June 2011
Feature Story

Karen Lusky

Harvard pathologist Mark Boguski, MD, PhD, and colleagues had in 2009 what they thought was a futuristic vision of how personalized cancer care would work by 2020. They predicted that a pathologist would make a diagnosis of cancer and then “drop a little piece of the tumor in a magic sequencing machine,” as Dr. Boguski put it in a presentation he made in May to Executive War College attendees.

The sequencing would identify the tumor’s mutations and gene expression changes, “allowing us to reverse engineer the pathways and perhaps even do therapeutic simulations with virtual drugs,” said Dr. Boguski, of the Department of Pathology and Center for Biomedical Informatics at Boston’s Beth Israel Deaconess Medical Center and associate professor, Harvard Medical School.

When their vision was published in 2009 in an article titled “Customized care 2020: how medical sequencing and network biology will enable personalized medicine” (Boguski MS, et al. F1000 Biol Rep. 2009; 1:73), they felt “safe” predicting that it would not be so for more than a decade, he told the audience. One year later they saw a case study in Genome Biology that showed “in principle” it was already happening. “We woke up,” he said.

At the Executive War College, Dr. Boguski reported on that case as well as a more recent one, and provided an overview of clinical cancer genomics, for which Harvard is building the infrastructure.

“At Harvard, we are looking at the few published exceptional cases and thinking we could apply genomic sequencing of tumors to many different kinds of patients with various kinds of cancer.” They are convinced, he told CAP TODAY, that it could become the standard of care for certain cancers. “We are therefore building a system that would allow us to do genomic sequencing of cancer on a semiroutine basis rather than as an exceptional research project.”

The case study reported in Genome Biology was that of a 78-year-old patient who sought medical evaluation for a sore throat (Jones S, et al. Genome Biol. 2010; 11(8): R82). The doctor found a lump on the back of the man’s tongue; a biopsy showed it was a papillary adenocarcinoma that probably originated in a minor salivary gland, Dr. Boguski said. The patient underwent laser resection and lymph node dissection, which revealed metastatic disease in three of 21 nodes. The patient then underwent a course of adjuvant radiation.

At the time of treatment, the patient showed no signs of cancer below the neck, Dr. Boguski said. But when the patient returned for followup staging four months later, a PET-CT scan found numerous bilateral pulmonary metastases.

Although there was no standard chemotherapy protocol available to treat the rare cancer of the tongue, Dr. Boguski said, the tumor did show “mildly elevated expression” for EGFR. So the oncologists started the patient on the EGFR inhibitor, erlotinib. That turned out to be like “sprinkling fertilizer on a garden,” Dr. Boguski said. “The tumors just continued to grow.”

The hospital then obtained consent to send the man to its genome center for a diagnostic genomic analysis of a biopsy of one of the pulmonary lesions.

The genomic analysis of the tumor, which required comparison with the genome of the patient’s peripheral lymphocytes as a control, showed that the drug erlotinib could not help because the tumor possessed mutations in other genes known to override the effect of inhibition.

The analysis found that overexpression of the RET oncogene seemed to be fueling the cancer, the authors of the report in Genome Biology wrote. So the oncologist ordered a RET inhibitor, sunitinib, which caused the patient’s cancer to shrink for four months. The drug is normally used for renal cell carcinoma, Steven Jones, PhD, lead author of the Genome Biology report and head of bioinformatics at the Genome Sciences Centre, British Columbia Cancer Agency, told CAP TODAY.

When the tumors began to grow again, the oncologist prescribed sorafenib, a RET inhibitor made by another manufacturer and believed to have a slightly different mechanism of action than sunitinib, Dr. Jones explains.

In addition to sorafenib, the patient took sulindac, a nonsteroidal antiinflammatory agent. The initial genomic analysis indicated that both of these drugs might help. The patient’s cancer stabilized again for about another four months.

In his War College talk, Dr. Boguski postulated that by doing the genomic analysis on the “original biopsy of the tongue, perhaps the most effective therapy would have been administered at the outset.” One also has to wonder, he said, whether the radiation therapy played a part in causing the new mutations in the tumor.

“The point is that you could have predicted up front that the drug was ineffective and saved the expense of the drug and potentially prevented the tumor from progressing,” he told CAP TODAY. Which makes a relatively expensive test like genomic analysis of the tumor start to look sensible.

