MP Clinical Data PDF

MarginProbe® Peer-reviewed Literature Significant and consistent reduction in positive margins and re-excision Allweis Thill Schnabel Sebastian Blohmer Coble Kupstas Gooch Hermann Geha Cen...

MarginProbe® Peer-reviewed Literature Significant and consistent reduction in positive margins and re-excision Allweis Thill Schnabel Sebastian Blohmer Coble Kupstas Gooch Hermann Geha Cen (Israel) (Germany) (US/Pivotal) (US) (Germany) (US) (US) (US) (IL) (US) (US) Arch Ann Surg Am J Surg The Breast Ann Surg SpringerPlus Gynecol Am J Surg Am J Surg Breast J Breast J Ann Surg Published Oncol (2008) (2014) Oncol (2014) (2015) Obstet (2017) (2018) (2020) (2020) Oncol (2021) (2019) (2016) Prospective, Prospective, Retrospective, Retrospective, Retrospective, Prospective, Prospective, Single Arm, Prospective, Single Arm, Study Single Arm, Single Arm, Single Arm, Retrospective Retrospective Double Arm, Retrospective Double Arm, Controlled, Double Arm, Controlled, Comparison to Comparison to Comparison to Chart Review, Chart Review, Randomized, Chart Review, Design Randomized, Comparison to Randomized, Historical Comparison to Historical Historical Single Arm Single Arm Controlled Single Arm Controlled Historical Controlled Historical Control Control Control Control Control Number of 293 109 596 351 322 256 240 341 106 46 42 Patients Re-excision: 12.7% 39% 25.8% 25.8% 30% 15% 18.3% N/A 10.4% 34.8% 30.8% Control Re-excision: 5.6% 17% 19.8% 9.7% 14.5% 6.6% 9.2% 9.7% 2.8% 4.3% 6.3% MarginProbe® Relative 23% re-ex 56% 56% 62% 52% 56% 50% 75% margins 73% 88% 80% Reduction 62% margins Consistent Columbia site performance Gross Full Cavity Post No Tumor of Schnabel across all Assessment Shave SOC on Ink Gross Pivotal Trial Pure DCIS historical Performance 8-10cc SOC Guideline Assessment Thicker shaves Cosmesis re-ex rates 32% reduction maintained Notes favorable Cosmesis additional regardless of SOC resulted in Neoadjuvant favorable tissue 44-17% = 61% No change in in tissue No change in density significantly 30-10% = 67% tissue volume volume tissue volume IORT more positive 16-8% = 50% margins 12-4% = 67% cleared The American Journal of Surgery (2008) 196, 483– 489 The American Society of Breast Surgeons A prospective, randomized, controlled, multicenter study of a real-time, intraoperative probe for positive margin detection in breast-conserving surgery Tanir M. Allweis, M.D.a,*, Zvi Kaufman, M.D.b, Shlomo Lelcuk, M.D.c, Itzhak Pappo, M.D.d, Tami Karni, M.D.d, Shlomo Schneebaum, M.D.e, Rona Spector, M.D.c, Asher Schindel, M.D.c, Dan Hershko, M.D.f, Moshe Zilberman, M.D.g, Joel Sayfan, M.D.h, Yuri Berlin, M.D.h, Amram Hadary, M.D.i, Oded Olsha, M.D.j, Haim Paran, M.D.b, Mordechai Gutman, M.D.b, Moshe Carmon, M.D.j a Hadassah Hebrew University Medical Center, Jerusalem, Israel; bMeir General Hospital, Kfar-Saba, Israel; cRabin Medical Center, Belinson Campus, Petah-Tikva, Israel; dAssaf-Harofeh Medical Center, Zrifin Beer-Yaakov, Israel; eTel Aviv Sourasky Medical Center, Tel-Aviv, Israel; fRambam Medical Center, Haifa, Israel; gThe Baruch Padeh Medical Center, Poriya, Israel; hHaemek Medical Center, Afula, Israel; iSieff Government Hospital, Safed, Israel; jShaare Zedek Medical Center, Jerusalem, Israel KEYWORDS: Abstract Breast cancer; BACKGROUND: This randomized, double-arm trial was designed to study the benefit of a novel Breast-conserving device (MarginProbe, Dune Medical Devices, Caesarea, Israel) in intraoperative margin assessment for surgery; breast-conserving surgery (BCS) and the associated reduction in reoperations. Intraoperative margin METHODS: In the device group, the probe was applied to the lumpectomy specimen and additional assessment; tissue was excised according to device readings. Study arms were compared by reoperation rates and Surgical margins; by correct surgical reaction confirmed by histology. Lumpectomy; RESULTS: Three hundred patients were enrolled. Device use was associated with improved correct Positive margins surgical reaction, defined as additional re-excision in all histologically detected positive margins, with tumor within 1 mm of inked margin. The repeat lumpectomy rate was significantly reduced by 56% in the device arm: 5.6% versus 12.7% in the control arm. There were no differences in excised tissue volume or cosmetic outcome. CONCLUSIONS: Intraoperative use of the MarginProbe for positive margin detection is safe and effective in BCS and decreases the rate of repeat operations. © 2008 Elsevier Inc. All rights reserved. Complete removal of tumor with clear margins during breast-conserving surgery (BCS) for breast cancer is one of the major goals of the operation. Failure to obtain clear Supported by Dune Medical Devices, Caesarea, Israel. * Corresponding author. Tel.: ⫹972-2-5844111; fax: ⫹972-2-5844232. margins is one of the leading risk factors for local recur- E-mail address: [email protected] rence.1–7 However, since microscopic involvement of the Manuscript received April 23, 2008; revised manuscript May 29, 2008 margins is not readily assessable intraoperatively, finding 0002-9610/$ - see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.amjsurg.2008.06.024 484 The American Journal of Surgery, Vol 196, No 4, October 2008 tumor close to or at the margin of a lumpectomy specimen Subjects were randomized into 2 arms: with or without by permanent histology is not an unusual event. Reopera- device use. In both arms surgeons were allowed to use any tion for positive margins after BCS may be necessary in up standard of care (SOC) intraoperative methods to evaluate to 50% of cases,8 –11 and is associated with patient discom- margin status such as palpation, specimen imaging, and fort, increased costs, and poorer cosmetic outcome. Na- intraoperative gross and/or microscopic pathology assess- tional Comprehensive Cancer Network guidelines12 suggest ment. Pathology data were collected for the primary that positive lumpectomy margins require reoperation, and lumpectomy and all repeat ipsilateral surgical procedures standard practice is that it is unacceptable to have tumor within 6 months. Pathology data included specimen dimen- cells directly at the margin of the excised specimen.13 How- sions, oriented histological margin distance for each of the ever, there is no consensus regarding the required distance 6 specimen aspects, as well as microscopic evaluation for of tumor from the cut edge1,14 and positive/close margins tumor presence and margin width of all re-excised margin may be defined anywhere between tumor at the inked mar- specimens. Cosmetic evaluation was performed at baseline, gin to tumor within 5 mm of the inked margin.10,13,15 7–30 days, and 3– 6 months following surgery by an observer Patients not amenable to margin-free lumpectomy should unaware of the patient’s randomization assignment. A sym- undergo mastectomy.12 metry-based scale2 was used for assessment as follows: As many as 54% of main lumpectomy specimens will 1. Excellent: treated breast almost identical to untreated have positive or close margins,16 –18 requiring re-excision. breast Intraoperative information regarding the presence of tumor 2. Good: minimal difference between the treated and un- at or near the margin enables the surgeon to excise addi- treated breasts tional tissue immediately, minimizing the need for reopera- 3. Fair: obvious difference between treated and untreated tion. Intraoperative assessment based on palpation alone is breasts highly inaccurate and frozen section or touch prep cytology 4. Poor: major functional and esthetic sequelae in treated for margin assessment are time-consuming and impre- breast. cise.2,19 A new device (MarginProbe, Dune Medical De- vices, Caesarea, Israel) that can detect tumor cells at or close to the margin of resection has been developed. When Device description previously studied intraoperatively in a blinded fashion,20 the device detected all positive margins in 86% (19/22) of The MarginProbe is used for intraoperative margin as- pathologically positive lumpectomy specimens from 57 pa- sessment. The device components include a console and a tients. sterile, disposable hand-piece, which are connected by ca- The device is intended to be applied to the main lumpec- bles. Measurement is performed by applying the probe tip to tomy specimen immediately upon excision, which is when a point on the resected lumpectomy specimen (Figure 1). At the surgeon decides if any additional tissue needs to be each point measured, radiofrequency signals are transmitted re-excised. We sought to assess real-time intraoperative use from the probe to the