ImClone's Next Blockbuster
By Adam Feuerstein
Senior Writer
3/19/2007 7:16 AM EDT
URL: thestreet.com
ImClone Systems (IMCL) and its cancer drug Erbitux get another big test with results expected soon from a pivotal study in pancreatic cancer.
Erbitux is currently approved to treat colon and head and neck cancers. ImClone and partner Bristol Myers Squibb (BMY) will add pancreatic cancer to that list if the data is positive from this phase III trial. Results are expected in late March or early April.
OSI Pharmaceuticals (OSIP) also is watching the Erbitux trial closely because the results could possibly hurt its cancer drug Tarceva, which doctors use now to treat pancreatic cancer.
An especially aggressive malignancy, pancreatic cancer is difficult to diagnose and manage. Because most patients with pancreatic cancer are diagnosed late, current treatments aren't especially effective, and life expectancy is short.
If Erbitux is shown to prolong survival in pancreatic cancer patients, the upside to the drug's peak sales could range from $100 million to $250 million, according to a sensitivity analysis done by JP Morgan biotech analyst Geoff Meacham.
On the flip side, about $50 million in current off-label Erbitux sales in pancreatic cancer are at risk if the study fails.
Meacham doesn't see much upside for OSI -- and Tarceva sales -- coming from the Erbitux trial, but there is a lot of downside risk. Tarceva sales could fall by as much as $100 million if the Erbitux data shows the drug outperforming Tarceva. Meacham rates ImClone neutral and OSI underweight. His firm has banking relationships with both companies.
ImClone shares closed Friday up a penny to $30.03. OSI shares closed up 17 cents to $31.92.
The Southwest Oncology Group, a research cooperative of cancer doctors, is conducting the Erbitux study, which enrolled 704 patients with newly diagnosed pancreatic cancer. Patients in the study were randomized evenly to receive a combination of Erbitux plus the chemo drug gemcitabine, or a placebo plus gemcitabine. The primary efficacy endpoint of the study is overall survival.
In 2005, the Food and Drug Administration approved OSI's Tarceva for use in pancreatic cancer based on a phase III study that showed that a combination of Tarceva and gemcitabine prolonged median survival by a mere 13 days compared with patients who received a placebo plus gemcitabine. Results from this study are in the chart below:
Tarceva plus gemcitabine in pancreatic cancer
Treatment-naive patients
Tarceva+gemcitabine Placebo+gemcitabine p value
No. of patients 260 261
Response rate 8.60% 7.90% p=0.87
Median survival 6.4 mos. 6.0 mos. p=0.028
Progression-free survival 3.8 mos. 3.5 mos. p=0.006
Source: FDA
Erbitux won't need to beat Tarceva to gain FDA approval, but it will have to equal or better its rival in order to gain traction among doctors and patients. Erbitux's challenge also is complicated by the fact that it's more expensive than Tarceva.
As it stands now, Tarceva has had some difficulty cracking the pancreatic cancer market, because many doctors and patients don't believe the benefit of the drug (another 13 days of life) is worth its cost -- or its side effects. The majority of pancreatic cancer patients eligible for chemotherapy today receive gemcitabine alone as their primary treatment.
There is positive data from an earlier study of Erbitux in pancreatic cancer that does give some optimism to the upcoming phase III trial.
In 2004, 41 patients were enrolled in a single-arm phase II study and given a combination of Erbitux plus gemcitabine. Median survival in the study was 7.1 months. But these results need to be interpreted carefully since the study lacked a co-arm.
But the ability of pancreatic cancer to stymie even the most advanced cancer drugs should not be underestimated. Last year, a similar phase III study of Genentech's (DNA) Avastin plus gemcitabine failed to show any significant survival benefit, despite earlier clinical evidence pointing toward Avastin's efficacy in the disease. The list of failed studies in pancreatic cancer is quite long, in fact. |
