Gene Profiling Reveals Novel Subtype of Liver Cancer

Missed Something? The NCI Cancer Bulletin Archive allows you to search every issue of this online publication since January 2004. That's over 100 weeks' worth of articles on a variety of cancer research topics and updates.

Researchers have used gene expression profiling to identify a subtype of liver cancer that had not been recognized previously by conventional diagnostic methods. The team adopted an experimental strategy that involved comparing gene activity across three species. The newly identified tumor subtype has patterns of activity in common with liver stem cells, and the tumors may arise from this population of cells, called adult hepatic progenitor cells.

The subtype, hepatoblast B (HB), is associated with a poor prognosis. It is distinguished from other types of liver cancer by the differential expression of hundreds of genes. The researchers compared gene activity in liver tumors from humans and mice, and validated the HB subtype gene expression signature using an independent cohort of liver cancer patients, according to an article published early online in Nature Medicine.

The variability of liver cancer suggests that the disease may comprise biologically distinct subtypes, including different cells of origin. Both adult hepatocytes and liver stem cells can be a source of tumors, and the new findings indicate that gene expression profiling can provide information about a tumor's cellular origins. Previous studies suggested that cancer cells retain some of the gene expression of the cellular lineage from which the cancer originated.

The study included 61 cases of liver cancer from Chinese individuals, and 14 were classified in the HB subgroup. An analysis of activated pathways in the tumors suggests that the AP-1 complex of transcription factors may be "the driving force in tumorigenesis" of the HB subtype.

"This subtype could not be distinguished from other liver tumors by pathologists, and this further demonstrates the usefulness of gene expression profiling in classifying tumors," notes lead researcher Dr. Snorri Thorgeirsson of NCI's Center for Cancer Research. "As we move toward more precise classification of liver tumors into homogeneous subtypes, we hope to translate these data into useful clinical applications."

Gene-Expression Signature Predictive in Node-Negative Breast Cancer

An international group of investigators has validated a 76-gene prognostic signature for lymph node-negative (LNN) primary breast cancer that showed promising results in an initial single-center study in the Netherlands. The results of the validation study appear online on the Journal of Clinical Oncology Web site.

The investigators used frozen tissue samples from 180 women with LNN breast cancer who had been followed for more than 5 years after diagnosis or who developed distant relapse within 5 years. RNA isolated from the samples was hybridized to custom-made gene chips containing the 76 genes whose expression was hypothesized to predict relapse and survival. Differences in distant metastasis-free survival and overall survival between groups of women predicted as high risk or low risk by the signature were calculated.

The gene signature accurately predicted 27 of 30 relapses that occurred within the first 5 years of follow-up, and correctly identified approximately 50 percent of patients who did not relapse. Results from the predictive gene profile were compared with risk classifications using the St. Gallen and NIH criteria. Using either set of criteria, 29 of the 30 metastatic relapses would have been predicted correctly, but 94 to 98 percent of patients who did not relapse would also have been recommended adjuvant systemic therapy. "Therefore," state the investigators, "application of this gene signature could result in a substantial reduction of the number of LNN patients who would otherwise be recommended for unnecessary adjuvant systemic therapy, in particular avoiding overtreatment by chemotherapy."

Folate in Food Cuts Pancreatic Cancer Risk

Pancreatic cancer is often asymptomatic while it develops; once diagnosed, overall survival is poor, making effective chemoprevention especially important. Researchers in Sweden have found population-based evidence that dietary folate reduces the risk of pancreatic cancer by up to 75 percent, according to a study published in the March 15 Journal of the National Cancer Institute (JNCI).

Folate - also known as folic acid and vitamin B-9 - is found naturally in some leafy green vegetables and fruits, especially citrus. Since 1998, FDA requirements have prompted the addition of synthetic folic acid to many prepared foods. Folic acid affects DNA methylation, which has broad implications for human health.

Information on diet came from a 1997 questionnaire given to nearly 82,000 women and men who took part in two large prospective population studies in Sweden. Results after 6.8 years compared those who consumed the most folate from any source with those who consumed the least, and found a protective effect of 67 percent, but this did not extend to folic acid from dietary supplements and multivitamins. "Although our results suggest that increased consumption of foods naturally rich in folate may be beneficial, they do not encourage increased use of supplements for the prevention of pancreatic cancer," said Dr. Susanna C. Larsson of the Karolinska Institutet in Stockholm.

The dose-response findings suggest the more dietary folate, the better the protection against the disease. Alcohol consumption and smoking have been linked to pancreatic cancer risk in other studies, but neither showed any interaction with dietary folate in these data.

Lung Cancer Screening Trial Tests Silenced Genes as Biomarkers

A clinical trial reports that testing the sputum of individuals at high risk of lung cancer for silenced genes shows promise as a potential screening tool for detecting early signs of the disease. The experimental test, which is not yet ready for clinical use, screens 14 genes that are inactivated at different stages of lung cancer for the presence of chemicals called methyl groups that can attach to genes and silence them. The process is called hypermethylation; previous studies have suggested that hypermethylated genes could be biological markers for cancer.

Six of the 14 genes - p16, PAX5-ß, MGMT, DAPK, GATA5, and RASSF1A - were associated with a 50 percent increased risk of lung cancer. Participants who had three or more of these methylated, silenced genes in sputum that was collected within 18 months of diagnosis had a 6.5-fold increased risk for lung cancer. The test identified 65 percent of individuals who later developed symptoms of lung cancer, but it also identified 35 percent of cancer-free control participants, according to findings in the March 15 Cancer Research.

Dr. Steven Belinsky, who directs the Lung Cancer Program at the Lovelace Respiratory Research Institute in Albuquerque, led the research. Participants came from the University of Colorado Cancer Center Sputum Screening Cohort Study, an ongoing prospective study initiated in 1993 to determine whether mucus that coats all parts of the lung might contain genetic evidence of cancer cells when expelled in sputum.

The test performed poorly when the sputum was collected more than 18 months before lung cancer was diagnosed. A person who tests positive would receive a follow-up diagnostic bronchoscopy or x-ray to determine if tumors exist, according to Dr. Belinsky. If tumors are not evident, patients could be retested in several months.

Childhood Leukemia Incidence and Influenza in Great Britain

Researchers from the University of Oxford in England have identified small peaks in the incidence of childhood acute lymphoblastic leukemia (ALL) in Great Britain in 1976 and 1990 that immediately followed winter influenza epidemics. While their time-trends analysis was not designed to measure causation, the observed association supports the hypothesis that some types of childhood leukemia may be triggered by a challenge to the immune system. The results were published in the March 15 JNCI.

The investigators collected ALL incidence data from the National Registry of Childhood Tumours (NRCT) for children younger than 15 in England, Scotland, and Wales between 1974 and 2000. Cases were categorized as either CD10-positive precursor B-cell ALL (cALL) or non-cALL.

An increasing trend in the incidence of ALL between 1974 and 2000 was identified. Between 1980 and 1996, this trend was apparently due to a specific increase in the cALL subtype, suggesting that the cause of cALL differs from that of other childhood leukemias. The peaks in the incidence of ALL in 1976 and 1990, observed after winter influenza outbreaks, were also caused by a specific increase in cALL.

Their results, state the investigators, "are consistent with hypotheses suggesting that some childhood leukemia may be triggered by infection occurring close to the time of diagnosis of leukemia…and they raise the possibility that influenza may sometimes be involved."