tissue, reflected back, and collected by of the device in a randomized controlled manner. the console. The reflected signals are algorithmically ana- The study was designed to assess the safety of the device lyzed and the device readings are displayed as “positive” or and its effectiveness versus a control group in increasing the likelihood of a correct intraoperative surgical reaction dur- ing lumpectomy procedures. In addition, the impact of de- vice use on reoperation rates, excised tissue volume, and cosmetic outcome were studied. Materials and Methods Study design and patient selection Subjects were enrolled at 11 institutions in a prospective, multicenter, randomized, controlled, institutional review board–approved study. All patients were scheduled to un- dergo BCS for treatment of invasive and/or pre-invasive breast cancer. Patients were excluded if they had received neoadjuvant chemotherapy, had prior surgical procedures or implants in the ipsilateral breast, or were participating in other clinical trials that might interfere with the protocol or device measurements. All patients were at least 18 years of Figure 1 Intraoperative application of device to breast lumpec- age, and were capable of giving informed consent. tomy specimen. T.M. Allweis et al. Intraoperative breast margin assessment device 485 “negative.” The probe has a footprint diameter of 16 mm, an lumpectomy specimen, the surgeon re-excised tissue from effective measurement area diameter of 7 mm, and a detec- the lumpectomy cavity in the margins indicated as positive tion depth of about 1 mm. Each measurement is completed by the device (in the “device” arm), as well as any margins in 1.5 seconds, allowing for sampling of multiple points indicated by standard intraoperative methods in both arms. over the specimen surface in a short period of time. The device was only applied to the main lumpectomy spec- imen. Applying the device to intraoperatively re-excised margin specimens was precluded by protocol. Likewise, the Intraoperative procedure device was not used in reoperations. Lumpectomy was performed in the standard fashion. Only once the excision and suture orientation of the main Pathology lumpectomy specimen were completed, the patient was ran- domized to 1 of 2 study arms: “device” or “control.” In All specimens were suture-oriented in the operating patients randomized to the “device” arm, the device was room and sent for permanent pathological analysis. Speci- applied by the surgeon to the main lumpectomy specimen, mens were inked, sectioned, and embedded in paraffin as as follows: each of 6 margins (medial, lateral, superior, per each institution’s standard procedure. Slides were inferior, deep and anterior) was sampled by the probe at 5– 8 stained with hematoxylin and eosin and evaluated for the points. Output was displayed on the screen, and results were presence of carcinoma at or near the margins. grouped and labeled by margin (Figure 2). A short blue bar indicated a negative reading, while a long red bar indicated Repeat surgical procedures a positive reading. A margin was considered positive if 1 or more readings were indicated as positive by the device. In Postoperative patient management was not dictated by both study arms the surgeons used standard intraoperative the protocol. Surgeons recommended and performed reop- margin assessment at their discretion, and performed re- erations according to institutional treatment paradigms and excision of cavity margins as deemed necessary. In addition patient preferences. For example, various institutions apply to palpation, the intraoperative modalities used included different criteria for insufficient margin width requiring specimen imaging and intraoperative pathological assess- additional surgery. Furthermore, reoperation is sometimes ment (gross and microscopic). Both specimen imaging and avoided despite insufficient margin width, eg, when all any intraoperative pathological evaluation, if used, were breast tissue has already been excised in the relevant as- performed after device application away from the operating pects. Performance of additional breast-conserving surger- room, by a professional blinded to the patient’s randomiza- ies, as well as conversions to mastectomy, were followed tion arm. Results of these assessments were reported back to and documented within 6 months of primary surgery. the surgeon. Following sampling of all margins on the Data analysis Effectiveness in both study arms was determined accord- ing to final histology data. The ability to correctly and intraoperatively identify all of the involved margins on the main specimen, and re-excise them was defined as correct intraoperative surgical reaction (ISR). Correctness or incor- rectness of ISR was defined based on permanent histology data. ISR was defined as correct only when all main spec- imen margins detected as positive by histology were re- excised intraoperatively. A margin was considered positive (for the purpose of ISR calculation) if tumor was present within 1 mm of the inked surface. Study arms were com- pared for correct ISR, reoperation rates (including and ex- cluding mastectomy), long-term cosmetic outcome, and to- tal volume of tissue excised. Figure 3 describes the calculation of correct or incorrect ISR. Thus all of the histologically positive margins had to have corresponding Figure 2 Device output display for a typical patient. Data ac- re-excised specimens (as documented in the histology re- cumulate on the screen from left to right and from top to bottom. port) in order for the case to be considered one with correct Most recent measurement is highlighted at the top left. Blue and ISR. Total tissue volume was calculated from the specimen red bars are negative and positive readings, respectively. Yellow dimensions as recorded in the histology report by applying frames and labels mark the margins from which readings were the spheroid volume formula (including main specimen and obtained. all additional intraoperatively re-excised specimens). Sub- 486 The American Journal of Surgery, Vol 196, No 4, October 2008 vice” and “control” arms, respectively. Of these, 7 patients did not fit the inclusion/exclusion criteria for the study or had protocol violations and were excluded (1 received neo- adjuvant therapy, 1 underwent a mastectomy as the primary procedure, 2 had prior surgery in the same quadrant, 1 was not preoperatively diagnosed with malignancy, and in 2 patients the device was not applied to all specimen mar- gins). Analysis was performed on 293 patients, 143 and 150 in the “device” and “control” arms, respectively. For the NPL subgroup, 168 patients were analyzed, 82 in the “de- vice” arm and 86 in the “control” arm. Demographic and tumor characteristics were well balanced between both arms for the entire cohort and for the NPL subgroup (Table 1). The 2 study arms were also balanced with regards to use and effect of intraoperative methods for margin assessment (Table 2). Entire patient cohort Randomization was performed following excision of the main lumpectomy specimen to ensure uniformity of the initial specimen excision in both study arms. Main lumpec- tomy specimen positivity rates, ie, tumor detected within 1 mm of a margin, were similar in both groups: 41% (58/143) in the “device” group and 41% (61/150) in the “control” group. Correct ISR rate was significantly higher in the “device” arm than in the “control” arm, 60% (35/58) versus 41% (25/61), respectively (P ⫽.044). Seventeen patients under- went 18 repeat operations (7 re-excisions and 10 mastecto- mies in a second operation, 1 re-excision in a third opera- Table 1 Patient characteristics Figure 3 Description of intraoperative surgical reaction (ISR). Entire cohort NPL subgroup A case is considered as: (a) correct ISR if all histologically posi- tive specimen margins have corresponding re-excised specimens; “Device” “Control” “Device” “Control” or (b) incorrect ISR if at least 1 histologically positive specimen margin was not intraoperatively re-excised. n 143 150 82 86 Age (y) 59 60 60 62 Lesion size, mean (mm) 18.7 17.1 14.8 15.2 group analysis was performed for patients with nonpalpable Histology (%) lesions (NPL) who underwent preoperative image-guided Invasive 37 38 36 37 localization. ISR and cosmetic outcome were compared Ductal carcinoma between study arms using the Fisher exact test. Re-opera- in situ (DCIS) 12 8 18 14 tion rates and excised tissue volume were compared using Invasive lobular 5 6 6 2 Poisson regression and Wilcoxon rank-sum test, respec- Mixed 46 48 40 47 Tumor grade (%) tively. 