Cancer Drugs, Survival, Research Priorities, and the Human Condition

It is fundamental to human nature to hope, and one manifestation of this aspect of human nature is the desperate but understandable desire of cancer patients with very grave prognoses to look for a glimmering possibility of beating the odds. In this post I will look closely at the data on an anti-cancer antibody may have incrementally raised the bar in the treatment of one particularly grave cancer and discuss how this and other similar cancer drug development fits within the current framework of healthcare delivery and reimbursement.  

Download PDF

It was recently announced that the RAINBOW trial of the effectiveness of Eli Lilly’s anti-VEGFR-2 antibody Ramucirumab as a second-line treatment for gastric cancer has yielded positive “top-line” results.  Gastric cancer is essentially untreatable with very limited survival times. In a study of 665 patients the “RAINBOW” study found that Ramucirumab together with paclitaxel resulted in an increase in median overall survival compared with paclitaxel alone: 9.6 months for the combination and 7.4 months for paclitaxel alone.

My purpose in this post is not to debate the value of the drug. My purpose is also not to question the meaning of a two-month survival advantage in a terminal disease, an important issue, but there are other forums and other kinds of experts to address it. Instead, I want to focus on the absolute significance of the data for pharmaceutical policy.  It is not at all clear to me that the results of the RAINBOW study settle the question of whether or not Ramucirumab has a significant effect on overall mortality. In reaching this tentative conclusion, I have looked at other studies of paclitaxel’s effect on gastric cancer as well as a gastric cancer clinical trial of another antibody, Avastin, that also works by blocking the effect of VEGF (albeit by binding to VEGF rather than a VEGF receptor).  Taken together, the data from the Avastin trial and other studies of paclitaxel show just how the gold-standard endpoint of overall survival varies from trial to trial and raise the question of whether the effectiveness of Ramucirumab in this trial is a durable, replicable clinical effect or a statistical artifact. 

First, comparing the RAINBOW study to the Avastin study shows a remarkably similar effect of the two drugs, though one was deemed a “failure” (because of a p-value >.05) and the other hailed as a “success” (because of a p-value of <.05). Ramucirumab impedes the formation of new blood vessels by blocking one of the receptors for VEGF (vascular endothelial growth factor).  Thus Ramucirumab works on the same basic pathway as Avastin, an antibody which binds directly to VEGF to prevent it from binding to its receptors. Avastin was tested as a first-line therapy for gastric cancer in a study of 774 patients.  In that trial, the experimental group received Avastin in combination with fluoro-pyrimidine (FP) and cisplatin, while the control group received a placebo in combination with FP and cisplatin. The Avastin group achieved a median survival of 12.1 months compared to a median survival of 10.1 months in the control group, a result that produced a p-value of .1002.  On that basis, Avastin was NOT proven effective as a first-line treatment for gastric cancer.  The two-month increase in survival in the slightly larger Avastin trial as a first-line therapy was not statistically significant while the 2.2 month difference in the slightly smaller RAINBOW trial, looking at a later-stage population with a correspondingly shorter median survival, was statistically significant.  Both drugs added about two months to the median survival of their experimental groups, but that two months was a statistically significant gain in the shorter relative time frame (7.4 to 9.6 versus 10.1-12.1).

Also clouding the question of the effectiveness of Ramucirumab in the RAINBOW trial is the data from other studies of paclitaxel in gastric cancer. A 2009 review article by Jinichi Sakamoto et al provided the results of several trials of paclitaxel for gastric cancer, as a first-line treatment or as a second-line treatment, and as a single agent or as part of a combination therapy.  Sakamoto’s review article cited a small study (36 patients) by Cascinu et al in which paclitaxel used in second-line treatment as a single agent provided a median survival time of 8 months, which of course is markedly longer than the paclitaxel group in the RAINBOW trial.  Sakamoto’s article also included another small study of paclitaxel as a second-line therapy for gastric cancer in combination with another agent (which was how it was used in the RAINBOW trial).  In the 2007 32-patient study by Lee et al, paclitaxel plus cisplatin achieved a median survival of 9.1 months (very close to the same result achieved in combination with Ramucirumab in the RAINBOW trial). 

There are numerous instances in clinical trials where the failure of a treatment to demonstrate effectiveness may be attributed to the fact that the control group did better than expected: a stent to prevent strokes, an antiviral for prophylaxis of CMV,  an experimental agent for lowering triglycerides, and even a trial of Lipitor versus Zocor.  Of course no drug company would ever announce that its drug was shown to be effective because the control group did worse than expected, but of logically must sometimes be the case.  It may or may not be the case here.

What is clear is that in developing Ramucirumab for gastric cancer, Eli Lilly could not possibly have been aiming to achieve a dramatic breakthrough in the treatment of gastric cancer.  Rather, Eli Lilly spent a fair amount of money (certainly more than $20 million) in the hopes of adding to the market for the drug.  That is certainly a rational choice, and I hardly fault Eli Lilly for its decision.  After all, even though generic cisplatin and paclitaxel may perform as well (as in the smaller 32-patient study cited by Sakamoto), cisplatin is not labeled for use in gastric cancer.  This makes reimbursement for Ramucirumab much more likely if the FDA is persuaded to add the second-line treatment of gastric cancer to the drug’s label. However, it does raise the question of whether or not the current system of healthcare and healthcare reimbursement overly incentivizes the development of incrementally effective drugs as opposed to more high-risk innovative drugs that present the potential for significant increases in therapeutic effect.

I saw an advertisement recently in the Los Angeles Times that carried the heading “Every Cancer Treatment Begins With Can.”  The advertisement was placed by the Los Alamitos Medical Center, but similar ads, with similar positive messages, are run by many major medical centers and cancer treatment centers.  I understand why those ads are used and I also understand why people with diagnoses of serious cancers respond to them and want that message.  But the reality is that for patients with the advanced stage of pancreatic cancer, or gastric cancer, or glioblastoma multiforme, the chances of surviving five years are extremely low, and the adverse effects of treatment are relatively severe.  But if it is not a fundamental aspect of human nature, it is certainly a fundamental aspect of the American character, to look for any possible ray of hope and, when reality is a 5% chance of survival, to fight for that 5% chance.  I am not, by any means, advocating euthanasia, or advocating that we deny cancer patients viable medical treatment options.  I am simply questioning the structure of a healthcare system that provides large incentives for small gains and somewhat lower incentives for the research that might result in large gains.  It is the human condition to hope, and I hope we do indeed win additional great victories in the “War on Cancer.”  However, we may have to tinker some with our healthcare system to get there.