations. Levels above target were allowed at the discretion of the treating physician only when there were no significant side effects and seizures persisted at target concentrations. At all study sites, physicians had to be able to use TDM results within 7 days of sampling. In the second group (control group), blood samples for the determination of AED levels were collected in a similar way, but TDM results were not made available to the treating physician. In this group, dosage was adjusted on purely clinical grounds, aiming at achieving optimal seizure control over the shortest reasonable period. If no satisfactory response was achieved after 6-1 2 months and the physician thought that continuation of treatment without knowledge of serum drug concentration was no longer ethically acceptable, the patient could be crossed Epilqxia, Vul. 41, Nu. 2, 2000
G. JANNUZZI ET AL.
224
over to the TDM group and managed subsequently according to the protocol described for that group. Except for the feedback (or lack of feedback) derived from TDM data, precise modalities of dosage adjustments within each group were left to the clinicians' judgment, and patients were followed up as in routine clinical care. Duration of follow-up was 2 years unless exit criteria were met. The latter were (a) need to switch the patient to another drug, (b) need to add a second drug, or (c) for the control group only, unsatisfactory response after 6-12 months, requiring evaluation of TDM data collected up to that time (in the latter case, patients were still followed up for 5 2 years according to the procedures outlined for the TDM group). Seizures were recorded daily on appropriate cards by the patients or their guardians. All patients were seen in the clinic approximately every month during the first 3 months, every 3 months in the subsequent 9 months, and at least twice during the second year of follow-up. Apart from serum AED level monitoring, evaluations included a medical examination and generic questioning for possible side effects. Other investigations were carried out if clinically indicated. All data were recorded in especially designed case report forms.
Determination of serum AED levels Serum AED levels were determined in the laboratory of individual participating centers by using commercial
fluorescence polarization (FPIA) or enzyme multiplied immunoassay (EMIT).
End points and statistical analysis The proportion of patients achieving complete seizure remission during the last 12 months of follow-up was considered the primary efficacy end point. Other efficacy measures included proportion of patients remaining seizure free since initiation of treatment, time to first seizure, and time to reach 12-month seizure remission. Tolerability was assessed by determining the proportion of patients with side effects at any time during the study, and proportion of patients with specific side effects. Retention of patients in the study and retention in a seizurefree state were evaluated by Kaplan-Meier survival curves and Logrank test for statistical comparison. Between-group comparisons for baseline parameters and for efficacy and tolerability end points were made by using the x2 test for frequency data and Student's t test for comparison of means. A p value 4 . 0 5 was considered statistically significant. RESULTS Characteristics of the patients Details of the patients randomized to the groups are summarized in Table 1. The two groups were comparable with respect to age distribution and baseline seizure
TABLE 1. Characteristics of the patients included in the trial TDM (n = 93) Male gender (%) Age (yr, means f SD) Median number of seizures (range) in the 4 mo before treatment Number of patients with partial epilepsy" Number of patients with generalized epilepsy" Seizure type",b Simple partial (+. secondary generalization) Complex partial (f secondary generalization) Secondarily generalized tonic-clonic Primarily generalized tonic-clonic Absence Myoclonic Atonic Clonic Not specified Allocation to treatment Partial epilepsy (%) Carbamazepine Phenobarbital Phenytoin Sodium valproate Generalized epilepsy (%) Carbamazepine Phenobarbital Phenytoin Sodium valproate Numbers in brackets refer to percentage of patients.
' Some patients had more than one seizure type. Epilepsia, Vol. 41, NO. 2, 2000
Controls 87)
(11 =
50 27+. 17 3 (1-240) 57 (61) 36 (39)
55 2 9 + 18 3 (1-360) 58 (67) 29 (33)
17 (18) 35 (38) 29 (3 1) 30 (32) 3 (3) 6 (6) 2 (2) 2 (2) 0
25 (29) 40 (46) 37 (42) 26 (30) 4 (5) 5 (6) 1(1) 0
50 2 1 8
47 6 6 8
6 16 1 15
8 16 0 9
1(1)
225
EVALUATION OF THERAPEUTIC DRUG MONITORING frequency. Compared with controls, the TDM group showed a slightly higher proportion of women (50 vs. 45%) and patients with generalized epilepsy (39 vs. 33%), but the difference was not significant. In both groups, the vast majority of patients with partial epilepsy was allocated to treatment with CBZ, whereas PB and VPA were the most commonly used drugs in patients with generalized epilepsy.
mean serum drug levels outside the target range (mostly below range) during the first 6 months of the study was 8% in the TDM group compared with 25% in the control group (p < 0.01). After the last dosage adjustment at follow-up, serum drug levels outside the target range were found in 6% of TDM patients compared with 22% of controls (p < 0.01). Of the 10 patients in the TDM group who continued to have seizures during the last 12 months of follow-up, only one had serum drug levels below target range. Conversely, of 1 I uncontrolled patients in the control group, four had levels below target.
