Comparison of SNB results
Table 1 summarizes the outcomes of SNB in each method. Of 338 patients, 337 patients had metastatic SNs and subsequently underwent ALND. One patient had no reaction with either RI or ICG, and ALND was performed to confirm metastatic status. SN(RI), SN(ICG), or SN(RI+ICG) was observed in 337 patients (99.7%). SN(RI) was identified in 331 (97.9%) patients and SN(ICG) was identified in 334 (98.8%) patients. There was no significant difference between the three methods in SN identification rate.
The average number of resected SNs per patient was 2.78 using SN(ICG) and 2.05 using SN(RI) (p<0.01). The metastatic SN detection rate per patient was the highest with SN(RI+ICG) (99.7%) compared with the other methods (RI, 91.7%; ICG, 96.4%; p<0.01).
Table 2 summarizes the concordance and discordance of metastatic SN detection per patient between SN(RI) and SN(ICG) methods. Of 338 patients, 299 patients with metastatic SNs had positive SN reactions to both RI and ICG, while no reaction was observed in 1 patient. In contrast, 11 patients had metastatic SNs that reacted to only RI and 27 patients had metastatic SNs that reacted to only ICG. The concordance and discordance rates of detecting metastatic SNs between SN(RI) and SN(ICG) were 88.8% ([299+1]/338) and 11.2% ([27+11]/338), respectively.
Details of discordant cases
Table 3 summarizes the rates of SN(RI) and SN(ICG) in discordant cases and findings of lymphoscintigraphy. Of the 27 patients in whom metastatic SNs could not be detected by RI, 8 patients had unconfirmed SN reaction to RI. Furthermore, 7 of these 8 patients had no findings of 99mTc accumulation (hot spots) in the axillary area on lymphoscintigraphy (Fig. 1a), while only a weak hot spot was observed in 1 patient. All other 19 patients in whom metastatic SNs could not be detected using RI had strong hot spots in the axillary area on lymphoscintigraphy (Fig. 1b), and all the patients in whom SNs showed a reaction to RI were confirmed to have metastatic SNs. Of the 11 patients in whom metastatic SNs could not be detected using ICG, SN(ICG) was not confirmed in 3 patients, and all 11 patients had hot spots in the axillary area on lymphoscintigraphy.
Correlation between nodal staging and clinicopathological features
The clinical and pathological characteristics of all patients are summarized in Table 4. Among 338 patients, 276 were diagnosed with pN1 (total metastatic LNs, 1–3) and 62 with pN2-3 (total metastatic LNs, 4 or more). The overall median age was 54.1 years. Furthermore, 287 patients had invasive ductal carcinoma and 51 had special types of invasive carcinoma (invasive lobular carcinoma, ILC, 40; invasive micropapillary carcinoma, IMPC, 6; metaplastic carcinoma, 3; and apocrine carcinoma, APO, and mucinous carcinoma, MUC, 1). To investigate the difference between pN2-3 status and pN1 status, univariate analyses using the χ2 test were performed with each explanatory variable. Although discordance between SN(RI) and SN(ICG) was not associated with pN2-3 in this analysis (p=0.647), the absence of SN(RI) was significantly associated with pN2-3 status (p<0.01). Additionally, cT2-3 stage (p<0.01), histological type (p<0.01), higher histological grade (G3) (p=0.015), and type of surgery (mastectomy) (p<0.01) were significantly associated with pN2-3 status.
Multivariate logistic regression analyses of pN2-3
The variables in univariate analysis that significantly (p<0.05) correlated with pN2-3 status were analyzed using multivariate logistic analysis. To avoid multicollinearity, we excluded the type of surgery in this analysis. Similar to the findings of univariate analysis, the absence of SN(RI) (odds ratio, 8.64; p<0.01), cT2-3 stage (odds ratio, 2.549; p<0.01), histological type (odds ratio, 3.276; p<0.01), and higher histological grade (G3) (odds ratio, 1.921; p=0.039) were significantly associated with pN2-3 status (Table 5).