DOI: https://doi.org/10.21203/rs.3.rs-2003609/v1
Background: Contrast-enhanced dual-energy computed tomography produces iodine maps (i-maps) based on tissue iodine concentration (IC). We analyzed the features of i-maps in operable breast cancer.
Methods: I-maps made from patients with operable breast cancer were retrospectively reviewed. The mean IC of the whole tumor and visual patterns (sharp/obscure) were analyzed with respect to pathological features. The tumor extent was retrospectively verified with dynamic contrast-enhanced magnetic resonance of mammary gland
(MRM) and pathological specimens.
Results: The median IC of 858 cases was 4.3 (interquartile range [IQR]: 4.0–5.1) mg/mL. The IC of the luminal A-like subtype of invasive breast cancer was significantly higher than that of the human epidermal growth factor 2 (HER2) and triple-negative (TN) subtypes (luminal A-like: 4.5 [IQR: 4.3–5.5] mg/mL vs. HER2: 3.9 [IQR: 3.5–4.4] mg/mL and TN: 3.8 [IQR: 3.6–4.2] mg/mL; both p < 0.05). The IC was significantly correlated with the histological grade and Ki-67 labeling index. Sharp visual patterns correlated with the estrogen receptor and Ki-67 labeling index, while obscure patterns correlated with the HER2 subtype. I-maps underestimated tumor extent in 84 (9.8%) of the 532 partial resection cases, especially in lobular carcinoma and mucinous carcinoma.
Conclusions: The IC and visual patterns correlated with the pathological features of operable breast cancer. Most breast cancers are identifiable on i-maps; however, the adjunction of MRM is preferred for tumors with low IC on i-maps to evaluate tumor extent.
Breast cancer is the most prevalent malignancy in women. In 2021, approximately 2.2 million women were diagnosed with breast cancer, of whom almost 700,000 died from the disease [1]. Breast cancer is a widely heterogeneous illness, and its histological type and tumor grade correlate with malignancy and prognosis [2]. Biomarkers expressed in breast cancer, such as the estrogen receptor (ER), progesterone receptor (PgR), human epidermal growth factor 2 (HER2), and Ki-67 labeling index, classify different subtypes of cancer and guide treatment strategies and prognosis [2]. Response to treatment and prognosis depend on the subtype of breast cancer [2–4]. Mammography and ultrasound are the most common examination methods for breast cancer [5–8], and screening using these approaches has contributed to the early detection and timely treatment of breast cancer, which are associated with a better prognosis [8–10].
Computed tomography (CT) is frequently used to detect metastasis in the axillary lymph nodes or distant organs in patients with breast cancer; although it is helpful for screening whole tumors at the time of diagnosis or during the treatment period, conventional CT is not suitable for precisely characterizing breast tumors because of its relatively low resolution of soft tissue, limited imaging mode, and limited qualitative parameters.
Since the development of dual-energy CT (DECT) in recent years, this technology has attracted widespread interest as a method used in the context of multiple diseases [11–14]. DECT technology is based on the attenuation of materials when they are subjected to two different levels of energy, usually high (140–150 kVp) and low energy (80–100 kVp); using these two datasets of separate energy levels for each part of the body, DECT can obtain virtual monochromatic X-ray images and substance discrimination (density) images. This produces a better image quality by improving contrast, reducing artifacts, and enabling highly accurate substance discrimination.
Contrast-enhanced DECT (ceDECT) using an iodine-based agent allows the visualization of various body parts on an iodine map (i-map), while a dedicated application can calculate iodine concentration (IC; mg/mL) at the region of interest [15]. Moreover, ceDECT has proved useful for diagnosing and staging several types of cancers because i-maps can clearly show the iodine distribution of the tumor itself, excluding the contrast of normal tissue [16–20]. Recent studies have also demonstrated the utility of ceDECT for detecting breast cancer [21–23]. However, discussion is limited about the features of various types of breast cancer visualized on i-maps derived from ceDECT. We performed ceDECT for the preoperative evaluation of breast cancer. This study aimed to analyze the imaging characteristics of i-maps derived from ceDECT for various types of operable breast cancer, mainly from the breast surgeon’s perspective.