1 18 15 22 20 In order to assess the device’s contribution to surgical 2 48 44 44 47 decisions under varying surgical paradigms, ISR was recal- 3 34 41 34 33 culated for both study arms using a range of positive his- ER/PR status (%) tological margin definitions of 1–5 mm. Both positive 67 65 61 64 Both negative 19 19 18 15 One positive 14 17 21 21 Her2 neu status (%) Positive (2⫹, 3⫹) 24 32 28 24 Results Negative (0, 1⫹) 76 68 72 76 No statistical difference between arms. From November 2006 through November 2007, 300 pa- ER ⫽ estrogen receptor; PR ⫽ progesterone receptor. tients were enrolled in the study, 149 and 151 in the “de- T.M. Allweis et al. Intraoperative breast margin assessment device 487 Table 2 Use of intraoperative modalities for margin and “control,” respectively). The main lumpectomy speci- assessment in both study arms men positivity rate was similar in both groups: 35% (29/82) in the “device” arm and 38% (33/86) in the “control” arm. Entire cohort NPL subgroup Correct ISR rate was significantly higher in the “device” arm than in the “control” arm, 69% (20/29) versus 39% “Device” “Control” “Device” “Control” (13/33), respectively (P ⫽.024). Eight patients underwent 8 n 143 150 82 86 reoperations in the “device” arm and in the “control” arm 13 Specimen imaging patients underwent 18 reoperations. Thus, reoperation rates (%) 51 48 85 83 were significantly lower with device use compared to the Intraoperative control arm, at 9.8% (8/82) and 20.9% (18/86), respectively pathology (%) 22 26 10 20 (P ⫽.02), a 53% decrease. A similar significant 52% decrease in re-excision rates is observed when excluding mastectomy procedures. tion) in the “device” arm, and 23 patients underwent 28 The average tissue volumes excised during the first sur- repeat operations (17 re-excisions and 6 mastectomies in a gical procedure in the NPL group were 100 cm3 and 96 cm3 second operation, 2 re-excisions and 3 mastectomies in a (P ⫽.30) for “device” and “control,” respectively. The third operation) in the “control” arm. Reoperation rate was proportion of patients with long-term “excellent” or “good” lower in the “device” arm compared with the “control” arm, cosmetic evaluation was similar in both arms, 78% and 71% 12.6% (18/143) versus 18.6% (28/150) respectively; how- (P ⫽.31) for “device” and “control,” respectively. Results ever, despite this 32.3% reduction in repeat operation rate, are displayed in Table 3. this difference did not achieve statistical significance (P ⫽.098). When excluding patients who ultimately underwent Surgical reaction using device output mastectomy, re-excision rate was significantly lower in the “device” arm, 5.6% (8/143) compared with 12.7% (19/150) Device output should have led to removal of additional in the “control” arm (P ⫽.0027), amounting to a significant tissue in the aspects that were indicated as positive. This 56% decrease in re-excision rates. was not always feasible, eg, if tissue was excised down to The average total tissue volumes excised during the first the pectoralis major muscle or anteriorly up to the subder- procedure were 107 cm3 and 94 cm3 (P ⫽.066) for the mal plane. In this study, 448 margins were indicated as “device” and “control” groups, respectively. The proportion positive by the device in 143 patients. Only 76% (342/448) of patients with long-term “excellent” or “good” cosmetic and 78% (186/237) of these margins were intraoperatively evaluation was similar in both arms (71% and 69% for re-excised by the surgeons in the entire cohort and NPL “device” and “control,” respectively, P ⫽.71). Results are subgroup, respectively. Of the 106 margins in 65 patients displayed in Table 3. that were not re-excised in the entire cohort, 57% (60/106) were in the anterior or deep aspect where no breast tissue NPL patient subgroup remained to be resected. The surgeons could have re-ex- cised the remaining 43% (46/106) of device-detected posi- The proportion of patients with NPL undergoing preop- tive margins but elected not to do so. Had these latter erative image-guided localization was balanced between margins been re-excised, correct ISR for the “device” arm study arms (57% [82/143] and 57% [86/150] for “device” would have increased to 67% (39/58). Similar proportions Table 3 Summary of results in entire patient cohort and NPL patient subgroup “Device” “Control” P value Entire cohort n 143 150 Correct ISR rate (%) 60 (35/58) 41 (25/61).044 Re-excision rate (%) 5.6 (8/143) 12.7 (19/150).0027 Reoperation rate (%) including mastectomy 12.6 (18/143) 18.6 (28/150).098 Excised tissue volume, mean (cm3) 107 94.066 Good or excellent long-term cosmetic outcome 71 69 1 NPL cohort n 82 86 Correct ISR rate (%) 69 (20/29) 39 (13/33).024 Re-excision rate (%) 6.1 (5/82) 12.8 (11/86).039 Reoperation rate (%) including mastectomy 9.8 (8/82) 20.9 (18/86).020 Excised tissue volume, mean (cm3) 100 96.300 Good or excellent long-term cosmetic outcome 78 71.311 488 The American Journal of Surgery, Vol 196, No 4, October 2008 were observed in the NPL subgroup with 51 margins indi- and 33% for a 5-mm margin. A similar effect was noted cated as being positive by the device in 31 patients not when this analysis was performed for the NPL subgroup. re-excised, 37% (19/51) of them which could have been Re-excision rates did not change since they were not de- re-excised, and correct ISR would have increased to 79% pendent on the study definition of positive margins. (23/29) had all these margins been re-excised. Positive margin thresholds Comments Results of the benefit of device use presented above are for a positive margin defined as histologically detected This prospective, randomized, controlled study demon- cancer within 1 mm of the lumpectomy margin. However, strates the safety and effectiveness of a new device for in light of the variability in patient management paradigms, real-time intraoperative margin assessment. After excision we decided to compare the study arms with other definitions of the main lumpectomy specimen during BCS, the surgeon of what constitutes a positive margin. must decide whether additional tissue needs to be re-excised The positivity threshold was raised to 2, 3, and 5 mm and from the lumpectomy cavity, and if so, from which aspects. correct ISR was recalculated for both study arms. When In this study, patients were randomized at this point of the positivity thresholds increase and become more conserva- procedure to 2 study arms, with or without device use. In tive, correct ISR rates decrease in both arms, as can be both arms surgeons equally used other available intraoper- expected; however, the benefit of “device” over “control” is ative modalities for margin assessment, as deemed neces- maintained (Figure 4). Correct ISR rates for the “device” sary. These included palpation, specimen imaging, and arm shifted from 60% when the margin was defined as 1 intraoperative gross and/or microscopic pathology assess- mm to 52% when the margin was defined as 5 mm. For the ment. We followed the surgical decisions to further remove “control” arm, the correct ISR was 41% for a 1-mm margin tissue for every patient and judged them based on the final histological report. Results indicate that use of the device at this point in the procedure, in conjunction with other intra- operative modalities, will provide additional margin assess- ment data, and effectively improve the surgeon’s ability to correctly react by further intraoperative re-excision. The ability to correctly re-excise all histologically positive mar- gins was enhanced by nearly 50%, from 41% to 60% when device use was added to other intraoperative modalities. The need for reoperation was not defined or dictated by the study protocol. Every surgeon used his/her routine criteria for taking a patient back to the operating room. Sometimes patients with histological margins ⱕ1 mm were not taken back to the operating room for one of the following reasons: all breast tissue in the relevant aspect was already excised in the first procedure (eg, the deep margin was positive but had been excised down to the pectoral muscle); institutional criteria for margin positivity was different than the study defined 1 mm threshold; and patient preference. This ac- counts for the difference between incomplete ISR rates and reoperation rates in both arms. With all those real-life ele- ments and considerations affecting results, actual re-exci- sion rates decreased accordingly by 56%. Device benefit is somewhat more pronounced in the subgroup of patients with NPL. Although it was initially presumed that surgeons may find these NPL