Effectiveness of Carbon Localization for Invasive Breast Cancer: An Institutional Experience

Introduction The final oncological and aesthetic results of breast-conserving surgery (BCS) are influenced by the precise localization of breast cancer (BC) tumors and by the quality of the intraoperative margin assessment technique. This study aimed to assess the effectiveness of the carbon localization (CL) technique by determining the success rate of BC identification and the proportion of adequate complete resection of BC lesions. Methods We conducted a cross-sectional retrospective study of patients treated with primary BCS for invasive BC who underwent CL of their BC lesion at the Jules Bordet Institute between January 2015 and December 2017. Descriptive statistics with categorical and continuous variables were used. The success rate of tumor identification and the rate of adequate excision were calculated using the test of percentages for independent dichotomous data. Results This study included 542 patients with 564 nonpalpable BC lesions. The median pathological tumor size was 12 mm. Of these, 460 were invasive ductal carcinomas. Most of the tumors were of the luminal subtype. CL was performed using ultrasound guidance in 98.5% of cases. The median delay between CL and surgery was 5 days, with 46% of the patients having CL one day before surgery. The lumpectomy weighed 38 g on average, with a median diameter of the surgical sample at 6 cm and a median volume of 44 cm3 (6–369). One-stage complete resection was successfully performed in 93.4% of cases. In 36% of cases, an intraoperative re-excision was performed, based on intraoperative macroscopic pathological margin evaluation. The tumor was identified in 98.9% of cases in the breast surgical specimen. Conclusion This study demonstrated high success rates for BC tumor identification (99%) and one-stage complete resection (93.4%) after BCS and CL. These results show that CL is an effective, simple, and inexpensive localization technique for successful excision of BC lesions during BCS.


Introduction
Breast cancer (BC) has become a concern for all populations, and its increasing incidence is alarming [1]. Te incidence of BC increases by 0.5% each year [2] and is expected to reach more than 28 million cases in 2040 [1]. Te incidence of BC varies greatly from one region to another depending on various risk factors, increases in life expectancy, and improvements in BC screening programs [3]. Improved screening strategies and early diagnosis of BC have increased the detection of hidden nonpalpable lesions [4], which are mostly often detected at an early stage and eligible for breastconserving surgery (BCS) [5]. Since the 1990s, BCS followed by radiotherapy has become a new standard surgical procedure for these tumors, and, concomitantly, the management of BC has evolved towards the de-escalation of surgical treatment. Recent studies have demonstrated that BCS with radiotherapy not only decreases the risk of ipsilateral recurrence but is also associated with enhanced overall survival and breast-specifc survival in patients undergoing BCS compared to mastectomy (Mx) [6].
To perform BCS, all cancerous tissues should be resected while preserving as much healthy tissue as possible to maintain good aesthetic outcomes and enhance patient satisfaction [5]. However, to respect these two criteria, precise localization of the tumor is necessary and a good intraoperative margin assessment technique is needed to ensure that the entire tumor is resected to avoid additional interventions [4]. Several methods for assessing intraoperative margins have been developed, based on histology, conventional radiology, and unconventional imaging techniques. In general, a negative margin for invasive BC is defned as "no ink on the tumor," which follows the SSO-ASTRO 2014 guidelines [5].
Regarding the localization of BC tumors, wire-guided localization (WGL) has long been considered the gold standard technique for BC localization [6]. WGL involves inserting a self-retaining wire into the tumor using ultrasound (US) or stereotactic guidance [4]. Tis method has several drawbacks: the timing, the guidewire should ideally be positioned on the day of surgery or the day before to avoid possible wire migration and infection, the fact that the procedure requires a qualifed radiologist to perform the procedure, and the potential for relatively serious postprocedural complications of wire migration and diathermy burns conducted to the skin through the wire [7]. Moreover, in addition to its complexity, reported 17% of nonpalpable cases have positive margins that require further intervention, thereby increasing the risk of recurrence [4].
In addition, the carbon localization (CL) procedure is another older localization technique that was frst reported by Svane in 1983 [10] and has been increasingly used in recent times as an alternative procedure to WGL. CL involves the use of a sterile suspension of charcoal that is injected around the outer edges of the tumor, leaving a fne track in the parenchyma and under the skin. It is a simple, fast, and safe technique, with rare associated complications, has lower cost than that of other methods, and ofers the possibility of marking at longer intervals [11].
Belgium is one of the countries with the highest incidence of BC, and for quality care, a multidisciplinary approach is always ofered to patients [3]. Hence, due to the need for high-quality management of this relatively frequent pathology, CL for nonpalpable lesions has been used in our institution since the early 1990s. Te aim of our study was to assess the value of CL in the preoperative workup of women diagnosed with early-stage BC undergoing BCS. We also sought to assess the efectiveness of this technique by determining the success rate of surgical excision and the proportion of adequate and complete resection of BC lesions.