The ceDECT images of female patients who were diagnosed with operable breast cancer and treated at Hiroshima City North Medical Center Asa Hospital from 2012 to 2021 were retrospectively reviewed. Cases of severe, advanced, unresectable metastatic breast cancer, non-primary breast cancer, and male breast cancer were excluded. Patients were diagnosed with operable breast cancer using mammography, ultrasound, and pathological diagnosis by core needle biopsy. Informed consent was obtained from the patients who participated in the examinations, surgeries, and treatments. Sentinel lymph node biopsy was performed in node-negative patients, and axial dissection was performed in node-positive patients. Breast reconstruction using a tissue expander was performed in eligible patients who requested it.
ceDECT was performed using the SOMATOM Definition Flash (Siemens Healthineers, Forchheim, Germany). Iopamidol (300 and 370 mgI/mL; Bayer, Tokyo, Japan) or iohexol (300 mgI/mL; Daiichi Sankyo, Tokyo, Japan), which are non-ionic iodine contrast media, were administered at a dose of 540 mg/kg of bodyweight at a rate of 2.0 mL/s over 50 s using an automatic contrast medium injector (Dual Shot GX, Nemoto, Tokyo,Japan). Imaging was initiated 50 s after commencing contrast media administration. The other settings were as follows: tube voltage, 100/140 kV; tube current CT automatic exposure control and quality reference mAs, 250/193; circulating time, 0.5 s; detector, 128 slice × 0.6 mm; pitch factor, 0.8; scan length, 450 mm; reconstruction function, D30f; reconstruction slice thickness, 5 mm; and scanning time, 16 s. ceDECT data were analyzed using viewer software on a syngo.via workstation (Siemens Healthineers, Forchheim, Germany). Standard linear-blended images were reconstructed by applying a blending factor of 0.5 (M_0.5; 50% of the low-kV spectrum and 50% of the high-kV spectrum).
A 2-mm-thick i-map of each case was created from liver virtual non contrast imaging. The region of interest of the i-map was placed on the tumor, and the IC was calculated [15]. The IC of the entire tumor was expressed as the median and interquartile range (IQR). The tumor extent was measured, including intraductal spread, using a contrasted lesion on the i-map.
To the best of our knowledge, the visual patterns of tumors on i-maps have not yet been investigated in the literature; therefore, the visual patterns of breast cancer on i-maps were classified as follows: tumors in which > 50% of the whole tumor shape was visible were labeled as having a “sharp pattern,” while tumors in which < 50% of the whole tumor shape was visible and that appeared heterogeneous and obscure were labeled as having an “obscure pattern.” Unclear cases were designated as “not defined.” The patterns were determined by at least one breast surgeon and one radiologist who specialized in diagnosing and treating breast cancer; if the judgments did not match, a majority vote was taken with other radiologists to reach a consensus.
Dynamic magnetic resonance imaging of mammary gland (MRM) was performed on patients undergoing breast-conserving surgery using Signa HDxt 3.0T (GE Healthcare, Chicago, IL, USA). Images were acquired in the axial plane with the following sequences: axial, T2-weighted, fat-suppressed, fast spin-echo imaging and pre- and post-contrast, axial, T1-weighted, 3-dimensional fast spoiled gradient-recalled echo sequence with parallel volume imaging (VIBRANT, GE Healthcare, Oslo, Norway). The tumor extent on MRM was measured mainly via the visualization of a contrasted lesion in the early-phase VIBRANT images. The contrast media gadobutrol (0.1 mL/kg) or gadodiamide/gadopentetic acid (0.2 mL/kg) was administered, followed by an injection of saline (30 mL). The contrast determination time was 90 s/300 s [24].
