Validation study for the hypothesis of internal mammary sentinel lymph node lymphatic drainage in breast cancer

According to axilla sentinel lymph node lymphatic drainage pattern, we hypothesized that internal mammary sentinel lymph node (IM-SLN) receives lymphatic drainage from not only the primary tumor area, but also the entire breast parenchyma. Based on the hypothesis a modified radiotracer injection technique was established and could increase the visualization rate of the IM-SLN significantly. To verify the hypothesis, two kinds of tracers were injected at different sites of breast. The radiotracer was injected with the modified technique, and the fluorescence tracer was injected in the peritumoral intra-parenchyma. The location of IM-SLN was identified by preoperative lymphoscintigraphy and intraoperative gamma probe. Then, internal mammary sentinel lymph node biopsy (IM-SLNB) was performed. The fluorescence status of IM-SLN was identified by the fluorescence imaging system. A total of 216 patients were enrolled from September 2013 to July 2015. The overall visualization rate of IM-SLN was 71.8% (155/216). The success rate of IM-SLNB was 97.3% (145/149). The radiotracer and the fluorescence tracer were identified in the same IM-SLN in 127 cases, the correlation and the agreement is significant (Case-base, rs=0.836, P<0.001; Kappa=0.823, P<0.001). Different tracers injected into the different sites of the intra-parenchyma reached the same IM-SLN, which demonstrates the hypothesis that IM-SLN receives the lymphatic drainage from not only the primary tumor area but also the entire breast parenchyma.

According to the axilla sentinel lymph node (ASLN) lymphatic drainage pattern (i.e., ASLN receives lymphatic drainage from not only the primary tumor area, but also the entire breast) [9,10], we hypothesized that IM-SLN receives lymphatic drainage from not only the primary tumor area, but also the entire breast parenchyma [8]. Based on the hypothesis, a modified radiotracer injection technique has been established which increased the visualization rate of the IM-SLN significantly [8]. The hypothesis needs to be demonstrated. To avoid more injury by complete IMLN dissection following IM-SLNB, an alternative validation study for the hypothesis was performed: two different tracers were injected in different sites of the intra-parenchyma to observe whether they could reach to the same IM-SLN.

Characteristics of IM-SLNB
The clinically pathological characteristics of the 216 enrolled patients are presented in Table 1. The detection rate of ASLN was 98.6% (213/216). The overall visualization rate of IM-SLN detected by preoperative lymphoscintigraphy and gamma probe was 71.8% (155/216). 96.1% (149/155) of them received IM-SLNB. The success rate of IM-SLNB was 97.3% (145/149). The data on clinical outcome of the patients underwent IM-SLNB show in Table 2. In 12 patients underwent breast conserving surgery, 5 cases who were identified the location of primary tumor could not reach IM-SLNB had to be made an extra incision in the skin to reach IM-SLNB.
In patients who performed IM-SLNB successfully, a total of 279 lymph nodes were removed, the median number of IM-SLNs was 2 (range 1-4 nodes). The IM-SLNs were located in the first ( Table 3. The IM-SLN involvement rate was 8.1% (7/86) in patient with clinically axillary node negative patients and 18.6% (11/59) in positive patients respectively. All patients with positive IM-SLN received regional nodal irradiation to IMLNs. The clinical, pathological and treatment details of these patients were shown in Table 4. In patients with ≥4 positive axillary lymph nodes, regional nodal irradiation to IMLNs had been avoided in 50.0% cases (9/18) with negative IM-SLN. In patients with 1-3 positive axillary lymph nodes, regional nodal irradiation to IMLNs might be avoided in 91.2% cases (52/57) with negative IM-SLN.

Correlation and agreement between the radiotracer and the fluorescence tracer
In the validation study, 145 patients underwent IM-SLNB successfully, of which 127 patients identified the radiotracer and the fluorescence tracer reached to the same IM-SLN, 18 patients were detected only the radiotracer positive IM-SLN (Table 5). Accordingly, the radiotracer and the fluorescence tracer in the same IM-SLN showed a strong correlation coefficient at 0.836 (Case-base, r s ≥0.5, P<0.05). The degree of agreement between the radiotracer and the fluorescence tracer was Kappa=0.823 (very good), showing high degree of agreement between the two tracers (Kappa>0.8, P<0.05).

