The Lancet March 24, 2007

The SANAD study of effectiveness of carbamazepine, gabapentin, lamotrigine, oxcarbazepine, or topiramate for treatment of partial epilepsy (Arm A) and valproate, lamotrigine, or topiramate for generalised and unclassifiable epilepsy (Arm B)

Anthony G Marson et al. (SANAD Study Group)

Bottom Line

The SANAD trial established lamotrigine as a preferred first-line treatment for partial-onset epilepsy due to better tolerability over carbamazepine, while confirming valproate as the most effective first-line option for generalized and unclassifiable epilepsies.

Key Findings

1. In Arm A (partial epilepsy), lamotrigine demonstrated significantly better time to treatment failure compared to carbamazepine (HR 0.78, 95% CI 0.63–0.97), gabapentin (HR 0.65, 95% CI 0.52–0.80), and topiramate (HR 0.64, 95% CI 0.52–0.79).
2. In Arm A, carbamazepine had a non-significant advantage for time to 12-month remission compared to lamotrigine (HR 0.91, 95% CI 0.77–1.09), but lamotrigine was deemed non-inferior in the per-protocol analysis.
3. In Arm B (generalised and unclassifiable epilepsy), valproate showed significantly better time to treatment failure than topiramate (HR 1.57, 95% CI 1.19–2.08) and a non-significant advantage over lamotrigine (HR 1.25, 95% CI 0.94–1.68).
4. For the subgroup of patients with idiopathic generalised epilepsy in Arm B, valproate was significantly superior to both lamotrigine (HR 1.55, 95% CI 1.07–2.24) and topiramate (HR 1.89, 95% CI 1.32–2.70) for time to treatment failure.
5. In Arm B, valproate was significantly better than lamotrigine for time to 12-month remission overall (HR 0.76, 95% CI 0.62–0.94).

Study Design

Design
Randomised Controlled Trial
Open-Label
Sample
2,437
Patients
Duration
Up to 7 years
Median
Setting
Multicenter, UK
Population Patients aged 5 years and older with untreated or previously treated (but failing) epilepsy. Arm A (n=1,721) included patients for whom carbamazepine was standard treatment (predominantly partial/focal epilepsy). Arm B (n=716) included patients for whom valproate was standard treatment (predominantly generalized or unclassifiable epilepsy).
Intervention Arm A: Gabapentin, lamotrigine, oxcarbazepine, or topiramate. Arm B: Lamotrigine or topiramate.
Comparator Arm A: Carbamazepine (standard treatment). Arm B: Valproate (standard treatment).
Outcome Time to treatment failure (defined as drug withdrawal due to inadequate seizure control, unacceptable adverse events, or both) and time to 12-month seizure remission.

Study Limitations

The unblinded (open-label) pragmatic design may have introduced investigator and patient bias, particularly regarding subjective adverse event reporting and the clinical threshold for declaring 'treatment failure'.
The initial titration schedule for topiramate was relatively rapid, which likely inflated its adverse effect profile and compromised its time to treatment failure results.
The inclusion of patients with 'unclassifiable' epilepsy in Arm B introduced clinical heterogeneity that may have diluted specific syndromic efficacy signals.
The trial did not include levetiracetam, which subsequently became a widely used antiepileptic drug (later evaluated in the SANAD II trial).

Clinical Significance

The SANAD study provided definitive, long-term comparative evidence that reshaped clinical practice globally. For focal (partial) epilepsy, it established lamotrigine as the new first-line therapy of choice due to its superior tolerability and non-inferior efficacy compared to carbamazepine. For generalized and unclassifiable epilepsies, it reinforced valproate as the most effective agent, significantly outperforming newer drugs like topiramate and lamotrigine. However, the study highlighted the ongoing clinical challenge of balancing valproate's superior efficacy against its known teratogenicity in women of childbearing potential.

Historical Context

Before the SANAD trial, evidence for antiepileptic drug (AED) efficacy was primarily derived from short-term, placebo-controlled, regulatory add-on trials. There was a critical void of long-term, head-to-head comparative effectiveness data between older standard AEDs (carbamazepine, valproate) and the newer generation of drugs (lamotrigine, gabapentin, topiramate, oxcarbazepine). Commissioned by the UK's National Institute for Health Research (NIHR), the SANAD trial was a landmark pragmatic study that shifted the clinical paradigm from solely measuring 'seizure freedom' to emphasizing 'time to treatment failure'—a composite metric reflecting real-world efficacy, tolerability, and drug retention.

Guided Discussion

High-yield insights from every perspective

Med Student
Medical Student

How do the mechanisms of action differ between carbamazepine, which the SANAD study evaluated for partial seizures, and valproate, which was evaluated for broad-spectrum coverage in generalized epilepsies?