more challeng- ing to surgically locate and fully excise, results showed that correct intraoperative reaction and reoperation rates were maintained in the control arm at about 40% and 20%, respectively, in this subgroup. While the SOC modalities for margin assessment have similar performance in NPL com- Figure 4 Comparison of “device” and “control” over a range of pared to palpable lesions, device use improves and with it positive margin definitions in (a) entire patient cohort and (b) a the benefit of the device relative to SOC. In nearly 70% of subgroup of patients with nonpalpable lesions (NPL) who under- patients in the “device” arm the surgeon was able to intra- went preoperative image guided localization. operatively re-excise all histologically positive margins, T.M. Allweis et al. Intraoperative breast margin assessment device 489 compared to nearly 40% in the “control” arm, an increase of References more than 75%. For this subgroup, over 50% fewer patients were taken back to the operating room for additional sur- 1. Newman LA, Washington TA. New trends in breast conservation gery when the device was used. therapy. Surg Clin North Am 2003;83:841– 83. Device use does not lead to excision of significantly 2. Morrow M, Strom EA, Bassett LW, et al. Standard for breast conser- vation therapy in the management of invasive breast carcinoma. CA larger tissue volume for both the entire cohort and the NPL Cancer J Clin 2002;52:277–300. subgroup. Cosmetic outcome is maintained relative to SOC. 3. Gentilini O, Intra M, Gandini S, et al. Ipsilateral breast tumor reap- Lacking a consensus on what constitutes a positive margin, pearance in patients treated with conservative surgery after primary surgeons use different criteria. Some consider a margin to chemotherapy. The role of surgical margins on outcome. J Surg Oncol be positive only if tumor cells are located at the specimen 2006;94:375–9. 4. Smitt MC, Nowels K, Carlson RW, et al. Predictors of reexcision edge while others use more conservative paradigms requir- findings and recurrence after breast conservation. Int J Radiat Oncol ing a clear margin of 1 mm, 2 mm or more. Initially, Biol Phys 2003;57:979 – 85. positive margins were defined in this study to be those with 5. Huston TL, Simmons RM. Locally recurrent breast cancer after con- tumor detected within 1 mm from the margin, and correct servation therapy. Am J Surg 2005;189:229 –35. ISR was judged accordingly. However, data from histopa- 6. Leong C, Boyages J, Jayasinghe UW, et al. Effect of margins on ipsilateral breast tumor recurrence after breast conservation therapy for thology reports was also available for margin widths which lymph node-negative breast carcinoma. Cancer 2004;100:1823–32. were greater than 1 mm. Thus, device benefit over SOC was 7. McCready DR. Keeping abreast of marginal controversies. Ann Surg analyzed for positive margins that were more conservatively Oncol 2004;11:885–7. defined. Correct ISR was recalculated for a positive margin 8. Tafra L, Fine R, Whitworth P, et al. Prospective randomized study definition that was gradually reset at 1 mm to 5 mm. As the comparing cryo-assisted and needle-wire localization of ultrasound visible breast tumors. Am J Surg 2006;192:462–70. required clear margin width increases, more margins are 9. Menes TS, Tartter PA, Bleiweiss I, et al. The consequence of multiple rendered positive by definition, and it becomes more diffi- re-excisions to obtain clear lumpectomy margins in breast cancer cult to detect and re-excise all of them. It is thus not patients. Ann Surg Oncol 2005;12:1–5. surprising that correct ISR in both study arms decreases 10. McLaughlin SA, Ochoa-Frongia LM, Patil SM, et al. Influence of when the wider margin definition is applied. However, it is frozen-section analysis of sentinel lymph node and lumpectomy mar- noteworthy that the benefit of using the device in conjunc- gin status on reoperation rates in patients undergoing breast-conser- vation therapy. J Am Coll Surg 2008;206:76 – 82. tion with other modalities is maintained throughout the 11. Mullenix PS, Cuadrado DG, Steele SR, et al. Secondary operations are range. This holds true for both the entire cohort and the frequently required to complete the surgical phase of therapy in the era subgroup of patients with image guided localization. In an of breast conservation and sentinel lymph node biopsy. Am J Surg earlier study performed with this same device, Karni et al20 2004;187:643– 6. reported similar results. While in Karni et al’s work device 12. NCCN Clinical Practice Guidelines in Oncology 2007. National Com- prehensive Cancer Network. output was recorded but surgeons were blinded and did not 13. Singletary SE. Surgical margins in patients with early-stage breast respond to device output, data from the current work yield cancer with breast conservation surgery. Am J Surg 2002;184:383–93. similar results in actual use of device output in intraopera- 14. Taghian A, Mohiuddin M, Jagsi R, et al. Current perceptions regarding tive decision-making. The implication of these results is that surgical margin status after breast-conserving therapy: results of a whatever the preferred paradigm of clear margin width, survey. Ann Surg 2005;241:629 –39. 15. Silverstein MJ, Lagios MD, Groshen S, et al. The influence of margin patients are likely to benefit if surgeons add device use to width on local control of ductal carcinoma in situ of the breast. N Engl their other available intraoperative modalities and re-excise J Med 1999;340:1455– 61. all margins indicated by the device as positive. 16. Chagpar A, Yen T, Sahin A, et al. Intraoperative margin assessment The device is safe and effective in increasing the ability reduces reexcision rates in patients with ductal carcinoma is situ of the surgeon to detect positive margins and to react cor- treated with breast conserving surgery. Am J Surg 2003;186:371–7. 17. Keskek M, Kotharia M, Ardehalia B, et al. Factors predisposing to rectly during the primary procedure. The device thus con- cavity margin positivity following conservation surgery for breast tributes to a significant 56% reduction in reoperation rates cancer. Eur J Surg Oncol 2004;30:1058 – 64. for BCS. 18. Cabioglu N, Hunt KK, Sahin AA, et al. Role for intraoperative margin In several cases in this study surgeons chose not to assessment in patients undergoing breast conserving surgery. Ann further excise when a positive margin was detected in a Surg Oncol 2007;14:1458 –71. 19. Klimberg VS, Harms S, Korourian S. Assessing margin status. Surg margin that was amenable to further excision (ie, not the Oncol 1999;8:77– 84. anterior or deep margins). Future studies should seek to 20. Karni T, Pappo I, Sandbank J, et al. A device for real-time, intraop- measure the effect of these decisions, and the optimal mode erative margin assessment in breast-conservation surgery. Am J Surg of use of this device. 2007;194:467–73. Journal of Surgical Research 160, 277–281 (2010) doi:10.1016/j.jss.2009.02.025 RESEARCH REVIEW Diagnostic Performance of a Novel Device for Real-Time Margin Assessment in Lumpectomy Specimens Itzhak Pappo, M.D.,*,1 Rona Spector, M.D.,‡ Asher Schindel, M.D.,‡ Sara Morgenstern, M.D.,§ Judith Sandbank, M.D.,† Leonor Treho Leider, M.D.,{ Shlomo Schneebaum, M.D.,k Shlomo Lelcuk, M.D.,‡ and Tami Karni, M.D.