Complications
No serious bleeding and pain was found after using the modified radiotracer injection technique in all patients. A total of 2 patients were found with minor generalized skin reactions which occurred after injection of indocyanine green (ICG) during the surgery. A small pleural lesion (≤ 2 mm) was noted intraoperatively in 1.4% cases (2/145) and no pneumothorax was seen postoperatively on chest X-rays. Intraoperative bleeding from the internal mammary artery occurred in 3.4% cases (5/145), and was successfully resolved. There were no postoperative complications and reactions after the two-week following up, and no increase of days in hospital stay from this procedure.

DISCUSSION
IMLN metastases have been demonstrated to occur in 28-52% of ALN positive patients and 5-17% of ALN negative patients [2,11,12]. In patients with a negative axilla, a positive IMLN portends a similar prognosis as ALN positive patients, impacting both recurrence and survival [1][2][3]. The results of the MA.20 showed that the addition of regional nodal irradiation (including IMLNs) to whole-breast irradiation reduced the rate of breast-cancer recurrence in patients with node-positive or high-risk node-negative breast cancer [13]. The EORTC 22922/10925 study found that regional nodal irradiation improved the rates of disease-free and distant diseasefree survival and reduced the rate of death from breast cancer among patients with early-stage breast cancer [14]. Furthermore, the DBCG-IMN study identified that IMLNs irradiation increased overall survival in patients with early-stage node-positive breast cancer [15]. The 2016 National Comprehensive Cancer Network Breast Cancer Clinical Practice Guidelines recommend radiotherapy to IMLNs for patients with ≥4 positive ALNs (category 1), and strongly consider radiotherapy to IMLNs for patients with 1-3 positive axillary nodes (category 2A), both after mastectomy and lumpectomy [16]. However, lowrisk did not mean IMLN negative and high-risk did not mean IMLN metastases [17]. Studies of extended radical mastectomy reported that 36.8%-46.2% patients with ≥4 positive ALNs and 18.8%-26.7% patients with 1-3 positive ALNs identified IMLN metastases, and negative IMLN was found in about 70% patients with ≥4 positive ALNs [1,[18][19][20]. Thus, these inclusion criteria might induce over-and under-treatment. Up to now, there have no reasonable methods to detect out metastasis in IMLN exactly. But IM-SLNB via intercostal space could make it possible-tailored IMLNs radiotherapy and minimally invasive staging [17]. Even though breast cancer staging has incorporated IM-SLNB concept since the 6 th edition of the American Joint Committee on Cancer, IM-SLNB has not been performed routinely [21]. The studies of IM-SLNB showed that the success rate of IM-SLNB has reached 60-100% with minimal or no changes in operative time [4][5][6][7], but the visualization rate of IM-SLN was low [6][7][8], which has been the restriction for both clinical study and daily practice of IM-SLNB (Table 6).
Clinical studies found that superficial injection (intradermal, subdermal, periareolar, and subareolar) of radiotracer was hard to identify IM-SLN but intraparenchymal injection (peritumoral, intratumoral, or subtumoral) was more reliable [33][34][35][36][37][38]. These results suggest that the dermal and subdermal lymphatic flow is rarely directed to IMLNs, whereas some intraparenchymal lymphatic flow is directed to IMLNs. Anatomy study described that the breast parenchyma has extensive lymphatic network and has rich anastomoses with the superficial cutaneous lymph plexus of the developing skin [39]. It is considered that radiotracer, wherever injected, could flow to the same ASLN. That means ASLN receive the lymphatic drainage from not only the primary tumor area, but also the entire breast organ [10]. Based on Abbreviations: ER estrogen receptor status, PR progesterone receptor status, HER-2 human epidermal growth factor receptor-2, WBI whole breast irradiation, RNI regional node irradiation, PMRT postmastectomy radiotherapy.
this concept, the hypothesis of IM-SLN was supposed, which IM-SLN receives lymphatic drainage from not only the primary tumor area, but also the entire breast parenchyma. According to the hypothesis, a modified technique (periareolar intra-parenchyma, high volume, and ultrasonographic guidance) was formulated, which could significantly increase the preoperative visualization rate of the IM-SLN without lowering ASLN visualization rate [8,38].  [40]. However, the radiotracer was not injected in peritumoral intra-parenchyma but in periareolar intraparenchyma with the modified technique based on the hypothesis. The question arises as to whether all nodes detected by the method should be considered as "true" IM-SLN or whether some of them are actually "secondtier" IMLN. The accuracy of the modified technique has been confirmed by our team at the previous study [41]. The hypothesis of ASLN lymphatic drainage pattern was proved with subsequent axillary lymph node dissection (ALND). As the extended radical mastectomy (included all IMLNs resection) has been abandoned since 1960s [42,43], the hypothesis of IM-SLN lymphatic drainage pattern cannot be validated by this way. Now, another method was used to validate the IM-SLN lymphatic drainage hypothesis. The ICG fluorescence tracer is a safe and effective method for SLNB in breast cancer with acceptable sensitivity and specificity comparable to conventional methods [44][45][46]. In our breast cancer center, it has been compared with the combined method (blue dye with radiotracer) in identifying ASLN. It showed that all ASLN identified by combined method also were ICG fluorescence positive and non-sentinel lymph nodes were ICG negative after ALND (n= 69, P<0.05). The internal mammary lymph nodes commonly receive less than 25% of the total lymph from the breast [47]. Due to little volume of ICG tracer is difficulty to detect by the fluorescence imaging system, it is hard to find IM-SLN by this tracer. But IM-SLN can be detected by radiotracer with the modified radiotracer injection technique and performed biopsy. In the validation study, the ICG fluorescence tracer was injected intraparenchymally at the peritumoral and the radiotracer was injected with the modified technique. By this method, to identify different tracers injected in different sites could reach to the same IM-SLN. After IM-SLNB the status of IM-SLN was identified by intraoperative gamma probe and fluorescence imaging system. The radiotracer and the fluorescence tracer in the same IM-SLN showed a strong correlation coefficient at 0.836 (Case-base, r s >0.5, P<0.05). The degree of agreement between the radiotracer and the fluorescence tracer was Kappa=0.823   sentinel node in the internal mammary chain. In the study, the results of the metastases site and the number of IM-SLNs were in accordance with the past study of extended radical mastectomy, which could reflect the accuracy of IM-SLNB indirectly [48][49][50]. There were no serious adverse events or reactions after the radiotracer injected guiding by the modified injection technique. In sum, different tracers injected into the different sites of the intra-parenchyma reached the same IM-SLN, which demonstrates the hypothesis that IM-SLN receives the lymphatic drainage from not only the primary tumor area but also the entire breast parenchyma. www.impactjournals.com/oncotarget