Clinical data, including age and menopausal status, were obtained from the medical records. Body mass index was calculated using the height and weight of each patient (kg/m2). The breast imaging-reporting and data system (BI-RADS) breast density category was also evaluated by mammography [25]. The postsurgical specimen’s final pathologic diagnosis was considered the reference standard. Histopathologic diagnoses complied with the World Health Organization classification [26], including tumor size, pathologic T and N categories, histologic type, tumor grade, ER status, PgR status, HER2 status, and Ki-67 index, which were obtained from surgical pathology reports. Biomarker subtypes of invasive breast carcinoma were classified as luminal A (ER + HER2 - Ki-67 < 20%), luminal B (ER + HER2 - Ki-67 > 20%), HER2 (ER ± HER2 + Ki-67 any), and triple-negative (TN) (ER - HER2 - Ki-67 any), according to previous guidelines [27]. The positive margin of the resected specimen was determined to be < 5 mm from the invasive or non-invasive component [28, 29].
The determination of tumor extent with the i-map was validated using MRM and pathological examination of the surgical specimen of patients who underwent partial resection. These cases were classified as follows: cases were defined as “proper estimation” when the tumor extent determined by i-map was consistent with that by MRM and surgical specimens with safe margins; cases were defined as “underestimation” when the tumor extent determined by i-map was narrow compared to that by MRM and surgical specimens; and cases were defined as “overestimation” when the tumor extent determined by i-map was wide compared to that by MRM and surgical specimens. This judgment was made by at least one breast surgeon, one radiologist, and one pathologist who specialized in the diagnosis and treatment of breast cancer. In cases where there was no consensus, the majority vote was considered.
Statistical analysis was performed using EZR, version 1.54 [30]. The analysis of variance was used to compare different clinicopathological factors between IC groups. The Wilcoxon rank-sum test was used to compare the IC of each group. Spearman’s correlation test was used to evaluate the correlation between visual patterns and pathological factors. The significance was set at p < 0.05.
Patient characteristics are described in Table 1. A total of 858 invasive breast cancer images were assessed. Various representative i-maps of benign breast tumors and cancers are shown in Fig. 1, along with the IC and visual pattern (sharp/obscure/not defined) (Fig. 1a and b). Small ductal carcinomas in situ (DCIS) and infiltrating ductal carcinomas (IDC) with a diameter of < 5 mm could frequently be detected by their sharp pattern, but some were undetectable (Fig. 1c).
| Characteristics | Number (%) |
|---|---|
| Age, years (range) | 57 (33–86) |
| Menstruation Premenopause Postmenopause | 309 (36.0) 549 (64.0) |
| BMI < 20 20–25.0 25.1–30.0 ≥ 30 | 154 (18.1) 360 (41.9) 310 (36.0) 34 (4.0) |
| BI-RADS breast density of mammography | |
| Extremely dense | 64 (7.4) |
| Heterogeneously dense | 628 (73.1) |
| Fatty | 162 (18.9) |
| Unknown | 4 (< 0.1) |
| Operation | |
| Breast Partial resection Mastectomy | |
| 532 (62.3) | |
| 326 (37.7) | |
| Axilla Sentinel lymph node biopsy Axial lymph node dissection | |
| 565 (65.8) | |
| 293 (34.2) | |
| T status (%) Tis T1 T2 T3 T4 | |
| 73 (8.5) | |
| 414 (48.2) | |
| 201 (23.4) | |
| 100 (11.7) | |
| 70 (8.1) | |
| Histology | |
| Non-invasive carcinoma DCIS LCIS | 73 66 7 |
| Invasive carcinoma | 785 |
| Infiltrating duct carcinoma Lobular carcinoma Mucinous carcinoma Tubular carcinoma Carcinoma with apocrine differentiation Metaplastic carcinoma Cribriform carcinoma Invasive micropapillary carcinoma Glycogen-rich carcinoma Papillary carcinoma with invasion Secretory carcinoma Neuroendocrine carcinoma | 684 65 15 8 5 3 3 2 5 5 1 3 |