* *Department of General Surgery, Assaf Harofeh Medical Center, Zrifin, Israel; †Department of Pathology, Assaf Harofeh Medical Center, Zrifin, Israel; ‡Department of General Surgery, Rabin Medical Center, Beilinson Campus, Petach-Tikva, Israel; §Department of Pathology, Rabin Medical Center, Beilinson Campus, Petach-Tikva, Israel; kDepartment of General Surgery, Sorasky Medical Center, Tel-Aviv, Israel; and {Department of Pathology, Sorasky Medical Center, Tel-Aviv, Israel Submitted for publication December 9, 2008 Conclusions. The device has high sensitivity and Background. Margin status in breast lumpectomy specificity in distinguishing between normal and can- procedures is a prognostic factor for local recurrence cer tissue even down to small cancer features. Ó 2010 and the need to obtain clear margins is often a cause Elsevier Inc. All rights reserved. for repeated surgical procedures. A recently developed Key Words: breast cancer; dielectric tissue proper- device for real-time intraoperative margin assessment ties; margin status. (MarginProbe; Dune Medical Devices, Caesarea, Israel), was clinically tested. The work presented here looks at the diagnostic performance of the device. Methods. The device was applied to freshly excised lumpectomy and mastectomy specimens at specific tis- INTRODUCTION sue measurement sites. These measurement sites were accurately marked, cut out, and sent for histopatho- The combination of breast conservation surgery with logic analysis. Device readings (positive or negative) radiation therapy was shown to be equivalent, in terms were compared with histology findings (namely malig- of survival, to mastectomy for treatment of breast can- nant, containing any microscopically detected tumor, cer. However, this requires achieving microscopi- or nonmalignant) on a per measurement site basis. The sensitivity and specificity of the device was com- cally clean resection margins in the excised lump. puted for the full dataset and for additional relevant While clean margins can be achieved in the initial sur- subgroups. gery, re-excision rates reported are between 21% and Results. A total of 869 tissue measurement sites 50% [3, 4]. In addition to increased costs, morbidity, were obtained from 76 patients, 753 were analyzed, of and poorer cosmetic results, the rate of local recurrence which 165 were cancerous and 588 were nonmalignant. in patients who were re-excised was found to be higher Device performance on relatively homogeneous sites than in those with initial clean margins [5–7]. An inves- was: sensitivity 1.00 (95% CI: 0.85–1), specificity 0.87 tigational device for real-time intraoperative assess- (95% CI: 0.83–0.90). Performance for the full dataset ment of margin status (MarginProbe; Dune Medical was: sensitivity 0.70 (95% CI: 0.63–0.77), specificity Devices, Caesarea, Israel) has been shown to reduce 0.70 (95% CI: 0.67–0.74). Device sensitivity was esti- re-excision rates in breast conservation surgeries by mated to change from 56% to 97% as the cancer feature size increased from 0.7 mm to 6.6 mm. Detection rate of more than 50% by detecting positive margins during samples containing pure DCIS clusters was not differ- the initial excision. ent from rates of samples containing IDC. This device is based on near-field radio frequency (RF) spectroscopy and is designed to detect differences 1 To whom correspondence and reprint requests should be ad- between dielectric properties of malignant and normal dressed at The Comprehensive Breast Care Institute, Assaf Harofeh breast tissue adjacent to the probe’s sensor. These dif- Medical Center, Zerifin 70300, Israel. E-mail: [email protected]. ferences have been studied since the last century and 277 0022-4804/09 $36.00 Ó 2010 Elsevier Inc. All rights reserved. 278 JOURNAL OF SURGICAL RESEARCH: VOL. 160, NO. 2, MAY 15, 2010 FIG. 1. Tissue handling during the study: (A) breast tissue specimen marking with numbered pins at the center of measurement sites; (B) cutting of a tissue disk around the pin with tissue punch; (C) placement of tissue disks in numbered cassettes. (Color version of figure is avail- able online.) are well established [9-12]. The differences are a result site for subsequent analysis (see description in next section) required of changes in ion concentration and mobility, metabo- more time—approximately 1 min per site. The signals are analyzed in real time, and each measurement is given a positive (malignant), neg- lism, water content, membrane and cellular structure, ative (nonmalignant), or a failed result (criteria described below) and other characteristics. The device is a single use based on an algorithm that was optimized on a different dataset. hand-held tool, connected via cables to a console. The hand-held probe is applied to the lumpectomy specimen Tissue Measurements and Processing surface and provides the status of the measurement site At the pathology lab, following inking, freshly excised specimens in contact with the device tip in real-time. The device were bread-loafed in order to facilitate access to tumor peripheral does not require intraoperative pathologic analysis. measurement sites. The probe was placed in contact with 7-mm In this study, we establish the basic diagnostic perfor- wide measurement sites (up to 25 per specimen) on internal surfaces mance of the device by comparing device output with of slices of the specimen, and measurements were taken and recorded. Both grossly cancerous-appearing and normal-appearing measure- permanent histopathologic evaluation of the measured ment sites were sampled. Specific attention was given to the grossly tissue sites, sampled on resected breast specimens. normal appearing sites located close to the tumor. The device readings were logged in the console. All personnel remained blinded to the re- METHODS sults during the procedure. In order to ensure accurate correspon- dence between the tissue site to which the device was applied and that which was histopathologically evaluated, immediately after General Design each measurement the tissue site was marked by a numbered pin Tissue measurements were performed on freshly excised lumpec- placed at the center of the measured area (Fig. 1A). After routine fix- tomy and mastectomy specimens. All medical staff members were ation of the specimen (for permanent histopathology analysis), a coin- blinded to device output. Measured tissue sites were individually shaped tissue specimen was cut out from each pinned site using marked, precisely cut out from the main specimen using a biopsy a 10-mm-diameter surgical punch (Fig. 1B), creating 10-mm-diameter punch, routinely processed, and histologically analyzed. tissue disks approximately 2 to 3 mm thick. These tissue samples Eighty patients consented and were enrolled from March 2006 were placed in cassettes, numbered according to the pin numbers, through June 2007 at three medical centers. Inclusion criteria speci- with the measured side facing a predefined side of the cassette fied female patients over 18 y of age, with diagnosed infiltrative or (Fig. 1C). Tissue ink was applied in the hole left by the pin in order in situ carcinoma of the breast, undergoing lumpectomy or mastec- to enable the analysis of the measured area (a 7-mm diameter circle tomy. Excluded from this study were patients who underwent neoad- concentric to the center pin hole). A slide was prepared for each cas- juvant treatment, had prior surgery in the same breast quadrant, or sette, parallel to the measured plane. Special care was taken in order participated in another study. to make sure that tissue samples were flat and that each slide con- tained the full 10-mm circle of tissue, as close as possible to the mea- sured surface (less than 0.5 mm deep). Device Description Data Analysis The device used in this work was an investigational diagnostic device (MarginProbe; Dune Medical Devices). The MarginProbe is Tissue a near-field RF spectroscopy based real-time detection device, which is designed to distinguish between malignant and non-malignant Each histologic section was microscopically evaluated by a core breast tissue during surgery. It consists of a hand-held single use ster- pathology lab. In addition, slides were digitally scanned and the ob- ile probe, connected by cables to a console. When the probe tip is served tissue types were marked by the pathologist on the digital im- placed against tissue it automatically attaches a sensor to the tissue ages, based on the microscopic analysis of the slide. Within each slide, (using a vacuum-based mechanism). The lightweight, hand-held the pathologist marked all tissue types characterized. These included probe applies gentle suction to the tissue for controlled, user-indepen- cancerous types (e.g., invasive ductal carcinoma, ductal carcinoma in dent attachment. The console sends RF waves over a large frequency situ, invasive lobular carcinoma) and numerous nonmalignant types range, which are transmitted to the tissue through the sensor. The (e.g., mammary, adipose, fibrotic, adenosis, stroma, vascular, lym- RF signals are reflected from the tissue through the sensor, and are phatic). Following the digitization of the slides’ histopathology data, acquired by the console. The effective measurement volume of the slides with incomplete histologic data on the full 7-mm circular area sensor is a 7-mm-diameter disk, up to 1-mm thick. The probe’s me- measured by the device were disqualified by personnel blinded to de- chanical footprint is 15 mm in diameter. Each measurement takes vice output. Disqualification occurred due to two reasons (defined approximately 1 to 5 s to complete; thus the probe can be effectively prior to initiation of the study): (1) an incomplete 7-mm circular shape applied to dozens of measurement sites within a few minutes. In around the inked pin-hole; (2) unidentifiable inked