Study design and patients
From September 2013 to July 2015, 216 patients with core biopsy proved invasive breast cancer scheduled to receive preoperative tracers injection, who agreed with performing IM-SLNB as part of their breast cancer surgery, were recruited to the IM-SLNB study. Patients with previous invasive breast cancer, hypersensitivity to iodine or ICG, hyperthyroidism and patients who were either pregnant or lactating were excluded from the study.
Two different kinds of tracer were injected at different sites in this validation study. The radiotracer (1.0-1.2ml 9.25-18.5MBq 99m Tc-labeled sulfur colloid) was injected with the modified radiotracer injection technique. It was injected into intra-parenchyma at the 6 and 12 o'clock positions 0.5-1.0cm from the areola guided by ultrasound (ALOKA-SSD-5000, ALOKA, Tokyo, Japan) 3-18h before surgery [8,38]. The fluorescence tracer (1.0ml 0.5% ICG) was injected in the peritumoral intra-parenchyma guided by ultrasound just before the beginning of the peration. The radioactive IM-SLNs were detected by preoperative lymphoscintigraphy (Toshiba GCA 901AHG, Toshiba Corporation, Tokyo, Japan) ( Figure 1) 30min before the surgery and gamma probe (Neoprobe, Neo2000 gamma detection system, Johnson & Johnson New Brunswick, NJ, USA) during the surgery. IM-SLNB was performed for patients with the radioactive IM-SLNs. After IM-SLNB, the fluorescent status of IM-SLN was identified with the fluorescence imaging system (Ming De, MD fluorescence imaging system, Langfang, People's Republic of China) (Figure 2). The number and the status of IM-SLNs were recorded to identify the IM-SLN visualized rate and the concordance rate of the radiotracer and the fluorescence tracer. All IM-SLNs were assessed by routine pathology.
The study was conducted within a single institute (Breast Cancer Center, Shandong Cancer Hospital Affiliated to Shandong University). All patients gave informed consent to participate in the study which had approval from the Shandong Cancer Hospital Affiliated to Shandong University Research Ethics Board (No. SDTHEC20130324). Any immediate toxicity following injection of ICG and any adverse events during the study period were recorded during surgery and at the twoweek follow up appointment.