Key Response

Carbamazepine primarily blocks voltage-gated sodium channels, making it effective for focal (partial) onset seizures. Valproate has multiple mechanisms including sodium channel blockade, T-type calcium channel modulation, and increasing GABA levels, providing the broad-spectrum efficacy needed for generalized and unclassifiable epilepsies.

Resident
Resident

The SANAD trial established valproate as the most effective first-line agent for generalized epilepsy. How does this finding complicate the management of a 22-year-old female patient newly diagnosed with generalized epilepsy, and what alternatives should be considered?

Key Response

Valproate is highly teratogenic, causing neural tube defects and cognitive impairment in offspring. For women of childbearing potential, balancing the superior efficacy of valproate shown in SANAD against its teratogenic risks requires considering alternatives like lamotrigine or levetiracetam, despite them potentially having lower efficacy for certain generalized seizure types.

Fellow
Fellow

In SANAD Arm A, lamotrigine was preferred over carbamazepine due to a longer time to treatment failure. However, how did the two drugs compare regarding the secondary outcome of time to 12-month remission, and how does this nuance affect drug selection?

Key Response

Carbamazepine actually showed a slight trend toward better efficacy for absolute seizure control (time to 12-month remission) compared to lamotrigine, but lamotrigine won the primary outcome of 'time to treatment failure' due to significantly better tolerability. A fellow must weigh whether to push through initial carbamazepine side effects for a potentially higher chance of absolute remission in highly motivated patients.

Attending
Attending

SANAD was an unblinded, pragmatic effectiveness trial rather than a double-blind efficacy trial. How does this design choice more accurately reflect real-world clinical practice, and how should it influence patient counseling regarding antiepileptic drug initiation?

Key Response

Pragmatic trials measure 'effectiveness' (real-world performance heavily influenced by tolerability and adherence) rather than pure pharmacological 'efficacy'. This teaches us that counseling on side-effect management is as critical to long-term seizure freedom as the molecular efficacy of the drug, as most treatment failures are due to adverse effects rather than lack of efficacy.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD
PhD

How might the unblinded nature of the SANAD trial have introduced differential ascertainment bias into the primary outcome of 'time to treatment failure', and what statistical sensitivity analyses could be employed to estimate the magnitude of this bias?

Key Response

Lack of blinding can influence both the patient's and physician's threshold to withdraw a drug due to perceived side effects, directly impacting the 'time to treatment failure' survival curve. Analyzing objective versus subjective reasons for withdrawal and using competing risk models would help disentangle true pharmacological failure from bias-driven withdrawal.

Journal Editor
Journal Editor

Arm B of the SANAD trial included a significant proportion of patients with 'unclassifiable' epilepsy. From an editorial standpoint, how does pooling this heterogeneous cohort with generalized epilepsy threaten the internal validity of the conclusion that valproate is superior?

Key Response

Unclassifiable epilepsy might include undiagnosed partial epilepsies where valproate performs differently. Pooling these dilutes the true efficacy signal for purely generalized epilepsies. A rigorous reviewer would demand stratified analyses to ensure the valproate superiority was not driven by misclassified focal cases, ensuring robust internal validity.

Guideline Committee
Guideline Committee

Given the SANAD trial's pragmatic, unblinded design, how should guideline committees grade the quality of evidence when recommending lamotrigine as a first-line agent for partial epilepsy, and how does this highlight discrepancies between NICE and AAN/AES guideline methodologies?

Key Response

AAN/AES guidelines typically require Class I evidence (demanding double-blinding) for a Level A recommendation, whereas NICE guidelines heavily favor pragmatic trials reflecting real-world effectiveness. Because SANAD was unblinded, it might be downgraded in rigorous efficacy-based grading systems, highlighting the tension between evaluating real-world utility versus strict pharmacological efficacy in guideline development.

Clinical Landscape

Noteworthy Related Trials

1985

VA Cooperative Study 118

n = 622 · NEJM

Tested

Carbamazepine, phenobarbital, primidone, or phenytoin

Population

Adults with partial and secondarily generalized seizures

Comparator

Active head-to-head comparison

Endpoint

Retention in the trial (composite of efficacy and toxicity)

Key result: Carbamazepine and phenytoin were the most effective and best tolerated agents.
2005

MESS Trial

n = 1,443 · Lancet

Tested

Immediate antiepileptic drug treatment

Population

Patients with a single unprovoked seizure or early epilepsy

Comparator

Deferred treatment

Endpoint

Time to first seizure and time to 2-year remission

Key result: Immediate treatment delayed the next seizure but did not alter long-term remission rates.
2021

SANAD II Trial

n = 990 · Lancet

Tested

Levetiracetam or zonisamide

Population

Patients with newly diagnosed focal epilepsy

Comparator

Lamotrigine

Endpoint

Time to 12-month remission

Key result: Lamotrigine was found to be more effective and cost-effective than both levetiracetam and zonisamide.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis