Table 5.
Associations between Clusters #1–3 and individual genes using the NKI295 sample set
| Cluster 1 | Cluster 2 | Cluster 3 | ||||
| % | p-val | % | p-val | % | p-val | |
| EGFR | 39% | 0.1783 | 43% | 0.0091b | 38% | 0.15 |
| HER2 | 26% | 0.0017 | 25% | <0.0001c | 24% | <0.0001a |
| HER4* | 21% | <0.0001 | 12% | <0.0001 | 18% | <0.0001 |
| TGFA | 40% | 0.0665 | 48% | 0.0002 | 47% | 0.0021 |
| AREG | 22% | 0.0007c | 23% | <0.0001a | 28% | 0.064f |
| EGF | 35% | 0.1380 | 25% | 0.0691 | 27% | 0.033d |
| CRYAB | 35% | 0.3214f | 38% | 0.0524 | 38% | 0.0013 |
| KRAS amplicon | 38% | 0.1973e | 52% | <0.0001c | 63% | <0.0001a |
| KRAS gene | 27% | 0.0022a | 31% | 0.8795 | 36% | 0.14e |
| HRAS | 48% | <0.0001c | 51% | <0.0001 | 47% | 0.0018 |
| NRAS | 45% | 0.0362 | 56% | <0.0001c | 59% | <0.0001a |
| PIK3ca | 22% | 0.0032b | 27% | 0.1415e | 30% | 0.33e |
| PIK3R1 | 24% | 0.0009a | 20% | <0.0001a | 19% | <0.0001 |
| AKT1 | 41% | 0.0112 | 39% | 0.0899 | 34% | 0.36 |
| AKT2* | 40% | 0.0519 | 37% | 0.3524 | 33% | 0.94 |
| AKT3 | 26% | 0.0004 | 33% | 0.1569 | 35% | 0.64f |
| MEK1 | 39% | 0.0335 | 47% | 0.0032d | 48% | <0.0001 |
| MEK2 | 58% | <0.0001a | 44% | 0.0113d | 36% | 0.55f |
| ERK1 | 37% | 0.0718e | 23% | 0.0009c | 19% | <0.0001a |
| ERK2 | 39% | 0.0238 | 37% | 0.3457e | 36% | 0.46e |
Chi-squared analyses were used to identify associations between the high expression of the individual EGFR-activation profiles for each cluster (top 1/3) and the expression of individual genes categorized as high (top 1/3). The % of tumors with the high expression of each cluster and that show the high expression of the individual gene is shown.
*Note: HER4 could not be assessed in UNC data due to too many missing values; HER3 was not present in the NKI data set; AKT2 was not present in the UNC dataset.
a the statistically significant association was also significant in the UNC data set (p < 0.0025).
b the association was nominally significant in the NKI dataset (p < 0.05), but significant in the UNC dataset (p < 0.0025).
c the association was significant in the NKI dataset (p < 0.0025), but nominally significant in the UNC dataset (p < 0.05).
d the association was nominally significant in both datasets (p < 0.05).
e the association was significant in UNC dataset (p < 0.0025).
f the association was nominally significant in the UNC dataset (p < 0.05).
Bonferroni corrected level of significance α = 0.0025