Internal mammary sentinel lymph node biopsy
During the operation, IM-SLNB was performed for patients with radioactive IM-SLNs via the intercostal space. Through the mastectomy incision or modified radical mastectomy incision access to the location of IM-SLNs under the guidance of intraoperative gamma probe. Then the pectoral major fascia muscle fibers were separated to expose the relevant intercostal space. Next, the external and internal intercostal muscles were   divided transversely in its middle. To avoid injury to the anterior intercostal vessels, the division should be located between the two costal cartilages or along the superior costal border. IM-SLNs located in other intercostal spaces were removed by the same method. A postoperative chest X-ray was performed in case of accidental pleural lesion. In patients who accepted breast conserving surgery, if the location of primary tumor could not reach IM-SLNB, an extra incision (2.5-3.0cm) in the skin has to be made.

Histopathology of sentinel lymph nodes
All SLNs underwent pathological evaluation according to local protocol including serial sectioning at 2.0-3.0mm followed by routine staining with Haematoxylin and Eosin.

Statistical analysis
The data were analyzed with the SPSS 17.0 software package. Chi-square-test or Fisher's-exact-test was performed to compare the visualization rates among the groups. The correlations between the radiotracer and the fluorescence tracer in the same IM-SLN were calculated using the Spearman rank correlation coefficient. The criteria for judging the size of the correlation coefficient were applied: correlations<0.30 are considered minor, correlations between 0.3-0.49 are considered medium, and ≥0.5 are considered strong. Cohen's kappa statistic was used to determine inter-examiner agreement. According to Altman's guidelines, it is poor when kappa scores ≤0.20, fair when kappa between 0.21-0.40, moderate when kappa between 0.41-0.60, good when kappa 0.61-0.80, and very good when kappa ≥0.80. Reported P values represent twosided tests. Significance was defined as P<0.05.

ACKNOWLEDGMENTS
Wang YS had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Design and conduct of the study: Wang YS, Yu JM. Collection: Cong BB, Qiu PF, Cao XS. Management: Wang YS, Liu YB, Yang GR. Analysis: Cong BB, Wang YS. Interpretation of the data: Wang YS, Yu JM, Cong BB. Preparation, review, or approval of the manuscript: Wang YS, Yu JM, Cong BB. We thank Tong Zhao, MA, Zheng-Bo Zhou, MA, Peng Chen, MA, Chun-Jian Wang, MA, Zhao-Peng Zhang, MA and Xiao Sun, MA from our breast cancer center. All six of these individuals contributed to study conduct, manuscript review, or both; none received compensation for their contributions. We thank Rogers W, PhD, an English teacher from University of Jinan, who contributed to revise the language of the manuscript. We also thank all of the investigators and their site research teams. Lastly, we wish to thank the brave patients with breast cancer who participated in this study and their caregivers. The work was supported by grants from the Graduate Innovation Foundation of University of Jinan (YCXB15006) and the Natural Science Foundation of Shandong Province (2014ZRC03036).

CONFLICTS OF INTEREST
There is no conflict of interest in the study.