Table 4.
Margin threshold and IBTR: Bayesian network meta-analysis
| Analysis | Threshold distance for negative margins relative to positive (mm) | |||
|---|---|---|---|---|
| Mean OR (95% CrI)* | ||||
| >0 or 1 | 2 | 3 | 10 | |
| Main model, no. of patients | 2,230 | 2,412 | 289 | 1,963 |
| All studies | 0.45 (0.32 to 0.61) | 0.32 (0.21 to 0.48) | 0.30 (0.12 to 0.76) | 0.32 (0.19 to 0.49) |
| Sensitivity analysis, no. of patients | 1,957 | 1,851 | 272 | 1,079 |
| RT cohorts only | 0.45 (0.34 to 0.61) | 0.33 (0.23 to 0.47) | 0.22 (0.08 to 0.53) | 0.37 (0.24 to 0.57) |
| Sensitivity analysis, no. of patients | 1,781 | 1,524 | 289 | 616 |
| Van Zee et al22 excluded | 0.43 (0.31 to 0.57) | 0.29 (0.19 to 0.45) | 0.32 (0.14 to 0.75) | 0.27 (0.16 to 0.47) |
| Sensitivity analysis, no. of patients | 2,230 | 2,412 | — | 1,963 |
| 3 mm excluded | 0.47 (0.34 to 0.63) | 0.34 (0.23 to 0.49) | — | 0.36 (0.23 to 0.56) |
| Sensitivity analysis, no. of patients | 2,692 | 2,555 | 322† | 2,160 |
| Adding studies with no summary age data‡ | 0.44 (0.30 to 0.63) | 0.31 (0.19 to 0.51) | 0.32 (0.14 to 0.73) | 0.20 (0.11 to 0.35)§ |
| Adjustment for covariates (based on main model) | ||||
| Age | 0.46 (0.33 to 0.63) | 0.34 (0.22 to 0.51) | 0.33 (0.13 to 0.83) | 0.33 (0.20 to 0.51) |
| Median recruitment year | 0.45 (0.31 to 0.62) | 0.31 (0.19 to 0.46) | 0.29 (0.12 to 0.68) | 0.32 (0.20 to 0.49) |
| Proportion with RT | 0.46 (0.33 to 0.63) | 0.33 (0.22 to 0.49) | 0.29 (0.12 to 0.74) | 0.32 (0.20 to 0.50) |
| Proportion with endocrine therapy⊤ | 0.45 (0.29 to 0.70) | 0.33 (0.18 to 0.57) | 0.29 (0.10 to 0.79) | 0.31 (0.17 to 0.57) |
| Proportion with high-grade DCIS⊤ | 0.45 (0.32 to 0.62) | 0.33 (0.21 to 0.48) | 0.31 (0.12 to 0.74) | 0.39 (0.25 to 0.59) |
Data adapted.16
Abbreviations: CrI, credible interval; DCIS, ductal carcinoma in situ; IBTR, ipsilateral breast tumor recurrence; OR, odds ratio; RT, radiation therapy
* Adjusted for follow-up
†Two studies using a 5 mm threshold were included with the 3 mm threshold group
‡These studies were ineligible for inclusion in the meta-analysis from Marinovich et al because of lack of summary age data (see eligibility criteria); hence sensitivity analysis reports estimates if these were included in models
§95% CrI for relative odds ratio of 10 v > 0 or 1 mm did not cross 1 (Methods Meta-Analysis)
⊤Because of missing covariate information, these analyses were undertaken in a reduced number of studies (19 for endocrine therapy; 16 for high-grade DCIS)