Slightly more than a year ago, I got a call from a recruiter. She was looking for someone who could help a company file its NDA for a new anti-inflammatory compound. Having been involved in two phase III programs for osteoarthritis (OA), and having presented data to the FDA, I was pretty sure I could help the company, NicOx. However, I never heard from the company, and I moved on. So did the NicOx, but apparently not for the better. In September 2009, it filed an NDA for its lead compound, Naproxcinod, the indication being OA. And on May 15, it heard from the advisory committee that voted 14 to 1 (abstention) not to recommend approval of Naproxcinod for the treatment of OA. What happened, and what can we learn from it?
Science first. Naproxcinod is supposed to be a nitric oxide-donating anti-inflammatory compound and the first-in-class CINOD (Cyclooxygenase-Inhibiting Nitric Oxide Donator). The whole point of the nitric oxide donation is that nitric oxide causes vaso-dilatation, thereby decreasing the blood pressure and nullifying the cardiac side effects of non-steroidal anti-inflammatory compounds (NSAIDS) such as Naproxen which do their damage by their effect on prostaglandins. NicOx also wanted to show that Naproxcinod had better gastrointestinal tolerability and safety than Naproxen, thanks to the same nitric oxide. If this was to work out, NicOx would be have a truly competitive product.
In order to show just that, NicOx completed three clinical phase III studies – two for knee OA (301 and 302 studies) and one for hip (303 study), all for 13 weeks only. Cardiovascular outcomes data was collected till 52 weeks. The primary goal was the treatment of signs and symptoms of OA, but secondary goals included observing an effect on BP and endoscopic evaluation for ulcers and erosions. In study 301, two doses of Naproxcinod (750 mg bid and 375 mg bid) were superior to placebo but not to Naproxen (500 mg bid) in the WOMAC scores for OA. In study 302, Naproxcinod 750 mg bid was superior to placebo and Naproxen but the 375 mg dose was only superior to placebo. In study 303, Naproxcinod 750 mg bid was superior to placebo and essentially equivalent to Naproxen (the 375 mg bid dose was not studied). Since only the 750 mg bid dose was studied in two OA populations, it can be considered for filing. Looking at the FDA documents provides another look at the efficacy. I won’t go into the statistical mumbo jumbo, but there are several questions on the parameters the company chose to define efficacy. Total phase III exposure of the 750 mg bid dose was only, and I repeat only, 789 patients.
On the safety side, where the effect on BP and GI irritability comes in, the AE profile tells its own story. Among serious adverse events, or SAEs, Naproxcinod had an incidence of 0.3% (GI) and 0.2% (cardiac), worse than the 0.3% (GI) and less than 0.1% (cardiac) for Naproxen. Among all adverse events (not just serious), Naproxcinod 750 mg bid had an incidence of 56.4%, barely better (certainly not statistically better) than 58% for Naproxen.
This leads the FDA advisory committee to say “Based on findings from the two studies, there does not appear to be replicated evidence to support that naproxcinod has similar efficacy to naproxen” and conclude
1. Naproxcinod, at doses of 750 mg and 375 mg twice a day, is efficacious in the relief of signs and symptoms of osteoarthritis.
2. Naproxcinod has not been demonstrated to be non-inferior to naproxen.
3. The general safety profile of naproxcinod is consistent with that of the NSAID drug class.
In other words, maybe just as good as Naproxen for the treatment of OA, and then maybe not even as good. The cardiovascular division also had its say when it concluded:
1. The BP effect due to naproxcinod was not consistently less than baseline through the dosing interval. Naproxcinod is therefore not approvable as an antihypertensive agent.
2. Naproxen had a predominantly pressor effect through the dosing interval. Relative to equimolar doses of naproxen, with naproxcinod there appears to be a consistent lowering effect on SBP and DBP at peak (i.e. 1-4 hours post-dosing) but not at the end of the dosing interval.
3. More than two-fold changes in peak-trough effects were noted in some ABPM recordings suggesting that the proposed dosing regimen does not have a consistent effect throughout the dosing interval.
In other words, there was a short term beneficial effect on the blood pressure (1-4 hours after taking drug) but the benefit did not last through the day. Ironically, the committee was concerned about hypotension manifested as dizziness. Therefore, not only did they not approve a label claiming a cardiovascular benefit, but wanted a warning for hypotension, similar to that seen for nitroglycerin.
Finally, the division of gastroenterology commented on endoscopic studies 0002, 0005, and 0027 which were, at most, 6 weeks in length, comparing Naproxen, Naproxcinod, and placebo. Unfortunately, healthy volunteers were studied for two weeks in two studies, and erosions were the primary endpoint (the division likes six months in patients and ulcers, respectively). Only one study was in patients with ulcers as the primary endpoint and the study was for 6 weeks. In this study, the Naproxcinod 750 mg bid/Naproxen 500 mg bid ratio of ulcers was 0.7, with a p value of 0.07 (i.e. not statistically significant).
