ARTICLE IN BRIEF
Two teams of researchers independently identified robust molecular subgroups that are far more effective at differentiating glioma types and outcomes than neuropathologists can glean from histology, suggesting a more precise treatment.
Clinicians involved in treating diffuse low-grade and intermediate-grade gliomas have always been stumped by the fact that people with the same grades of tumors have such different outcomes. Now, researchers have offered an explanation that could potentially alter the way these patients are diagnosed and treated, according to reports published in the June 25 issue of The New England Journal of Medicine (NEJM).
Two different teams of scientists independently identified robust molecular subgroups that are far more effective at differentiating glioma types and outcomes than neuropathologists can glean from histology, suggesting a more precise treatment.
In fact, these tumor biomarkers — mutations in isocitrate dehydrogenase (IDH), mutations in the telomerase reverse transcriptase (TERT) promoter, and co-deletion of chromosome arms 1p and 19q — are so important and game-changing in the glioma field that neuropathologist Daniel J. Brat, MD, PhD, a professor and vice chair of pathology and laboratory medicine at Emory University and a lead investigator of one of the NEJM studies, went to the World Health Organization's Classification of Brain Tumors meeting in mid-June to discuss the inclusion of these biomarkers and molecular profiling into the definition and diagnosis of the disease.
“These findings are far-reaching,” said Dr. Brat, who is a member of The Cancer Genome Atlas (TCGA) network, a national consortium of research and technology teams funded by the National Cancer Institute and the National Human Genome Research Institute. The data from TCGA are nearly identical to those from another study from the Mayo Clinic and the University of California, San Francisco (UCSF) published in the same NEJM issue.
“Molecular profiles based on biomarkers will be the primary way to diagnose gliomas, rather than making a diagnosis primarily from histology,” Dr. Brat said. “This is a new era.”
MAYO CLINIC/UCSF STUDY
Daniel Lachance, MD, FAAN, an associate professor of neurology and head of the Section of Neuro-oncology at the Mayo Clinic in Rochester, MN, and a co-first author of one of the NEJM studies on genetic biomarkers in grade II and III gliomas, agreed that these findings “will change the practice of medicine.”
Grade II and grade III gliomas, the authors explained, arise most often in the cerebral hemispheres of adults and include astrocytomas, oligodendrogliomas, and oligoastrocytomas.
It is not uncommon in cancer biology to use the latest genetic technology to better define tumor types, but it had not yet been done in diffuse gliomas. Dr. Lachance and his colleagues have been mining data from their glioma patients to look for common genetic markers for almost a decade. In the latest study, they defined glioma molecular groups using IDH mutations, 1p/19q co-deletion, and mutations in the TERT promoter.
They scored tumors as negative or positive for each of these biomarkers in 1,087 gliomas. They then compared patient characteristics — age, course of disease, treatment response, and outcome — among the top five molecular groups: triple-positive, TERT- and IDH-mutated, IDH-mutated-only, TERT-mutated-only, and triple-negative. They validated the data in 351 gliomas from the UCSF Adult Glioma Study and 419 gliomas from the Cancer Genome Atlas Study.
They observed that patients within each group had a similar age of onset and overall survival. For example, the average age of diagnosis was youngest (37 years old) in patients with IDH mutations only, compared with people diagnosed later in life (at around 59 years old) who had gliomas with only TERT mutations. (For more detail on the molecular groups, see the sidebar, “Five Molecular Groups of Gliomas and Associated Outcomes.”)
Information on germline risk variants from these patients enabled the Mayo and UCSF researchers to show that “each of the five molecular groups were associated with specific variants,” said Dr. Lachance. In particular, the group had previously reported in Nature Genetics in 2012 that a single nucleotide polymorphism on chromosome 8q24 (rs55705857) is associated with a greater than six-fold risk of any IDH-mutated glioma, and this was affirmed in the present study.
Findings from all three data sets — Mayo, UCSF, and TCGA — matched up. According to the study, there were eight possible combinations of the three different biomarkers, and five of them were powerful enough to classify most of the 1,087 gliomas. Mayo's Jeanette E. Eckel-Passow, PhD, was the lead statistician on the study.
Dr. Lachance said that the use of these three molecular tests is a more reliable way to manage glioma diagnosis and treatment. “By using a combination of these genetic markers, we can group gliomas that share important characteristics,” he said.
“A tumor acquires a series of genetic mistakes that allows cells to function in an abnormal way,” he explained. “The findings from all of these different cohorts affirm that there is an important biology here that may well be distinct in each of the molecular groups.”
He said the findings have immediate implications for treatment. For instance, patients with IDH mutations and the 1p/19q co-deletion appear to respond better to radiation and chemotherapy. “Many of these patients have survived 10 to 15 years and longer. If people with specific molecular signatures live longer and respond better to therapies, we have to begin to think about how to avoid or delay the toxicity of those treatments,” he said. “We need to start targeting our therapies based on the biology of the tumor,” he added.
“Having this new classification system will allow us to better treat our patients,” added Robert B. Jenkins, MD, PhD, the senior author of the Mayo/UCSF study.
