Thanks to GRACE member speedpuppy for highlighting the link to a New York Times article on the one of the key obstacles to meaningful progress in cancer management being the difficulty of getting clinical research completed. It discusses challenges from both the patient and physician perspective. Only about 3% of patients in the country enroll on clinical trials, and in lung cancer it’s lower than that. Not suprisingly, clinical trial participation is heavily skewed toward academic centers, so many community practices participate in little or no clinical research.
There are certainly challenges from the physician perspective. The leading one is that clinical research is generally a financially disadvantageous option for a physician and a practice. Clinical trials tend to take much more time per patient than managing patients outside of clinical research. They often provide the same or less money than the cost of running the trial for each patient. They also require more attention to following the precise treatment plans of a protocol. Larger research centers require extra nurses and administrative staff to oversee the details of clinical trials, and this is a significant investment. And the paperwork involved is a royal pain, requiring time that many docs just don’t feel that they have.
From the patient perspective, some people are very wary of being a “human guinea pig” in clinical research (a mindset fostered by some media outlets trying to generate controversy), and in the US in particular, patients are often extremely wary about participating in clinical trials with a placebo for some patients, even though a placebo-controlled trial is often the best way to really learn how effective vs. toxic a new treatment is. In truth, many of the most important trials in cancer are now done outside of the US, where more patients are willing to accept some uncertainty, and partly because the alternatives may be less favorable. Importantly, trials may require considerable effort to find, and they often entail travel to participate, for unclear benefits. Finally, it’s also important to note that many patients who want to participate in trials are unable to do so because they aren’t eligible, often due to prior treatments or other medical problems. A recent study that did a chart review in the Kaiser system in California found that only 15-30% of patients there were eligible for various clinical trials in advanced NSCLC.
The idea for restricting eligibility is to provide a more homogenous population of patients with an anticipated similar prognosis. However, trials often include restrictions that are unnecessarily limiting. This can leave patients without options and slows the pace of our progress in the field.
The article also highlights the problem that some trials instigated by pharmaceutical companies don’t ask research questions that move the field forward as much as promote a marketing question, such as whether a weekly or an every three week schedule for drug X is better for patients. It’s absolutely true that some studies are tantamount to a drug company paying a hospital or practice to have the oncologists begin to use and gain experience with (and perhaps a preference for) their agents.
Other posts have also highlighted the burden of regulation and paperwork involved with opening a trial. This hyper-regulation doesn’t end when the trial is opened but continues throughout the research process. Though well-intended to guarantee safeguards, the system is a patchwork of ever-increasing legal interests, and it leads to our near paralysis in opening and conducting trials.
To summarize, it’s absolutely true that one of the major reasons our progress in cancer management has been so slow is that it’s difficult to conduct clinical research, especially in the US, and especially now. Oncologists don’t have a good incentive to do it — they consume more time and more money, and they entail burdensome paperwork. Patients are all too often restricted from being permitted to participate, often have to go to great lengths and distances to find a potential clinical rial. And many good candidates for clinical trials decline to participate because of a wariness of “clinical research” and the uncertainty of the value of a treatment on a protocol. Meanwhile, the regulatory environment of our current clinical trials system, while well-meaning, is soul-crushingly bureaucratic and time-consuming.
At the end of the day, when clinical research is what moves the field forward and only 1-2% of lung cancer patients participate in trials, it’s clear that we need to do better.
Posted on August 4, 2009 at 6:48 pm
Okay…I have been holding back on this so as not to introduce a real tangent into the serious business we are concerned with here, but the topic is ultimately inescapable and this is as good a place and time as any. So, here goes:
The current clinical testing protocol for approving new medicines is rigidly disciplined by experimental statistics for good reasons that we all grudgingly respect. BUT, where is the statistical proof that the current testing protocol is the best protocol? The existing protocol was developed “inside-the-box” and the statistics applied to it are also trapped in that box.
We have the mathematical techniques to discover the best protocol and prove it is best – using the same rules of statistics, but allowing them to roam a bit. This type of analysis started in the year I was born (1950) with the development of General Systems Theory.
http://en.wikipedia.org/wiki/Ludwig_von_Bertalanffy
Systems models are mathematical simulations of real world processes. I recall seeing some interesting applications in the fields of population and evolutionary biology. In these applications researchers try to unravel the reason that some species survive and others fail by testing different assumptions about key traits that lend survival advantages. The way it works is that rules of behavior, competitive advantage and genetic change are defined mathematically to a computer and the computer can then test them over and over through hundreds or thousands of generations (called “trials”) to see which species survive best. If the same one wins 95% of the time, it is a pretty clear winner. If it wins only 67% of the time, it’s not so clearly a winner.
http://en.wikipedia.org/wiki/Evolutionary_computation
These computational techniques (often called system dynamics or computational modeling) have spread to many areas.
http://en.wikipedia.org/wiki/Computational_biology
In economics, the field of agent-based computational economics has emerged in which long-held theories of economic behavior can be challenged by trying various rules for economic agents (consumers and firms) to follow and seeing if they match actual economic outcomes over a number of economic transactions (“trials”).
http://en.wikipedia.org/wiki/Agent-Based_Computational_Economics
Applications of this type of modeling in the social sciences, have the intriguing additional feature of allowing the modeling rules to incorporate goal seeking behavior by intelligent critters who also have the capacity to share knowledge and collaborate with one another.
With me so far? See where this goes? What if we modeled the following challenge:
One group of simulated cancer patients is compelled to wait for medicines to be approved through the current process. Another group is allowed to roam free – gathering information and choosing treatments as they please. Which group will discover the best medicines first and advance the science fastest?
I am not suggesting we switch to a free-for-all approach tomorrow. That would be reckless – thousands of people would receive questionable treatments and many would die. BUT, these same things happen under the current testing protocol while many others die waiting.
I am suggesting only that it might be worthwhile to ask a supercomputer to explore this challenge and see what can be learned by looking outside the box – and doing so in a way that produces only computer simulation outcomes as opposed to real ones. It may be that some hybrid approach turns out best – say an approach that frees up the choices after following protocol for some part of the way.
I apologize for such a long post that may only make sense to a biologist/economist nerd like me. I recently ran across encouraging evidence of others on this trail.
http://www.medicalnewstoday.com/articles/157114.php
I would love to hear that such research has already been done somewhere and the statistical validity of the current testing and discovery protocol has already been firmly established.