"72 plus months "
72 plus weeks obviously. I also buy into Ernie's assertion that Cabo will have plenty of patients to treat even if/when Nivolumab were to enter market. Not everyone responded to Nivolumab.
"HCC should be a relatively uncontested indication for Cabo. I don't think the results were nearly so good for Tivantinib"
Right, it will be interesting to see how this market shapes up. Even if Tivantinib were successful, Cabo has a decent financial opportunity on a global scale. I think Cabo early renal data was very strong in heavily pretreated renal cancer population. Maybe underappreciated, and I would like to see eventual OS data. I think Cabo data may be closer to Nivolumab data than you may think. Those bone responses and pain relief going out 72 plus months were pretty compelling. I noticed the recent Cabo patent on bone pain relief and bone involvement.---For lung,melanoma,prostate,thyroid, and renal.cancers.
" The only guidance is that Roche expects to file an NDA in 2014. They need the results in hand soon to make that deadline."
Guidance from Roche hasn't changed since recent phase 1 update demonstrating close to 14 months pfs with Cobi/Zelboraf combination? What is the number of events required to trigger interim analysis?
per clinicaltr ialdotg ov
RCC(METEOR) 198 out of 211 sites actively recruiting
HCC(CELESTIAL) 89 out of 139 actively recruiting
Full recruitment for METEOR is probably right around the corner. CELESTIAL will probably get a much greater focus soon.
No problem. I've been looking in MAPK direction for awhile now.
Here is a good paper
" A Novel SND1-BRAF Fusion Confers Resistance to c-Met Inhibitor PF-04217903 in GTL16 Cells though MAPK Activation"
"I don't think the impact of cabozantinib treatment has been defined as yet in terms of MAPK activation, at least I don't recall seeing it the lit..."
The EXAM trial responses in RAS population is the best we have thus far.
"The results showed that patients with RAS mutations also experienced a significant benefit in PFS with cabozantinib, with patients in the experimental arm achieving a median PFS of 47 weeks compared with 8 weeks in the placebo arm (HR = 0.15; 95% CI, 0.02-1.10; P = .0317)."
Here is background information to support Exelixis's 24 month OS estimation for EXAM. I can't get a read on how many of those 5 patients long term treated patients had metastatic disease upon study entry.
Background: Metastatic medullary thyroid cancer (MTC) carries a median survival of about two years (Modigliani et al. 1998; Roman et al. 2006). Cabozantinib inhibits three primary pathways implicated in MTC pathogenesis and progression: MET, vascular endothelial growth factor receptor-2 (VEGFR2), and RET. A phase I study was initiated in September 2005 in patients (pts) with advanced solid tumors. Methods: 85 pts, including 37 with MTC were enrolled to evaluate safety, pharmacokinetics, and to determine maximum-tolerated dose (MTD). Once the MTD of 140 mg freebase (equivalent to 175 mg malate salt) was determined, 20 metastatic or locally advanced MTC pts were enrolled in an MTD expansion cohort. 10/37 MTC pts achieved a confirmed partial response (cPR) (Kurzock et al. 2011). We report here on long-term disease control (DC), defined as cPR or progression-free at ≥ 24 months (m), in the MTC cohort.
"98 of the original 219 cabo arm patients were still receiving treatment. That was 3 years ago. It is unlikely (but certainly possible) that any are still on treatment"
5 of 11 responders in phase 1 MTC population that included a total of 37 patients remained on treatment for median of 4.5 yrs. At the time of ASCO 2013 phase 1 MTC abstract, you can see from quote below that 1 patient was on drug for 73 months(6 years and 1 month) and counting.
"5 of these 11 pts remain on treatment for a median time of 55 m (53-73m)"
Here is the rest
Results: After a minimum follow-up period of 52 m, 11 of 37 pts (30%) experienced DC, and have remained progression-free for 24 m. Time from diagnosis and time since development of metastatic disease was similar in patients with (11) or without (26) long term DC. As of 04 Dec 2012, 5 of these 11 pts remain on treatment for a median time of 55 m (53-73m). A best response of cPR was associated with long-term benefit: 4/5 pts continuing on treatment achieved cPR, one achieved SD. The adverse event profile of cabozantinib was generally predictable and adverse events were managed with symptomatic treatment and dose modifications. The median final dose for pts with long-term DC was 60 mg (range 20-140), with a median number of dose reductions of 2 (range 0E4). Reasons for dose reductions in two or more pts experiencing long-term DC included diarrhea (7), palmar-plantar erythrodysaesthesia syndrome/rash (6), mucositis/stomatitis (3), anorexia (3), nausea (2), and vomiting (2).
"70 HR means there is a 30% risk of death in control arm vs placebo, but is not solely based on median."
30% reduction in risk of death.
"I just wonder if - ultimately - final approval considerations for Cabo in mCRPC may require additional ADCOM guidance with Comet-2 trial results in hand.
I don't think we will have to worry about an ADCOM. If COMET 1 is stat sig for OS, I believe that alone will be sufficient for approval. After all, COMET 1 has an OS primary endpoint. ADCOM's would be a bigger consideration under PFS primary endpoints. I believe Cabo's safety data thus far will not warrant ADCOM upon a stat sig OS signal in COMET 1..