| Nodal metastasis | |
| Negative | 486 (61.5) |
| Positive | 298 (38.5) |
| Histological grade | |
| 1 | 220 (28.1) |
| 2 | 306 (39.1) |
| 3 | 259 (32.8) |
| LVI | |
| Negative | 416 (53.0) |
| Positive | 369 (47.0) |
| Subtype | |
| Luminal A-like | 231 (26.9) |
| Luminal B-like | 292 (34.0) |
| HER2+ | 237 (27.6) |
| Triple-negative | 98 (11.6) |
| Ki-67 labeling index | |
| < 24% | 354 (45.2) |
| ≥ 24% | 431 (54.8) |
| BI-RAD, breast imaging-reporting and data system; BMI, body mass index; DCIS, ductal carcinoma in situ; HER2, human epidermal growth factor receptor 2; LCIS, lobular carcinoma in situ; LVI, lymphovascular invasion. | |
The IC of each whole tumor was reviewed. Breast cancers had a higher IC than benign breast tumors (4.3 [IQR: 4.0–5.1] mg/dL vs. 2.2 [IQR: 2.3–3.1] mg/dL; p = 0.02). The IC of normal breast tissue was 1.8 (IQR: 1.5–2.2) mg/mL, which was consistent among patients with various physical factors and tumors (Fig. 2a). There were 491 sharp-pattern tumors (57.2%) and 302 obscure-pattern tumors (35.1%) detected, while 65 tumors were not defined (7.5%). Sharp-pattern tumors showed a higher IC than obscure-pattern tumors (4.7 [IQR: 4.4–5.7] mg/dL vs. 3.8 [IQR: 3.5–4.7] mg/dL; p = 0.24) (Fig. 2b). Among non-invasive carcinomas, DCIS showed a significantly higher IC than lobular carcinoma (LC) in situ (4.2 [IQR: 4.4–5.7] mg/dL vs. 3.2 [IQR: 3.0–3.5] mg/dL; p = 0.03) (Fig. 2c). Among the subtypes of invasive carcinoma, IDC and carcinoma with apocrine differentiation showed high ICs. Invasive LC and mucinous carcinoma (MC) showed low ICs compared to other histological subtypes, including tubular carcinoma, metaplastic carcinoma, and invasive micropapillary carcinoma, among other types of cancer (Fig. 2d). The luminal A-like subtype (4.5 [IQR: 4.3–5.5] mg/dL) showed a significantly higher IC than the HER2 (3.9 [IQR: 3.5–4.4] mg/dL; p = 0.03) and TN subtypes (3.8 [IQR: 3.6–4.2] mg/dL; p = 0.02) (Fig. 2e).
The correlations between IC and clinicopathological features of various breast cancer subtypes were examined. The histological grade and Ki-67 index were significantly correlated with IC (Table 2). In terms of visual features, sharp patterns positively correlated with ER + and Ki-67 index, and obscure patterns positively correlated with the HER2 value (Table 3).
| Parameter | IC of tumor, mean (IQR) | p |
|---|---|---|
| Hormone status | ||
| Premenopausal | 4.3 (3.9–5.5) | 0.913 |
| Postmenopausal | 4.4 (4.1–5.6) | |
| BMI, kg/m2 | ||
| < 25 | 4.5 (4.0–5.6) | 0.256 |
| ≥ 25 | 4.2 (4.1–5.4) | |
| MMG density | ||
| Extremely/heterogeneous | 4.2 (4.0–5.5) | 0.652 |
| Fatty | 4.4 (4.1–5.6) | |
| T status | ||
| T1 | 4.4 (4.0–5.6) | 0.401 |
| T2–4 | 4.3 (4.2–5.3) | |
| Nodal metastasis | ||
| Positive | 4.4 (4.0–5.5) | 0.445 |
| Negative | 4.2 (3.9–5.4) | |
| Histological grade | ||
| 1–2 | 4.5 (4.3–5.8) | 0.015 |
| 3 | 4.0 (3.8–5.0) | |
| LVI | ||
| Negative | 4.4 (4.0–5.6) | 0.227 |
| Positive | 4.2 (3.9–5.2) | |
| ER | ||
| Negative | 4.4 (3.9–5.2) | 0.115 |
| Positive | 4.6 (4.0–5.9) | |
| HER2 | ||
| Negative | 4.3 (3.9–5.5) | 0.456 |
| Positive | 4.4 (3.8–5.4) | |
| Ki-67 labeling index | ||
| < 20% | 4.6(4.3–5.8) | 0.012 |
| ≥ 20% | 4.1 (3.8–5.1) | |
| BMI, body mass index; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; IC, iodine concentration; IQR, interquartile range; LVI, lymphovascular invasion; MMG, mammography. | ||
| Parameter | R value | p |
|---|---|---|
| Sharp pattern and ER value | 0.123 | 0.003 |
| Sharp pattern and HER2 value | -0.141 | 0.159 |
| Sharp pattern and LVI value | -0.993 | 0.445 |
