Relationship between Clinicopathologic Variables in Breast Cancer Overall Survival Using Biogeography-Based Optimization Algorithm
Table 3
Multivariate Cox regression analysis of clinicopathologic variables in breast cancer overall survival.
Variable
Comparison
Fully adjusted
Stepwise selected
BBO selected
Model 11
Model 22
Model 33
C-statistic
76%
78%
78%
77%
80%
p-value
<0.001
<0.001
<0.001
<0.001
<0.001
HR
p-value
HR
p-value
HR
p-value
HR
p-value
HR
p-value
Age
50 years
50 years
0.515
0.015
0.589
0.036
0.523
0.003
Grade
III
I, II
1.524
0.132
1.830
0.026
1.642
0.073
Her 2
Positive
Negative
0.103
0.103
Estrogen receptor
Positive
Negative
0.506
0.195
0.451
0.003
Progesterone receptor
Positive
Negative
0.801
0.595
Tumor size (AJCC stage)
2.4 cm
2.4 cm
2.430
0.005
3.017
<0.001
2.435
0.004
2.436
0.004
2.372
0.006
Lymph node (AJCC stage)
Positive
Negative
1.511
0.103
1.301
0.038
Lymphovascular invasion
Positive
Negative
1.776
0.046
1.878
0.016
1.642
0.068
1.753
0.038
1.606
0.096
Dermal invasion
Positive
Negative
1.549
0.298
2.152
0.048
1.722
0.188
1.548
0.028
Perineural invasion
Yes
No
1.745
0.081
1.994
0.028
1.827
0.060
1.546
0.168
Surgery method
Total mastectomy
Partial mastectomy
1.735
0.087
1.869
0.031
1.848
0.066
1.633
0.092
Radiotherapy
Without RT
With RT
1.299
0.378
Chemotherapy
Without CT
With CT
1.338
0.379
1.234
0.491
Hormone therapy
Without HT
With HT
1.022
0.962
2.231
0.002
1.918
0.014
2.178
0.003
Target therapy
Without TT
With TT
0.859
0.697
Stepwise selection met the p = 0.05 level for entry into the model; 2 Stepwise selection met the p = 0.1 level for entry into the model; 3 Stepwise selection met the p = 0.2 level for entry into the model. p-value is estimated using Cox PH regression. HR: hazard ratio; HER2: human epidermal growth factor receptor 2.