Study Design and Patients.
We conducted a crosssectional retrospective descriptive study, covering all patients who underwent BCS for early-stage invasive BC and CL before surgical tumor resection, from January 2015 to December 2017 at the Jules Bordet Institute. Patients with a palpable tumor, those who received neoadjuvant therapy (NAT), those with only in situ disease, or those treated with mastectomy were excluded from the study. Tis study was conducted after review and approval by the institutional ethics committee (CE3268). Te patients were identifed from our prospectively maintained database review, and patient demographics, tumor characteristics, localization technique details, and treatment-type data were recorded. Te characteristics of the surgical specimen (weight, maximal diameter of the specimen, volume of the excised breast specimen, complete one-stage resection, macroscopic tumor identifcation, and resection margins) were also collected.

Localization Technique: An Institutional Procedure.
A few days before surgery, US was performed by one of the four institutional breast interventional radiologists to determine the location of the tumor. After identifcation of the lesion(s), 2-3 cc of 1% lidocaine was injected into the subcutaneous tissue overlying the lesion. A syringe with an 18-gauge needle was inserted vertically around the lesion, usually with freehand US guidance (antiparallel to the US probe), and approximately 1 mL of sterile 4% charcoal suspension (CARBONE 4%; carbo activatus 200 mg-Natrii chloridum 45 mg, Aqua p.i. ad 5 ml; Laboratoires Sterop, Brussels, Belgium) was slowly injected, while the needle was withdrawn towards the skin entry point. Care was taken to avoid excessive carbon injection just under the skin and unnecessary residual skin tattooing. After CL, a description of the tumor location, detailing the distance to the skin and the position in relation to the nipple length, was added to the radiological report, and a schematic representation of the BC lesion and trace marking was made. When the lesions were not detectable by US, CL localization was performed under stereotaxic guidance with an 18-gauge needle inserted around the lesion and/or the previously inserted marking clip. A schematic representation of the procedure is shown in Figure 1.

Surgical Technique and Histopathological Analysis.
During surgery, an incision was made over the carbon label injection site, whenever possible, and surgery was performed using standard BCS techniques (Figures 2(a)-2(g)). Axillary surgery was performed by sentinel lymph node biopsy (SLNB) with or without axillary lymph node 2 Te Breast Journal dissection (ALND) according to institutional guidelines. Intraoperative margin assessment of breast surgical specimens was performed by pathological macroscopic evaluation, following current institutional practice ( Figure 2(h)). A complete histopathological evaluation was then performed according to the international guidelines for BC tumor assessment.

Defnitions.
Te total volume of the excised specimen was calculated with the ellipsoid formula, using the three dimensions measured by the pathologist (volume � 4/3π(1/2length × 1/2widt h × 1/2height)). Adequate excision was defned as the complete removal of the tumor with tumor-free margins (i.e., no invasive carcinoma or ductal carcinoma in situ on inked margins). Secondary re-excision was defned as a second surgery for fnal positive margins , and approximately 1 mL of sterile 4% charcoal suspension is slowly injected, while the needle is withdrawn towards the skin entry point (d, f, g). Care is taken to avoid excessive carbon injection (g) just under the skin and unnecessary residual skin tattooing (h). A schematic representation (d) of the BC lesion and trace marking is made and added to the radiological report. (i.e., relumpectomy or mastectomy) of the same breast within 3 months. Success or failure of the localization procedure was defned by the presence or absence of a tumor in the surgical specimen after histopathological evaluation.
To assess the incidence of immediate and delayed complications after CL, physician-reported data on periprocedural (immediate) and frst postoperative imaging (delayed), to detect any abnormalities correlated with the carbon injection, were collected.