| Sharp pattern and Ki-67 labeling index | 0.293 | 0.002 |
| Obscure pattern and ER value | -0.336 | 0.526 |
| Obscure pattern and HER2 value | 0.291 | 0.002 |
| Obscure pattern and LVI value | 0.112 | 0.239 |
| Obscure pattern and Ki-67 labeling index | -0.249 | 0.345 |
| Sharp pattern, 491 (57.2%); Obscure pattern, 302 (35.1%); Not defined, 65 (7.5%) | ||
| IC, iodine concentration; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; LVI, lymphovascular invasion. | ||
It was verified whether the i-map could safely evaluate tumor extent by comparing it with MRM. Representative images are shown in Fig. 3a–d. The i-map, MRM, and resection margins of pathological specimens of patients who underwent partial mastectomy (n = 532) were retrospectively evaluated, and 414 (91.0%) cases were judged as proper estimation, which means that the evaluation of tumor extent was the same as that of MRM. There were 84 (9.8%) cases judged as underestimation. These assessments included patients with LC (n = 35, 41.6%), IDC (n = 33, 39.2%), and MC (n = 10, 11.9%). The proportions of underestimation in the MC and LC groups were significantly higher than that in the IDC group (IDC: 8.2% vs. MC: 66.7% [p = 0.02] and LC: 53.8% [p = 0.04]). These results showed that the histological type affected the evaluation results in the diagnosis of tumor extent using the i-map (Fig. 4).
This study investigated the i-maps of 858 operable breast cancers derived from ceDECT, focusing on IC and visual patterns, to detect and evaluate various subtypes of breast cancer. The i-maps produced sharp and obscure patterns depending on the visibility of the tumor, and the ICs of tumors were associated with the histological subtype and some malignant features of breast cancer; therefore, these indicators may be valuable, especially for surgeons, radiologists, and pathologists engaged in the diagnosis and treatment of breast cancer.
Few studies have evaluated the utility of ceDECT in breast cancer. Wang et al. reported associations between DECT parameters and immunohistochemical biomarkers of invasive breast cancer [21]. Volterrani et al. reported the feasibility of evaluating loco-regional staging by DECT in breast cancer [22]. Zhang et al. reported the efficacy of DECT for detecting lymph node metastases in preoperative breast cancer [23]. Our retrospective evaluation of i-maps in breast cancer confirmed that DCIS, an ER+/luminal A-like subtype with a low risk of malignancy, correlated with a high IC and displayed a sharp pattern; in contrast, the TN/HER2 subtype, which is associated with a high risk of malignancy, and special types like LC correlated with low ICs and showed obscure patterns. Furthermore, small-size cancers with DCIS, an ER+/luminal-A like subtype, and invasive breast tumors with a low risk of malignancy were highly detectable on i-maps, with high ICs and sharp patterns. In contrast, some special histological types (such as LC and MC), the TN and HER2 + subtypes, and some highly malignant cancers were obscure on i-maps, with a low IC or obscure pattern; therefore, the number of patients with these types of malignancies was underestimated by i-map before partial breast resection. We speculate that the difference in IC between various breast cancer subtypes depends on several factors, including tumoral cellular density, fibrosis, sclerosis, necrosis, and tumor vascularization.
I-maps may be useful in breast cancer patients because, during ceDECT, the patient lies in the supine position, which is the same position used during surgery; therefore, an i-map could be advantageous for the surgeon to estimate tumor location and breast thickness. In contrast, the main contraindication of ceDECT is allergy to contrast medium.
