Volume 139, Issue 6, Pages 1995-2004.e15 (December 2010) A 4-Gene Signature Predicts Survival of Patients With Resected Adenocarcinoma of the Esophagus, Junction, and Gastric Cardia Christopher J. Peters, Jonathan R.E. Rees, Richard H. Hardwick, James S. Hardwick, Sarah L. Vowler, Chin–Ann J. Ong, Chunsheng Zhang, Vicki Save, Maria O'Donovan, Doris Rassl, Derek Alderson, Carlos Caldas, Rebecca C. Fitzgerald Gastroenterology Volume 139, Issue 6, Pages 1995-2004.e15 (December 2010) DOI: 10.1053/j.gastro.2010.05.080 Copyright © 2010 AGA Institute Terms and Conditions
Figure 1 Heat maps of gene expression associated with (A) survival and (B) the number of involved lymph nodes with the corresponding histograms of Pearson correlation coefficients. On the heat maps samples are ranked on the y-axis in order of (A) increasing survival and (B) increasing number of involved lymph nodes, and the genes associated with this feature are ranked along the x-axis. The blue bars on the histograms represent the distribution of the data and the red lines represent the distribution expected from the Monte Carlo simulation. Gastroenterology 2010 139, 1995-2004.e15DOI: (10.1053/j.gastro.2010.05.080) Copyright © 2010 AGA Institute Terms and Conditions
Figure 2 Examples 0, 1+, 2+, and 3+ scoring from immunohistochemistry staining of TMAs for DCK, PAPSS2, SIRT2, and TRIM44. Gastroenterology 2010 139, 1995-2004.e15DOI: (10.1053/j.gastro.2010.05.080) Copyright © 2010 AGA Institute Terms and Conditions
Figure 3 Kaplan–Meier plots of survival comparing internal validation patients with dysregulation and no dysregulation of (A) DCK (χ2LR, 4.46; 1 df; P = .035), (B) PAPSS2 (χ2LR, 12.11; 1 df; P = .001), (C) SIRT2 (χ2LR, 3.97; 1 df; P = .046), and (D) TRIM44 (χ2LR, 3.35; 1 df; P = .063). Notches represent censored data; P values were calculated using the log-rank (Mantel–Cox) test. Gastroenterology 2010 139, 1995-2004.e15DOI: (10.1053/j.gastro.2010.05.080) Copyright © 2010 AGA Institute Terms and Conditions
Figure 4 Kaplan–Meier plots of survival comparing outcomes for patients with 0 of 4, 1–2 of 4, and 3–4 of 4 genes dysregulated for (A) the complete external validation dataset (χ2LR, 13.582; 2 df; P = 0.001), (B) external validation set patients who were chemotherapy-naive (χ2LR, 4.323; 2 df; P = .038), and (C) external validation dataset patients who had received neoadjuvant chemotherapy (χ2LR, 5.013; 2 df; P = .025). Notches represent censored data; P values were calculated using the log-rank (Mantel–Cox) test. Gastroenterology 2010 139, 1995-2004.e15DOI: (10.1053/j.gastro.2010.05.080) Copyright © 2010 AGA Institute Terms and Conditions
Figure 5 Kaplan–Meier plots of survival comparing external validation patients with dysregulation and no dysregulation of (A) DCK (χ2LR, 0.038; 1 df; P = .85), (B) PAPSS2 (χ2LR, 1.91; 1 df; P = .17), (C) SIRT2 (χ2LR, 1.97; 1 df; P = .16), and (D) TRIM44 (χ2LR, 6.86; 1 df; P = .009). Notches represent censored data; P values were calculated using the log-rank (Mantel–Cox) test. Gastroenterology 2010 139, 1995-2004.e15DOI: (10.1053/j.gastro.2010.05.080) Copyright © 2010 AGA Institute Terms and Conditions
Supplementary Figure 1 Box and whisker plots of the average delta CT for the good and poor prognosis groups for (A) TRIM44 (up-regulated in poor prognosis tumors), (B) ADCY9 (up-regulated in poor prognosis tumors), (C) NEU4 (down-regulated in poor prognosis tumors), and (D) DCK (down-regulated in poor prognosis tumors). Gastroenterology 2010 139, 1995-2004.e15DOI: (10.1053/j.gastro.2010.05.080) Copyright © 2010 AGA Institute Terms and Conditions