Rational polytherapy: Myth or reality?

A clinical significant “rational” combination of AEDs should provide suppraadditive, synergistic anticonvulsivant effects and additive/infraadditive toxicity. Indeed, when the suppraadditive anticonvulsivant effects are associated with a high toxicity (whenever adverse effects of a synergistic combination also leads to suppraadditive summation) the protective index (defined as a comparison of the amount of a drug causing a therapeutic effect to the one producing toxicity, and determined as the ratio given by the toxic dose divided by the therapeutic dose) may be unchanged or even reduced. Hence, high values of the protective index are preferable to low values, given that a higher dose of the AED would have to be taken to reach the toxic threshold compared with the dose to obtain a therapeutic effect (Brigo et al., 2013).

The concept of “drug loading” (initial higher dose of a drug that is given at the beginning of a course of treatment before dropping down to a lower maintenance dose (equivalent dosages i.e., equal drug loads) may also bring some insight into whether the increased effectiveness of a combination of AEDs (either with the same or different MoAs) is due to an improvement in efficacy or in tolerability. A robust study looking at the monotherapy versus polytherapy focussing primarily on tolerability did not show evidence of differences in overall efficacy and neurotoxicity between the two regimes in which AEDs were started in equal dose loads (Deckers et al., 2001).

Clinical efficacy has been demonstrated by the use of AEDs with the same MoA (Stephen and Brodie, 2002). However, undesirable and avoidable neurotoxic associations may occur or emerge in patients treated with combinations of AEDs with the same MoAs, as between sodium blockers such as PHT, carbamazepine (CBZ), lamotrigine (LTG), or oxcarbazepine (OXC) (Brodie and Yuen, 1997), although results were conflicting. Indeed, an OXC placebo-controlled, dose ranging trial (Barcs et al., 2000) showed a dose-dependent increase in the incidence of premature discontinuations due to AEDs regardless of concomitant AED use. Different MoA for OXC and CBZ were also proposed in this study due to the fact that a large number of patients were already on CBZ and benefited with the addition of OXC. Another more recent, retrospective and observational study with lacosamide (LCM) as adjunctive therapy in uncontrolled epilepsy (Flores et al., 2012) showed that adding LCM to drugs acting on the sodium channel was in general, but not statistically significant, more efficacious compared to adding it to AEs with other MoAs.

From the early experience with SGAEDs, it became clear that the majority of the patients could be controlled by monotherapy, with only less than 4% of them requiring polytherapy (Kwan and Brodie, 2000a; Brodie, 2013). It is uncertain whether this small percentage of success could be attributed to a synergistic effect or to an additive effect of the multiple drugs used.

Previous attempts to prove a synergistic effect in more limited situations have not shown consisting success (Kwan and Brodie, 2000a; Besag et al. 1998). According to the work of Kwan and Brodie (2000a), none of the 11 patients who received add-on treatment after a second drug became seizure-free. In the study of Besag et al. (1998), the authors concluded that toxicity is more likely to occur when LTG is added to CBZ. A reduction of the dose of CBZ usually reduces the toxicity, allowing the dose of LTG to be increased to achieve its maximum effect. Isobolographic analysis from a seizure model in mice also showed lack of synergism between LCM with VPA or PHT (Brigo et al., 2013). Other combinations that revealed no synergism include those between topiramate (TPM) or sodium valproate (VPA) with PHT or CBZ (St Louis, 2009).

Moreover, the results of pivotal studies to commercialise some of the SGAEDs, namely OXC or eslicarbazepine acetate (ESL), failed to prove the principle of RP. In these studies, the addition of CBZ to another AEDs with the same or different MoA, lead to additive efficacy and more side effects (Halasz et al., 2010). In a well-designed, multicentre, parallel-group, open label study looking at this problem, patients with partial epilepsy not controlled by a single AED or sequential AED monotherapy were randomised to alternative monotherapy or to adjunctive therapy. However, again, there was no significant difference between the two groups regarding seizure freedom or retention rate (Beghi et al., 2003).

By looking at the problem in a different way, we might assume that using AEDs with multiple MoAs would be more efficient than using AEDs with only one known MoA. However, recent data on the use of zonisamide (ZNS) versus CBZ (Baulac et al., 2012), or the older study comparing VPA and CBZ (Mattson et al., 1992), showed that CBZ was as effective, or even more effective, than the AEDs with multiple MoAs.