Statistical
Analysis. Statistical analysis was performed using Statistical Analysis Software (SAS). Descriptive statistics are used to summarize the patient, tumor, and clinical characteristics of the cohort. Categorical variables are reported as numbers and proportions. Continuous variables are reported as the number of subjects, mean values, standard deviation, or median and interquartile range, depending on the distribution. Te efectiveness of CL in patients with nonpalpable BC was determined by evaluating the success rate of tumor identifcation and complete surgical removal of the tumor with adequate margins and by assessing the volume of the surgically excised breast specimen. Te success rate of tumor identifcation and the rate of adequate excision were calculated using the test of percentages for independent dichotomous data, such as the ratio of the number of patients with a tumor (pathological) identifed after BCS and the number of patients with negative margins, respectively, to the total number of cT1 patients included.

Patients and Tumor
Characteristics. Tis retrospective, single-institution study included 542 patients, with 564 nonpalpable invasive BC tumors, who underwent BCS after CL at the Jules Bordet Institute during the study period. Te median age of the patients was 62.1 years (21.3-88.6), and the majority were postmenopausal at the time of diagnosis (84.5%). Most patients had a healthy weight, with 80.6% having BMI <25. Both sides were similarly afected (51.1% on the right side and 48.9% on the left side). Forty (7.4%) patients had multiple BC tumors, which led to excision of 564 BC tumors. Table 1 shows the clinical and pathological characteristics of the study population. Te average pathological tumor size was 12 mm (1.5-40) with only 1.8% exceeding 2 cm. Among these tumors, 460 (81.6%) were invasive ductal carcinomas (IDCs) of nonspecifc type, 79 (14%) were invasive lobular carcinomas (ILCs), and the majority were positive for estrogen receptor (ER) and progesterone receptor (PR) (91.1% and 78.9%, respectively). According to the intrinsic subtype classifcation, most of the tumors were luminal subtype, while only 9.5% were HER2enriched breast tumors, and 34 (6%) were triple-negative tumors. Axillary lymph nodes were involved in 15.4% of cases. Tese results guided adjuvant treatment following a multidisciplinary decision, consisting of intraoperative radiotherapy (IORT) and irradiation of the whole breast in 63.3% and 36.7% of cases, chemotherapy in 29.2% of cases, and endocrine therapy in 92.1% of cases.

Summary of the Carbon Localization Technique.
Tumor localization was performed one day or more before surgery, with an average delay of 5 days between CL and surgery, using US guidance in 98.5% of cases or stereotaxic mammography in 1.5% of patients. In general, the tumors were not deep with a median tumor depth measured using US of 8 mm (Table 2).
Lumpectomy was performed for all 564 tumors mentioned above, followed by SLN biopsy in 81.4% or ALND in 17% of cases. Te lumpectomy specimens weighed 38 g on average and varied from 4 g to 185 g. Te average larger diameter of the sample was 6 cm and reached 15.5 cm in one case; thus, the volume of the sample varied from 6 cm 3 to 369 cm 3 with a median of 44 cm 3 . One-stage complete resection was successfully performed in 527 of 564 tumors, leaving behind 37 cases (6.6%) in which the fnal margin status was positive, leading to secondary re-excision. Additional intraoperative re-excision was performed in 36% of cases based on intraoperative macroscopic pathological

Procedural and Postprocedural CL Complications.
No allergic reactions or other important carbon-related complications were reported in our study population. Eleven patients (0.02%) developed localized hematoma. No interference was reported for pathological examination or for the frst postoperative (one year) breast imaging evaluation.

