New Generation of Clinical Trials Could Improve Patient Care

[gpawelski]

Donald Berry, Ph.D., professor and chair of the Department of Biostatistics and Applied Mathematics at M.D. Anderson Cancer Center stated in the January 2006 issue of Nature Reviews Drug Discovery, the statistical method used nearly exclusively to design and monitor clinical trials today (the frequentist method) is so narrowly focused and rigorous in its requirements that it limits innovation and learning.

He advocates adopting the Bayesian methodology, a statistical approach that is more in line with how science works. It is used routinely in physics, geology and other sciences. And he has put the approach to the test at M.D. Anderson, where more than 100 cancer-related phase I and II clinical trials were being planned or carried out using the Bayesian approach.

The main difference between the Bayesian approach and the frequentist approach to clinical trials has to do with how each method deals with uncertainty, an inescapable component of any clinical trial. Unlike frequentist methods, Bayesian methods assign anything unknown a probability using information from previous experiments. The Bayesian methods make use of the results of previous experiments, to do continuous updating as information accrues, whereas the frequentist approaches assume we have no prior results.

Doctors want to be able to use biomarkers to determine who is responding to what medication and look at multiple potential treatment combinations. They want to be able to treat a patient optimally depending on the patient's disease characteristics. Cancer is a diverse disease and what works to treat one person's disease may not work for another.

Clinical trials test the efficacy (not the accuracy) of a drug. The efficacy of a drug is to produce a desired effect, which is tumor response (shrinkage). Single arm clinical trials provide the tumor response evidence that is the basis for approving new cancer drugs. The Bayesian methology can bring some much-needed "accuracy" to the forefront of clinical trials.

Clearly, more effective cancer therapies are desperately needed, and after 30 years of investigation aimed at intensified multi-agent chemotherapy, we should look for other avenues of study. In an era of ever-increasing numbers of partially effective cancer therapeutics, there is an obvious need for more accurate methologies. We cannot afford any more 'trial-and-error' treatments.

http://www.mdanderson.org/departments/n ... 508b603a14

[gpawelski]

The Bayesian method is no stranger to the technology of Cell Culture Assay Testing, a "functional" biomarker. In fact, it is what gives credit to the accuracy of assay tests. The method has to do with "conditional probability." The probability that event E (an effect) and C (a cause) will both occur is the product of the event C occurring, times the conditional probability of an event E occuring (remember that in elementary statistics?). An example: The chances of being hit by a truck and bleeding to death is the product of the probability of being hit by a truck and the probability of bleeding to death if you get hit by a truck. Well, so what?

The Bayesian method turns this calculation around. That is, it tries to calulate the probability of C, given that E has occurred. Baye's Theorem is useful and reasonably well accepted for some applications such as testing whether the assumptions of probability are valid. For instance, if you flip 100 coins in the air at once, and only get tails 5 times, you have to assume that they aren't "fair" coins. The whole idea of it all, is to get more accuracy out of analysis.

The absolute predictive accuracy of cell culture assay tests varies according to the overall response rate in the patient population, in accordance with Bayesian principles. The actual performance of assays in each type of tumor precisely match predictions made from Bayes' Theorem. The theoretical expectations for cell death assays, based on Bayes' Theorem, are overall specificity for drug resistance of 0.92 and an overall sensitivity of 0.72.

Thus, the absolute probability of response with assay "sensitive" and "resistant" drugs varies according to the overall prior response probability in the patient population. Which means assay "resistant" patients have a below average probability of response and assay "sensitive" patients have an above average probability of response. Treatment with assay "sensitive" drug(s) is more likely to be associated with a favorable outcome than treatment with assay "resistant" drug(s).

Cell death assays are broadly applicable to a wide range of human neoplasms, ranging from low response rate tumors (like pancreatic cancer and Cholangiocarcinoma) to high response rate tumors (like acute lymphoblastic leukemia, breast cancer and ovarian cancer). In cases where more than one acceptable regimen exists, the physician can select the regimen containing the most favorable drugs and avoid the regimen containing the most unfavorable drugs.

Bayes' theorem is a tool for assessing how probable evidence makes some hypothesis. It is a powerful theorem of probability calculus which is used as a tool for measuring propensities in nature rather than the strength of evidence (Solving a Problem in the Doctrine of Changes).

Bayes' theorem describes the relationship between the accuracy of a predictive test (post-testing probability) and the overall incidence of what is being tested (pre-testing probability).

Bayes' theorem indicates that laboratroy assays will be accurate in the prediction of clinical drug resistance in tumors with high overall response rates in assays that are extremely specific for drug resistance (>99% specificity).

Post-test probability of response is independent of pre-test (expected) probability of response. Once identified, post-test response probabilities vary according to both assay results and pre-test reponse probabilities, precisely according to predictions based on Bayes' theorem. This allows the construction of a monogram for determining assay-predicted probability of response.

Assay Results and Bayes' Theorem: http://weisenthal.org/figure06.htm

[MurielK]

Well, this should wake up a few people this morning! I love it! Brings back memories of my 10+ stat. and measurement courses in grad. school. Thanks for the explanation.

