It's hard to say whether FDA looks on at follow-on indications more favorably than the initial one because the standards for acceptable benefit risk ratio differ from disease types, and changes over time.
On the risk side, later submission enjoy the wealth of post-marketing safety data. Unless serious safety concern is revealed, the longer the drug is on the market, the better risk management by treating physicians, and better understanding by the FDA. The lowering of the risk denominator effectively raises the relative benefit-risk ratio for later indications.
But on the benefit side, the threshold for clinical benefit is really determined by the tumor type or disease setting and the risk profile. The efficacy hurdle will be lower for more advanced disease, for which there are more limited treatment options, thus greater unmet medical need, for drugs with favorable risk profile, for drugs that have already beeb demonstrated for the same tumor type albeit in a different seeting. The reverse is also true. Although it may be argued that more clinical data may be available to the oncology community due to off-label use,and company sponsored or investigator initated trials, that advisory committee's perception may be influenced, but unless those data were officially included in the filing package, the influence should be small.
Take Gemzar for an example, in 1996, it was approved for refractory pancreatic cancer based on a controlled trial of less than 80 subjects which merely showed symptomatic relief without evidence of prolonging survival or delay disease progression. But the only option then was 5-FU, and the symptomatic relief is perceived real. With an ODAC vote of 7 to 3, it was approved. Once it became the standard of care, it raised hurdles for future drugs aiming at the same indication. Can you imagine Tarceva being approved based on anything less than a significant PFS or OS? No way.
Then in 1998, Gemzar was approved for lung based on solid overall survival, and tumor response data. It's fair to say, whether it was a supplemental NDA or not, its approval is untainted.
2004 came, Gemzar was approved for breast based on TTP, and ORR, with a single statement on the label on OS. The interim OS analysis was impressive (from Kathy Albain's presentation at ASCO 2004 :The OS hazard ratio (HR) was 0.775 (.627, .959) in favor of GT, p=.018. Median OS for GT was 18.5 mos (16.5-21.2) vs T, 15.8 mos (14.4-17.4). One-year survival was 70.7% (65.1-76.3%) for GT and 60.9% (54.8-66.9%) for T (p=.019).
The only reasonable explanation that it did not make it to the label was because of either an overly conservative alpha was used for OS, or the final OS result barely missed whatever statistical alpha that was allocated to it. You be the judge, was the hurdle lowered - would any reasonable soul (statistician or not) deny that Gemzar is effective in breast cancer, given the benefit seen in this large randomized trial, and the humongous amount of publication? No. In this case, statistical perfection is not necessary. Did FDA go out of its way to help Gemzar, which had a demonstrable benefit risk ratio to reach the market? You bet! Never mind it missed the magic alpha for OS, never mind in 2004, it's the prevailing thought that OS is a must have for an approval. The evidence is strong enough.
Was Gemzar given any special treatment? No. FDA allowed avastin to be filed based on cooperative group trial, in absence of a significant OS advantage. Unfortunately, the TTP endpoint documentation was not up to regulatory standard. So if you don't have OS advantage, fine, but what if you don't even have a TTP due to technical reason, sorry, can't help you on that.
Then 2006 came, Gemzar approved for ovarian. Lucky them, PFS now is the gold standard. OS is obviously elevated to the platinum standard, but gold standard is just fine. No, this has nothing to do with it being a supplemental NDA.
Final thought - I am getting tired of my writing - so hopefully we agree that alpha control is only meaningful when the benefit risk ratio is not convincing in eyes of FDA oncologists. Provenge has superior benefit risk ratio - it had the platinum OS advantage, the titanium safety advantage. There is no reason that it should not be approved.
