Purpose: Enhanced prognostication power is becoming more desirable in clinical oncology. In this study, we explored the prognostic potential of multigene hypermethylation profiling in non-small-cell lung cancer.
Experimental design: We evaluated a panel of eight genes (p16, APC, ATM, hMLH1, MGMT, DAPK, ECAD, and RASSF1A) using methylation-specific PCR in 105 archived specimens of non-small-cell lung cancer representing all stages of the illness. We analyzed the effect of gene methylation status on outcome individually in a cumulative manner and in a combinatorial approach using recursive partitioning to identify methylation profiles, which affect overall survival.
Results: In this data set, tumors harboring promoter hypermethylation at two or more genes exhibit similar survival trends to others in the cohort. Using recursive partitioning, three genes (APC, ATM, and RASSF1A) emerged as determinants of prognostic groups. This designation retained its statistical significance even when disease stage and age were entered into a multivariate analysis. Using this approach, patients whose tumors were hypermethylated at APC and those hypermethylated at only ATM (not also at APC or RASSF1A) enjoyed substantially longer 1- and 2-year survival than patients in the remaining groups. In 32 adjacent histologically normal lung tissue specimens, we detected similar methylation abnormalities.
Conclusion: Assessment of promoter hypermethylation aberrations may facilitate prognostic profiling of lung tumors, but validation in independent data sets is needed to verify these profiles. This system uses material that is abundantly available with linked outcome data and can be used to generate reliable epigenetic determinants.