Predictor | | Coefficient (SE) | Odds ratio (95% CI) | p value |
|---|
Intercept | | − 1.9686 (0.9268) | – | – |
Age (years) | Per year | − 0.0108 (0.0069) | 0.99 (0.98–1.00) | 0.12 |
100/BMI | Per 100 m2/kg | 0.2951 (0.1141) | 1.34 (1.07–1.68) | < 0.01 |
Tumor stage | Per stage | − 0.4824 (0.1212) | 0.62 (0.49–0.78) | < 0.0001 |
Grade | Per grade | 0.4206 (0.2274) | 1.52 (0.98–2.38) | 0.06 |
Lymph node status | cN0 | 0 | 1 | – |
| | cN + | − 0.2430 (0.1701) | 0.78 (0.56–1.09) | 0.15 |
Tumor type | Other | 0 | 1 | – |
| | Ductal | − 0.1559 (0.1924) | 0.86 (0.59–1.25) | 0.42 |
| | Lobular | − 0.4229 (0.5363) | 0.66 (0.23–1.87) | 0.43 |
ER | Per percent | − 0.0137 (0.0031) | 0.99 (0.98–0.99) | < 0.0001 |
PgR | Per percent | − 0.0190 (0.0053) | 0.98 (0.97–0.99) | < 0.01 |
Ki-67 | Per percent | 0.0170 (0.0044) | 1.02 (1.01–1.03) | < 0.01 |
- Regression coefficients with standard errors from the final regression model, associated odds ratios with 95% confidence intervals, and p values for Wald tests are shown. The p values and confidence intervals should be regarded as measures of importance within the regression model rather than measures of significance, especially since the conditions for statistical testing may not be fulfilled after a model selection process. The predicted probability Prob for pCR can be calculated using the following formula: Prob = exp(Z)/(1 + exp(Z)) with Z = − 0.0190 + 0.9714 X and X = − 1.9686 − 0.0108 age − 29.51/BMI − 0.4824 tumor stage + 0.4206 grade − 0.2430 cN − 0.1559 ductal − 0.4229 lobular − 0.0137 ER − 0.0190 PgR + 0.0170 Ki-67. Note that cN, ductal, and lobular values are one when present and zero when absent. 0.9714 is the shrinkage factor, and − 0.0190 is a correction term. Multiplying the result by 100 provides percentage values
- BMI, body mass index; ER, estrogen receptor (expression); pCR, pathological complete response; PgR, progesterone receptor (expression); SE, standard error; CI, confidence interval
