Objective:
Lung cancer is the leading cause of cancer-related mortality worldwide. Disease stage still remains the best prognostic factor for patients with localized non-small cell lung cancer. The TNM staging system, however, does not address the heterogeneity of this disease. Sub-classification and identification of distinct prognostic sub-groups within each stage may allow the optimization of clinical trial design and potentially improve outcome. This is a retrospective pilot study, in which we attempt to identify genomic biomarkers predictive of recurrence in stage I lung cancer by analysing copy number (CN) data obtained by next-generation sequencing.
Materials and Methods:
Ninety eight patients with stage I NSCLC, who underwent elective radical surgery were identified from a tissue bank of 323 tumour samples. Their demographic and surgical data, including their recurrence status were collected and an extensive database compiled. The cases were split into two cohorts depending on their histology (adenocarcinoma vs. squamous cell carcinoma). Formalin-fixed paraffin-embedded blocks were retrieved from the local pathology archive and DNA was extracted from macrodissected tumour tissue using the QiAmp DNA microkit. DNA libraries were prepared and samples were sequenced using Illumina Genome Analyzer II. The frequency of CN gain and loss along the entire genome was compared between the recurrent and non-recurrent cancers.
Results:
Comparative whole genome maps of the recurrent and non-recurrent cohort did not show any significant differences. Attempts to distinguish the recurrent from the non-recurrent cohorts with previously published algorithms, based on whole genome CN variation were also unsuccessful. However, a newly devised logistic regression model based on pan-genomic assessment of CN variation was able to differentiate recurrent from non-recurrent cancers in both histological subtypes.
Conclusion:
Although no single chromosomal region was associated with cancer recurrence, the two groups were distinguishable with an algorithm that assesses total genomic change. Analysis of a larger cohort will be required for validation.
