'We're beginning to understand that there's much more to predicting what kinds of treatments a particular tumor might respond to than just what we can figure out from looking at traditional pathology specimen.'

This knowledge should allow doctors to tailor therapies more appropriately to individual patients, and also open the door to devising new therapies for tumors that have different genetic make-ups.

The specific tumors in question are glioblastomas, a difficult-to-treat, aggressive form of cancer with a low survival rate.

"A subset of patients are more likely to respond, and by understanding why other subsets are failing, we can begin to think about customizing combinations of therapies to help treat those" patients, said Dr. Paul Mischel, senior author of the study that appears in the Nov. 10 issue of the New England Journal of Medicine.

"This drives a wedge into the dirt and begins to expose the possibility of figuring out what is really happening in patients who aren't responding," added Mischel, a researcher at the David Geffen School of Medicine and Jonsson Comprehensive Cancer Center at the University of California, Los Angeles.

Treating Glioblastomas

The findings may lead to an immediate shift in how brain-tumor treatments are directed.

"We already are starting to change the way we look at treatments and narrow down the treatments we give to patients," said Dr. Susan Pannullo, director of neuro-oncology at New York Presbyterian Hospital-Weill Cornell Medical College in New York City. "This data is very important because there is so little that works right now."

Two drugs, Tarceva (erlotinib) and Iressa (gefitinib), have had some success in treating glioblastomas and have been approved for treating advanced lung cancer that has not responded to other treatments, the study authors said.

Both drugs belong to a class of medications called EGFR (epidermal growth factor receptor) kinase inhibitors. The EGFR gene is frequently dysfunctional in glioblastomas, yet only 10 percent to 20 percent of patients with this type of tumor respond to Iressa or Tarceva.

As an accompanying editorial in the journal pointed out, there seems to be no correlation between a patient's response to the drugs and the function of EGFR. The drugs were obviously affecting a different mutation in patients with glioblastoma.

The question was, which one? Glioblastomas exhibit abnormal activity in a number of genes, including the EGFR gene, and the loss of PTEN, a tumor-suppressor protein.

"We hypothesized that two proteins, EGFRvIII [a variant of EGFR] and PTEN played a critical role and thus their interaction might be important for developing a response" to Tarceva and Iressa, Mischel explained.

Prognosis Is Still Poor

To test the hypothesis, Mischel and his colleagues performed a genetic analysis of tissue from 26 glioblastoma patients who had either clearly responded or clearly failed to respond to Tarceva or Iressa.

As it turned out, glioblastomas that produced both EGFRvIII and PTEN were 51 times more likely to shrink when treated with one of these drugs than tumors with different genetic makeups. And patients also went five times longer (243 days versus 50 days) before their tumors progressed.

The results were verified in tissues from another group of patients who had participated in a clinical trial of Tarceva.

Although the prognosis is still poor for these patients, Pannullo said, these results may represent a big step toward eventually finding a cure.

"We're beginning to understand that there's much more to predicting what kinds of treatments a particular tumor might respond to than just what we can figure out from looking at traditional pathology specimen," Pannullo said. "This study, along with another one, really shows that there is hope that looking at genetics may help to improve the way that we treat these patients."

More information: To learn more about brain tumors, visit the National Institute of Neurological Disorders and Stroke.