The only evidence for a possible synergism between two AEDs was established for VPA and LTG (Brodie and Yuen, 1997; Pisani et al., 1999), but these two drugs lead to complex pharmacokinetics interactions that may jeopardize a consistent interpretation. Furthermore, the study of Pisani et al. (1999) had a small sample of 20 patients and changes in seizure frequency were analysed descriptively, without formal statistical analysis. Another study (Moeller et al., 2009) with similar limitations due to its retrospective design in which patients were prescribed LTG plus VPA at the discretion of the epileptologist and not in a randomized way, this combination proved to be clinically effective. Furthermore, the adverse effects that occurred were resolved by decreasing the dosage of one of the drugs without seizure exacerbation.

Experiments in the preclinical phase have not conformed to the literature on synergistic clinical efficacy. A review of more than one hundred studies evaluating 536 experimental drug interactions concluded that no single combination of AEDs consistently displayed true synergism (Jonker et al., 2007).

Several reviews have addressed this issue, with the overall conclusion being that RP “remains speculative concerning better efficacy based on the use of AEDs with differing MoAs” (Ben-Menachem, 2014), and that “experimental and clinical evidence in support of it is sparse” (Brodie and Sills, 2011).

As for RP and side effects, there is evidence that adding AEDs with the same MOAs may lead to an enhancement of side effects. “In particular, care should be taken to avoid an excessive drug load, which can be associated with decreased tolerability” (Brodie and Sills, 2011).

It is difficult to drawn reliable information from studies with AEDs with multiple MoAs given the fact that the same MoA might not be responsible for drug efficacy, or toxicity, in different patients. Furthermore, the mechanistic minutia of many available AEDs, with hypothetic contributions to their clinical effects, is still not fully understood. This brings more confusion to the concept of RP as it is mostly based on MoAs (Brodie and Sills, 2011).

Most patients with RE take two to three AEDs, some even four or more. Therefore, some rules should be introduced to rationalise their use in order to achieve an optimal outcome. Can RP be a good strategy? In other words, is there evidence supporting RP? The correct response is no. So far, there is no evidence unequivocally validating this form of polytherapy. Nonetheless, strong arguments favouring RP may be found.

According to Kwan and Brodie (2000a), in 2000, 64% of the patients with epilepsy de novo were achieving freedom from seizures with an appropriate monotherapy regimen. Nine years later, this percentage increased by 4.4% (68.4%) through the use of and at the expense of polytherapy regimen (3, 4, and even 5 AEDs) with SGAEDs that had new or different MoAs (Brodie and Bamagous, 2009). Although small, this amount should not be disregarded when addressing RE, and it seems worthwhile considering polytherapy always using at least one SGAED.

Although sparse, there is experimental (Table 1) and clinical (Table 2) evidence in support of RP. Animal models of seizures, epilepsy and drug-related neurotoxicity are ideal to evaluate the efficacy and toxicity of drug combinations in large groups of genetically homogeneous animals. The latter is relevant even if direct transition from experimental data to humans is questionable (Barker-Haliski et al., 2014; Brodie and Sills, 2011). These studies should include: a) efficacy and toxicity models in which both AEDs are at least minimally effective; b) AED ratios reflecting those used in clinical setting; c) concentration analysis of AEDs in both plasma and brain, in order to exclude confounding pharmacokinetic interactions; and d) the use of appropriate methods of analysis, such as isobolography, because they measure effectiveness and determine infra-additive, additive, or supra-additive interactions (Brigo et al., 2013).

Some examples of experimental synergistic combinations of AEDs will follow. The anticonvulsant interaction between PB and PHT was found to be supradditive against maximal electroshock (MES) seizures in mice, but neurotoxicity was not studied (Masuda et al., 1981). Another combination, VPA and PHT, displayed synergism in terms of anticonvulsant effect against MES in the rodent and the neurotoxic activity was simple additive (Chez et al., 1994). A clear-cut synergy was again evident against MES in mice when TPM was co-administered with CBZ or PB (Shank et al., 1994). A distinct synergism also occurred for the combinations of gabapentin (GBP) with CBZ, VPA, PHT and PB (Borowicz et al., 2002). Synergistic combination in terms of efficacy was shown between LTG and TPM (Luszczki et al., 2003), and TPM with felbamate (FBM) and OXC, but subadditivity or antagonism, with respect to neurotoxicity with OXC and FBM or with OXC and LTG was also found in the MES-induced seizures and chimney test in mice (Luszczki and Czuczwar, 2005). The same was found for LCM with CBZ, LTG, TPM, GBP and levetiracetam (LEV) in the 6-Hz seizure model in mice (Shandra et al., 2013).

