TABLE II.
Results from Univariate Linear Regression Models for Prediction of MRI Diagnostic Accuracy Based on the Absolute Difference Between MRI and Pathological Residual Tumor Size
| Number of subject |
MRI-pathology tumor size difference (cm) |
Regression model coefficient of determination |
Regression model slope parameter |
|
|---|---|---|---|---|
|
|
||||
| Clinical and tumor characteristic | N (%) | Mean ± Stdev | R2 | P-value |
| Tumor type | 0.16436 | <0.0001 | ||
| IDC | N = 85 (87%) | 0.69 ± 1.20 | ||
| ILC and mixed | N = 13 (13%) | 3.07 ± 4.13 | ||
| Tumor morphology | 0.08448 | 0.0039 | ||
| Mass lesion | N = 74 (76%) | 0.69 ± 1.00 | ||
| Non-mass-like | N = 23 (24%) | 2.06 ± 3.59 | ||
| Tumor grade | 0.0385 | 0.0541 | ||
| 4–7 (low–medium) | N = 60 (62%) | 1.32 ± 2.44 | ||
| 8–9 (high) | N = 37 (38%) | 0.51 ± 0.79 | ||
| Hormonal receptor | 0.05961 | 0.0154 | ||
| Positive | N = 57 (58%) | 1.42 ± 2.49 | ||
| Negative | N = 41 (42%) | 0.43 ± 0.68 | ||
| HER-2 receptor | 0.07846 | 0.0055 | ||
| Positive | N = 40 (41%) | 0.34 ± 0.50 | ||
| Negative | N = 57 (59%) | 1.48 ± 2.50 | ||
| MR scanner | 0.04395 | 0.0383 | ||
| 1.5 T | N = 51 (52%) | 0.60 ± 1.48 | ||
| 3.0 T | N = 47 (48%) | 1.44 ± 2.39 | ||
| Chemotherapy regimen | 0.01281 | 0.2672 | ||
| AC + Taxane | N = 63 (64%) | 0.83 ± 2.16 | ||
| Taxane without AC | N = 35 (36%) | 1.31 ± 1.66 | ||
| Days to operation | 0.00249 | 0.6259 | ||
| 0–30 days | N = 43 (44%) | 0.89 ± 2.22 | ||
| >30 days | N = 55 (56%) | 1.09 ± 1.84 | ||