Retention in the study Overall, 116 (64%) patients completed 2 years of follow-up. Proportion of completers was almost identical in the TDM and the control groups (62 vs. 67%, respectively), with retention curves for the two groups being virtually superimposable (Fig. 1). Of the 64 patients who exited the study prematurely, 43 (67%) were lost to follow-up, 10 (16%) exited because of inefficacy and/or adverse events, and 11 (17%) exited for other reasons. Proportion of patients exiting the study for any specific cause was comparable in the two groups. Of the patients randomized to the control group, only one was crossed over to the alternative group after 1 month.
Seizure control The proportion of patients achieving 12-month remission was 60% in the TDM group and 61% in the control group. More patients with generalized epilepsy achieved 12-month remission compared with patients with partial epilepsy. However, neither proportion of patients achieving remission nor rates at which remission was achieved differed between the TDM and the control groups (Fig. 3). Overall, 56% of patients in the TDM group were seizure free during the last 12 months of follow-up compared with 58% of controls (NS). Corresponding figures were 61% and 62%, respectively, for the subgroup of patients with generalized epilepsy and 53% and 55%, respectively, for those with partial epilepsy. None of these differences was statistically significant. Overall, 38% of patients in the TDM group remained seizure free since initiation of treatment, compared with 41 % of those in the control group (NS). Median time to first seizure in the TDM group (51 days, 25-75% percentile; 14-151 days) tended to be longer than in the control group (3 1 days, 25-75% percentile, 15-90 days), but the difference was not statistically significant. Moreover, there were no significant differences in time to occurrence of a first seizure as estimated from curves showing proportion of patients remaining seizure free as a function of time (Fig. 4).
Serum drug levels Mean serum concentrations of CBZ, VPA, and PB at different time periods during the trial are shown in Fig. 2. At all time points, mean values were within target range for both groups (except for a borderline low mean PB concentration of 14.5 kg/ml at 6 months in the control group), and there were no significant differences between groups. As shown in Table 2, an appreciable proportion of patients with serum drug levels outside target range was found only with PB, and serum PB levels below target were twice as common in the control group as in the TDM group. Of the seven patients given PHT (not included in Table 2), one of two in the TDM group and three of five in the control group had levels below target. Overall, the proportion of assessable patients with
FIG. 1. Kaplan-Meier curves showing retention of patients in the trial as a function of time. Comparison of survival curves showed no statistically significant differences (Logrank test, x2,0.270, p = 0.60). Hazard ratio was 1.05 (95% confidence limits, 0.701.83).
,?
-lDM
.G 50-
_----- Controls
E
a
25-
o
! 0
, 50
,
,
,
,
,
,
,
,
,
,
,
,
,
100 150 200 250 300 350 400 450 500 550 600 650 700
Days Epilepsiu, Vol. 41, No. 2, 2000
G. JANNUZZI ET AL.
226
Carbamaaepine 15
F 4 10
n39
T T
c
0TDM
Controls
0
. 7
F 3
c
s
5
0
0-6
7-12
13-18
19-24
Months
Phenobarbital n=7
T
n=l1
0-6
n=ll T
n41
T
Li’”
0 TDM Controls
FIG. 2. Mean serum concentrations of carbamazepine, phenobarbital, and valproate (&D) at different time periods during the trial. Values recorded during the indicated period were averaged for each patient, and the resulting value was then used to calculate means for each group. There were no statistically significant differences between any of the pairs.
19-24
7-12
Months
Valproic Acid 100 h
E
9 v
0 TDM
75
Controls
c
# 3
6
50
25
0 0
0-6
7-12
13-18
19-24
Months Adverse effects Adverse effects were reported in 45 (48%)of patients in the TDM group and in 41 (47%) of those in the control group. None of these effects was considered serious. SomEpilepsia, Vol. 41, No. 2, 2000
nolence or sedation was the most commonly reported adverse experience, being observed in one fourth of patients in both groups. The frequency of the most common adverse effects was similar in the two groups (Table 3).