pin-hole. Beyond this protocol, however, the accurate marking of each measurement these characteristics of disqualified slides no other common features  PAPPO ET AL.: REAL-TIME MARGIN ASSESSMENT IN LUMPECTOMY SPECIMENS 279 were present. The coverage area and feature size (denoted as the fea- ture diameter in mm) of the different tissue types were then analyzed automatically by use of dedicated software. Most of the tissue samples contained several tissue types. Sixty-nine percent of the samples con- tained four or more types and as many as eight types exist in several samples. Following this analysis, each tissue sample was categorized as malignant (if any amount of cancer cells was identified by the pa- thologist within the 7-mm-diameter circle centered at the inked pin- hole) or nonmalignant. Measurements Measurements performed by the device passed automatic qualifica- tion tests in real-time by the device software. Two tests are performed: (1) attachment test, based on vacuum levels in the probe acquired dur- ing the measurement; and (2) integrity test, checking for any malfunc- tion of the probe in acquiring the RF signal. All qualified measurements were compared with histology of the tissue samples on a per measurement site basis. Sensitivity and specificity were calculated for the full dataset, as well as for various relevant subsets acquired during the trial. Receiver operating charac- FIG. 2. ROC curves of 3 different datasets: a) samples with >75% teristic (ROC) curve analysis was performed on these sets by calculat- content of one tissue type - squares; b) samples with >50% content of ing device output off-line for various algorithm cut-off points and one tissue type - circles; and c) full dataset containing all samples - recalculating sensitivities and specificities for each cut-off. In addi- solid line. tion, an analysis of sensitivity for various subgroups was performed. Positive measurement sites were divided according to both cancer fea- a curve plots the sensitivity versus the false positive rate ture size and according to pathology finding, and the detection rate (100–specificity) at different cut-off points. Figure 2 per subgroup was calculated and compared. presents ROC curves of the device performance derived by analyzing the data offline and scanning the cutoff RESULTS point. The figure includes three datasets: (A) tissues containing at least 75% of a single tissue type; (B) all Eighty patients from 3 medical centers were included tissues containing at least 50% of a single tissue type; in the study. Four patients were excluded: two under- and (C) the full dataset collected in the experiment, con- went neoadjuvant treatment and for two, a device taining cancers of all sizes (down to 0.15-mm-diameter malfunction occurred and valid data were not obtained. features). In addition, Table 1 summarizes the ‘‘areas A total number of 869 measurement sites, from 76 under graph’’ (AUG) for each curve with their 95% con- patients, were measured and sampled. Of these, 116 fidence interval. The dataset containing the most homo- (13%) were excluded from this analysis due to incom- geneous tissues, presented in Fig. 2 by squares, has an plete histologic analysis (as noted above). Overall, 753 AUG of 0.96 (95% CI: 0.94–0.98). measurement sites were analyzed. To assess the device’s basic technologic ability, a sub- Sensitivity Analysis set of relatively homogeneous measurement sites was analyzed. The 447 measurement sites in this subset In this section, we analyze various subgroups of the had a single tissue type present in at least 75% of the cancerous measurement sites. Grouping by cancer fea- probe’s 7-mm-diameter measurement footprint. Of ture size and grouping by different cancer histopathol- these 22 (from 15 patients) were cancerous and 425 ogy were studied. Note that this analysis has no effect were nonmalignant. The performance of the device on on specificity. In order to assess the performance of this dataset was: sensitivity 1.00 (95% CI: 0.85–1), spec- the device as the cancerous feature size drops, all the ificity 0.87 (95% CI: 0.83–0.90). For measurement sites cancerous tissue samples (165) were divided into seven containing at least 50% of a single tissue type (29 cancer- groups according to cancer feature size, and the sensi- ous, from 18 patients, and 567 nonmalignant) the perfor- tivity of each group was calculated (Fig. 3A). An mance was: sensitivity 1.00 (95% CI: 0.88–1), specificity 0.72 (95% CI: 0.68–0.76). Classification performance on TABLE 1 the full dataset (165 cancerous sites from 50 patients, AUG of the ROC Curves in Fig. 3 and 588 nonmalignant sites) was: sensitivity 0.70 (95% Dataset description AUG (95% CI: lower–upper) CI: 0.63–0.77), specificity 0.70 (95% CI: 0.67–0.74). Tissue samples with >75% content 0.96 (95% CI: 0.94–0.98) ROC Curve Analysis of one tissue type (squares) Tissue samples with >50% content 0.95 (95% CI: 0.92–0.97) ROC curve presentation of a device’s performance can of one tissue type (circles) Full dataset (solid line) 0.74 (95% CI: 0.70–0.79) be helpful in the analysis of its diagnostic abilities. Such 280 JOURNAL OF SURGICAL RESEARCH: VOL. 160, NO. 2, MAY 15, 2010 TABLE 2 Device Sensitivity for Different Histopathology Sub- groups: Invasive Ductal Carcinoma (IDC), Ductal Car- cinoma In Situ (DCIS), Invasive Lobular Carcinoma (ILC), and Combinations of Types. The ‘‘Other’’ Group Includes Rarer Types Such as Mucinous and Papillary Carcinomas Cancerous samples Cancer histo- Number of samples Detected Detection rate pathology IDC 87 59 0.68 DCIS 35 22 0.63 ILC 7 5 0.71 IDC þ DCIS 25 21 0.84 ILC þ DCIS 3 3 1.00 Other 8 6 0.75 Full dataset 165 116 0.70 tween the subgroups. The two most common groups, invasive ductal carcinoma (IDC) and ductal carcinoma in situ (DCIS), have sensitivities of 0.68 (95% CI:57– 77) and 0.63 (95% CI:45–79), respectively. DISCUSSION This study shows that the MarginProbe distin- guishes between normal and malignant breast tissue FIG. 3. (A) Detection rate of samples with varying cancer feature size; actual rate in blue and estimated rate (using logistic regression) with high sensitivity and specificity (100% and 87%, re- in red, with estimation error bars denoting 95% confidence levels; (B) spectively, for relatively homogeneous measurement histogram of samples per group with average feature size denoted for sites). Device performance decreases for tissues with each group. (Color version of figure is available online.) more heterogeneous compositions. The cancerous sam- ples with high heterogeneity correspond to the small estimation of sensitivity as a function of cancer feature cancer feature sizes. The detection of measurement size was performed by logistic regression, shown in sites containing very small amounts of malignant tis- Fig. 3A, and fits the data well. The number of measure- sue was also studied here and the results show a depen- ment sites in each group is presented in Fig. 3B and is dence of sensitivity on cancer feature size—as cancer quite evenly spread. The estimated sensitivity changes feature size present in the measurement site decreases, from 97% to 56% as the average feature size per group so does sensitivity. For features above 5 mm in diame- drops from 6.6 mm to 0.7 mm (the group with the aver- ter, the sensitivity is 100%, while taking into account age size of 0.7 mm contains cancers of sizes as small as all the malignant tissue measured in the study, down 0.15 mm). All but one of the cancerous samples (28 out to the smallest feature size (0.15 mm), the sensitivity of 29) with more than 50% homogeneity belong in the decreases to 70%. Homogeneity of the nonmalignant 6.6 mm feature size group, which was detected with tissue samples affects the specificity. While most sam- sensitivity of 100%. The less homogeneous cancer sam- ples contain several tissue types, in the nonmalignant ples are spread between the various smaller feature homogeneous samples containing more than 75% of size groups. The lower sensitivity in these subgroups one tissue type, the coverage area of the other types is brings the overall sensitivity of all the cancer samples much smaller in comparison. However, in the rest of to 70%. the samples, the coverage area of the various types is It is also of interest to analyze the performance for evenly spread. The measurements were sampled at different histopathology types, irrespective of their fea- locations similar to those that will be encountered by ture size. The sensitivity of subgroups divided accord- a surgeon during use of the device in the operating ing to histopathology