Seizure 2000; 9: 464–468

doi: 10.1053/seiz.2000.0442, available online at http://www.idealibrary.com on

Epilepsy after the first drug fails: substitution or

add-on?

PATRICK KWAN & MARTIN J. BRODIE

Epilepsy Unit, University Department of Medicine and Therapeutics, Western Infirmary, Glasgow,
Scotland, UK

Correspondence to: Professor M. J. Brodie, Epilepsy Unit, Department of Medicine and Therapeutics, Western
Infirmary, Glasgow G11 6NT, Scotland, UK. E-mail: Martin.J.Brodie@clinmed.gla.ac.uk

When and how a combination of antiepileptic drugs (AEDs) should be used in patients unresponsive to monotherapy is not
known. We followed up prospectively 248 patients in whom treatment with the first AED was unsuccessful. When treatment
failed due to intolerable adverse events, a second (substituted) drug was prescribed. When failure was due to lack of efficacy,
either AED substitution or combination (add-on) was undertaken. Patients were considered to be seizure-free if they had no
seizures for at least 1 year. Among patients with inadequate seizure control on the first well tolerated AED, those who received
substituted monotherapy (n = 35) and those who received add-on treatment (n = 42) had similar seizure-free rates (substitution
vs. add-on: 17% vs. 26%) and incidence of intolerable side effects (substitution vs. add-on: 26% vs. 12%). Based on the drugs’
perceived primary mode of action, more patients became seizure-free when the combination involved a sodium channel blocker
and a drug with multiple mechanisms of action (36%) compared to other combinations (7%, P = 0.05). None of the 11 patients
who received add-on treatment after a second drug had also failed became seizure-free, compared to 26% in those who received
add-on as soon as the first tolerated AED proved to be ineffective (n = 42, P = 0.05). These preliminary observations have
generated verifiable hypotheses regarding the early management of epilepsy. A randomized study comparing substitution and
combination after the failure of the first AED is underway.
c 2000 BEA Trading Ltd

Key words: epilepsy; antiepileptic drugs; outcome; monotherapy; polytherapy; mechanism of action.

INTRODUCTION series of studies in the late 1970s and early 1980s12–16 ,

Up to one third of patients continue to have seizures
despite antiepileptic drug (AED) treatment1, 2 . Uncon-
trolled epilepsy is associated with excess mortality3 ,
cognitive and behavioural dysfunction4 , and social and
educational disadvantage5 for which society pays a
high price6 . Long-term outcome studies2 and random-
ized trials7–10 suggest that fewer than 50% of patients
will become seizure-free on the first AED. An alterna-
tive drug (substitution) is unavoidable when the patient
develops intolerable adverse events, but when seizures
persist despite a sufficient dose of the first AED, it is
unclear whether or not substitution should be tried be-
fore combining drugs11 .
This controversy arises largely because the ‘natural
history’ of newly diagnosed epilepsy in response to
treatment is not well understood. Since the advantages
of single- over multiple-drug treatment, particularly
in terms of side effect profile, were demonstrated by a
monotherapy has rightly become the established prin-
ciple for managing new onset epilepsy. However, the
explosion of new AEDs in the past decade with their
mechanistic diversity17, 18 and generally better tolera-
bility19 has raised the possibility of effective and safe
combinations for patients unresponsive to monother-
apy20 . We examined the effectiveness of substitution
and add-on therapy after treatment with the first AED
failed in a prospective cohort of newly diagnosed
patients followed up since January 1984 at a single
centre.
MATERIALS AND METHODS
This study included unselected patients in whom
epilepsy was diagnosed and treatment initiated at the
Epilepsy Unit in the Western Infirmary in Glasgow,
Scotland between 1 January 1984 and 31 December

