Breast-Conserving Surgery After Neoadjuvant Chemotherapy Guided by a Supine Magnetic Resonance Imaging Projection Mapping System Visualizing Original Cancer Extent

Abstract

It is ideal to remove areas where breast cancer existed before neoadjuvant chemotherapy (NAC) in breast-conserving surgery (BCS) after NAC to ensure a pathological response. However, this becomes challenging, particularly when NAC is effective. It is also important to determine the appropriate resection line to maintain breast cosmesis in the absence of residual cancer cells. We aimed to identify magnetic resonance imaging (MRI) findings that enable the removal of the original cancer extent without additional margins outside the cancer boundary in BCS after NAC using an MRI projection mapping system (PMS). We enrolled 36 women with breast cancer in the study. Contrast-enhanced MRI was performed in the prone and supine positions before NAC. Supine unenhanced MRI with skin markers for registration was performed after NAC, one day before surgery. A maximum intensity projection image visualizing both the tumor extent before NAC and skin markers was projected onto the breast using PMS, and BCS was performed. We compared the clinical, pathological, and MRI findings between patients who required additional margins and those who did not require additional margins during BCS using MRI-PMS. Nineteen of the 36 patients did not require additional margins for BCS after NAC using MRI-PMS. These patients had a higher tumor diameter reduction rate and a lower signal ratio between the tumor bed and normal tissue after NAC than those who required additional margins. This study indicates that MRI-PMS allows BCS without additional margins and may facilitate the conservation of normal breast tissue in patients with a greater diameter reduction rate or a lower signal ratio between the tumor bed and normal tissue after NAC.

INTRODUCTION

Patients with invasive breast cancer can receive neoadjuvant chemotherapy (NAC) before surgery to reduce tumor volume and minimize lymph node or occult metastases. NAC can convert the surgical procedure from total mastectomy to breast-conserving surgery (BCS) [1,2,3,4]. Choi et al. [5] demonstrated that “no tumor on ink” (i.e., either a negative or close margin) is associated with good outcomes of patients who undergo BCS after NAC, whereas a meta-analysis suggests an increased risk of local or distant recurrence in patients with close margins [6]. As no direct comparison has been made between excision of the original cancer extent before NAC and that of the residual tumor after NAC, the optimal BCS strategy after NAC remains controversial [7]. Although contrast-enhanced magnetic resonance imaging (MRI) is the most valuable tool for assessing the response of breast cancer to NAC [8], it is challenging to remove areas where breast cancer existed after NAC without guidance, as the tumor may shrink irregularly or leave scattered cancerous cells with reduced angiogenesis [3,9,10]. Therefore, the original tumor extent before NAC should be primarily removed when performing BCS after NAC [3,10,11].

We developed a prototype MRI projection mapping system (PMS) to project a maximum-intensity projection (MIP) image generated from a supine MRI of a patient with breast cancer in the operating room [12]. MRI-PMS is effective for determining the surgical margin of BCS using dye-guided cylindrical resection of breast cancers [12,13,14,15]. BCS without additional margins reduces surgical burden and more effectively preserves breast appearance. We hypothesized that BCS after NAC, without additional margins free of cancerous cells (i.e., a negative margin), could be performed in selected patients if MRI-PMS were used to visualize the original breast cancer extent. This study aimed to identify the MRI findings that enable BCS after NAC without additional margins using MRI-PMS.

METHODS

Patients

Ethical approval was obtained from Institutional Review Board of Nihon University Hospital (approval No. 20220102), and all patients gave informed consent to participate in this prospective study. We enrolled 36 women (mean age 55.2 ± 11.5 years) with invasive breast cancer (35 with invasive ductal carcinoma and 1 with apocrine carcinoma), who underwent BCS after NAC between September 2021 and March 2025. The exclusion criteria were patients who did not complete NAC or who underwent total mastectomy after NAC.