THE CANCER GENOME ATLAS STUDY
The Cancer Genome Atlas research network has been working since 2006 to identify genetic changes associated with different types of cancers. Genes that play a role in glioblastoma, ovarian, and lung cancers were the first to be mapped. The success of the initial $50 million investment from the National Cancer Institute and the National Human Genome Research Institute led to additional funding to add another 20 tumor types to the cancer genome atlas.
Scientists in the network conducted genome-wide analyses on 293 grade II and III gliomas from adult patients. They conducted clustering of mutations and data from DNA copy number, DNA methylation, and microRNA sequencing. From the analysis came three “robust, non-overlapping prognostically significant subtypes of lower-grade glioma that were captured more accurately by IDH, 1p/19q, and TP53 status than by histological class.”
Among their findings, those patients with low-grade gliomas (classified as either astrocytomas, oligodendrogliomas, or oligoastrocytomas) with an IDH mutation and 1p/19q co-deletion had the best outcomes.
These low-grade gliomas with an IDH mutation had either a 1p/19q co-deletion or a tumor protein p53 (TP53) mutation. Those with an IDH mutation and no 1p/19q co-deletion had shorter overall survival than those with the IDH mutation and the 1p/19q co-deletion. A large majority of lower-grade gliomas without the IDH mutation had “genetic aberrations and clinical behavior strikingly similar to those found in primary glioblastoma,” and these patients had the worst clinical outcomes. The three main groups identified by TCGA were nearly identical to the three most common groups identified by the Mayo Clinic/UCSF study.
The scientists said that the analysis was “unsupervised,” meaning the categories were not known before the actual computation. There was a strong correlation between gliomas with an IDH mutation and the 1p/19q co-deletion and the oligodendroglioma histological class (69 out of 84 samples), while mutations of IDH without the 1p/19q co-deletion included a mix of histological classes. The IDH wild-type (no mutation) samples were mostly astrocytomas (31 of 55 samples) and grade III gliomas (42 of 55 samples).
The findings held up across all four platforms (DNA copy number, DNA methylation, mRNA, and microRNA) used in the analysis.
The statisticians used an algorithm to count all of the mutations found in the 289 tissue samples. They found 9,885 mutations, with an average of 29 mutations per sample. Samples with wild-type IDH had more mutations (an average of 45 per sample) compared with those samples with an IDH mutation and the 1p/19q co-deletion, which had an average of 28 mutations per sample. (There were many other genetic mutations identified in the three different genetic subtypes identified in the study.)
Dr. Brat said that these molecular classifications are already helping neurologists refine their diagnosis and outline a treatment plan guided by what the tests reveal about the tumor. “Molecular subtyping of glioma will be required for diagnosis and for treatment,” he said. “It will also be necessary to collect this information in clinical trial designs so that scientists can select appropriate candidates to study.”
“For 120 years, we relied on morphological characteristics to help in the diagnosis of gliomas,” said Mark Gilbert, MD, a senior investigator and chief of the Neuro-Oncology Branch, which is part of both the National Cancer Institute and the National Institute of Neurological Disorders and Stroke. “But in the practice of neuro-oncology we have patients with glioblastoma who have lived well beyond what we estimated, and we have patients with low-grade gliomas who have done terrible. There have always been outliers, and we had no idea why.
“Now we know that there are differences in the biology of the tumor, and these are becoming more predictable based on genetic findings. This is a whole conceptual change,” he said. “It also shows the power of doing broad scale collaborations.
“We can now decide what treatment is best for a particular genetic sub-group,” he added. “We also need to ask how aggressive we should be in treating someone who might have a 15-year survival. Do we want to put them at risk of the side effects of the treatments we use now?”
Patrick Y. Wen, MD, FAAN, the director of the Center for Neuro-Oncology at the Dana Farber Cancer Institute and a professor of neurology at Harvard Medical School, agreed that “the classification of these tumors will be completely changed. In addition to sending a tumor sample to neuropathology, we will be ordering genetic tests on the tumor. This will make a huge difference. Each subgroup comes with a different prognosis.
“We will be able to tell patients who have an IDH mutation that they have a slow-growing tumor,” he said. “And we will also be able to tell patients without an IDH mutation that their prognosis is more serious and we should be more aggressive with treatment from day one.”
EXPERTS: ON MOLECULAR SUBGROUPS ASSOCIATED WITH GLIOMA OUTCOMES
NEUROLOGY IN THE NEWS: Personalized medicine for gliomas? Investigators identify molecular biomarkers for clinical outcomes of grade II and grade III gliomas. Tune in to a podcast interview with Daniel Lachance, MD, FAAN, an associate professor of neurology and head of the Section of Neuro-oncology at the Mayo Clinic in Rochester, MN, as he discusses how new data on molecular biomarkers for gliomas could change the way gliomas are diagnosed and treated. Dr. Lachance is a coauthor of a June 25 paper in the New England Journal of Medicine. Listen to this edition of the Neurology Today podcast series, Neurology in the News, here: http://bit.ly/NT-glioma-podcast.