Discussion
Te increasing number of nonpalpable BC lesions detected by screening programs has highlighted the need for rapid and accurate localization techniques for these tumors. Our study was conducted on 564 invasive BC tumors and confrmed the efcacy of CL in correctly locating these tumors. Macroscopic identifcation of the tumor was successfully achieved in 98.9% of cases, including one-stage complete resection in 93.4%. In other words, in only 6.6% of cases, the fnal margin status was positive, slightly better than that previously reported by Svane, the godfather of this technique, who reported a rate of 7.2% of cases with positive margins [10]; however, no study has shown an improvement in this rate. Tis is also a clear improvement over the percentage of positive margins reported using WGL (17%) [4]. However, this low rate of fnal positive margins must be interpreted with caution regarding whether it is solely due to the efect of CL, considering the fact that we regularly use macroscopic pathological examination of the breast surgical specimen for the assessment of intraoperative margins. However, this high rate of successful tumor identifcation in the breast surgical specimen after CL, and, consequently, the high rate of one-stage complete resection that was observed in our study, is one of the best reported rates compared to the other techniques used to localize nonpalpable BC tumors. In contrast, Davey et al. discussed in their recent review that several techniques have exhibited frustrating results with, for example, CAL, WGL, ROLL, and SMRI demonstrating positive margins in 28.2%, 20.1%, 17.2%, and 11.8% of cases, respectively [4].
Breast surgical specimen volume is another concern with regard to BCS for nonpalpable BC tumors due to its impact on postoperative breast aesthetic appearance. It is expected that BC localization techniques may help reduce resection volume and increase the fnal aesthetic outcome. Again, the volume of breast surgical specimens observed in our study of CL (median 44 cm 3 ) is one of the smallest described, as some other techniques for BC localization have been reported to excise twice what we have reported here [12].
Carbon has long been used in skin tattooing because it is biologically inert [13], and since its application in the localization of occult breast tumors, very few publications have studied its efectiveness (Table 4 shows all PubMed-indexed English articles on CL of nonpalpable BC treated by primary surgery), and researchers have been busy fnding other modernized techniques.
Tis was the largest study in the PubMed-indexed English literature, as reported in Table 1 [17]. Localization was performed by injecting a charcoal suspension using ultrasound in the vast majority of cases (98.5%), confrming the fact that USGL could be a BC localization technique in the future with a complementary advantage for the patient because it avoids an invasive procedure as during CL [4,6]. However, this perioperative USGL requires special training for the breast surgeon, something which is currently missing from most basic training programs [8].
BC localization was performed the day before surgery in almost half of the cases and for up to 60 days with no tracking problems. Te possibility of marking the tumor a few days before the operation facilitates preoperative preparation for both the patient and the hospital [18,20]. Mullen et al. described successful localization performed 83 days after injection and also described the phenomenon of phagocytosis, which permanently fxes carbon particles [13]. Tis has also made it possible to use the CL of BC tumors even in NAT settings, with good results reported. Lannin et al. reported excellent rates of excision with free margins in patients who underwent surgery after NAT reaching 91% of cases [21]. Mathieu et al. also demonstrated that CL guided resection to the initial site of the tumor in 91% of cases without residual tumor after NAT and demonstrated carbon migration into the axillary lymph nodes in 38% of cases [22]. Although the CL approach is the localization technique currently used for BC tumors treated with NAT, these patients were excluded from the present study for more reliable results.
Te pathological tumor size in our study ranged from 1.5 mm to 40 mm, which is consistent with what has been published on CL before, where the tumor size has varied between 0.7 mm and 45 mm (Table 4). It is also consistent with the tumor size reported by other existing techniques in the literature, where the mean size ranged from 8 mm to 51.3 mm [4].
No interference or disruption was reported by the pathologists at our institution during the preparation and examination of the specimens (Figure 2(h)). Cavalcanti et al. previously described the formation of an infammatory reaction, in which they demonstrated the presence of carbon Te Breast Journal particles in the cytoplasm of macrophages in all cases, but they also confrmed that this infammatory reaction and the presence of carbon posed no difculty, and the pathological diagnosis was made directly [23].