Compare the quality and quantity of data presented with that for Celecoxib, a drug that has the label the NicOx wanted. Specifically look at the duration of the studies, the primary endpoints and the patient numbers involved (source: Pfizer webpage)
Celecoxib Long-Term Arthritis Safety Study (CLASS) was a prospective long-term safety outcome study conducted in approximately 5,800 OA patients and 2,200 RA patients. Median exposures for Celexoxib (n = 3,987) and Diclofenac (n = 1,996) were 9 months while ibuprofen (n = 1,985) was 6 months. The primary endpoint of this outcome study was the incidence of complicated ulcers (gastrointestinal bleeding, perforation or obstruction).
Endoscopic Studies: A randomized, double-blind study in 430 RA patients was conducted in which an endoscopic examination was performed at 6 months. The incidence of endoscopic ulcers in patients taking Celecoxib 200 mg twice daily was 4% vs. 15% for patients taking Diclofenac SR 75 mg twice daily. The incidence of endoscopic ulcers was studied in two 12-week, placebo-controlled studies in 2157 OA and RA patients in whom baseline endoscopies revealed no ulcers.
What can we learn? Could things have been done differently? If NicOx had asked me a year ago, here is what I would have told them:
1. Naproxcinod would be approved if they went in purely as an NSAID since the data do support such a claim.
2. The endoscopic studies were too short, used the wrong kinds of subjects, and the wrong endpoints. There are enough studies showing the way (which is why I showed the Celecoxib data), and clear FDA guidelines, so I’m surprised that the company proceeded the way they did. It was clear that they were unlikely to get a claim for better GI tolerability. What they could have done was to get approval as an NSAID and to undertake these studies in phase IV. Better than getting an NDA rejected.
3. The cardiovascular effect was not sustained label was not likely. Once again, there are FDA guidelines, and precedence of other compounds, so I don’t see why the company chose the path it did. Especially with the data showing lack of sustained lowering of blood pressure.
Monday, May 24, 2010
Friday, May 7, 2010
A clinical look at Human Genome's lupus drug
On Tuesday, April 20, Human Genome Sciences (HGSI) released longer term data from its Bliss 76 Phase III study in Lupus, leading to a stock drop. First some back ground.
HGSI is developing Belimumab, an antibody to immature B cells, for the treatment of SLE by targeting the B Lymphocyte Stimulator (BLyS). Treatment with the antibody prevents B cells from maturing into auto-antibody secreting plasma cells. The Phase II data had been insipid, but data analysis by the company showed that patients who were sero-positive (i.e. HEp-2 ANA ≥ 1:80 and/or anti-dsDNA ≥ 30 IU/mL) were better responders than sero-negative patients. The Phase III studies were designed with that in mind. This is an important point that the reader should understand since not all SLE patients are sero-positive as defined in this trial - about 50% of patient in my practice are, and others report rates between 50 and 80% (which shows the heterogenous nature of the disease).
Two Phase III trials were conducted, one a 52 week trial (BLISS 52) in Asia, South America, and Eastern Europe, and the other a 76 week trial (BLISS 76) in North America and Western Europe. Doses looked at were 10 mg/kg and 1 mg/kg on a background of standard of care. At 52 weeks, both studies reported positive data with the 10 mg/kg dose. The response rates are below:
STUDY 10 MG/KG PLACEBO DIFF
BLISS 52 57.6 43.6 14
BLISS 76 43.2 33.8 9.4
Based on the above, it is clear that there is enough evidence to approve the drug. This was truly great news for us who have had no new treatment for our lupus patients in the past 50 years.
Now to the news from April 20. The 76 week data from BLISS 76 was released, and the company said 38.5% of patients responded to the 10 mg/kg dose (down from 43.2% at 52 weeks, a drop of 4.7%) compared to 32.4% of patients treated with a placebo (down from 33.8% at 52 weeks, a drop of 1.4%). The difference between the 10 mg/kg dose and placebo was not statistically significant. More curiously, the previously less appealing dose of 1 mg/kg was in many measures better than the 10 mg/kg dose. So, what is going on? And what are its implications?
It is unclear why the response rates dropped with time. Was it due to antibodies to the antibodies? After all, lupus patients are very adept at making antibodies. Or was it due to tachyphylaxis? Or some other reason?
The implications of this phenomenon are significant. Lets do the maths. Of my 20 patients who can be treated with Belimumab, lets say I put them all on it at (speculatively) $20,000 a year. That nets the company $400,000. After a year, I have to stop treating them, and can only use Belimumab on new patients (lets say about 4). Now the company is netting only $80,000, down from $400,000. But there is another, more important point. This relates to the question of why the effect wears off - does the effect, in fact, go into reverse? In other words, will there be a rebound? Will the disease, once the Belimumab is stopped, become worse than it was originally? Some may argue that no such rebound was seen in the clinical studies, but one must keep in mind that patients are not followed beyond a week or so after their last visit, so a rebound would not be captured by the study. So while the PI (principal investigator at the site) would know about it (assuming s/he knew the randomization), the companies would not.