227
EVALUATION OF THERAPEUTIC DRUG MONITORING TABLE 2. Proportion of patients with mean serum drug concentrations below or above target range during the first 6 months and after the last dosage adjustment
First 6 mo Carbamazepine Valproate Phenobarbital After last dose adjustment Carbamazepine Valproate Phenobarbital
% with levels below target
% with levels above target
TDM
TDM
Controls
Controls
2(n O(n 31 (n
= = =
47) 21) 16)
5 (n = 39) 0 (n = 9) 64(n = 14)
0 (n 0 (n O(n
47) 21) 16)
8 (n = 39) O(n = 9) O(n = 14)
O(n O(n 27 (n
= = =
41) 20) 15)
5 (n O(n 53 (n
0 (n = 41) 0 (n = 20) 0 ( n = 15)
5 (n = 38) O ( n = 11) O(n = 15)
DISCUSSION As pointed out recently (1l), “despite a good deal of anecdotal testimony, surprisingly little has been published demonstrating the benefits of anticonvulsant therapeutic drug monitoring in epileptic populations.” In one of the most commonly quoted studies, Lund (3) followed up prospectively for 3 years a group of 32 patients with partial or generalized tonic-clonic seizures uncontrolled by long-term PHT treatment. When PHT dosage was adjusted to achieve serum drug levels of 15.0 k 2.5 pg/ml, 14 patients became seizure free, and the remaining 18 also showed an important reduction in seizure frequency. Although these data are often considered to document the value of measuring serum drug levels, the study did not include a control group, and therefore it cannot be excluded that similar benefits could have been achieved by adjusting dosage on purely clinical grounds. Moreover, the population was a preselected group of previously refractory patients, and the optimal serum drug concentration for this group may not be applicable to patients with less severe seizure disorders. The importance of this point is highlighted by evidence that the serum concentration of AEDs that is required to suppress seizures varies in relation to both seizure type and pretreatment seizure frequency (30). Before our investigation, only one other study used a randomized controlled design for the assessment of the impact of TDM on epilepsy treatment (23). In that study, 126 patients with refractory epilepsy were randomly allocated to have their serum AED levels monitored or to be treated without knowledge of drug concentrations. Among the 105 patients who completed the required 1 year of follow-up, seizure control improved to a similar extent in both groups. Although the study failed to produce evidence of improved outcome in the TDM group, the proportion of patients with serum AED levels outside the optimal range (about one third) did not change substantially during the monitoring process, suggesting that physicians did not use correctly the information provided by the laboratory. The authors also discussed the possibility that their negative findings could be related to selection of a population of patients refractory to treatment
= = =
38) 11) 15)
= = =
anyway. The importance of patients selection was highlighted more recently by a retrospective analysis of 25 years of experience with AED monitoring in a large Australian center (31). Although overall there was no evidence of improved outcome after application of TDM, outcomes were significantly better when TDM was used within 6 months of the onset of the patient’s seizure 1
1.00 I
s 0.75‘2 ._
E .-c
u)
c
Partial .-....Controls
0.50-
t 0)
2m = a
S 0.25-
0.00
I
I
I
I
I
I
, I
1 .oo
-TDM
Generalized
5 $ ._
..-___ Controls
0.75-
E
c ._ v)
0.50-
h S 0.25-
I
0.00 0
I
I
I
100
200
300
I
I
400
500
I
600
700
Days
FIG. 3. Actuarial percentages of patients reaching 12-month seizure remission during follow-up. Comparison of data showed no statistically significant differences (Logrank test, x2,0.1 15, p = 0.74). Hazard ratio was 0.96 (95% confidence limits, 0.50-1.63).