Dr. Boguski also presented a more recent case in the U.S. that led to an unexpected change in treatment for a 60-year-old man with esophageal cancer. Diagnosed with metastatic disease when the cancer was discovered, the man underwent cytotoxic chemotherapy until a genomic analysis of his tumor showed the cancer might respond to Gleevec (imatinib).

“You would never think of giving Gleevec for an esophageal tumor,” Dr. Boguski told CAP TODAY, “but the target that the drug is normally effective for was present in [the man’s] particular tumor. It was an incidental finding.”

“I think there may be countless additional incidental findings when you start sequencing tumors,” he says.

The patient with esophageal cancer was started on Gleevec “and a prayer,” Dr. Boguski reported in his presentation, referring to an account of the case at www.medpagetoday.com/Blogs/ 25732. “The prayer wasn’t because we don’t have faith in the technology. The prayer came from the patient’s personal physician, Francis Collins, MD, PhD, director of the National Institutes of Health,” whom the medpagetoday article and many others describe as an evangelist.

In both case studies, Dr. Boguski points out, genomic sequencing identified FDA-approved drugs that might help the patients. “If you get a glimpse into the genetic underpinnings of various cancers, it may turn out that therapies on the shelves could be applied to a much broader range of tumors.”

As another example of how that may be the case, Dr. Boguski pointed to a recent Nature article by a Boston research group that sequenced the tumors of multiple myeloma patients (Chapman MA, et al. Nature. 2011. March 24;471:467–472). The researchers found that a small number of the 38 sequenced tumors had the BRAF mutation, which is found more commonly in melanoma, he says. “This indicates that perhaps those blood cancers could benefit from treatment by melanoma drugs.”

Jeffrey Saffitz, MD, PhD, chair of the Department of Pathology at Beth Israel Deaconess, notes that although drugs were originally designed to treat certain types of cancer, such as breast and colon, it’s now recognized that common pathway mutations “run across all kinds of cancer,” the idea being “you have a set of drugs targeted to a specific pathway mutation.”

In the field of cancer genomics, there are unknowns that will need to be resolved, among them which mutations to target in a tumor, says Dr. Boguski. That question came up when he gave his War College presentation to pharmaceutical industry representatives in Philadelphia where one of them pointed out that it’s not possible to distinguish a driver mutation from a passenger mutation. “That means you see a whole range of mutations [when sequencing] in a particular tumor and may not know which one is causing the malignancy, the driver, and needs treatment versus the passenger mutations that came along for the ride,” he says.

The point was an important one and well taken, he says, and adds, “A lot of times now, however, we know the molecular pathways that these mutations map to and sometimes we can predict which ones are the driver mutations based on our knowledge of the pathways.” As more data are accumulated, “we will discover the patterns that are driving [the cancer] which can be attacked by a drug.”

Also not yet known is whether DNA or RNA analysis will ultimately prove to be more valuable in oncology, he says. “A genomic analysis can be viewed broadly as a data package, which could be the genome or transcriptome or both. Only time and experience will tell us which is most cost-effective and useful for particular cancers and situations.”

Whether it’s important to genetically analyze various parts of a cancerous tumor is what Dr. Boguski calls another “fly in the ointment.” It’s suspected that the cellular heterogeneity observed in some tumors—where “some cells look a little nastier or scarier than others”—reflects an underlying genetic heterogeneity, he says. And “if you sample only one spot, you may miss something that should be considered for treatment.” At $10,000, it’s not economically feasible to sequence different parts of a tumor. “But when it becomes cheaper, I predict we will see sampling of tumors in more places to see if we need to do that routinely.”

What are the realities of using genomic technology in pathology? In terms of producing the genomic data, Dr. Boguski said, the send-out model used for esoteric tests “is the one that makes sense, as the technology is changing fast, and [the testing] is still very expensive to do.” But “you’re going to get a lot more data than you know what to do with.” It’s more important, he says, to teach pathologists how to interpret the data rather than how to produce it or the algorithms to annotate it.

Says Dr. Saffitz: “It’s all about the information technology that one will need to use to make sense out of the sequence data, which is where we have been investing heavily in our department with a group of bioinformatics experts.”

The bioinformatics experts, he adds, are developing software tools now that will look at a tumor’s genomics and all the available drugs and provide a personalized treatment plan for a [patient’s] cancer. “We are very optimistic about this. We think it’s the way you get rid of cancer without killing the patients,” Dr. Saffitz says.

So sure are they of its promise that they want to develop a process to make it as efficient as possible so they can imagine applying it fairly routinely within two years—far ahead of 2020.