Table 2. Clinical and demographic data for HER2-amplified luminal and ordinary-luminal patients.
METABRIC training | METABRIC validation | ROCK | |||||||
---|---|---|---|---|---|---|---|---|---|
Description | HER2-Amp.L. | Ord.-Lum. | p -value | HER2-Amp.L. | Ord.-Lum. | p -value | HER2-Amp.L. | Ord.-Lum. | p -value |
Size [mm] | 25.7 | 25.5 | 0.75 | 24.1 | 24.8 | 0.52 | 23.3 (73) | 21.2 (808) | 0.1 |
Grade | 2.6 | 2.2 | 3.4 ⋅ 10−4 | 2.6 | 2.2 | 8.9 ⋅ 10−5 | 2.7 (40) | 2.1 (482) | 2.9 ⋅ 10−6 |
NPI | 4.3 | 3.8 | 0.0055 | 4.2 | 3.8 | 0.006 | N/A | N/A | N/A |
Age at diagnosis [y.] | 62.3 | 63.6 | 0.67 | 61.7 | 63.5 | 0.31 | 52.4 (72) | 55.9 (772) | 0.02 |
Lymph nodes positive | 2.8 | 1.6 | 0.25 | 2.6 | 1.5 | 0.57 | N/A | N/A | N/A |
PR+/all [%] | 51% | 69% | 0.01 | 45% | 74% | 1.7 ⋅ 10−5 | 62% (37) | 80% (379) | 0.016 |
HER2+/all [%] | 84% | 0% | 5 ⋅ 10−122 | 86% | 0% | 4 ⋅ 10−122 | 86% (21) | 8% (163) | 5.5 ⋅ 10−18 |
Lum A/all [%] | 39% | 69% | 4 ⋅ 10−5 | 35% | 67% | 1 ⋅ 10−5 | 38% | 67% | 5.1 ⋅ 10−8 |
P53 mutation/all [%] | 25% (24) | 7% (275) | 0.007 | 26% (23) | 5% (280) | 8.5 ⋅ 10−4 | N/A | N/A | N/A |
10-year survival rate | 0.57 (47) | 0.76 (588) | 0.02 | 0.53 (44) | 0.73 (581) | 5.2 ⋅ 10−4 | 0.58 (64) | 0.7 (763) | 0.0035 |
Lower limit (2.5%) | 0.42 | 0.72 | 0.38 | 0.69 | 0.47 | 0.67 | |||
Upper limit (97.5%) | 0.77 | 0.81 | 0.74 | 0.78 | 0.72 | 0.74 | |||
Population | 51 | 629 | 51 | 629 | 90 | 975 |
The mean values of tumour size, grade, NPI, the average numbers of positive lymph nodes, and patients’ mean age for the HER2-amplified luminal (HER2-Amp.L.) and ordinary-luminal (Ord.-Lum.) subgroups are listed in this table. The ratios of IHC PR- and HER2-positively measured cases, the prevalence of a p53 mutation, and the constitution of luminal A labels within each group, are provided as percentages. P-values indicating the significance of the difference between the results are also provided for each data set, where the Wilcoxon test was applied to numeric parameters and the Proportion test to Boolean parameters given as percentages. The survival rates correspond to prognostic probabilities, where 1 is the perfect prediction and 0 stands for a certain death; the lower and upper limits correspond to the 95% confidence interval of each Kaplan-Meier curve; the overall Log-rank test p-values of survival curves stratification are listed for each data set. In the METABRIC data set the disease-specific survival information was considered, while in the ROCK data set it corresponds to the relapse-free survival. The number of samples in each subgroup is denoted in the last row. Where the number of samples available for calculation of each characteristic was lower than 95% of the population, it is provided in parenthesis. “N/A” stands for “non-applicable” due to missing data.