That's because the "positive" data from a small phase II clinical trial involving Corbus' only drug resunab are nowhere near as robust or convincing as the company claimed them to be Monday, said a health care investor short Corbus.
(You met this investor and Corbus bear in October, when I wrote a story explaining why Corbus was wasting its time developing resunab. He believes resunab doesn't work. Nothing presented by the company Monday changes his mind.)
Corbus shares closed Monday 51% higher to $8.85 on the announcement of the resunab phase II results in systemic sclerosis, an autoimmune disease that causes the accumulation of fibrotic tissue under the skin and in the lungs. At one point Monday, the market value of Corbus almost doubled but the stock sold off later in the day.
The sclerosis patients treated with resunab reported a median 33% response rate over the 16 weeks of the trial compared to a median 0% response rate for placebo patients, achieving the primary endpoint with statistical significance, Corbus said.
On a conference call Monday morning, the doctor who acted as the principal investigator of the study described the resunab results as "breathtaking."
Why shouldn't these results be trusted? Because the sclerosis patients who received resunab were healthier at the start of the trial compared to the patients treated with placebo, Corbus' own data show.
The imbalances in patient baseline characteristics provided an assist to the resunab treatment response and undermines the credibility of the trial results, my investor source said.
You don't have to believe him, of course. Corbus, to its credit, provided investors with enough data from the trial for them to make up their own minds. Slides from Corbus' resunab data presentation can be downloaded here. Links below are to individual slides in the Corbus resunab data presentation. Follow along for a detailed and critical look at the resunab data.
The Corbus study enrolled 43 sclerosis patients, randomized 2:1 to receive resunab or placebo.
On average, the patients in the resunab arm had sclerosis for 38 months, a shorter duration than the 41 months for placebo patients. Ninety-three percent of patients in the resunab arm were being treated with beneficial immunosuppressive drugs before the study began compared to 80% of the placebo patients.
Systemic sclerosis thickens the skin and worsens lung function of patients. On both measures, patients randomized to the resunab arm were healthier than those in the placebo arm at the outset of the study.
At baseline, the modified Rodnan skin score (a measure of skin thickness) averaged 23.5 points for resunab patients and 26.2 points for placebo patients. A higher Rodnan skin thickening score describes more severe disease, so at the start of the trial, resunab patients enjoyed a 2.7-point advantage over the placebo patients.
After 12 weeks of treatment in the trial, where efficacy measures were supposed to be calculated, resunab patients reported an approximate Rodnan score improvement of 3.9 points compared to 2.7 points for placebo patients. That's a difference of 1.2 points favoring resunab at the end of the trial -- smaller than the 2.7-point cushion provided at baseline. (Corbus provided investors with a graphical depiction of certain efficacy data without providing actual numbers. The numbers here are estimated from those Corbus graphs.)
At baseline, the forced vital capacity, or FVC, a measure of lung function, was
and 79.6% for placebo patients. A higher FVC means better lung function. Resunab patients had a 6.4 percentage point FVC advantage over placebo patients before the trial started.
After 12 weeks of treatment in the study, the FVC of resunab patients improved by approximately 0.85 percentage points. The FVC of the placebo patients declined by 0.75 percentage points. The FVC difference of 1.6 percentage points favored resunab, but again was smaller than the 6.4 percentage point difference at the start of the trial. It's also interesting to note -- and concerning -- that FVC in the resunab arm started to move sharply lower after patients were transitioned to the higher dose of the drug during the trial. By week 16, resunab patients were also showing decreased FVC compared to baseline.
The treatment difference between resunab and placebo on reduction of skin thickening and improvements in lung function weren't statistically significant over the course of the trial.
If resunab failed to improve the two most objective measures of efficacy for sclerosis patients, how was Corbus able to show a 33% to 0% median treatment response difference and declare the study a success?
Corbus was aided by the use of a novel, unproven composite efficacy scale. The Combined Response Index in diffuse cutaneous Systemic Sclerosis (CRISS) is an algorithm that takes into account subjective quality of life improvements reported by patients and doctors along with measurements of objective disease improvements.
Corbus is the first company to use the CRISS scale to develop a sclerosis drug. CRISS has not been validated by the U.S. Food and Drug Administration as an approvable efficacy endpoint for a drug aimed at treating sclerosis.
Dig into Corbus' study results to look at the individual components of the CRISS score and you'll see resunab performed better when patients were asked to assess how they felt during the trial. The FDA, generally, doesn't approve drugs based on quality of life data.
Corbus measured treatment response over 16 weeks -- 12 weeks of treatment and four weeks of follow-up when resunab and placebo weren't administered. The study's original design called for efficacy to be measured at 12 weeks, according to ClinicalTrials.gov. Strangely, on some of the efficacy measures, response to resunab increased from 12 weeks to 16 weeks when patients weren't on the drug.
By the way, the 0% treatment response rate in the placebo arm doesn't mean that none of the placebo patients responded. Corbus reported median results. A more detailed view of responses on a per-patient level shows that some placebo patients showed 100% improvements in the CRISS score, as did resunab. There were also some patients who didn't respond to resunab, performing as poorly as some placebo patients.
At risk of going too deep in the weeds on statistics, Corbus' "p value" for statistical significance was 0.044, based on a one-sided test. That equates to a more traditional two-sided "p value" of 0.088, which would not be statistically significant.
The statistics underlying the Corbus study were stacked in resunab's favor, just like the baseline characteristics of the enrolled patients.
Few things in drug development are more fungible and fraught with risk than the interpretation of data from small phase II studies. The details are important because all too often drugs that appear to be tremendously beneficial to patients in phase II studies end up doing nothing in larger, more rigorous phase III studies.
OncoGenex (OGXI) , Alcobra (ADHD) , CytRx (CYTR) , Newlink Genetics (NLNK) - Get Report , Endocyte (ECYT) - Get Report and Celldex Therapeutics (CLDX) - Get Report -- and their shareholders -- learned this lesson the hard way. There are more such companies. It's a long list.
It will be awhile before the true activity of resunab in sclerosis patients is known. Nothing is going to stop Corbus from advancing the drug into a larger phase III study, but the outcome of that study is years away. On Monday, Corbus presented the resunab study data in a way that highlighted the positives. To be short the stock, like my health care investor source, was a painful experience. That doesn't mean resunab is in the clear. There are plenty of red flags that place the drug's future at risk. To see them takes work many investors don't want to do.
Adam Feuerstein writes regularly for TheStreet. In keeping with company editorial policy, he doesn't own or short individual stocks, although he owns stock in TheStreet. He also doesn't invest in hedge funds or other private investment partnerships. Feuerstein appreciates your feedback; click here to send him an email.