Epilepsia, Vol. 41, No. 2, 2000
G. JANNUZZI ET AL.
228
_________ ____ _________-____ _____.__ --.. ____..__ ___-_------_7_____.----.~
~
~
~
__.-
I
.-C
; . 0.50.-m
E
'
-TDM
K
------ Controls 0.25-
0.00-
m
.~ 0.75-
I
1
1
___-_ ____
?L--_
I
1
.____
7
a
1
1
I
I
I
I
I
1
I
1
1
I
s
.-t
.g.-
-TDM ------ Controls
0)
0.50-
2
2
'
0.25-
0.00
I
I
I
I
I
I
I
I
I
I
I
1
Days
disorder. Indeed, initiation of therapy was one of the situations listed by Eadie (1 1) among the indications for adjusting dosage based on serum AED determinations. In light of the information discussed, our study was especially designed to determine whether dosage individualization based on TDM can improve seizure control rates or decrease the risk of toxicity in a population with newly diagnosed epilepsy. Although recruitment criteria were kept sufficiently broad to ensure representative data, some of the factors that might have confounded the results of previous studies were controlled through exclusion of conditions associated with a particularly benign history (rolandic and absence epilepsies) or, conversely, poor outcome (progressive neurologic disorders, epileptic encephalopathies). The number of patients evaluated was also sufficiently large to exclude the possibility that potentially important differences would be Epilepsia, Vol. 41, No. 2, 2000
missed because of inadequate statistical power. Finally, a 24-month duration of follow-up was appropriate to determine response to treatment over a clinically meaningful time scale. In spite of all these precautions, the study did not reveal major differences in any outcome measure between the monitored and the nonmonitored groups. Therefore under the conditions of the study, the hypothTABLE 3. Side effects reported in 2.5%qfpatients in any group TDM = 93)
Controls (n = 87)
23 (25%) 6 (6%) 7 (8%) 7 (8%) 4 (4%)
22 (25%) 4 (5%) 4 (5%) 5 (6%) 4 (5%)
(n Somnolence/sedation Dizziness Headache Irritability Fatigue
EVALUATION OF THERAPEUTIC DRUG MONITORING esis that TDM may facilitate achievement of seizure control or reduce adverse effects could not be confirmed. Several explanations may have contributed to the apparent lack of usefulness of TDM in this study. First, although there is general agreement that serum drug level monitoring is especially valuable with PHT (due to its narrow therapeutic range and poorly predictable nonlinear kinetics) (8,9,19), most of our monitored patients were started on treatment with CBZ and VPA. Although the latter drugs reflect the preference of most European physicians for the management of partial and generalized epilepsies, respectively (32), the relation between their serum concentration and clinical response is rather variable, and their dosage can easily be individualized on purely clinical grounds. Another explanation for our failure to demonstrate a significant impact for TDM is that, with the passing of time, clinicians have absorbed many of the lessons to be learned from drug monitoring, and now they use effective dosing schemes without monitoring drug concentrations as frequently as in the past. The findings that most patients in whom dosage was adjusted without feedback from TDM achieved serum drug concentrations within optimal range is consistent with this interpretation. Similar conclusions were recently reached by Eadie (1 1 ), who suggested, “the effect of drug concentration monitoring on the cost effectiveness of anticonvulsant therapy is probably not as significant now as it originally was.” The fact that the majority of patients in both groups became seizure free with treatment is in line with results of other prospective studies in comparable populations (33). Although it could be argued that TDM may be of special value in the subgroup of patients who are more difficult to treat (such as those receiving polytherapy, with the attendant risk of drug interactions), the fact that TDM-assisted dosage individualization did not improve outcome in a previously untreated population managed on rnonotherapy still represents an interesting finding. Overall, a review of available evidence, including the results of our study, does suggest that application of TDM to selected patients is likely to be more rewarding than routine measurements in a large clinic population (19). In our own study, for example, TDM might have been of special value in the four uncontrolled patients who were later found to have serum drug levels below target range in the control group. Other conditions in which TDM may be especially rewarding include the check for compliance and the management of patients in whom AED pharmacokinetics is likely to be altered by developmental factors, associated disease, or drug interactions. The setting of precise indications for TDM is important in view of the evidence that serum drug levels are often requested inappropriately (6,12,13,18). Misinterpretation of the results (for example, increasing dos-
229
age despite evidence that a patient is well controlled at a level below the usually quoted optimal range) also seems to be a common problem leading to potentially harmful clinical decisions (14-16). Knowledge on how to use correctly a TDM service is clearly more important than the availability of the service itself, as clearly shown by a randomized controlled study in which implementation of an educational and auditing program for serum AED level monitoring resulted in a significant decrease in the number of patients requiring readmission to hospital (24). In conclusion, we have shown that the majority of patients with newly diagnosed epilepsy receiving monotherapy can be optimally treated without need for monitoring serum AED drug levels. Because PHT was used in only a small minority of cases, these results may not be applicable to the subgroup treated with this drug, for which TDM is likely to be most useful due to its dosedependent kinetics. Furthermore, our results should not be interpreted as a demonstration that TDM has no value in the management of epilepsy, but simply as evidence that widespread application of TDM to large, unselected populations is probably not cost-effective. In our view, the same consideration should apply to the many studies that have failed to demonstrate a clear value of measuring serum levels of newer AEDs. By contrast, everyday clinical practice is often confronted with situations (for example, when there is reason to suspect noncompliance or unusual kinetics) in which determination of concentrations of older and newer AEDs will undoubtedly help in the process of making rational therapeutic decisions. Because withdrawing the service in such situations may be ethically unacceptable, demonstration of the usefulness of TDM in these selected cases may not be possible in randomized controlled trials.