Muriel

[Don M]

I think I sort of get it.. traditional testing is narrowly defined to very specific circumstances. The bayesian method is more broad? If you know that a drug does well in a controlled experiment, you still don't know how it will work in the general population.

You start out with the statement that a drug will work well in the general population and then refine that statement with data?

And what does this have to do with Schrodinger's cat anyway?

I didn't do all that well at sadistics.

Don m

[Bill]

Ironically, this technology may have an unintended detrimental impact on cancer patients. Namely, insurance companies, as part of on-going cost-cutting efforts, mandating as a condition of reimbursement that med oncs treat their cancer patients based on chemosensitivity testing results to the exclusion of other tx agents that may very likely have tx value but don't meet the testing criteria. Depending on the specific cancer and patient particulars, some patients may end up with a very short list of agents approved and available for their tx and, possibility, a reduced survival time to match. IMO, the bottom line for cancer patients is if you sift thru the mass of literature on chemosenstivity testing you will find no credible data that such testing will produce a clinically significant improvement in patient survival time. IMO, once this technology is commercially established the biggest beneficiaries will be the insurance companies.

JMO

B

[gpawelski]

Patient tumors with the same histology do not necessarily respond identically to the same agent or dose schedule of multiple agents. Laboratory screening of samples from a patient's tumor can help select the appropriate treatment to administer, avoiding ineffective drugs and sparing patients the side effects normally associated with these agents.

It can provide predictive information to help physicians choose between chemotherapy drugs, eliminate potentially ineffective drugs from treatment regimens and assist in the formulation of an optimal therapy choice for each patient. This can spare the patient from unnecessary toxicity associated with ineffective treatment and offers a better chance of tumor response resulting in progression-free survival.

Identifying patients with resistant neoplasms may not only spare them toxicity but may prolong their lives, by sparing them from the life shortening effects of ineffective chemotherapy.

Patients would certainly have a better chance of success had their cancer been chemo-sensitive rather than chemo-resistant, where it is more apparent that chemotherapy improves the survival of patients, and where identifying the most effective chemotherapy would be more likely to improve survival above that achieved with empiric chemotherapy.

As Dr. Berry argues in the MD Anderson news release, "...cancer is a diverse disease, and what works to treat one person's disease may not work for another..."

[RandyW]

Thanks for the iNfo Greg

[gpawelski]

Using a Cell Culture Assay (in vitro apoptosis) for choosing cancer drugs is no different than a marker like estrogen receptor or CD20 or a gene expression pattern. They are all markers. One is a structural marker, the other is a functional marker. A Cell Culture Assay is a "functional" biomarker. A functional biomarker provides information about the biomarker uptake rate in tumor cells or on tumor cell surfaces through fluorescence intensity changes.

As with any other laboratory test, the determination of the efficacy of cell culture assay tests is based on comparisions of laboratory tests with patient response (clinical correlations). The hypothesis to be tested with clinical correlations is that above-average drugs effects in the assays correlate with above-average drug effects in the patient, as measured by both response rates and patient survival.

Patients with test results in the "sensitive" range were more likely to respond than the total patient population as a whole. Conversely, patients with test results in the "resistant" range were less likely to respond than the patient population as a whole. On average, patients with assays in the "sensitive" range were 3.5-fold more likely to respond than patients with assays in the "resistant" range.

Targeted treatments take advantage of the biologic differences between cancer cells and healthy cells by "targeting" faulty genes or proteins that contribute to the growth and development of cancer. Many times these drugs are combined with chemotherapy, biologic therapy (immunotherapy), or other targeted treatments.

Understanding targeted treatments begins with understanding the cancer cell. Every tissue and organ in the body is made of cells. In order for cells to grow, divide, or die, they send and receive chemical messages. These messages are transmitted along specific pathways that involve various genes and proteins in a cell.

Targeted treatments fight cancer by correcting or modifying defective pathways in a cancer cell. In healthy cells, each pathway is tightly controlled. For instance, healthy cells are allowed to divide into new cells, and damaged cells are destroyed. However, in cancerous cells, certain points in the pathway become disrupted, usually through a genetic mutation (change in form).

Serious consequences to the cell may result from these mutations, depending on which pathway is affected. For example, suppose a cell develops a mutation that causes it to continue dividing into new cells? In other words, the signal is always on. If the signal never turns off, the cells that keep growing may eventually form a tumor.

The most appealing idea behind targeted drug therapy is that cancerous cells are destroyed and healthy cells are spared, resulting in fewer side effects of treatment. In contrast, traditional chemotherapy destroys both the cancer cells and the healthy cells, and does not have any mechanism to distinguish between them.

Because many cancer cells use similar pathways, the same drug could be used to treat one person's breast cancer and another person's lung cancer, as long as each tumor contained similar targets. This is why many of these treatments are being used in a variety of cancer types. Gleevec is used to treat both leukemia and a rare stomach tumor, called gastrointestinal stromal tumor (GIST).

Although targeted therapy is appealing, it is more complex than meets the eye. Cancer cells often have many mutations in many different pathways, so even if one route is shut down by a targeted treatment, the cancer cell may be able to use other routes. In other words, cancer cells have "backup systems" that allow them to survive. The result is that the drug does not shrink the tumor as expected. One approach to this problem is to target multiple pathways in a cancer cell.