No allergic reactions or carbon-related complications were reported in our study population, apart from simple complications related to the insertion of the needle itself. A few cases of carbon granuloma formation have been reported in the literature, many of which were injected and kept in place in the breast for certain reasons (e.g., benign etiology at biopsy and patient refusal) [11,24,25]. Tese have been reported to be benign formations, and no malignant transformations have been reported. In one case, granuloma formation was discovered 2 years after CL (24), confrming the inert nature of the product.
In addition to its safety and durability, one of the important reasons to use CL is its low price compared to other existing and developing BC localization modalities that require advanced instruments and technologies. Langlois and Carter demonstrated that the price of a carbon ampoule was only $2.8, while the price of wire fuctuated between $20 and $40 [14]. Similarly, Rose et al. showed that ultrasound CL after breast biopsy costs only $26.66, while hook localization costs $140 [17]. Tis low cost per procedural price is confrmed by our in-house ultrasoundguided BC localization technique using CL, which costs €29 at our hospital.
Diferent modalities and techniques are being used to help detect BC tumors [6], and each has its advantages and disadvantages, while the efectiveness of these diferent techniques appears to be similar. Te only technique that seems to diferentiate in terms of efcacy seems to be USGL, as reported in a recent meta-analysis of randomized controlled trials (RCTs) for BC tumor localization [4]. Tis publication mentioned that, unfortunately, the CL technique was not discussed, given the fact that there are no RCTs that compare this technique to other techniques, particularly WGL, which was a comparator for most of the RCTs of the other localization techniques currently in use.
Another relatively easy and efective concept for BC tumor localization is the ROLL technique, but its radioactive component remains an important logistical problem for a large number of hospitals [26]. To reduce the risk of difusion of injected products, radioactive and nonradioactive "seeds" were designed: 125 I seeds frst reported in 2001, seeds with radar impulses introduced in 2014, magnetic seeds in 2016, and radiofrequency seeds in 2017. Several limitations have been described for each of these techniques [26], for which the cost remains one of the main limitations with 125 I seed cost of almost €220 [27], whereas magnetic seeds alone can cost up to €500 [28]. Although a multitude of BC localization technique are currently available, and they demonstrate relatively similar efectiveness (successful excision rate), the evidence is limited to small cohort studies for some of these techniques, and no real comparator between the diferent techniques exists as most of the new techniques have been compared to WGL [4,29].
In order to have more accurate information about the efectiveness of these techniques, the European Breast Cancer Research Association of Surgical Trialists (EUBREAST) recently launched the MELODY (methods for localization of diferent types of breast lesions) trial, a multinational prospective intergroup cohort study, to assess BC localization techniques and devices from several perspectives [29].
For future studies comparing these diferent localization techniques, careful attention must be paid to the rate of negative margins and to the standardization of intraoperative margin assessment techniques, as well as a clear defnition of negative resection margins with or without intraoperative breast cavity re-excision.
Our study has limitations related primarily to its retrospective nature and the fact that we did not have the ability SLNB, sentinel lymph node biopsy; ALND, axillary lymph node dissection. 6 Te Breast Journal Te Breast Journal 7 to compare the efectiveness of this procedure with another technique, either WGL or another technique, since we did not use another BC localization technique in our hospital. Another limitation of our study is related to the fact that, in our current procedure, we systematically used macroscopic pathological examination of the breast surgical specimen for intraoperative margin assessment, which by its nature will improve the rate of complete resection of the tumor, and not just the CL procedure itself.

Conclusion
We present this study to highlight the results of a multidisciplinary team work from our institute on the localization of nonpalpable breast tumors by CL. Tis was the largest study on this subject to date and demonstrated high success rates for BC tumor identifcation (99%) and one-stage complete resection (93.4%). Tese results demonstrate that CL is an efective localization technique which can guide the surgeon towards successful excision of BC lesions during BCS using a simple and inexpensive method.