Appendix 1 Study investigators (number of patients recruited by each center is shown in parentheses): G. Aimo, R. Cremo, M. De Mattei (Torino, 12); M.G. Alessandri, A. Battaglia, R. Guerrini, 0. Salvadori (Calambrone, five); M.D. Benedetti, G. Benoni, L.G. Bongiovanni, 0. Cappellaro (Verona, 27); P. Benedetti, M. Brinciotti, F. Giannotti, M. Matricardi, S. Morano (Roma, nine); P. Benetello (Padova, one); I. Bonacchi, A. Hakani, A. Romeo, F. Viani, M. Viri (Milano, 10); G. Bertani, G. Castellini, G. Gobbi, A. Pini (Reggio Emilia, 12); E. Boati, A. De Fanti, A. Franza, L. Manenti, P. Pinto (Bergamo, five); F. Bordone E. Bottacchi, L. Carenini, L. Sironi (Aosta, four); A. Borgheresi, T. Valenza, G. Zaccara (Firenze, four); D. Cassano, S. Gentile, I. Sacerdote, F. Zola (Torino, six); P. Conti, G. De Luca, A. Pansini (Firenze, three); G. Coppola, R. Romano, R. Scarpello, G. Trianni (Casarano, 11); G.C. Fabrini, R. Galli, C. Gneri, R. Massetani, L. Murri, C. Pizzanelli (Pisa, 16); Epilepsia, V d 41, No. 2, 2000
230
G. JANNUZZI ET AL.
N.P. Falcone, A.M. Lanzetti, G.B. Lomonaco, A. Nappo (Viterbo, eight); R. Fenoil, D. Leotta, U. Morino, M. Nobili, L. Vivalda (Torino, seven); C.A. Galimberti, R. Manni, G.V. Melzi d’Eril, I. Sartori, A. Tartara (Pavia, four); G. Grampa, V. Malacrida, D. Porazzi, P. Secchi (Busto Arsizio, six); M. Gianelli, P. Naldi, R. Mutani (Novara, 10); S. Mazza, E. Moro, M.L. Vaccario (Roma, four); F. Monaco, G.P. Sechi (Sassari, five); R. Rocchi, M. Ulivelli, G.P. Vatti (Siena, 11). Acknowledgment: This study was supported in part by a grant from the European Commission (Contract BMHl -CT941622)
I
REFERENCES 1. Buchtal S, Svensmark 0, Schiller PJ. Clinical and electroencephalographic correlations with serum levels of diphenylhydantoin. Arch Neurol 1960;2:624-30. 2. Kutt H, McDowell F. Management of epilepsy with diphenylhydantoin sodium. JAMA 1968;203:969-72. 3 . Lund L. Anticonvulsant effect of diphenylhydantoin relative to plasma level: a prospective three-year study in ambulant patients with generalized epileptic seizures. Arch Neurol 1974;31:289-94. 4. Spector R, Park GD, Johnson GF, Vesell ES. Therapeutic drug monitoring. Clin Pharmacol Ther 1988;43:345-53. 5. Mclnness GT. The value of therapeutic drug monitoring for the practising physician: an hypothesis in need of testing. Br J Clin Pharmacol 1989;27:28 1 4 . 6. Chadwick DW. Overuse of monitoring of blood concentrations of antiepileptic drugs. Br Med J 1987;294:7234. 7. Perucca E, Grimaldi R, Crema A. Interpretation of drug levels in acute and chronic disease states. Clin Pharmacokinet 1985:lO: 498-5 13. 8. Perucca E, Richens A. Antiepileptic drugs: clinical aspects. In: Richens A, Marks V, eds. Therapeutic drug monitoring. Edinburgh: Churchill-Livingstone, 1983:32048. 9. Richens A, Perucca E. Clinical pharmacology and medical treatRichens A, Chadwick D, eds. A textbook of Churchill-Livingstone, 1993:495-559. 10. Lammers MW, Hekster YA, Keyser A, et al. Neither dosage nor serum levels of antiepileptic drugs are predictive for efficacy and adverse effects. Pharmacol World Sci 1995;17:201-6. 11. Eadie MJ. The role of therapeutic drug monitoring in improving the cost-effectiveness of anticonvulsant therapy. Clin Pharmucokinet 1995;29:29-35. 12. Bussey HI, Hoffman EW. A prospective evaluation of therapeutic drug monitoring. Ther Drug Monit 1983;5:245-8. 13. Hallworth MJ. Audit of therapeutic drug monitoring. Ann Clin Biochem 1988;25:121-8. 14. Beardsley RS, Freeman JM, Appel FA. Anticonvulsant serum lev-
