Gene expression profiling in NSCLC--age- and sex-specific differences
Posted Feb 13 2010 7:34pm
This week's JAMA (February 1020100 has a paper by Mostertz et al."Age- and Sex-Specific Genomic Profiles in Non-Small Cell Lung Cancer," that examines pathway activation profiles in a retrospective series of patients with NSCLC to identify differences in underlying biology that might explain clinically relevant differences observed with respect to age and sex. Check it out--good stuff!
Mostertz WStevenson MAcharya Cet al. Age- and Sex-Specific Genomic Profiles in Non-Small Cell Lung Cancer. JAMA 2010;303:535-543.
This is a retrospective study from a single institution of 787 "patient tumor samples" from "early-stage" (1-3A) NSCLC and at least 60 months of follow-up combined from four independent microarray gene expression data sets. The majority of patients were male (53%) and smokers (77%) and adenocarcinoma was the most frequent histology (78%) (keep note of this). Three out of the four sets were arrayed using Affymetrix U133A GeneChipswhilst the fourth was arrayed with U95 chips. These data were combined and normalized. In order to study gene expression as a function of age and sexthe patient samples were divided into age younger than 70 (520) and 70 and older (267) and men (414) and women (373). Five-year recurrence-free survival (RFS) is used as the clinical endpoint.
The authors confirm the generally observed findings that patients younger than 70 and women have a better 5Y-RFS in their cohort.
When stratified by agebeta-catenininvasiveness and Src were more likely to be activated in patients younger than 70. The authors found distinct gene expression patterns in high-risk (shorter RFS) patients with compared to low-risk (longer RFS). In patients younger than 70, Src and TNF gene signatures were more likely to be activated in the high-risk subgroup. In older patientsgene signatures for wound healing and invasiveness were more associated with the high-risk group but Kaplan-Meier survival curves were not statistically different between low- and high-risk groups.
When stratified by sexmen showed a higher probability of activation of chromosomal instabilityepigenetic stem cellinvasivenessMycand wound healing pathwayswhilst women were more likely to show activation of E2F1 pathway. In womenthe high-risk cluster was associated with activation of invasiveness and STAT3 pathways compared to the low-risk cluster. In menthe high-risk cluster was associated with STAT3TNFwound healingEGFR pathways compared to the low-risk cluster. But in all pathways,
Finallymultivariate Cox proportional hazards analysis of pathway-based prognostic clusters showed the women and younger than 70 subsets to be statistically significant (womenHR 2.02; <70HR 1.83) but not men or 70 or older groups. Note that the pathway activation signatures in high-risk women and in patients younger than 70 were found to be prognostically independent of stage and histology.
This is a significant paper in the field of lung cancernot only for the authors findings in themselvesbut also because of the technical approach and the relating of expression profiles and already established clinically relevant phenotypes. It seems that the number of older patients are under-represented relative to the general population of lung cancer patients and one line of further research would be to study this population in more detail. SImilarlythere also seems an over-representation of non-smokers and adenocarcinoma histology. This just points to the needas the authors rightly statethe need for further validation of their findings in a largermore representation of population of patients.