Compare that to the anti-TNF story - one of the great advantages of anti-TNFs such as Enbrel (Pfizer and Amgen) and Humira (Abbott) used in the treatment of RA is that their efficay does not wane with time. No wonder they have 8 - 10 year safety and efficacy data. Similar long term safety and efficacy data would be very useful for Benlysta.
HGSI is developing Belimumab, an antibody to immature B cells, for the treatment of SLE by targeting the B Lymphocyte Stimulator (BLyS). Treatment with the antibody prevents B cells from maturing into auto-antibody secreting plasma cells. The Phase II data had been insipid, but data analysis by the company showed that patients who were sero-positive (i.e. HEp-2 ANA ≥ 1:80 and/or anti-dsDNA ≥ 30 IU/mL) were better responders than sero-negative patients. The Phase III studies were designed with that in mind. This is an important point that the reader should understand since not all SLE patients are sero-positive as defined in this trial - about 50% of patient in my practice are, and others report rates between 50 and 80% (which shows the heterogenous nature of the disease).
Two Phase III trials were conducted, one a 52 week trial (BLISS 52) in Asia, South America, and Eastern Europe, and the other a 76 week trial (BLISS 76) in North America and Western Europe. Doses looked at were 10 mg/kg and 1 mg/kg on a background of standard of care. At 52 weeks, both studies reported positive data with the 10 mg/kg dose. The response rates are below:
STUDY 10 MG/KG PLACEBO DIFF
BLISS 52 57.6 43.6 14
BLISS 76 43.2 33.8 9.4
Based on the above, it is clear that there is enough evidence to approve the drug. This was truly great news for us who have had no new treatment for our lupus patients in the past 50 years.
Now to the news from April 20. The 76 week data from BLISS 76 was released, and the company said 38.5% of patients responded to the 10 mg/kg dose (down from 43.2% at 52 weeks, a drop of 4.7%) compared to 32.4% of patients treated with a placebo (down from 33.8% at 52 weeks, a drop of 1.4%). The difference between the 10 mg/kg dose and placebo was not statistically significant. More curiously, the previously less appealing dose of 1 mg/kg was in many measures better than the 10 mg/kg dose. So, what is going on? And what are its implications?
It is unclear why the response rates dropped with time. Was it due to antibodies to the antibodies? After all, lupus patients are very adept at making antibodies. Or was it due to tachyphylaxis? Or some other reason?
The implications of this phenomenon are significant. Lets do the maths. Of my 20 patients who can be treated with Belimumab, lets say I put them all on it at (speculatively) $20,000 a year. That nets the company $400,000. After a year, I have to stop treating them, and can only use Belimumab on new patients (lets say about 4). Now the company is netting only $80,000, down from $400,000. But there is another, more important point. This relates to the question of why the effect wears off - does the effect, in fact, go into reverse? In other words, will there be a rebound? Will the disease, once the Belimumab is stopped, become worse than it was originally? Some may argue that no such rebound was seen in the clinical studies, but one must keep in mind that patients are not followed beyond a week or so after their last visit, so a rebound would not be captured by the study. So while the PI (principal investigator at the site) would know about it (assuming s/he knew the randomization), the companies would not.
Compare that to the anti-TNF story - one of the great advantages of anti-TNFs such as Enbrel (Pfizer and Amgen) and Humira (Abbott) used in the treatment of RA is that their efficay does not wane with time. No wonder they have 8 - 10 year safety and efficacy data. Similar long term safety and efficacy data would be very useful for Benlysta.
Health Care "Reform" Bill and Biotechnology companies
Buried in the pages of the recently passed Health Care "Reform" Bill are some worthy incentives for small biotechnology companies. They are mentioned under Therapeutic Discovery Project Tax Credit. The program is similar to the tax credits for investments in certain energy projects. While the guidelines have not been fully vetted, the program provides a credit in the amount of 50% of investments in qualified therapeutic discovery projects and qualifying investments made in 2009 and 2010. It is limited to small companies, as defined by having fewer than 250 employees and covers projects that are designed to:
1. Treat or prevent diseases or conditions by conducting pre-clinical studies activities, clinical trials, or carrying out research protocols, for the purpose of securing approval of a drug or a biologic.
2. Determine the molecular factors related to diseases or conditions by developing molecular diagnostics to guide therapeutic decisions, or
3. Develop process, technologies, or products to further the delivery or administration of therapeutics.
While the "Reform" Bill has much that is lamentable, this little tidbit is certainly good news for the small biotechnology and pharmaceutical companies.
1. Treat or prevent diseases or conditions by conducting pre-clinical studies activities, clinical trials, or carrying out research protocols, for the purpose of securing approval of a drug or a biologic.
2. Determine the molecular factors related to diseases or conditions by developing molecular diagnostics to guide therapeutic decisions, or
3. Develop process, technologies, or products to further the delivery or administration of therapeutics.
While the "Reform" Bill has much that is lamentable, this little tidbit is certainly good news for the small biotechnology and pharmaceutical companies.
Subscribe to:
Posts (Atom)