1059–1311/00/070464 + 05 $35.00/0 c 2000 BEA Trading Ltd

Epilepsy after the first drug fails
470 190
100
18%
29%
80
% patients
60 24% 47%
40
47%
20 35%
0
1 2
Number of different antiepileptic drug regimens
Fig. 1: Response to successive antiepileptic drug regimens.
1997. No patients had received AED therapy prior
to enrolment. Patients in whom treatment with the
first AED was unsuccessful entered the study. The
choice of initial treatment took into account the type
of seizures and epilepsy syndrome and other clini-
cal characteristics19, 21 . After commencement of med-
ication, patients were reviewed at the epilepsy clinic
every 4–6 weeks for the first 6 months and at least
every 4 months thereafter. A dedicated telephone line
was set up for the patients or their primary care
physicians to contact the Epilepsy Unit should prob-
lems arise between scheduled clinic appointments.
Compliance was monitored as necessary at the clinic
with the aid of on-site measurement of serum AED
concentration22 . Dosages were titrated according to
recommended schedules23, 24 and adjusted during
follow-up as clinical circumstances dictated paying
particular attention to efficacy and tolerability. Each
patient’s clinical information and response to treat-
ment were recorded in a prospective database25 .
Response to medication was classified as seizure-
free (absence of any type of seizures or auras for at
least 1 year); failure of treatment due to inadequate
seizure control despite ability to tolerate the med-
ication (lack of efficacy) or due to adverse events
(including intolerable side effects and idiosyncratic re-
actions); or withdrawal of treatment for reasons unre-
lated to efficacy or tolerability such as concern about
potential adverse events, planning a pregnancy, and a
change of mind. A change of drug regimen was de-
fined as either substitution by alternative monotherapy
or addition of another drug. Patients who developed
idiosyncratic reactions or intolerable side effects were
treated with an alternative drug (substitution). When
seizure control was inadequate, either substitution or
combination (add-on) was prescribed.
465
65 25
Adverse events
21%
36%
Inadequate
seizure control
Seizure-free
62%
64%
17%
3 4
Patients in whom the first AED was replaced by
an alternative were analysed according to the reason
for substitution. In order to address the clinical ques-
tion of appropriate treatment strategy for patients who
experienced persistent seizures despite being able to
tolerable the medication, we compared the response
to substitution and add-on therapy in those with in-
adequate seizure control on the first well tolerated
AED. AEDs were classified according to their pu-
tative primary mode of action17, 18 . Drugs that pri-
marily act on voltage-gated sodium channels include
carbamazepine, phenytoin and lamotrigine. Vigaba-
trin and tiagabine enhance the inhibitory function
of γ -aminobutyric acid (GABA). Sodium valproate,
gabapentin and topiramate are classified as having
multiple mechanisms of action. The chi-squared test
and Fisher’s exact test were used for comparisons of
categorical data. All statistical tests were two-tailed.
Calculations were made using Minitab for Windows
(Release 11.21) software.
RESULTS
Among 470 previously untreated newly diagnosed
patients referred to the clinic between 1 January 1984
and 31 December 1997, treatment with the first AED
was unsuccessful in 248 patients (53%), who consti-
tuted the present study cohort. One hundred and thir-
teen patients had inadequate seizure control on the first
drug, 98 withdrew treatment due to adverse events (69,
intolerable side effects; 29, idiosyncratic reactions),
and 37 due to other reasons2 . Fifty-two percent were
male. The median follow-up period was 5 years (range
2–16 years). The median age at referral was 31 years
(range 9–89 years), and the median age at onset of
466
100