Another challenge is to identify for which patients the targeted treatment will be effective. When Iressa is used in patients with lung cancer, researchers discovered that only patients whose tumors contained specific mutations responded to this drug.

Finally, tumors can become resistant to a targeted treatment. This means that the drug no longer works, even if it has previously been effective in shrinking a tumor. To solve this problem, new drugs are being designed or combined with existing ones to target the tumor more effectively.

The introduction of targeted drugs has not been accompanied by specific "predictive tests" allowing for a rational and economical use of these drugs. However, given the technical and conceptual advantages of cell culture analysis, together with its performance and the modest efficacy of therapy prediction on analysis of genome expression, there is reason for a renewal in the interest of cell culture assays (functional biomarker) for optimized use of medical treatment of malignant disease.

There should be an immediate recognition that matchmaking between cancer and cancer treatment is one area in cancer research and treatment which is deserving of much greater attention and utilization. There should be an inclusive effort to study and utilize technologies which are based on both the sub-cellular (molecular) level and at the cellular (cell function/cell culture) level.

[gpawelski]

The traditional meaning of Health 2.0, according to Jane Sarasohn-Kahn's "Wisdom of Patients" has been the use of social software and light-weight tools to promote collaboration between patients, their caregivers, medical professionals and other stakeholders in health.

http://www.chcf.org/topics/chronicdisea ... mID=133631

An example of this in cancer medicine is Individualized Online Clinical Trial Protocol Version 1.0 by the Weisenthal Cancer Group, a Phase II evaluation of individualized cancer treatment with traditional cytotoxic chemotherapy, targeted anti-kinase drugs and anti-angiogenic agents.

With most clinical trials, investigators never give out information as to how people are doing. Most trials are failures with respect to actually improving things. The world doesn't find out what happen until after a hundred or 500 or 2,000 patients are treated and then only 24 hours before the New England Journal of Medicine publication date.

Individualized Online Clinical Trial Protocol Version 1.0 is a totally transparent clinical trial. Every patient who decides to enter a study should know what happened to previous patients. Patients are treated in real time, on the Internet, with the whole world watching to see how they do. It includes weekly progress reports, and if individual patients want, their own blogs as to how they are doing.

Stages have been implemented for a rather innovative clinical trial with cell culture assays, "real time" on the Internet. The purpose of the study is to show that cell culture assay technologies for "targeted" agents really do work. The short-term future of cancer therapeutics is combinations of "targeted" agents.

http://weisenthalcancer.com:80/Study%20 ... alHome.htm

They are not "marketing" the test beyond the confines of a clinical trial, which will be the most transparent clinical trial in the history of oncology, as all results are going to be reported, in real time, on a week by week, patient by patient basis, on the website.

Call or e-mail Weisenthal Cancer Group and ask to speak with Connie Rueff, the study coordinator. She will answer any questions you may have and will help you to determine if you are a candidate for entry. (714) 596-2100 or connie@weisenthalcancer.com

[gpawelski]

A clinical trial design applying Bayesian adaptive randomization for targeted therapy development in lung cancer: A step toward personalized medicine.

A $9 million U.S. Department of Defense research grant (hence the name BATTLE) to help clinicians develop biomarker-integrated approaches of targeted therapy of lung cancer elimination. It is a novel clinical trial program to explore and prescribe treatments based on the genetic and molecular profile of a patient's tumor.

After a panel of tumor markers is identified and assessed, patients with advanced NSCLC are enrolled in one of several trials testing agents that target abnormalities in their cancer. A novel adaptive randomization statistical design will be applied to the clinical trials to accelerate the identification of highly individualized treatments for each individual patient.

Investigators will study the mechanisms of response or resistance to existing and newer targeted agents, explore novel signaling pathways for future trials and identify molecular features in tumor and adjacent normal tissues that correlate with treatment outcome.

However, the importance of mechanistic work around targets as a starting point could have been downplayed in favor of a systems biology approach were compounds are first screened in cell-based assays, with mechanistic understanding of the target coming after validation of its impact on the biology of the cancer cells.

Many of these drugs cry out for validated clinical biomarkers to help set dosage and select patients likely to respond. Optimal and reproducible targeted testing continues to evade the diagnostics of the disease. Numerous genes, tumor, and patient factors contribute to the risk of the cancer coming back and the effectiveness of chemotherapy.

It could be vastly more beneficial to measure the net effect of all processes (systems) instead of just individual molecular targets. The cell is a system, an integrated, interacting network of genes, proteins, and other cellular constituents that produce functions. One needs to analyze the systems' response to drug treatments, not just one or a few targets (pathways/mechanisms).

There are many pathways/mechanisms to the altered cellular (forest) function, hence all the different "trees" which correlate in different situations. Improvement can be made by measuring what happens at the end (the effects on the forest), rather than the status of the indivudal trees.

http://www.asco.org/ASCO/Abstracts+&+Vi ... ctID=36349

http://clinicaltrials.gov/archive/NCT00 ... 2007_01_19