Article Figures & Data
Figures
- Fig. 1.
Flow chart describing the number of samples and candidate genes through the stages for phase I. First, Affymetrix microarrays (Human Genome U133 Plus 2.0 Array) were used to select those genes differentially expressed in bladder cancer (12). Second, TAs were used to analyze samples. In this phase, the number of samples was progressively increasing, while the number of genes was progressively decreasing.
- Fig. 2.
Heat map representation of unsupervised hierarchical clustering of diagnostic and prognostic genes (phase I). Samples are separated by pathologic stage and grade; control, LG nonmuscle invasive bladder cancer (NMIBC; Ta and T1 LG), HG NMIBC (Ta and T1 HG and Tis), and muscle invasive bladder cancer (MIBC; T2, T3, and T4). Differentially expressed genes (P < 0.05) with a median log 2 ratio > ±1 were used as input in 384-TA and 96-TA clusters. All the 45 genes contained in the 48-TA were used as input in these clusters. A and B, 384-TA diagnostic and prognostic heat maps, respectively. C and D, 96-TA diagnostic and prognostic heat maps, respectively. E and F, 48-TA diagnostic and prognostic heat maps, respectively. It is of note that in the overexpressed gene set of the 48-TA diagnostic heat map (E), two gene subclusters can be appreciated: one cluster that contains genes overexpressed mainly in HG tumors and a second cluster composed of genes overexpressed in tumors. The 14 underexpressed genes in tumor samples are relatively overexpressed in HG tumors. Rows, individual genes; columns, experimental samples. High expression levels are in red and low expression levels in green.
- Fig. 3.
Heat map representation of unsupervised hierarchical clustering of diagnostic and prognostic genes (phase II). Samples are separated by pathologic stage and grade; control, LG NMIBC (Ta and T1 LG), HG NMIBC (Ta and T1 HG and Tis), and MIBC (T2, T3, and T4). All the 45 genes contained in the 48-TA-urine were used as input in this cluster. A, 48-TA-urine diagnostic heat map. B, 48-TA-urine prognostic heat map. Rows, individual genes; columns, experimental samples. High expression levels are in red and low expression levels in green.
Tables
- Table 1.
Clinicohistopathologic features of the 576 bladder cancer patients and controls included in this study
Phase I Phase I (n total) Phase II (n total) n total 384-TA 96-TA 48-TA n males 38 26 214 278 170 448 n females 1 9 77 87 41 128 Male/female ratio 38 2.89 2.78 2.78 4.15 3.5 Average age 69.82 65.68 65.73 66.16 68.73 67.1 Tumor urine samples Tis 4 2 3 9 2 11 Ta LG −CIS 6 — 97 103 39 142 +CIS 2 — 4 6 — 6 Ta HG −CIS 3 7 19 29 9 38 +CIS 2 4 3 9 1 10 T1 LG −CIS 3 — 4 7 5 12 +CIS — — 2 2 — 2 T1 HG −CIS 3 3 15 21 11 32 +CIS 2 4 10 16 4 20 T2, T3, T4 HG −CIS 7 1 20 28 22 50 +CIS 1 4 9 14 4 18 Total tumors 33 25 186 244 97 341 Control urine samples Total controls 6 10 105 121 114 235 n total 39 35 291 365 211 576 Abbreviation: CIS, carcinoma in situ.
- Table 2.
Description of the genes selected in each of the three steps of the first phase
Selected differentially expressed genes (P < 0.05) Other selected genes Diagnostic and prognostic genes Diagnostic genes Prognostic genes Outliers/Bibliography Endogenous controls n n overexpressed n underexpressed n n overexpressed n underexpressed n n overexpressed n underexpressed Affymetrix→ 216 Diagnostic 88 128 66 16 50 63 41 22 29 7 + 3 384 TA 384 Prognostic 191 25 TA 384→ 25 Diagnostic 23 2 48 31 17 9 5 4 11 3 96 TA 96 Prognostic 21 4 TA 96→ 14 Diagnostic 12 2 20 18 2 5 4 1 6 3 48 TA 48 Prognostic 11 3 NOTE: Genes are over- or under-expressed in tumor or HG samples with respect to control or LG tumors, respectively. Diagnostic genes: differentially expressed genes between tumor and control samples. Prognostic genes: differentially expressed genes between LG and HG tumor samples (those genes that predict tumor aggressiveness). Genes could be differentially expressed either for diagnosis only or for prognosis only, or for both.
- Table 3.
Diagnostic performance of the 12-gene set model in different sets of samples
Gene set model from: Phase I Phase I Phase II Samples from: Phase I Phase II Phase II Overall n tumors/n controls 244/121 97/114 97/114 SN 98 70 89 SP 99 96 95 PPV 100 93 93 NPV 95 79 91 Low-risk NMIBC (Ta or T1 LG without associated CIS) n tumors/n controls 110/121 44/114 44/114 SN 96 57 80 SP 99 96 95 PPV 99 83 85 NPV 97 85 92 High-risk NMIBC (Ta or T1 LG with associated CIS, Ta or T1 HG, Tis) n tumors/n controls 92/121 27/114 27/114 SN 100 81 93 SP 99 96 95 PPV 99 81 81 NPV 100 96 98 MIBC (T2, T3, T4 HG, with or without associated CIS) n tumors/n controls 42/121 26/114 26/114 SN 95 81 100 SP 99 96 95 PPV 98 81 81 NPV 98 96 100 Abbreviations: PPV, positive predictive value; NPV, negative predictive value.
- Table 4.
Prognostic performance of the 12+2 gene set model
Gene set model from: Phase I Phase I Phase II Samples from: Phase I Phase II Phase II n tumors 244 97 97 SN 79 42 79 SP 92 70 91 PPV 92 63 91 NPV 78 50 78
Supplementary Data
Files in this Data Supplement:
Volume 16, Issue 9