19%
29%
80
% patients
60
49%
55%
40
20
32%
16%
0
Inadequate Intolerable
control side effects
n = 31 n = 69
Fig. 2: Response to the second antiepileptic drug according to reason for failure of the first drug.
epilepsy was 28 years (range 1–87 years).
The percentage of successful treatment declined in
each successive drug regimen (monotherapy or combi-
nation), while that of patients with inadequate seizure
control rose progressively (Fig. 1). Among the 248
patients whose epilepsy was not controlled on the
first AED, 166 received a substituted drug, 61 (37%)
of whom became seizure-free on this second choice.
Among the 113 patients with inadequate seizure con-
trol on the first AED, 58 opted to continue on the
same drug, 24 received add-on treatment, and 31 were
treated with a substituted drug, only five (16%) of
whom became seizure-free (Fig. 2, P < 0.01). Eigh-
teen patients were not able to tolerate trials of two dif-
ferent AEDs, five patients three AEDs, and one was
not able to tolerate even the fourth choice.
When a tolerable AED was eventually identified, it
was still ineffective in 56 patients, 18 of whom then
received add-on treatment and four substitution, while
34 opted to continue on the same drug. Thus, among
all patients with inadequate seizure control on the
first tolerated AED, 42 (24 after the first drug and 18
after subsequent drugs lacked efficacy) received add-
on therapy and 35 (31 after the first drug and four
after subsequent drugs) received substitution. There
was no significant difference in seizure-free rates (add-
on: 11 out of 42, 26%; substitution: 6 out of 35, 17%)
or incidence of adverse events necessitating with-
drawal (add-on: 5 out of 42, 12%; substitution: 9 out
of 35, 26%) between the two treatments (P = 0.25)
(Fig. 3). Eleven patients received add-on therapy at
a later stage after the substituted drug failed to con-
trol the seizures. None of them became seizure-free,
compared to a seizure-free rate of 26% among those
who received add-on as soon as the first tolerated AED
failed (n = 42; P = 0.05).
P. Kwan & M. J. Brodie
14% Intolerable
17% adverse events
24%
Inadequate
seizure control
45% Seizure-free
P < 0.01
62%
38%
Idiosyncratic Other
reactions withdrawals
n = 29 n = 37
A variety of AED combinations were employed
(Table 1). Based on the drugs’ perceived primary
mechanisms of action, more patients became seizure-
free (Fig. 4) when the combination involved a sodium
channel blocker and an AED with multiple mecha-
nisms of action (36%) compared to other combina-
tions (7%, P < 0.05). None of the patients who re-
ceived a combination of a sodium channel blocker and
a ‘pure’ GABA-ergic agent (vigabatrin or tiagabine)
became seizure-free.
DISCUSSION
The argument against add-on therapy traditionally has
been its propensity to cause greater toxicity without
substantial improvement in outcome11 . However, ad-
verse events due to pharmacokinetic and pharmaco-
dynamic interactions between AEDs can be equally
problematic during the substitution phase20, 26 . Only
12% of our patients given combination therapy had
to withdraw treatment due to side effects, which was
a lower proportion, although not statistically signifi-
cantly so, than those who changed to a second drug
(26%). Ninety percent of the combinations employed
the newer drugs, some of which are better tolerated
than their older counterparts27, 28 .
Although no significant difference in efficacy
was observed between alternative monotherapy and
duotherapy in our cohort, synergistic (supra-additive)
effects have been demonstrated for specific combi-
nations in comparative studies, notably sodium val-
proate and lamotrigine20, 29 . This discrepancy is likely
to be due to the inclusion of a relatively small number
of patients taking a large number of different com-
binations in the present study. Indeed, when the var-
Epilepsy after the first drug fails 467