Projection mapping system

The prototype PMS has been described previously [12]. Briefly, the PMS consisted of a projector (350 lm WXGA DLP projector, ASUS P2E; ASUSTeK Computer, Taipei, Taiwan), a color digital camera (CMOS Color: JAI GO-5000C; JAI, Kanagawa, Japan), and a 16-mm FA lens with a fixed aperture of f/8.0 (RICOH FL-BC1618-9; RICOH, Tokyo, Japan). The PMS has a movable arm that tilts the head to project an MIP image perpendicularly onto the patient’s breast. The depth and surface topography of the breasts were analyzed using a structured light method. Subsequently, the PMS projected the MIP image generated from the MRI onto the patient’s breast.

MRI study protocol

MRI study protocol is presented in Supplementary Figure 1. MRI examinations were conducted using a 3.0-T system (Ingenia Achieva; Philips Healthcare, Best, Netherlands). Contrast-enhanced MRI was performed to assess the location and extent of breast cancer before NAC (Figures 1A and 2A). Contrast-enhanced supine MRI was obtained to delineate the cancer contour, with the patient’s position identical to that in the operating room (Figures 1B and 2B). Two NAC regimens, determined by immunohistochemical analysis of the biopsied specimens, were administered in four cycles for each regimen (Table 1). Contrast-enhanced MRI was performed at the end of each regimen to evaluate tumor response to NAC (Figures 1C and 2C). One day before BCS, an unenhanced supine MRI with skin markers (i.e., nifedipine capsules; Bayer Healthcare, Osaka, Japan) was acquired (Figure 1D). The imaging parameters for prone and supine MRI have been described in previous studies [12,13]. The patients’ arms were raised using a homemade armrest so that their position during the two supine MRI examinations matched that in the operating room. The locations of the skin markers were marked with a pen before the markers were removed from the patient’s breasts. Under the supervision of a breast radiologist, a radiologic technologist fused the contrast-enhanced and unenhanced MRI images in the supine position, aligning with the nipple, the edge of the fibroglandular tissue, and a metallic marker placed during needle biopsy. Image fusion was performed using a dedicated workstation (Ziostation; Ziosoftware, Tokyo, Japan) (Figure 1E). Thus, MIP images were generated from the fused supine MRI images showing breast cancer before NAC and skin markers (Figures 1F and 2D). The PMS projected the MIP image onto the patient’s breast in the operating room, where the patient was under general anesthesia and had their arms raised on a homemade armrest, as during the supine MRI. The cancer contour observed by MRI-PMS is shown in red on the breast surface (Figures 1G and 2E). The extent of the cancer and its boundaries before NAC were determined by a radiologist. The surgical margin was determined by a breast surgeon as the tumor boundaries plus additional margins, considering the tumor subtype and therapeutic response to NAC (Figures 1G and 2E). Indocyanine green was injected at equal intervals into the breast tissue immediately above the greater pectoral muscle, starting from the surgical margin. Next, the armrest was removed to position the patient’s arms sideways. Breast cancer was removed cylindrically, guided by dye staining of the surgical margin [13].

Figure 1. A 67-year-old woman with human epidermal receptor 2-positive invasive breast cancer.

MIP image from prone MRI before NAC showing mass and non-mass enhancement. The maximum tumor diameter is 76 mm (A). Supine MRI before NAC showing an enhanced mass (B). MIP image of the prone MRI after NAC shows residual non-mass enhancement (arrowheads). The diameter reduction rate is 21.1% (C). Supine unenhanced MRI with skin markers (green arrowhead) performed one day before surgery (D). Supine MRI images before and after NAC (B, D) are shown (E). MIP of the fused supine MRI showing both the cancer contour (red arrows) and skin markers (green arrowheads) (F). The MIP is projected onto the patient’s breast using a PMS (G). The surgical margin (dashed red line with dashed red arrow) was delineated as breast cancer, as shown by PMS (red line with red arrow) plus additional margins of 10 mm (G). The black circles and arrows represent the residual tumors delineated by ultrasound. Pathology revealed that the surgical margin was free from cancerous cells, but the cancer involved the breast tissue 3 mm from the surgical margin (red arrow) (H). Therefore, if an additional margin had not been used, the surgical margin would have been positive for cancer in this patient. Red dots, invasive carcinoma; black dots, noninvasive carcinoma.

