Biotech Values

[DewDiligence]

Musings on GILD’s ‘ION’ results:

All told, the data package from ION-1, ION-2, and ION-3 are very strong, of course. Following are what I consider the points worthy of highlighting from a competitive standpoint. (Please note that the discussion below pertains to genotype-1 patents only.)


ION-1

In the 12-week arms of ION-1, ribavirin didn’t add any benefit, so it clearly isn’t needed for non-cirrhotic treatment-naïve patients. Although GILD did not break out the ION-1 results for cirrhotics vs non-cirrhotics or for GT1a vs GT1b, the lack of any SVR12 delta in the 12-week ribavirin arm vs the 12-week non-ribavirin arm makes it reasonable to infer that ribavirin won’t be needed for any treatment-naïve patients. This is bullish for GILD because it means that treatment-naïve patients can take a bona fide qD regimen and can avoid ribavirin’s side effect of anemia in some patients.

It isn’t yet known whether ABBV/ENTA will require ribavirin in treatment-naïve patients; this is what the PEARL-3 and PEAEL-4 studies will determine for GT1b and GT1a patients, respectively; top-line results from PEARL-3 and PEARL-4 are due any day.

Even if it turns out that ABBV/ENTA need ribavirin in some or all treatment-naïve patients, it’s not a big negative, IMO, because SAPHIRE-1 and SAPHIRE-2, where 100% of the patients took ribavirin, had dropouts rates of <=1%.

In terms of overall efficacy and safety, the 12-week arms of GILD’s ION-1 were comparable to ABBV/ENTA’s SAPHIRE-1; SVR12 rates were 97.7% (209/214) for GILD and 96.2% (455/473) for ABBV/ENTA. GILD has a slight edge because the SVR12 rate was marginally higher and 15.7% of ION-1 patents had cirrhosis (vs 0% in SAPHIRE-1). However, GILD has not broken out the GT1a vs GT1b subgroups, so it’s possible that ION-1 had a higher proportion of (easier-to-treat) GT1b patients than SAPHIRE-1.

The 24-week arms of ION-1 haven’t been reported yet, but they are all but meaningless from a commercial standpoint because no third-party payers will fund 24 weeks of treatment when 12 weeks of treatment produces an SVR12 rate of more than 97%. (The possible exception is cirrhotic patients, but GILD hasn’t yet disclosed the subgroup data for cirrhotics.)


ION-2

Unlike ION-1, ribavirin showed benefit in ION-2. In the 12-week ION-2 arms, the ribavirin arm had an SVR12 delta of +2.8% relative to the non-ribavirin arm, which is a non-trivial boost, IMO. GILD’s FDA and EMA labels for treatment-experienced patients will presumably include ribavirin in the mix—or at least strongly recommend its use. Thus, the most bullish possible scenario for GILD—that Sovaldi + Ledipasvir by themselves could be used for all GT1 patient subgroups in all settings—is out the window.

In terms of overall efficacy and safety, GILD’s ION-2 is comparable to ABBV/ENTA’s SAPHIRE-2; SVR12 rates were 96.4% (107/111) for GILD and 96.3% (286/297) for ABBV/ENTA. GILD could have a slight edge because 20% of ION-2 patients has cirrhosis (vs 0% in SAPHIRE-2); however, GILD has not broken out the GT1a vs GT1b subgroups, so it’s possible that ION-2 had a higher proportion of (easier-to-treat) GT1b patients than SAPHIRE-2.

The 24-week arms of ION-2 are hugely impressive from a medical standpoint, but they are commercially meaningless for the reason mentioned above: third-party payers will not fund 24 weeks of treatment when 12 weeks of treatment are sufficient. (The possible exception is cirrhotic patients, but GILD hasn’t yet disclosed the subgroup data for cirrhotics.)


ION-3

Shortening the treatment duration to 8 weeks produced slightly lower SVR12 rates (with or without ribavirin) than 12 weeks of treatment (without ribavirin). My guess is that GILD will use the ION-3 data to justify dropping the pursuit of 8-week labeling for any subgroup. From a business standpoint, GILD didn’t really want a treatment duration as short as 8 weeks, but they had to test it to placate regulators and third-party payers, who could otherwise have argued that GILD was treating patients longer than necessary in order to make more money.

It’s counter-intuitive that the 12-week arm in ION-3 had a negative delta of 2.3% relative to the 12-week non-ribavirin arm of ION-1, despite the fact that ION-1 included some cirrhotics while ION-3 didn’t. Thus, the relatively low SVR12 rate in the 12-week arm of ION-3 may be an aberration. When the SVR12 rates from the 8-week arms of ION-3 are compared to the combined SVR12 rate of the (ION-1 + ION-3) 12-week non-ribavirin arms, it becomes more plausible to argue that lopping off 4 weeks of treatment isn’t a good idea.


Summary

All told, GILD’s data from the ION-1, ION-2, and ION-3 studies are excellent, but investors should have expected that based on the results of LONESTAR and other phase-2 studies conducted by GILD. Then why is ENTA’s share price down so much today? I think the main reason is that ENTA’s share price had several dollars of vig baked in for the possibility of GILD’s botching the ION studies, which didn’t happen.

Finally, nothing in GILD’s data release today warrants a change to my ENTA valuation model in #msg-94993406. In that model, I give GILD a 62/38 advantage in GT1 market share starting in 2015, and I think such a split is still reasonable, particularly is ABBV/ENTA compete aggressively on price.