40
*
100 * P < 0.05
Intolerable
12% adverse events
26% 30
Inadequate
80 seizure control

Seizure-free

% patients
60
% patients

62%
20
57%
40

20 10
26%
17%
0
Add-on Substitution
n = 42 n = 35 0
Sodium blocker + Other
multiple actions combinations
Fig. 3: Response to add-on or substitution in patients with in-
adequate seizure control on the first well tolerated antiepilep- n = 28 n = 14
tic drug.
Fig. 4: Response to different combinations of antiepileptic
drugs according to mechanisms of action.

Table 1: Combinations of antiepileptic drugs prescribed to Combination therapy was more effective when pre-
patients receiving add-on therapy with inadequate seizure
control on first well tolerated drug.
scribed immediately after the first drug failed due to
lack of efficacy than when it was delayed until treat-
No. of ment with a substitution also proved unsuccessful.
Combinations patients
This difference in efficacy might reflect the severity
Sodium channel blocker + multiple actions 28 of the underlying disease, i.e. patients unresponsive
LTG + VPA 12 to two successive AEDs might have more ‘drug re-
CBZ + GBP 6
CBZ + TPM 3 sistant’ epilepsy than those uncontrolled on just one
CBZ + VPA 2 drug. Whether pharmacoresistance is present de novo
LTG + TPM 3 or evolves over time is debatable2 . The concept of
LTG + GBP 1 seizures begetting seizures was first hypothesized by
PHT + VPA 1
Gowers30 in the last century and is supported by the
Two sodium channel blockers 5 experimental model of kindling, whereby electrical
CBZ + LTG 3
CBZ + PHT 1 stimulation at what is initially a subconvulsive level in
PHT + OXC 1 an animal subsequently becomes sufficient to induce
Sodium channel blocker + GABA-ergic drug 5 seizures after repeated application31 .
CBZ + VGB 2 Although clinical data in favour of the kindling hy-
LTG + VGB 2 pothesis in human epilepsy is lacking32 , there is good
CBZ + TGB 1 evidence to suggest that recurrent seizures, particu-
Two drugs with multiple actions 2 larly those involving the limbic structures, can cause
VPA + TPM 1 enduring disturbances in neuronal function33–35 . In
VPA + GBP 1
addition, long-term outcome studies of patients who
GABA-ergic + multiple actions 1 underwent temporal lobectomy for refractory tempo-
VGB + VPA 1
ral lobe epilepsy suggest that secondary epileptoge-
Sodium channel blocker + glutamate antagonist 1 nesis at sites distant to the lesion may develop with
PHT + REM 1
uncontrolled seizures36 . The self-perpetuating nature
Total 42 of seizures in some patients may account for the pro-
CBZ = carbamazepine; GBP = gabapentin; LTG = lamotrigine; gressive decline in the likelihood of seizure freedom
OXC = oxcarbazepine; PHT = phenytoin; REM = remacemide; achieved with successive treatment regimens.
TGB = tiagabine; TPM = topiramate; VGB = vigabatrin; VPA =
Findings from this observational study have rele-
sodium valproate; GABA = γ -aminobutyric acid.
vance to the formulation of a more strategic approach
in managing patients with newly diagnosed epilepsy.
ious combinations were grouped together according to When the first AED failed due to lack of efficacy, the
the AEDs’ primary mode of action, significantly more successful rate of an alternative monotherapy was only
patients given a sodium channel blocker and a drug 16%, compared to 47% in drug-naı̈ve patients. The
with multiple mechanisms became seizure-free than chance of seizure freedom with pharmacological treat-
those treated with other combinations. ment after failure of two consecutive AEDs due to in-
468
adequate efficacy (as opposed to poor tolerability) is
slim. These patients should be assessed for epilepsy
surgery, which can render up to 90% of appropriately
selected patients seizure-free37 . For the majority of
patients not suitable for ‘curative’ resective surgery,
these preliminary observations suggest that rational
combination therapy should be employed early since
its delay, like delayed surgical intervention37 , might
risk irreversible psychosocial consequences and re-
duce the chance of eliminating disabling seizures.
The comparisons between substitution and add-
on, and between the different AED combinations in-
evitably suffer from confounding factors, including
possible bias in patient selection and non-uniformity
of drug regimens. Nevertheless, observations from this
study have generated verifiable hypotheses regard-
ing the management of epilepsy after the first AED
fails. Randomized trials to evaluate the effectiveness
of different AED mechanistic combinations and that
of early versus late combination therapy are needed.
Such a study involving newly diagnosed patients is un-
derway in Glasgow.
REFERENCES
1. Sander, J. W. A. S. Some aspects of prognosis in the epilepsies:
a review. Epilepsia 1993; 34: 1007–1116.
2. Kwan, P. and Brodie, M. J. Early identification of refractory
epilepsy. New England Journal of Medicine 2000; 342: 314–
319.
3. Leestma, J. E., Annegers, J. F., Brodie, M. J. et al. Sudden un-
expected death in epilepsy: observations from a large clinical
development program. Epilepsia 1997; 38: 47–55.
4. Kwan, P. and Brodie, M. J. Cognitive and behavioural effects
of epilepsy and antiepileptic drugs. Epilepsia, in press.
