On 2014 Feb 17, Jesus Mendez-Gonzalez commented:

This is a very interesting study in terms of early diagnosis of lung cancer. DNA methylation markers are especially promising in this area, as they can be studied in multiple specimens, arise even in premalignant lesions and could be used to complement low-dose CT screening, reducing its associated and problematic high false positive rate. To date, most studies have focused on hypothesis-driven targets, thus reducing the chance of identifying the best (but hidden) candidates. However, and importantly, the authors chose a wide, blind and functional identification method and validated the best candidates taking advantage of the TCGA database and two additional cohorts of lung cancer samples. They, finally, come up to three methylation markers that show an impressive sensitivity with 100% specificity. The real value of this panel is still to be determined: no data in fluid samples (e.g. bronchoaspirates or sputum) is available, and MSP technique (which has to deal with the problem of false positives due to potential incomplete bisulphite conversion) was only performed in 7 normal samples. However, the approach is really encouraging. The potential prognostic value of these alterations was also explored, but no relation to survival was found. For the authors this is somehow expected, due to the absence of “an established role in the pathogenesis of lung cancer and/or an extremely high prevalence of methylation”. This is reasonable, but there are some points that would be worth to discuss:

• As the authors point, “TCGA samples are not annotated for therapies received, therefore no control for treatment in analysis is possible”. Surely, recurrence after surgery –especially in early-stages tumors, where no adjuvant therapy is usually administered- would be a better end point, but no information is available.

• Unfortunately, this paper was published shortly after this one: “A prognostic DNA methylation signature for stage I non-small-cell lung cancer” ( http://www.ncbi.nlm.nih.gov/pubmed/24081945 ). Thus, there was no possibility of discussing the commonalities and differences that both studies found. In this article we analyzed the DNA methylation of a wide amount of NSCLC samples through the Infinium 450k array, the same platform as TCGA used (these data are also available for public analysis). Consistently, CDO1 and HOXA9 were found as largely differentially methylated between tumors and normal tissues. Additionaly, we studied the relationship between DNA methylation and recurrence in stage I resected tumors, where no adjuvant and potentially confounding treatments were given. Contrary to Wrangle et al. we found that HOXA9 promoter hypermethylation correlated with a worse progression free survival. These data were validated by pyrosequencing in an independent cohort.

  Two main differences have to be underscored between the studies. The first one has to do with the different end-points analyzed (time to death vs progression free survival). The second one is that we set a higher threshold to classify a sample as hypermethylated (0.4 vs 0.2). If our results are validated in larger and independent cohorts, there are at least two possibilities to explain these results: i) HOXA9 exerts an unknown role in the pathogenesis of lung cancer (some small studies suggest this option: http://www.ncbi.nlm.nih.gov/pubmed/21757291) and ii) HOXA9 “heavy” hypermethylation is a marker of particular malignant lesions more prone to show an aggressive behavior.

On 2014 Feb 17, Jesus Mendez-Gonzalez commented:

This is a very interesting study in terms of early diagnosis of lung cancer. DNA methylation markers are especially promising in this area, as they can be studied in multiple specimens, arise even in premalignant lesions and could be used to complement low-dose CT screening, reducing its associated and problematic high false positive rate. To date, most studies have focused on hypothesis-driven targets, thus reducing the chance of identifying the best (but hidden) candidates. However, and importantly, the authors chose a wide, blind and functional identification method and validated the best candidates taking advantage of the TCGA database and two additional cohorts of lung cancer samples. They, finally, come up to three methylation markers that show an impressive sensitivity with 100% specificity. The real value of this panel is still to be determined: no data in fluid samples (e.g. bronchoaspirates or sputum) is available, and MSP technique (which has to deal with the problem of false positives due to potential incomplete bisulphite conversion) was only performed in 7 normal samples. However, the approach is really encouraging. The potential prognostic value of these alterations was also explored, but no relation to survival was found. For the authors this is somehow expected, due to the absence of “an established role in the pathogenesis of lung cancer and/or an extremely high prevalence of methylation”. This is reasonable, but there are some points that would be worth to discuss:

• As the authors point, “TCGA samples are not annotated for therapies received, therefore no control for treatment in analysis is possible”. Surely, recurrence after surgery –especially in early-stages tumors, where no adjuvant therapy is usually administered- would be a better end point, but no information is available.

• Unfortunately, this paper was published shortly after this one: “A prognostic DNA methylation signature for stage I non-small-cell lung cancer” ( http://www.ncbi.nlm.nih.gov/pubmed/24081945 ). Thus, there was no possibility of discussing the commonalities and differences that both studies found. In this article we analyzed the DNA methylation of a wide amount of NSCLC samples through the Infinium 450k array, the same platform as TCGA used (these data are also available for public analysis). Consistently, CDO1 and HOXA9 were found as largely differentially methylated between tumors and normal tissues. Additionaly, we studied the relationship between DNA methylation and recurrence in stage I resected tumors, where no adjuvant and potentially confounding treatments were given. Contrary to Wrangle et al. we found that HOXA9 promoter hypermethylation correlated with a worse progression free survival. These data were validated by pyrosequencing in an independent cohort.

  Two main differences have to be underscored between the studies. The first one has to do with the different end-points analyzed (time to death vs progression free survival). The second one is that we set a higher threshold to classify a sample as hypermethylated (0.4 vs 0.2). If our results are validated in larger and independent cohorts, there are at least two possibilities to explain these results: i) HOXA9 exerts an unknown role in the pathogenesis of lung cancer (some small studies suggest this option: http://www.ncbi.nlm.nih.gov/pubmed/21757291) and ii) HOXA9 “heavy” hypermethylation is a marker of particular malignant lesions more prone to show an aggressive behavior.