Transcriptional Changes in Blood May Predict Lung Cancer Onset

Researchers of University Hospital Cologne (Germany) designed the predictor transcript in a cohort of smokers with prevalent lung cancer consisting of 13 cases and 11 controls. They then validated it in a second cohort of smokers, also with prevalent lung cancer, consisting of 22 cases and 15 controls. The researchers then used data from the European Prospective Investigation on Cancer and Nutrition (EPIC) trial and looked at all smokers among 25,000 subjects who developed lung cancer within 2 years of the study, a total of 12 cases. The RNA-fingerprint established in the two prevalent cohorts was used to predict future onset of lung cancer in the incident cohort, and the results were validated on two different array platforms.

The study results showed that when using the human WG-6 array in the cohort of smokers with prevalent lung cancer sensitivity and specificity of the test were estimated to be 86% and 86%, respectively. Class prediction was highly significant using K-Nearest Neighbor (KNN) based algorithms or hierarchical clustering. When further applying the test in the cohort with incident lung cancer the researchers significantly predicted the clinical manifestation of lung cancer. Using identical test criteria on data generated on a second, different technical platform, similar results were obtained. Permutation analysis further supported the validity of the data. The study was presented at the American Society of Clinical Oncology (ASCO) annual meeting, held during May-June 2008 in Orlando (FL, USA).

"A lung-cancer-specific expression profile is present in blood,” said lead author Thomas Zander, M.D., of the hematology and oncology department. "Prevalent lung cancer can be detected with an accuracy of 88%, and incident lung cancer can be detected with an accuracy of 80% prior to clinical manifestation.”


Related Links:
University Hospital Cologne