Introduction
Breast cancer is considered one of the major cause of cancer-related mortality in women worldwide. Five percent of newly diagnosed breast cancers is locally advanced breast cancer (LABC) [
8]. Women with LABC commonly have poor outcomes and required neoadjuvant chemotherapy (NAC) before surgery. As one important therapy of multimodality LABC treatment, NAC can either decrease large tumor bulk before mastectomy or allow breast-conserving therapy. Furthermore, after NAC, those patients with pathologic response have better prognosis than those without a response [
2,
5]. Therefore, predicting the tumor response to NAC is critical in the multimodality LABC treatment and chemotherapy regimens could be changed in advance in case of a predicted poor response.
Nowadays, many methods are available to assess tumor response, such as clinical palpation, ultrasound [
18], X-ray mammography, magnetic resonance imaging (MRI) [
21], and positron emission tomography/computed tomography (PET/CT) [
3]. However, they still have their limitations. Clinical palpation and ultrasound are not sensitive enough to detect tumor response. X-ray mammography is not indicated in presence of high mammographic density [
12] or fibrotic tissue [
15], and it has poor correlation with pathology [
20]. Compared to other method, MRI and PET/CT have high sensitivity to monitor LABC treatment response. However, it lacks standard dynamic contrast-enhanced criteria and remains relatively costly.
Thus, an inexpensive and noninvasive diagnostic method is needed to monitor NAC response. Near-infrared diffuse optical spectroscopy (DOS) has been recently considered in breast cancer diagnosis. This imaging modality came from optical imaging [
4] and did not gain clinical interest because of poor differentiation between tumor and benign tissue. With advances in detection technology and computing, DOS could measure near-infrared absorption and scattering spectra across tissue. This method can reflect the change of deoxyhemoglobin (Hb) into oxyhemoglobin (HbO
2) in tissues, which is related to angiogenesis, tumor cell proliferation, and hypoxia in tumor. Therefore, at present, DOS is considered another option in predicting early tumor response to neoadjuvant treatment [
1,
7,
17,
19]. However, DOS test cycles in these studies are not representing all treatment cycles or the sample size is relatively small (patient number approximately ten).
The purpose of our study was to explore the change in blood-oxygen content using DOS to predict NAC response against breast cancer. We also evaluated the ability of these DOS parameters to analyze how early they can be detected or the accuracy they can achieve.
Materials and methods
Patients and NAC treatment
Written informed consent was obtained from all patients enrolled in this study and the study was approved by the institutional ethics committee of Affiliated Tumor Hospital of Guangxi Medical University. Twenty patients who had locally advanced breast cancer or planned to perform breast-conserving therapy were enrolled from June 2015 to August 2016. These patients received NAC 2–6 cycles before surgery. The chemotherapy regimen of seven patients in nonresponse group was changed to another effective regimen or they underwent surgery after two to six cycles. They underwent baseline clinical exam, mammogram, ultrasound, or MRI before treatment, and received at least two follow-up DOS scan. Oncologists and pathologists were blinded to DOS data. Cancer diagnosis was confirmed in all patients by core needle biopsy before NAC treatment. Ultrasound or MRI represented the baseline imaging for these patients. Ultrasound and clinical examination were performed at follow-up period. MRI was performed at presurgery (defined as the time within 1 week before surgery) for each patient. The tumor size and its change were recorded for each patient. Pathological characteristics were recorded after mastectomy, such as tumor size, histologic subtype, grade, and tumor response.
Chemotherapy was performed according to NCCN guideline(version 2.2016). Chemotherapy treatment regimen included AC (doxorubicin/cyclophosphamide) followed by docetaxel, TAC (docetaxel/doxorubicin/cyclophosphamide), FEC (fluorouracil/epirubicin/cyclophosphamide), FAC (fluorouracil/doxorubicin/cyclophosphamide), XEC (capecitabine/epirubicin/cyclophosphamide). Two patients with HER-2-positive tumor received trastuzumab with taxane. Doses and treatment regimen were adjusted according to toxicity and response.
DOS system instrument
One type of DOS system was used, named E-sea (Multimode) Mammary Gland Blood-oxygen Functional Imaging System (ES-DM-X) (Wuhan E-sea Digital Technology Co., Ltd, Wuhan, China). ES-DM-X consists of blood and oxygen parameter detection and imaging in the diseased region of the mammary gland, and it displays abnormal blood vessels inside and outside diseased region. Within the blood-oxygen detection program, ES-DM-X can analyze blood-oxygen content in the tumor region and obtain related blood-oxygen images showed in the blood and oxygen content figure after the analysis.
The handheld probe was placed under the breast while the patient was sitting in a chair. Near-infrared light transmitted the skin and the parenchyma of the breast from the light source, it was collected by the optical imaging device and converted into electric signal. The tumor location and size in the volume of interest for blood-oxygen content images were similar with that in clinical exam and other imaging. DOS images of the breast lesion were obtained and blood-oxygen parameters in the breast lesion volume of interest could be directly obtained with ES-DM-X softscan platform.
NAC response assessment
The first DOS scans were performed before the start of NAC and considered as baseline data (defined as pretreatment), and subsequent scans were performed at each circle after the start of NAC treatment or according to the actual follow-up situation of the patient. The follow-up DOS examination was defined as the measurements taken within 1 week before next treatment. The final scan was performed within 1 week before surgery (defined as presurgery).Two experienced radiographer (Yinghua Yu and Xiao Zhu) independently performed the DOS [(ES-DM-X)].
