Table 2.
Multivariate logistic regression analysis of predictors of positive SLN findings
| POSITIVE SLN FINDINGS | ||||
|---|---|---|---|---|
|
| ||||
| Factor | N | OR | (95% CI) | P value |
| Biopsy method | 0.007 | |||
| Percutaneous (Ref) | 975 | 1.00 | ||
| Excisional | 259 | 1.90 | (1.20–3.01) | 0.007 |
|
| ||||
| Papillary histologic subtype | 0.006 | |||
| No (Ref) | 905 | 1.00 | ||
| Yes | 329 | 1.77 | (1.18–2.67) | 0.006 |
|
| ||||
| DCIS size >2 cm a | 0.030 | |||
| No (Ref) | 602 | 1.00 | ||
| Yes | 632 | 1.55 | (1.04–2.31) | 0.030 |
|
| ||||
| Total no. of interventions b | 0.001 | |||
| 1–2 (Ref) | 681 | 1.00 | ||
| 3 | 384 | 1.43 | (0.93–2.22) | 0.107 |
| 4 | 112 | 2.04 | (1.11–3.75) | 0.022 |
| ≥5 | 57 | 3.87 | (1.88–7.95) | <0.001 |
|
| ||||
| Final pathologic diagnosis | <0.001 | |||
| Pure DCIS (Ref) | 907 | 1.00 | ||
| Microinvasive cancer | 68 | 3.44 | (1.66–7.12) | 0.001 |
| Invasive cancer | 259 | 6.21 | (4.08–9.45) | <0.001 |
|
| ||||
| ISOLATED TUMOR CELLS c | ||||
|
| ||||
| Biopsy method | 0.001 | |||
| Percutaneous (Ref) | 920 | 1.00 | ||
| Excisional | 248 | 2.55 | (1.46–4.44) | 0.004 |
|
| ||||
| DCIS size >2 cm | 0.002 | |||
| No (Ref) | 573 | 1.00 | ||
| Yes | 595 | 2.38 | (1.37–4.16) | 0.001 |
|
| ||||
| Total no. of interventions | <0.001 | |||
| 1–2 (Ref) | 646 | 1.00 | ||
| 3 | 367 | 2.82 | (1.54–5.14) | 0.001 |
| 4 | 101 | 2.65 | (1.12–6.32) | 0.027 |
| ≥ 5 | 54 | 7.62 | (3.27–17.74) | <0.001 |
|
| ||||
| Final pathologic diagnosis | 0.047 | |||
| Pure DCIS (Ref) | 895 | 1.00 | ||
| Microinvasive cancer | 63 | 2.28 | (0.90–5.79) | 0.084 |
| Invasive cancer | 210 | 1.95 | (1.04–3.66) | 0.037 |
|
| ||||
| METASTASES c | ||||
|
| ||||
| Papillary histologic subtype | 0.008 | |||
| No (Ref) | 866 | 1.00 | ||
| Yes | 302 | 2.11 | (1.21–3.66) | 0.008 |
|
| ||||
| Final pathologic diagnosis | <0.001 | |||
| Pure DCIS (Ref) | 863 | 1.00 | ||
| Microinvasive cancer | 62 | 6.37 | (2.16–18.78) | 0.001 |
| Invasive cancer | 243 | 18.30 | (9.52–35.16) | <0.001 |
DCIS, ductal carcinoma in situ; OR, odds ratio; no., number; SLN, sentinel lymph node; SLND, sentinel lymph node dissection.
a
We chose to use the preoperative categorical variable rather than the postoperative continuous variable to go into the final regression model.
b
We chose to use total number of interventions rather than total number of biopsies and total number of surgeries in the final model as the first variable is a composite of the other two.
c
Only variables found to be predictive of positive SLN findings were chosen for entry into the models for isolated tumor cells and metastases to determine their individual contributions.