MIP = maximum-intensity projection; MRI = magnetic resonance imaging; NAC = neoadjuvant chemotherapy; PMS = projection mapping system.

Figure 2. A 43-year-old woman with luminal B and human epidermal receptor 2-positive invasive breast cancer.

MIP image from prone MRI before NAC, showing an irregularly shaped mass enhancement in the upper right breast (red arrowhead) (A). The maximum tumor diameter was 38 mm. Supine MRI before NAC showing an enhanced mass (B). The signal defect in the mass indicates a marker placed during ultrasound-guided biopsy (pink arrow). MIP image of the prone MRI after NAC showing concentric shrinkage of the breast cancer (red arrowhead) (C). The diameter reduction rate is 86.1%. MIP of the fused supine MRI before and after NAC showing breast cancer (red arrow) and skin markers (D). Pink arrow: marker in the tumor bed. The surgical margin (dashed red circle) is delineated as the cancer contour shown by the projection mapping system (red line) plus additional margins of 10 mm (E). Pathological examination revealed that the surgical margins were negative, and the shortest distance between the edge of the residual cancer and the surgical margins was 25 mm (F). If no additional margin is obtained, the surgical margin would remain negative for cancerous cells. Red line, invasive carcinoma; black dot, noninvasive carcinoma.

MIP = maximum-intensity projection; MRI = magnetic resonance imaging; NAC = neoadjuvant chemotherapy.

Table 1. Characteristics of patients with invasive breast cancer.

Cancer characteristics		Values (n = 36)
Age (yr)
	Mean ± SD	55.2 ± 11.5
	Range	32–77
BMI (kg/m2)
	Median [interquartile range]	22.4 [4.5]
	Range	17.9–34.8
Histological pathology
	IDC	35 (97.2)
	Apocrine carcinoma	1 (2.8)
T-classification
	T1c	8 (22.2)
	T2	25 (69.4)
	T3	2 (5.6)
	T4b	1 (2.8)
Molecular subtypes of invasive cancers
	Luminal A	2 (5.5)
	Luminal B, HER2−	9 (25.0)
	Luminal B, HER2+	9 (25.0)
	HER2	6 (16.7)
	Triple negative	10 (27.8)
NAC regimen
	EC + T	16 (44.4)
	EC + T + H	15 (41.7)
	EC + T + Pembro + CBDCA	5 (13.9)
Clinical effectiveness of chemotherapy after NAC
	Complete response	14 (38.9)
	Partial response	19 (52.8)
	Stable disease	3 (8.3)

Data shown are number (%) not otherwise specified.

SD = standard deviation; BMI = body mass index; IDC = invasive ductal carcinoma; HER2− = human epidermal growth factor receptor 2-negative; HER2+ = human epidermal growth factor receptor 2-positive; NAC = neoadjuvant chemotherapy; EC = epirubicin and cyclophosphamide; T = taxanes (docetaxel or paclitaxel); H = HER2-targeted agents (trastuzumab and pertuzumab); Pembro = pembrolizumab; CBDCA = carboplatin.

MRI analysis

The radiologist measured the maximum diameter of breast cancer on the MIP image generated from contrast-enhanced supine MRI before and after NAC. The diameter reduction rate was calculated as (1 − The Maximum Diameter of the Breast Cancer after NAC Divided by That Before NAC) × 100 (%). The presence or absence of non-mass enhancement before NAC was also documented. Tumor shrinkage after NAC was classified as either concentric or scattered [14]. The signal intensity ratio between the tumor bed and normal breast tissue was calculated using contrast-enhanced prone MRI after NAC [15].

Pathological analyses

The pathological margins of the surgical specimens were classified into three categories: negative (no cancer cells within 2 mm of the surgical margin), close (cancer cells within 2 mm of the margin but not at the surgical margin), and positive (cancer cells at the margin) [5,6,16]. When a residual lesion was present, the shortest distance from the edge of the residual tumor to the surgical margin was measured (Figures 1H and 2F). Two pathologists performed the analyses by consensus.

