Table 6.
Multivariate regression analysis with pCR as dependent variable
| Variable | pCR | No pCR | P | Odds Ratio (95%CI) |
|---|---|---|---|---|
| Training cohort | ||||
| Tumor length | 0.037 | 4.99 (1.10–22.63) | ||
| ≦4 cm | 8 (73%) | 22 (37%) | ||
| > 4 cm | 3 (27%) | 38 (63%) | ||
| CEA clearance pattern | 0.013 | 8.30 (1.56–44.17) | ||
| Exponential decrease (R2≧0.9) | 9 (82%) | 22 (37%) | ||
| Non-exponential decrease (R2 < 0.9) | 2 (18%) | 38 (63%) | ||
| Validation cohort | ||||
| Tumor length | 0.484 | 0.59 (0.13–2.59) | ||
| ≦4 cm | 3 (30%) | 28 (43%) | ||
| > 4 cm | 7 (70%) | 37 (57%) | ||
| CEA clearance pattern | 0.047 | 5.22 (1.02–26.60) | ||
| Exponential decrease (R2≧0.9) | 8 (80%) | 28 (43%) | ||
| Non-exponential decrease (R2 < 0.9) | 2 (20%) | 37 (57%) | ||
Abbreviation: pCR pathologic complete response
The p value in boldface means statistically significant, that is, less than 0.05