reporting is summarized in room. Thus, the single measurement diagnostic perfor- Table 2. No significant differences were observed be- mance presented here is anticipated to hold during  PAPPO ET AL.: REAL-TIME MARGIN ASSESSMENT IN LUMPECTOMY SPECIMENS 281 regular use of the device. Since measurements were not and lumpectomy plus irradiation for the treatment of invasive taken from the specimen surface, patient-level positive breast cancer. N Engl J Med 2002;347:1233; Veronesi U, Cascinelli N, Mariani L, et al. Twenty-year follow-up margin analysis was not performed. of a randomized study comparing breast-conserving surgery The device detection is not limited to invasive cancer with radical mastectomy for early breast cancer. N Engl J Med malignancy types. More elusive types such as DCIS 2002;347:1227. were detected, with no significant differences in detec- 2. NCCN Clinical Practice Guidelines in OncologyÔ. National tion rates relative to IDC and other types or combina- Comprehensive Cancer Network, 2007. tions of cancer types. The ability to detect DCIS is 3. Tafra L, Fine R, Whitworth P, et al. Prospective randomized significant since today DCIS is considered a more chal- study comparing cryo-assisted and needle-wire localization of ultrasound visible breast tumors. Am J Surg 2006;192:462. lenging intraoperative tissue assessment target. Device use had no visible effect on the tissue mea- 4. Menes TS, Tartter PI, Bleiweiss I, et al. The consequence of mul- tiple re-excisions to obtain clear lumpectomy margins in breast surement sites nor did it affect the ability to correctly cancer patients. Ann Surg Oncol 2005;12:881. examine samples by permanent histopathology. Ulti- 5. Aziz D, Rawlinson E, Narod SA, et al. The role of re-excision for mately the device is intended to be used intraopera- positive margins in optimizing local disease control after breast- tively to detect positive margins on a specimen conserving surgery for cancer. Breast J 2006;12:331. surface. It is common practice by surgeons and pathol- 6. Tartter PI, Kaplan J, Bleiweiss I, et al. Lumpectomy margins, re- ogists to divide the specimen into 6 aspects, and to excision, and local recurrence of breast cancer. Am J Surg 2000; report margin status for each aspect, subsequently re- 179:81. excising the corresponding surfaces in the cavity during 7. Anscher MS, Jones P, Prosnitz LR, et al. Local failure and mar- a second surgery. The device’s single measurement site gin status in early-stage breast carcinoma treated with conser- performance, presented here, together with a method vation surgery and radiation therapy. Ann Surg 1993;218:22. for sampling an entire face and a graphical user inter- 8. Allweiss TM, Kaufman Z, Lelcuk S, et al. A prospective, random- face that groups the measurements of each specimen ized, controlled, multicenter study of a real-time, intraoperative probe for positive margin detection in breast-conserving surgery. aspect could enable clinical utilization of the device by Am J Surg 2008;196:483. enabling surgeons to accurately address these positive 9. Gabriel S, Lau RW, Gabriel C. The dielectric properties of biolog- margins during the first surgery, thus avoiding a second ical tissue II: Measurements in the frequency range 10 Hz to surgery. A recently published study of the device 20 GHz. Phys Med Biol 1996;41:2251. showed reduction of re-excision rates by more than 10. Foster KR, Schepps JL. Dielectric properties of tumor and nor- 50% by using the device for detection of positive mar- mal tissues at radio through microwave frequencies. J Microw gins during the initial excision. Further studies of the Power 1981;16:107. clinical utility of the device are underway. 11. Morimoto T, Kimura S, Konishi Y, et al. A study of the electrical bio-impedance of tumors. J Inv Surg 1993;6:25. REFERENCES 12. Joines WT, Zhang Y, Li C, et al. The measured electrical proper- 1. Fisher B, Anderson S, Bryant J, et al. Twenty-year follow-up of ties of normal and malignant human tissues from 50 to 900 MHz. a randomized trial comparing total mastectomy, lumpectomy, Med Phys 1994;21:547. Ann Surg Oncol DOI 10.1245/s10434-014-3602-0 ORIGINAL ARTICLE – BREAST ONCOLOGY A Randomized Prospective Study of Lumpectomy Margin Assessment with Use of MarginProbe in Patients with Nonpalpable Breast Malignancies Freya Schnabel, MD1, Susan K. Boolbol, MD2, Mark Gittleman, MD3, Tami Karni, MD4, Lorraine Tafra, MD5, Sheldon Feldman, MD6, Alice Police, MD7, Neil B. Friedman, MD8, Scott Karlan, MD9, Dennis Holmes, MD10, Shawna C. Willey, MD11, Moshe Carmon, MD12, Kristen Fernandez, MD13, Stephanie Akbari, MD14, Jay Harness, MD15, Lisa Guerra, MD16, Thomas Frazier, MD17, Karen Lane, MD18, Rache M. Simmons, MD19, Alison Estabrook, MD20, and Tanir Allweis, MD21 1 Department of Surgery, NYU Langone Medical Center, NYU Clinical Cancer Center, New York, NY; 2Department of Surgery, Beth Israel Medical Center, New York, NY; 3Department of Surgery, Breast Care Specialists, Allentown, PA; 4 Department of Surgery, Assaf Harofeh Medical Center, Zerifin, Israel; 5Department of Surgery, Anne Arundel Medical Center, Annapolis, MD; 6Department of Surgery, Columbia University Medical Center, New York, NY; 7Department of Surgery, Pacific Breast Care, Costa Mesa, CA; 8Department of Surgery, Mercy Medical Center, Baltimore, MD; 9 Department of Surgery, Cedars Sinai Medical Center, West Hollywood, CA; 10Department of Surgery, University of South California, Los Angeles, CA; 11Department of Surgery, Georgetown University Hospital, Washington, DC; 12 Department of Surgery, Shaare Zedek Medical Center, Jerusalem, Israel; 13Department of Surgery, Franklin Square Hospital Center, Baltimore, MD; 14Department of Surgery, Virginia Hospital Center, Arlington, VA; 15Department of Surgery, St. Joseph Hospital, Orange, CA; 16Department of Surgery, HOAG Hospital, Newport Beach, CA; 17Department of Surgery, Bryn Mawr Hospital, Bryn Mawr, PA; 18Department of Surgery, UC Irvine Medical Center, Orange, CA; 19 Department of Surgery, Weill Medical College of Cornell University, New York, NY; 20Department of Surgery, St. Luke’s Roosevelt, New York, NY; 21Department of Surgery, Hadassah Medical Organization, Jerusalem, Israel ABSTRACT device or control arms. In the device arm, MarginProbe Background. The presence of tumor cells at the margins of was used to examine the main lumpectomy specimens and breast lumpectomy specimens is associated with an increased direct additional excision of positive margins. Intraopera- risk of ipsilateral tumor recurrence. Twenty to 30 % of tive imaging was used in both arms; no intraoperative patients undergoing breast-conserving surgery require second pathology assessment was permitted. procedures to achieve negative margins. This study evaluated Results. In total, 596 patients were enrolled. False-nega- the adjunctive use of the MarginProbe device (Dune Medical tive rates were 24.8 and 66.1 % and false-positive rates Devices Ltd, Caesarea, Israel) in providing real-time intra- were 53.6 and 16.6 % in the device and control arms, operative assessment of lumpectomy margins. respectively. All positive margins on positive main speci- Methods. This multicenter randomized trial enrolled mens were resected in 62 % (101 of 163) of cases in the patients with nonpalpable breast malignancies. The study device arm, versus 22 % (33 of 147) in the control arm evaluated MarginProbe use in addition to standard intra- (p \ 0.001). A total of 19.8 % (59 of 298) of patients in the operative methods for margin assessment. After specimen device arm underwent a reexcision procedure compared removal and inspection, patients were randomized to with 25.8 % (77 of 298) in the control arm (6 % absolute, 23 % relative reduction). The difference in tissue volume removed was not significant. ! The Author(s) 2014. This article is published with open access Conclusions. Adjunctive use of the MarginProbe device at Springerlink.com during breast-conserving surgery improved surgeons’ First Received: 4 October 2013 ability to identify and resect positive lumpectomy margins F. Schnabel, MD in the absence of intraoperative pathology assessment, e-mail: [email protected] reducing the number of patients requiring reexcision.  