" I can't think of a precedent for such an approval scenario, can you?"
Provenge needed a second trial for approval with guidance. I don't have enough experience following too many biotechs to give a good answer despite giving one example. I have a feeling Provenge may be an outlier on trial with OS primary endpoint though.
"In earlier posts, you (and Ernie, I think) had discussed cabazitaxel approval based upon an HR =0.7, and in looking at the Sanofi website, I see this is also expressed as a range of 0.59 - 0.83. I take this to mean that some of the P3 treatment group outliers were likely not statsig, yet the median was acceptably significant."
The range of 0.59-0.83 represents lower and upper bound range of 95% confidence interval. It basically means that if you did this experiment 100 times, there is a 95% chance the HR will fall within .59-.83.----I think, I don't mind if someone wanted to correct me if I am mistaken.
Median survival is only one point in time on KM curve. The HR takes into account the whole range events on KM curve. There can be a wide discrepancy between control arm and placebo at 70% of events to 50% of events(median) to 30% of events. .70 HR means there is a 30% risk of death in control arm vs placebo, but is not solely based on median. The range of differences between control and placebo can be very different at different points in time. In TROPIC trial for example, the placebo arm actually looked better than control at 2 months. The control arm (Jevtana) was performing worse than placebo for first 3 months. If you look at KM curve of TROPIC, you will notice that Jevtana had the greatest benefit over placebo at approx. 28 months of treatment (tail end of KM survival curve)
"If that separation were uniform on all plots on KM curve, we would be closer to .7 HR at that interim"
To add further-- My guess is that early parts of KM curve are tighter demonstrating less benefit than further out on KM curve. I think part of this has to due with Cabo toxicity and earlier deaths in that arm as a result. As we get more events, they will represent more of the patients that responded much better than earlier patient events. I suspect the KM curves on final analysis will look similar to Jevtana in CRPC. First few months with no separation and the further out on curve, the better the separation. Despite a more tolerable dose in COMET 1, I suspect the same will be said when we finally get to peek at COMET 1 final KM curves.
"If that separation were that large on all parts of plot, the HR would be larger.---I'm thinking"
Not with it today. I meant if Cabo demonstrated 6 month advantage separation on other plots on KM curve, the HR would be much better. Lower HR not higher than .825. If that separation were uniform on all plots on KM curve, we would be closer to .7 HR at that interim.
"If this ratio holds to the required event, will the trial have displayed statsig OS?
A 6 month comparative mOS survival advantage seems adequate to me...
Am I looking at this correctly?"
The most important number is HR, which was .825. I suspect the 6 month Cabo advantage over placebo for the median had one of the largest separations on KM curve plot compared to other plots on graph. If that separation were that large on all parts of plot, the HR would be larger.---I'm thinking
I don't know what is the required HR for stat sig, but remember this is a very small study. The hurdle for stat sig on this study is very large in comparison to COMET 1 for example.
"Do the Lan & DeMets rules somehow apply also to unplanned, unscheduled IA's? As I see it, the interim you refer to was an unplanned unblinding of events by the FDA at a previously unintended event cycle. Is there another set of rules that applies?"
I don't think any alpha was spent on unplanned unblinding done by FDA on June 2012 cutoff. That said, I am not very knowledgable on Alpha spend subject. The FDA was referencing April 6, 2011 cutoff (I may have mistakenly referenced June 2011 rather than actual April 2011 cutoff) with this quote
"An interim analysis of OS was expected to be performed at the 0.00006 level per a Lan-DeMets
O’Brien-Fleming alpha spending function based on an expected 31% information level. The
actual alpha level was based on the actual information fraction at the time of the analysis. If the
result of the interim analysis for OS was not significant, the primary analysis of OS would be
performed when the required number of deaths (217) had been observed. The primary analysis
was expected to be conducted at the 0.04 significance level per the alpha spending function.
The pre-specified interim analysis of OS was performed at the time of the primary analysis of
PFS and included 217 (44%) deaths required for the final analysis of this endpoint. The criterion
for stopping early due to rejection of the null hypothesis at the interim analysis [based upon a
significance level (0.0009)] was not met. No difference between study arms in the estimate of
overall survival was observed"
"Though I realize EXAM has enlisted an advanced patient group, the fact remains that MTC is a slow killer and I suspect many afflicted patients that receive advanced therapy beyond surgery and RT actually manage to outlive the disease. Patience is a requirement here..."
In EXAM population that only included progressive disease, the OS was estimated at approx. 20 months if I recall. Final enrollment concluded on Feb of 2011. At June 2011 cutoff, 96 events occurred out of 330. At June 2012 cutoff, 66 more events occurred in 1 years time frame. Two years have passed needing only 55 more events. We are awfully close IMHO. I wouldn't be surprised if they are cleaning data as we speak.
FWIW median OS on Cabo arm was 26 months and Placebo at 20 months at 162 events of the total 217 needed for final analysis.
"Below is quote from safety review with EXAM "
typo meant first interim after 96 OS events. The updated second (unplanned interim cutoff 6/15/2012) had 162 OS events broken down as follows
Cabo N=219 Placebo N=111
Cabo events=103(47%) Placebo events=59 (53%)