Limits...
Combined clinical and genomic signatures for the prognosis of early stage non-small cell lung cancer based on gene copy number alterations.

Aramburu A, Zudaire I, Pajares MJ, Agorreta J, Orta A, Lozano MD, Gúrpide A, Gómez-Román J, Martinez-Climent JA, Jassem J, Skrzypski M, Suraokar M, Behrens C, Wistuba II, Pio R, Rubio A, Montuenga LM - BMC Genomics (2015)

Bottom Line: The signatures were validated in two independent datasets (n = 73 for ADC, n = 97 for SCC), confirming that the prognostic value of both clinical-genomic models is robust, statistically significant (P = 0.008 for ADC and P = 0.019 for SCC) and outperforms both the clinical models (P = 0.060 for ADC and P = 0.121 for SCC) and the genomic models applied separately (P = 0.350 for ADC and P = 0.269 for SCC).The present work provides a methodology to generate a robust signature using copy number data that can be potentially used to any cancer.Using it, we found new prognostic scores based on tumor DNA that, jointly with clinical information, are able to predict overall survival (OS) in patients with early-stage ADC and SCC.

View Article: PubMed Central - PubMed

Affiliation: Group of Bioinformatics, CEIT and TECNUN, University of Navarra, San Sebastian, Spain.

ABSTRACT

Background: The development of a more refined prognostic methodology for early non-small cell lung cancer (NSCLC) is an unmet clinical need. An accurate prognostic tool might help to select patients at early stages for adjuvant therapies.

Results: A new integrated bioinformatics searching strategy, that combines gene copy number alterations and expression, together with clinical parameters was applied to derive two prognostic genomic signatures. The proposed methodology combines data from patients with and without clinical data with a priori information on the ability of a gene to be a prognostic marker. Two initial candidate sets of 513 and 150 genes for lung adenocarcinoma (ADC) and squamous cell carcinoma (SCC), respectively, were generated by identifying genes which have both: a) significant correlation between copy number and gene expression, and b) significant prognostic value at the gene expression level in external databases. From these candidates, two panels of 7 (ADC) and 5 (SCC) genes were further identified via semi-supervised learning. These panels, together with clinical data (stage, age and sex), were used to construct the ADC and SCC hazard scores combining clinical and genomic data. The signatures were validated in two independent datasets (n = 73 for ADC, n = 97 for SCC), confirming that the prognostic value of both clinical-genomic models is robust, statistically significant (P = 0.008 for ADC and P = 0.019 for SCC) and outperforms both the clinical models (P = 0.060 for ADC and P = 0.121 for SCC) and the genomic models applied separately (P = 0.350 for ADC and P = 0.269 for SCC).

Conclusion: The present work provides a methodology to generate a robust signature using copy number data that can be potentially used to any cancer. Using it, we found new prognostic scores based on tumor DNA that, jointly with clinical information, are able to predict overall survival (OS) in patients with early-stage ADC and SCC.

No MeSH data available.


Related in: MedlinePlus

Kaplan Meier curves for the training (a, b) and validation (c, d) sets of ADC patients. For each case, patients were divided into two risk groups according to the predicted risk using either clinical (a, c) or clinical-genomic data (b, d). Survival curves were compared using log-rank test p-values
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC4595201&req=5

Fig2: Kaplan Meier curves for the training (a, b) and validation (c, d) sets of ADC patients. For each case, patients were divided into two risk groups according to the predicted risk using either clinical (a, c) or clinical-genomic data (b, d). Survival curves were compared using log-rank test p-values

Mentions: The prognostic role of the clinical-genomic signatures was evaluated in the training set. Risk scores were calculated according to the clinical-genomic signatures and dichotomized with the medians of the scores (therefore, in each histological subtype, the low and high risk groups included the same number of patients). Furthermore, the prognostic capacity of the signatures was validated in two independent datasets, one for each histological subtype. All prognostic significances of both the clinical-genomic and the clinical models are shown in Table 2 and Figs. 2 and 3. The clinical-genomic signatures outperformed the clinical signatures in both the training and validation sets, i.e. the p-values were smaller for the clinical-genomic models than for the clinical models and a wider separation of the Kaplan-Meier survival curves was clearly observed.Table 2


Combined clinical and genomic signatures for the prognosis of early stage non-small cell lung cancer based on gene copy number alterations.

