HOUSE_OVERSIGHT_024907.jpg

COWEN
COLLABORATIVE INSIGHTS
February 25, 2019

Epidiolex appeared to be well tolerated in the trial. 94% of patients on 20 mg/kg
Epdiolex and 84% of patients on 10 mg/kg Epidiolex had an adverse event, compared to
72% of placebo patients. 88% of 20 mg/kg patients and 89% of 10 mg/kg patients
deemed their adverse events to be mild or moderate. The most common AEs on 20
mg/kg were somnolence, decreased appetite, diarrhea, upper respiratory infection,
pyrexia, vomiting and nasopharyngitis. For 10mg/kg the most common AEs were
somnolence, decreased appetite, upper respiratory infection, diarrhea, and status
epilepticus. None of the cases of status epilepticus on 10 mg/kg were deemed
treatment-related. Thirteen patients on 20 mg/kg Epidiolex had an SAE, of which five
were considered treatment related; and 13 patients on 10 mg/kg Epidiolex had an SAE,
of which 2 were considered treatment related, compared to 8 patients on placebo.
Elevations in ALT/AST levels were observed in 11 patients in the 20mg/kg group and 2
patients in the 10mg/kg group; 10 of the 13 patients were also on valproic acid. Five
Epidiolex patients withdrew due to the elevations, but none of the patients met the
criteria for drug-induced liver injury. GW noted that overall 10 mg/kg seemed to be
somewhat better tolerated; 6 patients on 20 mg/kg Epidiolex and 1 patient on 10
mg/kg Epidiolex discontinued treatment due to adverse events, compared with 1
patient on placebo. There were no deaths in the trial. The results of the trial were
published in NEJM in May 2018.

Following the release of Phase III datasets from 1 Dravet and 2 LGS trials, a key area of
controversy among investors (though not physicians) had been the drug-drug
interaction between Epidiolex and clobazam, and in particular whether Epidiolex was
effective in patients who were not also taking clobazam. This was put to rest at AES
2017 when GW presented a pooled analysis of the two Phase III LGS trials evaluating
Epidiolex's efficacy with and without concomitant clobazam. Even without clobazam,
Epidiolex produced solid placebo-adjusted response rates. Response was characterized
in terms of "25% responders", "50% responders", and "75% responders", meaning the
proportion of patients who had a 25%, 50%, or 75% decrease in seizure frequency. For
patients randomized to Epidiolex's 20mg/kg dose, the placebo-adjusted 50% response
rate was 22% for patients on Epidiolex without clobazam, compared to 33% for patients
on Epidiolex and clobazam. We believe that investors had been hoping for a 12-15%
placebo-adjusted 50% response rate for patients on Epidiolex without clobazam in order
to be satisfied that Epidiolex was active without clobazam, and therefore the results
cleared this bar. For patients randomized to 10 mg/kg Epidiolex, the placebo-adjusted
50% response rate was 25% for patients on Epidiolex without clobazam, compared to
27% for patients on Epidiolex with clobazam. The results for other thresholds of seizure
frequency reduction were also generally solid. For 25% responders, the placebo-
adjusted response rate was 9% for patients on 20 mg/kg Epidiolex without clobazam,
compared to 24% for patients on 20mg/kg Epidiolex with clobazam. At the 75%
responder threshold, the placebo adjusted response rate was 8% for patients on
20mg/kg Epidiolex without clobazam, compared to 30% for patients on 20mg/kg
Epidiolex with clobazam. In addition to the pooled Phase III data, there were also
abstracts from the compassionate use experience of Massachusetts General Hospital
and the University of Alabama Birmingham. The analyses from these two institutions
also concluded that Epidiolex does not need to be combined with clobazam to be
effective.

This report is intended for michael.cella@cowen.com. Unauthorized redistribution of this report is prohibited.

COWEN.COM
91
HOUSE_OVERSIGHT_024907