No subjects had been enrolled when Amendment 1 was issued. This amendment included changes to the study entry criteria based on discussions at the Investigator’s Meeting. These changes were as follows: • Subjects who had received treatment with benzodiazepines on a chronic or routine Basis were allowed to enter the study if they discontinued These medications 4 weeks prior to Visit 1 • Excluded psychiatric diagnoses were specified more narrowly as serious, unstable psychiatric diagnoses that would confound the Investigator’s ability to conduct the trial or prevent the subject from completing the requirements of the protocol • Any ADHD medication intake was to remain stable during the study and had to be stable 1 month prior to Visit 1 • Subjects taking phenobarbital or primidone prior to Visit 1 were excluded • Subjects with seizures too close together to be accurately counted were excluded Other minor changes included addition or clarification of procedural details that were unlikely to significantly affect enrollment or study outcome.

Amendment 2 allowed for establishment of a greater number of study sites and allowed for sites to be established in Canada and South Africa. Minor changes were made in the study safety assessments in order to satisfy requirements of the EMEA. These changes included addition of height and TSH measurements. Amendment 2 incorporated information about the FDA approval of LEV for use in children (4 years of age and above) with partial onset seizures (approval occurred in June 2005), and included information about the efficacy and safety results of other ongoing studies of LEV. Amendment 2 allowed for subjects with previous exposure to LEV to enroll in the study under limited circumstances. One impact of these changes was to allow more subjects to enroll, particularly subjects outside the US.

Amendment 3 modified the Inclusion and Exclusion criteria in the following ways, to make the sample more representative of the target population: • The IQ criterion for study inclusion was reduced from 70 to 65 • The weight criterion was changed from ≤ 80 kg to ≤ 100 kg • Exclusions for subjects with current psychiatric diagnoses were relaxed from the Initial trict criteria after data from pivotal study N159 showed that LEV did not exacerbate ifficulties in children with history of psychiatric illnesses. Serious, unstable sychiatric illnesses were still excluded (such as suicide risk within the past 6 months, sychotic disorder, or acute mania) • The requirement for 2 partial onset seizures within the 4 weeks prior to Visit 1 was reduced to a requirement of only 1 partial onset seizure within the same time period. This criterion change, which is consistent with the intended population on the current LEC label, would terefore enable the collection of additional relevant safety data. There is no evidence that reducing the seizure number criterion would impair the sensitivity of the study to demonstrate potential psychiatric or neuropsychological effects of LEV Assumptions for the sample size estimation were adjusted in Amendment 3, so the sample size was recalculated and the total evaluable n was reduced by 13 subjects. Due to difficulties with enrollment, the sample size calculation was reviewed. During the review, it was determined that the Leiter-R Memory Screen un-inflated standard deviation of 6.94 that was used in the initial sample size was miscalculated. The actual un-inflated standard deviation should have been 10.19. To adjust for the younger age group in N01103, their epileptic disease state, and the unknown time between test and re-test, the estimate was inflated to 13.00 (a little more than 25%). Additionally, the type II error rate was relaxed from 15% to 20%, reducing statistical power to 80%.

The increase to the type II error rate results in a smaller subject requirement, an acknowledgement of the difficulty enrolling subjects from this population while staying within generally accepted constraints for sample size determinations. Finally, the drop-out rate/major protocol deviation rate assumption was increased from 20% to 30% to more accurately reflect the actual rate occurring in the study. These changes were discussed with the FDA and accepted prior to implementing the changes. Amendment 3 also increased the number of study centers.