Table 2. Multifactorial RFS and RFoT models; relative frequency of covariate inclusion (in 1000 bootstrap samples).
|
RFS model (112 events) Median follow-up=6.2 (IQR 4.4–8.8) years
|
RFoT model (84 events) Median follow-up=5.0 (IQR 4.0–6.0) years
|
|||||
|---|---|---|---|---|---|---|
| Variable | HRa (95% CI) | P-value | Inclusion frequency (%) | HR (95% CI) | P-value | Inclusion frequency (%) |
| Nodal status | 1.82 (1.38, 2.40) | <0.001 | 98.0 | 2.17 (1.57, 2.99) | <0.001 | 100 |
| Tumour Size (cm) | 1.21 (1.10, 1.31) | 0.001 | 95.2 | 1.20 (1.10, 1.30) | 0.001 | 87.0 |
| Cytoplasmic kRASb | 6.05 (2.23, 16.44) | <0.001 | 81.6 | b | 66.0 | |
| Tunel | 1.49 (1.23, 1.81) | <0.001 | 85.1 | c | ||
| Nuclear Akt1 | 0.54 (0.36, 0.82) | <0.001 | 92.3 | c | ||
| Phospho mTOR | 0.33 (0.19, 0.59) | <0.001 | 79.1 | 0.55 (0.33, 0.94) | 0.03 | 72.0 |
| Phospho Raf (ser338) cytoplasmic | 2.12 (1.07, 4.02) | 0.03 | 70.8 | a | 14.2 | |
| Phospho MAPK nuclear | 2.80 (1.72, 4.57) | <0.001 | 79.0 | c | ||
| PgR nuclear | a | 44.0 | a | 15.4 | ||
| PTEN Nuclear | b | 59.5 | b | 85.0 | ||
| Nuclear rKIP | b | 55.5 | a | 13.5 | ||
| Phospho Raf (ser338) nuclear | a | 15.6 | 2.43 (1.16, 5.13) | 0.02 | 59.4 | |
| Grade | a | 22.0 | a | 10.3 | ||
| mTOR | a | 18.4 | a | 12.6 | ||
| Mapk p42/44 cytoplasmic | a | 8.6 | a | 12.0 | ||
| Cytoplasmic AKT2 | a | 10.5 | c | |||
| Phospho HER2 nuclear | a | 10.0 | c | |||
| HER2 | c | 1.43 (1.04, 2.00) | 0.03 | 57.0 | ||
| Nuclear kRAS | c | a | 47.9 | |||
| Tescy | c | a | 44.0 | |||
| H4jrme | c | a | 6.2 | |||
| Nuclear Mapk | c | a | 12.0 | |||
| Bcl2 | c | a | 20.7 | |||
| Tace | c | a | 31.8 | |||
| Tacep | c | a | 16.5 | |||
Abbreviations: CI=confidence interval; HR=hazard ratio; IQR=inter quartile range; RFS=recurrence-free survival; RfoT=recurrence free while on tamoxifen treatment.
Key: a: Excluded as ‘unreliable’ – inclusion frequency was < 50% samples; b: excluded as ‘unstable’ – form not apparent in ⩾50% samples; c: screened out.
a
For biomarkers with linear or polynomial effect, reported HR shows the amount of increase in risk of recurrence per 100 unit changes in the independent variable.
b
Before applying cubic transformation to cytoplasmic kRAS, variable was divided by 100.