Statistical analysis

We classified the 36 patients into two groups: 1) patients whose pathological study revealed ypT0, or those for whom the shortest distance from the edge of residual cancer to the surgical margin minus the additional margins was ≥ 2 mm, indicating they would not have required any additional margins in BCS after NAC using MRI-PMS (Figure 2); and 2) patients with a close or positive margin, or those for whom the shortest distance noted above minus the additional margins was < 2 mm (Figure 1). The pathologically shortest distance minus the additional margins was the shortest distance from the edge of the residual cancer to the tumor contour, as shown by MRI-PMS.

We compared patient age, body mass index, body weight change during NAC, immunohistochemical subtype, and Ki-67 levels in the biopsied specimens between the two patient groups. We also compared the MRI findings between the two groups. Student’s t-test, Mann–Whitney U test, or Fisher’s exact test were used as appropriate, and p < 0.05 was considered significant.

RESULTS

The responses to NAC assessed by MRI were: complete response in 14 (38.9%), partial response in 19 (52.8%), and stable disease in 3 (8.3%) patients (Table 1). MRI-PMS successfully projected MIP images, visualizing both the extent of cancer before NAC and the skin markers on the breast in all patients. The additional margins were 0 mm (n = 4), 5 mm (n = 12), 10 mm (n = 12), and 15 mm (n = 8). Histopathological analysis of the surgical specimens revealed negative surgical margins in all patients. Among the 36 patients, 13 achieved ypT0 disease, 23 had residual disease (ypTis, 3; ypT1, 15; ypT2, 5), 21 had negative margins with residual tumors, and 2 patients had close margins. The shortest distance from the edge of the residual cancer to the surgical margin ranged from 0.3 to 22 mm (mean ± standard deviation, 8.2 ± 5.6 mm) among the 23 patients. In 6 of the 23 patients, the shortest distance from the edge of the residual cancer to the surgical margin, minus the additional margins, was ≥ 2 mm. Therefore, 19 of the 36 patients (52.8%), including the 6 patients noted above and 13 with ypT0 disease, did not require any additional margins in BCS following NAC when using MRI-PMS (Supplementary Figure 2).

Table 2 summarizes the clinical and MRI variables along with their association with surgical margin status. A significantly higher diameter reduction rate was observed in the 19 patients requiring no additional margins compared to the 17 patients requiring additional margins (median, 100%; interquartile range [IQR], 15.4% vs. median, 30.3%; IQR, 29.7%; p < 0.001). A significantly lower signal intensity ratio was detected between the tumor bed and normal breast tissue after NAC (median, 1.1; IQR, 0.8 vs. median, 2.5; IQR, 0.9; p < 0.001).

Table 2. Summary of clinical and MRI variables and association with surgical margin status.

Characteristics			Total (n = 36)	Surgical margin without additional margins		p-value
				Negative* (n = 19)	Close† or positive‡ (n = 17)
Patient’s baseline characteristics
	Age (yr)					0.523
		Mean ± SD	55.2 ± 11.5	53.5 ± 11.7	57.1 ± 11.3
		Range	32–77	32–76	33–77
	BMI (kg/m²)					0.057
		Median [IQR]	22.4 [4.5]	22.2 [4.1]	23.7 [7.4]
		Range	17.9–34.8	17.9–32.9	19.2–34.8
	Body weight change during NAC (kg)					0.754
		Mean ± SD or median [IQR]	1.6 ± 3.6	0.9 [2.7]	1.7 [6.4]
		Range	−5.5 to 10	−3.9 to 10	−5.5 to 9.1
Pathological findings on biopsy before NAC
	Immunohistochemical subtype					0.070
		Luminal A, B	11 (30.6)	3	8
		Luminal/HER2, HER2, triple negative	25 (69.4)	16	9
	Ki-67 (%)					0.138
		Median [IQR]	50 [47.7]	60 [50]	40 [50]
		Range	10–95	25–90	10–95
Prone MRI findings
	Maximum tumor diameter before NAC (mm)					0.778
		Median [IQR]	31.5 [26.0]	31 [29]	32 [22]
		Range	13–76	13–76	18–72
	Tumor enhancement pattern before NAC					0.611
		Mass	21 (58.3)	11	10
		Non-mass enhancement	15 (41.7)	8	7
	Diameter reduction rate after NAC (%)					< 0.001§
		Median [IQR]	69.8 [71.8]	100 [15.4]	30.3 [39.7]
		Range	0–100	27.5–100	0–100
	Shrinkage pattern					0.092
		Concentric	26 (72.2)	16	10
		Scattered	10 (27.8)	3	7
	Signal intensity ratio between tumor bed and BPE on MRI after NAC					< 0.001§
		Median [IQR]	1.9 [1.4]	1.1 [0.8]	2.5 [0.9]
		Range	0.8–3.9	0.8–2.6	0.9–3.9