Nevertheless, with the possible exception of the preclinical synergistic interaction between VPA and LTG (Cuadrado et al., 2002), synergistic combinations identified in animal models do not reliable extend to the clinic (Barker-Haliski et al., 2014). However, experimental studies may be the right way to overcome restrictions imposed by human clinical studies, such as fixed drug doses (which prevent optimal combinations of dosages with the most benefit and least toxicity) (French and Faught, 2009). The first combinations with two AEDs with different MoAs, PB with PHT, proved to be superior in terms of efficacy and/or toxicity when compared to AEDs of similar MoAs, CBZ and PHT (Brigo et al., 2013). These were the initial studies leading to the concept of RP, which evolved with the advent of SGAEDs. Overall, every experimental study stresses the need to use a combined effect of two different pathways rather than a single pharmacological pathway (Brigo et al., 2013). Furthermore, the most successful experimental combination of two AEDs appears to be the one of a single MoA of one AED with a multiple MoAs of the other AED (Deckers et al., 2000). Recently, two widely used SGAEDs – perampanel (PER), a noncompetitive AMPA receptor antagonist, and zonisamide (ZNS), a modulator of voltage-sensitive sodium channels and T-type calcium currents – have shown synergism in the rat amygdala kindling model of temporal lobe epilepsy. This suggests that these two AEDs could be successfully used in focal epilepsies (Russmann et al., 2016).

Combinatorial clinical studies are difficult to undertake. They require the investigation of efficacy and tolerability of both single AEDs and combinations in a homogeneous population of patients with a design powerful enough to be able to separate synergism from additivity alone. Moreover, they also need adjustments of the AEDs combinations to balance overall drug load (Barker-Haliski et al., 2014). The literature claims a particular efficacy for a combination of a sodium channel blocker with either a GABA-ergic (Brodie and Sills, 2011) or a multiple MoAs drug (Kwan and Brodie, 2000b). The only broad evidence of synergism is with VPA and LTG. This was achieved in a trial (Brodie and Yuen, 1997) in patients without seizure control, where an attempt was made to substitute CBZ, PHT or VPA for monotherapy LTG (and in which an adjustment in the LTG dosing schedules for the pharmacokinetic interactions among these AEDs resulting in the same circulating LTG concentrations for all three combinations was performed). The data has shown that the efficacy was strikingly higher in the VPA plus LTG group when compared with the LTG plus CBZ. Detailing a study that has already been mentioned, a trial performed in twenty patients with focal seizures stable on a combination of AEDs, showed that among the thirteen who did not display a good response to the consecutive addition of VPA and LTG, four became seizure free and another four experienced more than a 50% seizure reduction when both drugs were given in combination. This effect was achieved despite lower doses and circulating concentrations than those occurring if administered separately (Pisani et al., 1999).

Other apparent or proved synergistic combinations of AEDs with different MoAs, although based in small samples of patients, small clinical trials or short treatment periods, are VPA with ethosuximide for generalized absences (Rowan et al., 1983), PB with PHT for generalized tonic-clonic seizures (Cereghino et al., 1975), CBZ with vigabatrin (VGB) or VPA (Walker and Koon, 1988; Kwan and Brodie, 2000a; Brodie and Munford, 1999), VGB with tiagabin (seldom used nowadays) for focal seizures (Leach and Brodie, 1994), LTG with TPM for a wide range of seizures (Stephen et al., 1998), stiripentol with clobazam in the severe myoclonic epilepsies of the pediatric population (Perucca, 2006), and azogabine/retigabine with either sodium channel blocking AEDs or with AEDs with other MoAs (Brodie et al. 2014), and LCM with any concomitant AED regimen, including sodium channel blocking AEDs (Chung et al., 2010). However, there is also evidence of a higher reduction in seizure frequency when LCM is associated to AEDs not acting on the sodium channel (Sake et al., 2010). To investigate whether different MoAs-based AED combination therapies in focal epilepsies are more effective than same MoAs-based AEDs combinations, Margolis et al. (2014) used real-world data from a large sample and showed that the former combinations had greater effectiveness (as measured by treatment persistence and lower risk for hospitalization or emergency department visits).

Finally, “irrational” polytherapy can occur due to several reasons and should be avoided (Brigo et al., 2013). A wrong AED may be chosen for starting a monotherapy regimen leading to an unfavourable event (such as an increase in seizures or toxic side effects). The solution is to replace the first AED by an appropriate one, rather than adding yet another AED. Furthermore, an inadequate knowledge of the MoA of AEDs and their pharmacokinetic and or pharmacodynamic interactions between them may also lead to unexpected or unpleasant symptoms.