F. Schnabel et al. MarginProbe may aid performance of breast-conserving METHODS surgery by reducing the burden of reexcision procedures for patients and the health care system. This study was a prospective, randomized (1:1), double- arm, controlled trial involving 21 institutions and 53 sur- geons. Participating centers represented a variety of Breast-conserving surgery (BCS) has been an estab- practice settings, including academic, community-based, lished approach to surgery for early-stage breast cancer for and private practice sites. Institutional review board more than 30 years.1 Contemporary series report that 60– approval was obtained at each site. Inclusion criteria 75 % of American women with early-stage breast cancer included patients over 18 years with nonpalpable intra- are treated with BCS.2 ductal and invasive breast cancers. All patients had opted BCS for noninvasive and invasive cancer includes a for BCS. Patients with multicentric or bilateral disease, lumpectomy procedure, with sentinel node biopsy in cases those with prior radiotherapy or neoadjuvant chemother- of invasive disease, and postoperative radiotherapy in most apy, and those with a history of surgery in the ipsilateral cases. A successful lumpectomy requires complete removal breast were excluded, as were patients who were pregnant of the malignancy, including a margin of surrounding nor- or lactating. mal breast tissue. This can be challenging to accomplish Informed consent was obtained from all patients. because the microscopic extent of breast cancer can be Patients underwent preoperative localization of their difficult for the surgeon to discern. Multiple studies have lesions and removal of main lumpectomy specimens as per demonstrated the association of involved or positive lump- surgeons’ usual practices. All main lumpectomy specimens ectomy margins with an increased risk of ipsilateral breast were oriented to delineate the six surfaces of the tissue tumor recurrence, even in the presence of radiotherapy.3–6 (superior, inferior, medial, lateral, anterior, and posterior). Although there is no universally accepted definition of After main lumpectomy specimen removal, surgeons used negative surgical margins, at least 20 % of patients undergo their usual methods of intraoperative assessment, including more than one procedure to achieve acceptable margins as inspection and palpation. Intraoperative pathology assess- part of breast-conserving strategies.2,7,8 ment was precluded. If a margin was deemed to be positive The MarginProbe (Dune Medical Devices Ltd, Caesarea, or close, additional tissue was excised. Patients were then Israel) was developed to provide surgeons with real-time randomized to device or control arms (Fig. 1). In the intraoperative assessment of lumpectomy margins. control arm, surgeons completed the lumpectomies, Designed to be used as an adjunct to current surgical including utilizing information from intraoperative imag- methods, the device measures the local electrical properties ing, per their routine. In the device arm, the MarginProbe (in the radiofrequency range) of breast tissue. These prop- was additionally used by the surgeon to examine all six erties are dependent on membrane potential, nuclear surfaces of the main lumpectomy specimens, with 5–8 morphology, and cellular connectivity and vascularity that measurements per face. A single positive reading identified differ between normal and malignant tissue.9 The device’s a margin as positive. Device output was recorded. Surgeons sensing diameter is 7 mm, and it provides a positive/nega- were required to excise additional tissue from the corre- tive reading for each measurement taken. The threshold for a sponding surface of the lumpectomy cavity from every positive reading was set based on readings directly com- device-identified positive margin. Additional tissue pared to pathology results.10 The diagnostic performance removed from the lumpectomy margins was not examined was sensitivity 70–100 % and specificity 70–87 %, by the device, nor was the lumpectomy cavity. Because the depending on the cancer feature size. The performance was device should be used within 20 min after specimen exci- similar for all histology types, including ductal carcinoma- sion, device arm intraoperative imaging, with additional in situ. In a multicenter trial where patients were randomized excisions if indicated, was performed after device use. In to usual surgical technique versus usual technique with both study arms, main lumpectomy specimens were inked. adjunctive use of the MarginProbe, the rate of reexcision All specimens were evaluated by pathologists who were surgery was reduced by 56 % in the device arm of the trial.11 blinded to study arm. Tissue dimensions, margin status, There was no difference in cosmetic outcomes. and margin distance for all surfaces were recorded. Spec- The current study examined the contribution of imen volume was calculated based on the Ellipsoid adjunctive use of MarginProbe to identification of all formula: (p/6) 9 L 9 W 9 D. Subjects were followed involved lumpectomy margins, reduction in the number of (including additional surgical procedures) until the com- patients with positive margins at the completion of pri- pletion of surgical treatment. Data were collected until the mary lumpectomy surgery, and decrease in the necessity earliest of the following events: 2 months after the for repeat surgical procedures to achieve acceptable patient’s last operation; conversion of the subject to mas- margins. tectomy; or initiation of chemotherapy. There were no Lumpectomy Margin Assessment with the MarginProbe Enrollment Device use, excisions as directed Remove Palpation/ Specimen Excisions, Random- Excisions, main visual Imaging orientation if indicated ization if indicated specimen inspection Main specimen removal Intraoperative assessment, & orientation additional excisions if indicated FIG. 1 Lumpectomy procedure restrictions placed on surgeons in terms of the performance TABLE 1 Patient demographics and baseline characteristics of additional surgical procedures. For the purposes of this Characteristic Device Control study, a positive margin was considered to be disease (n = 298) (n = 298) identified at B1 mm from the inked edge of tissue. Diag- nostic measures, including false-negative and false-positive Age, years, mean (SD) 60.3 (11.4) 60.2 (11.1) rates, were evaluated by comparison of device readings to Ethnic origin, n (%) pathology gold standard on a margin-by-margin basis. Whitea 250 (83.9) 260 (87.2) All statistical analyses were performed by SAS software Black 22 (7.4) 17 (5.7) (SAS, Cary, NC, USA). Numerical variables were tabu- Asian 12 (4.0) 10 (3.4) lated using mean and standard deviations. Categorical Other 14 (4.7) 11 (3.7) variables were tabulated using number of observations and Body mass index, kg/m2, mean (SD) 27.9 (6.6) 28.6 (6.6) percentages. Statistics were performed at a = 0.05 two- Diagnosis, n (%) sided significance level. Rates between arms were com- Invasive ductal 181 (60.7) 202 (67.8) pared by Fisher’s exact test. Reexcisions were compared by Invasive lobular 26 (8.7) 13 (4.4) Poisson’s regression. No missing data were imputed. Mixed invasive 8 (2.7) 5 (1.7) Safety was evaluated by reports of serious adverse Ductal carcinoma-in situ 83 (27.9) 78 (26.2) events and adverse events. Safety reports were tabulated by Receptor status, n (%) group, body system, and relation to treatment. ER positive 251 (84.2) 258 (86.6) PR positive 223 (74.8) 217 (72.8) RESULTS Preoperative imaging, n (%) Mammogram 296 (99.3) 294 (98.7) A total of 596 patients were randomized, with 298 in MRI 184 (61.7) 174 (58.4) each arm of the trial. Patient demographics and baseline Ultrasound 228 (76.5) 289 (97.0) characteristics are listed in Table 1. Patients underwent Preoperative core biopsy, n (%) 287 (96.3) 289 (97.0) extensive imaging before surgery. The mean extent of Mean extent of disease, cm 1.7 1.6 disease was similar in the two groups. The main specimen Main lumpectomy specimen volume, ml 61 60 volume was similar in both groups, reflecting no difference SD standard deviation, ER estrogen receptor, PR progesterone in surgical procedure before randomization. receptor, MRI magnetic resonance imaging The disposition of patients in both arms of the trial is a Including Hispanics shown in Fig. 2. In similar proportions of patients in both arms, the main lumpectomy specimen contained at least positive main specimen margins, surgeons correctly iden- one positive margin (Fig. 2, phase I). In patients with tified all positive margins on the main specimen and positive margins on initial lumpectomy specimens, an removed additional tissue from those involved margins average of two margins was involved, with no difference (Fig. 2, phase II) in 33 of 147 cases (22 %) in the control between the two arms. With reference to the patients with arm, versus 101 of 163 (62 %) cases in the device arm