The Road to Personalized Medicine Is Strewn with Obstacles

October 2011, Vol 2, No 6

Stockholm, Sweden—With recognition of common tumor mutations and a pipeline full of biologic agents that target them, personalized medicine should be all but a fait accompli. But one expert told attendees at the European Society for Medical Oncology’s 2011 European Multidisciplinary Cancer Congress, “We may be overpromising our patients.”

Gordon Mills, MD, of the Zayed Institute for Personalized Cancer Therapy at the University of Texas M.D. Anderson Cancer Center, Houston, said, “We are coming closer to having a scientific underpinning for what we want to do, but integrating this remains a challenge. Can we achieve personalized therapy? Is it even doable?”

Personalized medicine—matching the right drug with the right patient— is “not the present, but it is clearly in our future,” but there are many challenges, Dr Mills said. “The biggest problem is the ‘N of 1,’” he emphasized, because each individual patient has a unique genetic tumor profile that requires a finely tuned drug regimen. “Some subgroups will be so small, it will be impossible to implement clinical trials [to evaluate such regimens] with sufficient power,” he said.

Intratumor Heterogeneity

Emerging data further suggest that tumors within one patient may be heterogeneous. Discordance between the primary (initial, nonmetastatic) tumor and the recurrent or metastatic lesion has been recently established. Although this has been evaluated for estrogen receptor and HER2 status, genetic mutations can also vary. Some discordance between primary and advanced disease has been observed in as many as 40% of patients in some studies, and this presents a big clinical problem.

“We found one patient with 2 metastases, with 2 different mutations in PI3K, and the primary tumor was called a wild type [ie, lacking the mutation]. If you treated according to the primary tumor, you would not give this patient a PI3K inhibitor. But it turns out there was heterogeneity in the primary tumor,” Dr Mills said. “The mutations were there. If this is the case, will we have to grind up the whole tumor and look deeply? Yes, or we will run the risk of treating the wrong disease. This is the future: biopsy, biopsy, biopsy.”

Better categorization will also be required of the mutations that are discovered, in particular, distinguishing mutations that are “drivers” and “druggable” from those that are “passengers” and thus exert no harm. Tumors will need to be sequenced in depth to capture important subclones, he said.

“We are finding hundreds, if not thousands, of mutations, and not all are created equal. How will we integrate this information to find drivers that can be targeted for each patient? And for the actionable aberrations we have a limited number of drugs,” Dr Mills pointed out.

Even when the tumor is well described, and a drug has been marketed to target the patient’s various receptors and mutations, resistance ultimately develops in metastatic disease, and responses—which can be dramatic with some targeted agents— are not durable. “We have to figure out how to make durable, combinatorial therapy to improve outcomes,” he said.

Pilot Study Results

Dr Mills and his colleagues at the Zayed Institute for Personalized Cancer Therapy “are finding actionable mutations and doing ‘N of 1’ clinical trials in an efficient manner to determine which populations could benefit from a particular drug,” he said.

They are observing very high responses in many patients for whom they are able to identify the “driver” mutations and match these to an effective agent. But they have also found that in their first 1000 patients, only about 40% have genetic aberrations in the tumor, and only 25% have “actionable” mutations. This means that the pool of patients who will truly be candidates for personalized medicine may be narrower than originally believed.

“There are still many patients who will benefit, but in terms of delivering ‘personalized medicine’ to all patients, we are way below the desired numbers,” Dr Mills said.

Cost Considerations in Personalized Cancer Therapy

The new era of personalized medicine in oncology raises economic concerns of the cost-benefit ratio. The use of molecular testing, some argue, is adding a significant burden to the total cost of therapy. But others have shown, for example in relation to KRAS, that although the cost of testing is considerable, a true economic analysis shows that genetic testing actually reduces the total costs, by decreasing the number of patients who receive a high-cost therapy that does not benefit them.

Dr Mills also discussed the potential cost implications of genomic medicine. “The cost of implementing personalized cancer therapy has added a massive burden to healthcare costs at a very minor improvement in outcomes for a small population,” said Dr Mills. He noted that although there has been modest success so far, there have been “spectacular failures” as well.

“We are far from understanding the challenges and overcoming the pitfalls to the implementation of personalized cancer therapy,” he said, and one of these is the cost of technology and treatment in this new era.

“The cost of assessing molecular markers such as genetic sequencing and the cost of targeted therapies that benefit only a few patients constitute incredible burdens on already strained healthcare resources. Next generation sequencing [NGS] approaches are held as the next great step forward in implementing personalized cancer therapy. Indeed, it is possible to sequence a human genome for under $10,000, and the costs continue to drop rapidly. However, NGS is currently fraught with major problems,” Dr Mills maintained.

“The cost of $10,000 for a human genome is presupposed on a depth of sequencing of 30- to 60-fold. Unfortunately, at this depth of sequencing, achieving a true-positive rate of 60% results in a false-positive rate approaching 60%. Both of these are unacceptable for patient management. Further, the $10,000 does not include the costs of bioinformatics analysis of data storage and handling. Indeed, it has been estimated that even the eventual $1000 genome will cost $100,000 to manage and interpret,” he said.

Accurate treatment calls for targeting the specific tumor characteristics, and for patients with metastatic disease this means rebiopsying, perhaps repeatedly, to reveal important subclones within each tumor. “This would massively increase the cost, as well as be associated with the morbidity of repeat biopsies. The goal of replacing biopsies with molecular imaging or characterization of circulating tumor cells or DNA remains an important dream, but a dream nevertheless,” he said.

Resistance to targeted therapies is another challenge in recurrent disease; therefore, combination regimens will be necessary to address compensatory signaling pathways and to help overcome resistance.

“This is fraught with its own challenges of increased costs, increased toxicity, and difficulty in getting multiple companies to cooperate in clinical trials,” Dr Mills said.

However, José Baselga, MD, PhD, Co director of the Massachusetts General Hospital Cancer Center, who is a clinician and a translational scientist, was more optimistic about the economics of this approach. He commented that, “The cost of looking for mutations should be far less than the cost of treating patients with expensive therapies that will not work in them.”

With a goal of advancing personalized medicine over the next 10 years, Dr Baselga called for dedicated facilities to study the treatment of patients with novel agents, expansion of tumor genotyping and next-generation sequencing, creation of tumor biomarker facilities, sharing of data, incorporation of novel technologic platforms, innovative clinical trial design, and recruitment of “the best pool of physician scientists into a culture of team work.”

Related Articles