Results of search strategy
After a meticulous search utilizing the specified MeSH terms and additional sources, a total of 485 articles were initially identified. Following a rigorous screening process, 215 abstracts were selected for further examination based on their relevance. Upon thorough review of the full texts of these 215 studies, 204 were excluded primarily due to the absence of a direct comparison between BCT and mastectomy in terms of LR rates or the lack of pertinent HR/OR/RR data comparing these two treatment modalities. Consequently, a comprehensive flowchart (Fig. 1) was constructed to illustrate the article selection process and ensure transparency in the methodology.
Description of studies
The characteristics of the studies encompassed in this meta-analysis are succinctly presented in Table 1, detailing the total case count, publication year and country, study design, age and tumor stage controls, and follow-up duration for each study.
The distribution of T stage and N stage among patients in the BCT and mastectomy groups is outlined in Table 2, revealing a higher prevalence of tumors >2 cm in the mastectomy group across five studies [32,34,35,37,38] and a greater percentage of positive axillary lymph nodes in the mastectomy group across four studies [32,33,35,36]. The meta-analysis revealed a significant difference in tumor size between BCT and mastectomy groups among under-40 patients, with BCT patients having a higher proportion of small tumors (≤T1) (OR=1.38, 95%CI: 1.23-1.56, P<0.00001). And forest plot illustrates that the BCT group has a higher proportion of patients without positive lymph node (N0) compared to the mastectomy group among patients under 40 years old (OR=1.32, 95%CI: 1.12-1.54, P=0.0007).
Table 3 records the administration of adjuvant and neoadjuvant chemotherapy, adjuvant endocrine therapy, adjuvant targeted therapy, and adjuvant radiotherapy in both treatment groups. Notably, all patients undergoing BCT in five out of six studies [32-37] received adjuvant radiotherapy, while only 10% to 69% of mastectomy patients did, resulting in an overall radiotherapy ratio of 54.5% for the mastectomy group. The BCT group exhibited a significantly higher rate of radiotherapy compared to the mastectomy group (OR=2.28; 95% CI: 1.69-3.07; P<0.00001). However, for those studies with detailed data, the ORs for patients receiving adjuvant chemotherapy and hormonotherapy following mastectomy and BCT were not statistically significant (chemotherapy OR=0.92; 95% CI: 0.81-1.05; P=0.12; hormonotherapy OR=1.06; 95% CI: 0.88-1.28; P=0.51). Due to insufficient data, an analysis of neoadjuvant chemotherapy and adjuvant targeted therapy was not feasible.
Table 4 summarizes the number of patients experiencing local recurrence (LR) in both groups, accompanied by the corresponding LR rates, HR/OR values, and P values. This compilation encompasses 11 population-based studies, totaling 9215 patients under 40 years old, with 4190 undergoing BCT and 5025 undergoing mastectomy. Four of these studies [31,32,33,37] reported statistically significant results, suggesting a potentially higher LR rate in the BCT group compared to the mastectomy group.
Lastly, Table 5 evaluates the methodological quality of the observational studies using the Newcastle-Ottawa scale, providing an assessment of their rigor and credibility. All 11 articles received a Newcastle-Ottawa Scale (NOS) score ranging from 7 to 9, indicating medium to high quality.
Meta-analysis results:
1 Subgroup analysis by follow up duration in young breast cancer patients (age≤40) comparing BCT and mastectomy.
In our comprehensive meta-analysis, we analyzed 11 studies contributing 15 risk estimates, with four studies presenting two estimates based on median follow-up durations. Notably, the BCT group exhibited a 63% higher LR rate compared to the mastectomy group. Our analysis revealed a significant association between BCT and an elevated risk of LR (OR=1.63; 95% CI: 1.25-2.13), as depicted in Figure 2. Heterogeneity was observed among the included studies, with I²=70.2% and chi-square P=0.000.
Subgroup analysis further elaborated that significant differences in LR rates existed between follow-up groups, specifically at 5 years (OR=1.86; 95% CI: 1.31-2.62) and 5_11 years (OR=1.50; 95% CI: 1.00-2.25). Remarkably, the 5-year LR rate in the BCT group was 86% higher than that in the mastectomy group.
Sensitivity analysis confirmed the stability of our findings, with consistent results across different studies, as evidenced in Figure 3. The Egger's test (P=0.087) indicated no substantial publication bias.
Regression analysis revealed that T stage (coefficient=-0.244, P=0.530), N stage (coefficient=-0.120, P=0.547), and follow-up time (coefficient=-0.306, P=0.333) did not significantly influence the OR. Means the differences in T stage, N stage, and follow-up among the included studies do not have a significant impact on the OR values of LR. However, the extremely low P-value obtained from the test of residual homogeneity hinted at a need for further model refinement or analysis to better elucidate the variations in LR rates, as illustrated in Figure 4.
2 Subgroup analysis by tumor stage duration in young breast cancer patients (age≤40) comparing BCT and mastectomy.
As shown in Figure 5, both T1-2 patients [OR=1.88, 95% CI: 1.20-2.94] and T1-4 patients [OR=1.46, 95% CI: 1.02-2.10] in the BCT group demonstrated a significantly elevated risk of LR compared to those in the mastectomy group. The heterogeneity tests for these subgroups revealed substantial heterogeneity, with I²=70.2%, P=0.005 for T1-2 patients and I²=73.5%, P=0.000 for T1-4 patients, indicating variations among the included studies. Nonetheless, the sensitivity analysis underscored the robustness and reliability of our findings, reinforcing the significance of the observed differences in LR risk between the BCT and mastectomy groups across various T stages.
3 Subgroup analysis by node stage duration in young breast cancer patients (age≤40) comparing BCT and mastectomy.
The meta-analysis unveiled a noteworthy disparity in LR risk among N0-1 patients, with those undergoing BCT experiencing a significantly elevated risk compared to mastectomy patients (OR=2.56,95% CI: 1.90-3.44), as depicted in Figure 6. However, this trend was not replicated in the N0-3 patient subgroup, where the BCT group did not exhibit a significantly higher risk of LR compared to the mastectomy group(OR=1.38,95% CI: 0.99-1.93).
The heterogeneity test for the N0-1 patient subset returned an I²=0.0%, P-value=0.542, indicating an absence of significant heterogeneity among the contributing studies. But the sensitivity analysis underscored the robustness and reliability of our findings, reinforcing the distinct patterns of LR risk observed across different lymph node status categories.
4 Subgroup analysis by age group duration in breast cancer patients comparing BCT and mastectomy.
As depicted in Figure 7, both age subgroups within the BCT group exhibited a significantly elevated risk of LR compared to the mastectomy group. Specifically, for patients aged ≤35 years(OR=2.04,95% CI: 1.48-2.81), for those aged ≤40 years(OR=1.53,95% CI: 1.10-2.13).
The heterogeneity test for the first age subgroup (≤35 years) , I²= 27.9% P=0.239, indicating no significant heterogeneity among the included studies. In contrast, the heterogeneity test for the second age subgroup (≤40 years) showed I²=73.4%,P-value= 0.000, suggesting substantial heterogeneity.
However, despite this heterogeneity in the second subgroup, the sensitivity analysis, confirmed the robustness and reliability of our findings, reinforcing the significantly higher risk of LR associated with BCT compared to mastectomy in both age subgroups.