Data shown are number (%) not otherwise specified.

SD = standard deviation; BMI = body mass index; IQR = interquartile range; NAC = neoadjuvant chemotherapy; HER2 = human epidermal growth factor receptor 2; MRI = magnetic resonance imaging; BPE = background parenchymal enhancement.

^*Negative: ≥ 2mm; ^†Close: < 2mm; ^‡Positive: tumor on ink; ^§Statistically significant.

Representative cases from the two patient groups are shown in Figures 1 and 2.

DISCUSSION

The surgical margins of patients with breast cancer after NAC, which were free of cancerous cells in all cases, were delineated using MRI-PMS. Our study suggests that BCS without additional margins can be achieved in 19 of the 36 patients (52.8%), who showed a greater diameter reduction rate or a lower signal intensity ratio between the tumor bed and normal breast tissue after NAC, as assessed by MRI.

MRI-PMS projected MIP images directly onto a patient’s breast, unlike methods using transparent sheets or 3D printing [3,10,11,17]. There was concern that changes in body weight during NAC might induce alterations in breast size and shape, whereas MIP generation was likely unaffected, possibly due to the relatively small changes (i.e., 1.6 ± 3.6 kg). In addition, we carefully referenced the aforementioned anatomical landmarks to project the MIP image perpendicular to the chest wall and minimize registration errors.

This study had some limitations. First, the threshold for the diameter reduction rate or signal ratio was not determined due to the small sample size. Second, two supine MRI examinations were necessary to determine the surgical margin of BCS after NAC, as we needed to visualize both the original cancer extent before NAC and the skin markers placed on the patient’s breast one day before BCS. If we had generated contrast-enhanced supine MRI images from prone imaging before NAC using artificial intelligence techniques, we could have simplified the study protocol. Instead, we did not use gadolinium in the second supine MRI to minimize the amount injected or to reduce gadolinium contamination [18]. Third, the additional margins were determined based on the surgeon’s experience in this study. The residual tumor size on MRI is not always identical to that observed on pathology due to non-mass enhancement, scattered shrinkage patterns, or cancers with less angiogenesis after NAC [9,19]. Nonetheless, this study indicated that MRI helps identify patients who do not require additional margins during BCS when using MRI-PMS. Finally, we reduced the excision area in patients who responded well to NAC. However, ten of the 36 patients in the present study showed scattered shrinkage patterns, suggesting the presence of scattered residual cancerous cells. MRI-PMS may be useful for delineating the surgical margins of BCS after NAC in cases of marked reduction in breast cancer, particularly with the use of gadolinium in the second supine MRI.

In conclusion, MRI-PMS is feasible for delineating surgical margins in patients undergoing BCS after NAC, as all our patients had negative surgical margins. This proof-of-concept study indicates that MRI-PMS can enable BCS without additional margins in patients with a greater diameter reduction rate or lower signal intensity ratio between the tumor bed and normal breast tissue after NAC, as assessed by MRI.

SUPPLEMENTARY MATERIALS

Supplementary Figure 1

Study protocol.

Supplementary Figure 2

Pathological assessment of the surgical specimen.