Aramburu A, Zudaire I, Pajares MJ, Agorreta J, Orta A, Lozano MD, Gúrpide A, Gómez-Román J, Martinez-Climent JA, Jassem J, Skrzypski M, Suraokar M, Behrens C, Wistuba II, Pio R, Rubio A, Montuenga LM - BMC Genomics (2015)

Kaplan Meier curves for the training (a, b) and validation (c, d) sets of ADC patients. For each case, patients were divided into two risk groups according to the predicted risk using either clinical (a, c) or clinical-genomic data (b, d). Survival curves were compared using log-rank test p-values
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC4595201&req=5

Fig2: Kaplan Meier curves for the training (a, b) and validation (c, d) sets of ADC patients. For each case, patients were divided into two risk groups according to the predicted risk using either clinical (a, c) or clinical-genomic data (b, d). Survival curves were compared using log-rank test p-values
Mentions: The prognostic role of the clinical-genomic signatures was evaluated in the training set. Risk scores were calculated according to the clinical-genomic signatures and dichotomized with the medians of the scores (therefore, in each histological subtype, the low and high risk groups included the same number of patients). Furthermore, the prognostic capacity of the signatures was validated in two independent datasets, one for each histological subtype. All prognostic significances of both the clinical-genomic and the clinical models are shown in Table 2 and Figs. 2 and 3. The clinical-genomic signatures outperformed the clinical signatures in both the training and validation sets, i.e. the p-values were smaller for the clinical-genomic models than for the clinical models and a wider separation of the Kaplan-Meier survival curves was clearly observed.Table 2

Bottom Line: The signatures were validated in two independent datasets (n = 73 for ADC, n = 97 for SCC), confirming that the prognostic value of both clinical-genomic models is robust, statistically significant (P = 0.008 for ADC and P = 0.019 for SCC) and outperforms both the clinical models (P = 0.060 for ADC and P = 0.121 for SCC) and the genomic models applied separately (P = 0.350 for ADC and P = 0.269 for SCC).The present work provides a methodology to generate a robust signature using copy number data that can be potentially used to any cancer.Using it, we found new prognostic scores based on tumor DNA that, jointly with clinical information, are able to predict overall survival (OS) in patients with early-stage ADC and SCC.

View Article: PubMed Central - PubMed

Affiliation: Group of Bioinformatics, CEIT and TECNUN, University of Navarra, San Sebastian, Spain.

ABSTRACT

Background: The development of a more refined prognostic methodology for early non-small cell lung cancer (NSCLC) is an unmet clinical need. An accurate prognostic tool might help to select patients at early stages for adjuvant therapies.

Results: A new integrated bioinformatics searching strategy, that combines gene copy number alterations and expression, together with clinical parameters was applied to derive two prognostic genomic signatures. The proposed methodology combines data from patients with and without clinical data with a priori information on the ability of a gene to be a prognostic marker. Two initial candidate sets of 513 and 150 genes for lung adenocarcinoma (ADC) and squamous cell carcinoma (SCC), respectively, were generated by identifying genes which have both: a) significant correlation between copy number and gene expression, and b) significant prognostic value at the gene expression level in external databases. From these candidates, two panels of 7 (ADC) and 5 (SCC) genes were further identified via semi-supervised learning. These panels, together with clinical data (stage, age and sex), were used to construct the ADC and SCC hazard scores combining clinical and genomic data. The signatures were validated in two independent datasets (n = 73 for ADC, n = 97 for SCC), confirming that the prognostic value of both clinical-genomic models is robust, statistically significant (P = 0.008 for ADC and P = 0.019 for SCC) and outperforms both the clinical models (P = 0.060 for ADC and P = 0.121 for SCC) and the genomic models applied separately (P = 0.350 for ADC and P = 0.269 for SCC).

Conclusion: The present work provides a methodology to generate a robust signature using copy number data that can be potentially used to any cancer. Using it, we found new prognostic scores based on tumor DNA that, jointly with clinical information, are able to predict overall survival (OS) in patients with early-stage ADC and SCC.

No MeSH data available.


Related in: MedlinePlus