Epilepsia, Vol. 41, No. 2, 2000
els are useful only if the physician appropriately uses them: an assessment of the impact of providing serum level data to physicians. Epilepsia 1983;24:330-5. 15. Jackson M, Dawson P, McCrea W. The hazards of prescribing from serum level. Seizure 1994:3:225-33. 16. Woo E, Chan YM, Yu YL, Chan YW, Huang CY. If a well stabilized epileptic patient has a subtherapeutic antiepileptic drug level, should the dose be increased? A randomized prospective study. Epilepsia 1988;29:129-39. 17. Gross A. Best practice in therapeutic drug monitoring. Br J Clin Pharmacol 1998;46:95-9. 18. Schoenenberger RA, Tanasijevic MJ, Jha A, Bates DW. Appropriateness of antiepileptic drug level monitoring. JAMA 1995;274: 1622-6. 19. Theodore WH. Rational use of antiepileptic drug levels. Pharmucol Ther I992;54297-305. 20. Mattson RH. Antiepileptic drug monitoring: a re-appraisal. Epilepsia 1995;36(suppl 5):22-9. 2 I . Beardsley RS, Freeman JM, Appel FA. Anticonvulsant serum levels are useful only if the physician appropriately uses them: an assessment of the impact of providing serum level data to physicians. Epilepsia 1983:24:330-5. 22. McKee PJW, Percy-Robb I, Brodie MJ. Therapeutic drug monitoring improves seizure control and reduces anticonvulsant side effects in patients with refractory epilepsy. Seizure 1992;1:275-9. 23. Froscher W, Eichelbaum M, Gugler R, Hildebrandt G, Penin H. A prospective randomised trial on the effect of monitoring plasma anticonvulsant levels in epilepsy. J Neurol 1981;224:193-201. 24. Wing DS, Duff HJ. The impact of a therapeutic drug monitoring program for phenytoin. Ther Drug Monit 1989;11:32-7. 25. Houtman PN, Hall SK, Green A, Rylance GW. Rapid anticonvulsant monitoring in an epilepsy clinic. Arch Dis Child 1990;65: 264-8. 26. Larkin JG, Herrick AL, McGuire GM, Percy-Robb IW, Brodie MJ. Antiepileptic drug monitoring at the epilepsy clinic: a prospective evaluation. Epilepsia 1991;32:89-95. 27. McKee PJW, Larkin JG, Brodie AF, Percy-Robb IW, Brodie MJ. Five years of anticonvulsant monitoring on site at the epilepsy clinic. Ther Drug Monit 1993;15:83-90. 28. Crist KD, Nahata MC, Ety J. Positive impact of a therapeutic drug monitoring program on total aminoglycoside dose and cost of hospitalization. Ther Drug Morzit 1987;9:306-10. 29. Fleiss J . Statisticul methods jbr rates and proportions. New York: John Wiley and Sons, 1973. 30. Schmidt D, Einicke I, Haenel F. The influence of seizure type on the efficacy of plasma concentrations of phenytoin, phenobarbital and carbamazepine. Arch Neurol 1986:43:263-5. 31. Eadie MJ. Plasma antiepileptic drug monitoring in a neurological practice: a 25-year experience. Ther Drug Monit 1994;16:458-68. 32. Baldy-Moulinier M, Covanis A, D’Urso S, et al. Therapeutic strategies against epilepsy in the Mediterranean countries: a report from an international collaborative survey. Seizure 1998;7:513-20. 33. Beghi E, Perucca E. The management of epilepsy in the 1990s: acquisitions, uncertainties and perspectives for future research. Drugs 1995;49:680-94.