5. Sillanpaa, M., Jalava, M., Kaleva, O. and Shinnar, S. Long-
term prognosis of seizures with onset in childhood. New Eng-
land Journal of Medicine 1998; 338: 1715–1722.
6. Begley, C. E., Famulari, M., Annegers, J. F. et al. The cost
of epilepsy in the United States: an estimate from population-
based clinical and survey data. Epilepsia 2000; 41: 342–351.
7. Mattson, R. H., Cramer, J. A., Collins, J. F., VA Epilepsy
Cooperative Study No. 264 Group. A comparison of val-
proate with carbamazepine for the treatment of complex partial
seizures and secondarily generalized tonic–clonic seizures in
adults. New England Journal of Medicine 1992; 327: 765–771.
8. Richens, A., Davidson, D. L. W., Cartlidge, N. E. F. and
Easter, D. J. A multicentre comparative trial of sodium val-
proate and carbamazepine in adult onset epilepsy. Journal of
Neurology, Neurosurgery and Psychiatry 1994; 57: 682–687.
9. Brodie, M. J., Richens, A. and Yuen, A. W. C. Double-blind
comparison of lamotrigine and carbamazepine in newly diag-
nosed epilepsy. Lancet 1995; 345: 476–479.
10. Beydoun, A. Monotherapy trials of new antiepileptic drugs.
Epilepsia 1997; 38 (Suppl. 9): S21–31.
11. Schmidt, D. and Gram, L. Monotherapy versus polytherapy in
epilepsy. A reappraisal. CNS Drugs 1995; 3: 194–208.
12. Shorvon, S. D. and Reynolds, E. H. Unnecessary polyphar-
macy for epilepsy. British Medical Journal 1977; 1: 1635–
1637.
13. Shorvon, S. D. and Reynolds, E. H. Reduction of polyphar-
macy for epilepsy. British Medical Journal 1979; 2: 1023–
P. Kwan & M. J. Brodie
1025.
14. Reynolds, E. H. and Shorvon, S. D. Monotherapy or polyther-
apy for epilepsy. Epilepsia 1981; 22: 1–10.
15. Schmidt, D. Two anti-epileptic drugs for intractable epilepsy
with complex partial seizures. Journal of Neurology, Neuro-
surgery and Psychiatry 1982; 45: 1119–1124.
16. Schmidt, D. Reduction of two drug therapy in intractable
epilepsy. Epilepsia 1983; 24: 368–376.
17. Macdonald, R. L. and Kelly, K. M. Antiepileptic drug mecha-
nisms of action. Epilepsia 1995; 36 (Suppl. 2): S2–12.
18. Meldrum, B. S. Update on the mechanism of action of
antiepileptic drugs. Epilepsia 1996; 37 (Suppl. 6): S4–11.
19. Brodie, M. J. Monostars: an aid to choosing an antiepileptic
drug as monotherapy. Epilepsia 1999; 40 (Suppl. 6): S17–22.
20. Brodie, M. J., Yuen, A. W. C. and the 105 Study Group.
Lamotrigine substitution study: evidence for synergism with
sodium valproate? Epilepsy Research 1997; 26: 423–432.
21. Brodie, M. J. and French, J. A. Management of epilepsy in
adolescents and adults. Lancet 2000; 356: 323–329.
22. McKee, P. J. W., Larkin, J. G., Brodie, A. F., Percy-Robb, I.
and Brodie, M. J. Five years of anticonvulsant monitoring on
site at the epilepsy clinic. Therapeutic Drug Monitoring 1993;
15: 83–90.
23. Brodie, M. J. and Dichter, M. A. Antiepileptic drugs. New
England Journal of Medicine 1996; 334: 168–175.
24. Dichter, M. A. and Brodie, M. J. New antiepileptic drugs. New
England Journal of Medicine 1996; 334: 1583–1590.
25. Tobias, E., Brodie, A. F. and Brodie, M. J. An outcome audit
at the epilepsy clinic: results from 1000 consecutive referrals.
Seizure 1994; 3: 37–43.
26. Brodie, M. J., Mumford, J. P. and the 012 Study
Group. Double-blind substitution of vigabatrin and valproate
in carbamazepine-resistant partial epilepsy. Epilepsy Research
1999; 34: 199–205.
27. Marson, A. G., Kadir, Z. A., Hutton, J. L. and Chadwick, D. W.
The new antiepileptic drugs: a systematic review of their effi-
cacy and tolerability. Epilepsia 1997; 38: 859–880.
28. Brodie, M. J. Management strategies for refractory
localisation-related epilepsy. Epilepsia, in press.
29. Pisani, F., Oteri, G., Russo, M. F., Di Perri, R., Perucca, E. and
Richens, A. The efficacy of valproate-lamotrigine comedica-
tion in refractory complex partial seizures: evidence for a phar-
macodynamic interaction. Epilepsia 1999; 40: 1141–1146.
30. Gowers, W. R. Epilepsy and Other Chronic Convulsive Dis-
eases: Their Causes, Symptoms, and Treatment. London, J &
A Churchill, 1881.
31. Löscher, W. and Schmidt, D. Which animal models should be
used in the search for new antiepileptic drugs? A proposal
based on experimental and clinical considerations. Epilepsy
Research 1988; 2: 145–181.
32. Berg, A. and Shinnar, S. Do seizures beget seizures? An as-
sessment of the clinical relevance in humans. Journal of Clin-
ical Neurophysiology 1997; 14: 101–110.
33. Glass, M. and Dragunow, M. Neurochemical and morpholog-
ical changes associated with human epilepsy. Brain Research
Review 1995; 21: 29–41.
34. Liu, Z., Mikati, M. and Holmes, G. L. Mesial temporal sclero-
sis: pathogenesis and significance. Pediatric Neurology 1995;
12: 5–16.
35. Blümcke, I., Beck, H., Lie, A. A. and Wiestler, O. D. Molec-
ular neuropathology of human mesial temporal lobe epilepsy.
Epilepsy Research 1999; 36: 205–223.
36. Eliashiv, S. D., Dewar, S., Wainwright, I., Engel, J. Jr and
Fried, I. Long-term follow-up after temporal lobe resection for
lesions associated with chronic seizures. Neurology 1997; 48:
1383–1388.
37. Engel, J. Jr. The timing of surgical intervention for mesial
temporal lobe epilepsy: a plan for a randomised clinical trial.
Archives of Neurology 1999; 56: 1338–1341.