MRM is a very valuable technique that is especially useful in patients undergoing conservative breast surgery to determine the proper surgical margins. Dynamic contrast MRM is remarkable in facilitating the detection of the intraductal spread of breast cancer [24, 31]. Previous studies indicated that parameters derived from MRM were associated with histological subtype and prognosis, including treatment response to chemotherapy [32–34]. However, dynamic contrast MRM has its disadvantages. First, MRM depends on the menstrual cycle in premenopausal women, so it may potentially overestimate the tumor extent. Second, MRM is contraindicated in patients with internal metals or claustrophobia. We found that some special histological types (such as LC and MC), the TN and HER2 + subtypes, and some highly malignant cancers were obscure on i-maps derived from ceDECT with a low IC or obscure pattern, and the proper surgical margin was therefore difficult to evaluate. Our results suggest that tumors with a low IC and low visibility on the i-map of ceDECT scans require MRM before breast-conserving surgery. We cannot make any conclusions regarding the superiority of i-map or MRM. Our results indicate that i-maps may prevent overestimation by MRM for some histological or pathological features. An appropriate knowledge of the use of i-maps to perform appropriate resections with safe margins is variable among surgeons.
F-fluorodeoxyglucose-positron emission tomography (FDG-PET) and breast-specific PET scans are promising modalities for breast cancer [18]. Previous studies reported excellent detection of breast cancer, with parameters derived from FDG-PET being associated with the histological subtype and prognosis, including treatment response to chemotherapy [35–37]. Disadvantages of FDG-PET are as follows: higher radiation dose, higher cost than conventional CT, and the contraindication of diabetes.
Despite the findings of this study, it has several limitations. First, this study is single institute retrospective analysis. Second, our findings cannot elucidate the mechanism underlying variation IC and visual pattern in the tumor microenvironment. Our findings may be linked to other studies evaluating DECT for patients with breast cancer to build on their significance.
We argue that ceDECT is not a complete substitute for MRM or FDG-PET, but understanding the differences between these diagnostic tools can lead to a more accurate diagnosis in patients with breast cancer, as the pathological and physical characteristics of patients affect breast cancer visualization.
Our study has several limitations. First, this was a single-institution retrospective study. Second, the reasons for the different IC and visual patterns among tumors are unknown.
The IC and visual patterns of tumors on i-maps derived from ceDECT were associated with the histological features of breast cancer. Most breast cancers are identifiable on the i-map; however, the adjunction of MRM is preferred for tumors with a low IC on the i-map to evaluate the tumor extent.
BI-RAD, breast imaging-reporting and data system; BMI, body mass index; ceDECT, contrast-enhanced dual-energy computed tomography; CT, computed tomography; DCIS, ductal carcinoma in situ; DECT, dual-energy computed tomography; ER, estrogen receptor; FDG-PET, F-fluorodeoxyglucose-positron emission tomography; HER2, human epidermal growth factor 2; i-map, iodine map; IC, iodine concentration; IDC, infiltrating ductal carcinoma; IQR, interquartile range; LC, lobular carcinoma; LVI, lymphovascular invasion; MC, mucinous carcinoma; MMG, mammography; MRM, magnetic resonance mammary gland; PgR, progesterone receptor; TN, triple-negative.
Ethics approval: This study was approved by the Hiroshima Asa City Hospital review board.
Consent to participate: The participant has consented to the study.
Consent for publication: All authors have consented to the publication of this manuscript.
Availability of data and materials: The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.
Competing interests: The authors declare that they have no competing interests.
Funding: No funding was received for conducting this study.
Authors’ contributions: NG carried out the study, collected and analyzed the data, and drafted the manuscript. MF, CO, HM, and MK coordinated and critically reviewed the manuscript. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work with respect to accuracy or integrity.
Code availability: Not applicable
Acknowledgments: We thank all members of the Department of Radiology and Pathology, Hiroshima Asa City Hospital for data collection and analysis and other helpful support.