Tumor size and pathological examination analysis were included in our NAC response assessment. Tumor size was measured by clinical exam, ultrasound, mammogram, or MRI. According to the World Health Organization (WHO) criteria, tumor responses assessed by tumor size were as follows [
10]: complete response (CR) meant that the tumor lesion disappeared completely and lasted for more than 4 weeks. If the tumor size decreased more than 50% and was stable for 4 weeks, it could be included in partial response (PR). If the tumor lesion decreased less than 50% or increased less than 25%, it was defined as stable disease (SD). Tumor responses assessed by pathological examination after NAC were classified in three categories [
17]: pathological complete response, pathological good response, or pathological minimal response. (1) Pathological complete response was defined as no residual malignancy or in situ tumor presence [
9]. (2) Pathological good response was defined as a tumor size reduction over 50% with the invasive cells less than 10%. (3) Pathological minimal response was defined as minimal invasive tumor reduction. Patients who obtained CR or PR in radiological definitions and complete or good pathological response as demonstrated by pathological assessment were included in the response group. The remaining patients were included in the nonresponse group.
Statistical analysis
Statistical analysis was calculated using SPSS 13.0 software (Chicago, IL, USA). Blood and oxygen content and tumor size were not normally distributed (Kolmogorov–Smirnov test, P < 0.05). Therefore, changes in blood and oxygen content value were compared using two-way non-parametric Mann–Whitney test between two groups such as pretreatment and presurgery, response and nonresponse groups. P < 0.05 was considered statistically significant. Diagnostic parameters such as sensitivity, specificity, and the area under the ROC curve (AUC) were applied to assess the diagnostic performance in differentiating tumor response from nonresponse.
Discussion
NAC is an effective treatment against LABC before mastectomy or breast-conserving therapy. However, it is of utmost importance an early response evaluation for predicting whether NAC should be continued before surgical intervention.
It had been accepted that the change in tumor size measured on ultrasound, X-ray or MRI, could predict pathological responses after NAC [
13]. Change in tumor size takes time (for example at least after two to three treatment cycles) [
1]. Therefore, some functional methods [
14], such as contrast-enhanced ultrasound [
11], MRI [
16] and fluorodeoxyglucose (FDG PET) [
6], have been researched to assess NAC treatment response. However, contrast-enhanced ultrasound was invasive and still pending for approval of microbubble contrast agents for this indication. MRI was invasive to inject contrast agent and relatively costly. Furthermore, no functional standard criteria on MRI have yet emerged to evaluate NAC treatment response. The cost of FDG PET limited its widespread application.
Measuring oxyhemoglobin content can reveal the content of oxygen in blood, thus revealing body tissue metabolism. Deoxyhemoglobin absorbance is different from that of oxyhemoglobin [
22] at 600–800 nm wavelength. The red light absorbed by hemoglobin in angiogenic regions, can reflect the difference between non-blood vessel areas and blood vessel areas. Hence, tumor response could be evaluated if the changes due to re-distribution of blood volume and oxygenation level are detected. DOS could assess tumor response by analyzing oxyhemoglobin (HbO2) and deoxyhemoglobin (Hb) change in tissue [
1,
17].
Our study adopted a new type of DOS system named ES-DM-X system (Wuhan E-sea Digital Technology Co., Ltd, Wuhan, China). The change in blood-oxygen content could be directly perceived from the screen image of ES-DM-X system. According to the blood-oxygen content image and the relative blood-oxygen content parameter after each NAC cycle, the doctor is able to estimate as early as possible if NAC regimen fits for the patient. This kind of examination is safe and noninvasive because of the absence of radiation damage and without contrast medium injection. Furthermore, it is relatively not expensive and requests less time to perform image examination than other systems. Thus, measurements can be repeated as many times as needed with no harm to patients, resulting in a more easily accepted scanning method by patients during the treatment response monitor period. Although the blood-oxygen content parameter measurement in ES-DM-X system is different from that of other DOS system, ES-DM-X system detection principle was similar with other systems and blood-oxygen image was more clearly visualized than the other DOS systems.
Our study used clinical or pathologic response (response versus nonresponse) as the endpoint, which was also adopted by previous studies [
1,
17]. The response and nonresponse group showed different response patterns in blood-oxygen content change. Comparing each cycle, the percentage change in blood content could distinguish responder from non-responder as early as after the first treatment cycle in response group (Fig.
3; Table
3). In other words, blood content could be a sensitive method to detect treatment response. However, oxygen percentage content value was not significantly modified until after three examining cycles, suggesting that change in oxygen content was not as sensitive as blood content. When comparing the lesion size and blood-oxygen change in response group, blood content parameter changed more dramatically than tumor size and oxygen content in response group, revealing its good tumor response prediction with NAC treatment compared with the information given by tumor size or oxygen content. The blood content percentage change was the highest among oxygen content and tumor size at the scan before surgery in response group (Fig.
2a), meaning that change in blood content could be more easily monitored than other parameter in response group. There is a need to mention that our pretreatment blood content and oxygen content value were not significantly different between response and nonresponse group (Table
3), while a new publication demonstrated that the pretreatment DOS texture features can predict breast cancer response to NAC [
19]. The reason may be that our study adopted a new type of DOS system named ES-DM-X system, and the blood-oxygen content parameter measurement in ES-DM-X system is different from that of other DOS system. The limited sample size is the limitation in both studies. Thus, a larger prospective study is needed to improve the present different results. After all, both studies confirms that DOS system could predict early therapeutic response.
We finally analyzed the diagnostic performance of blood content, oxygen content, and tumor size for predicting tumor response with NAC treatment (Table
5). The change in blood content percentage received an AUC of 0.921, which was the highest among other parameters such as tumor size and oxygen content, indicating that the blood content percentage change may be the optimal predictor of tumor response with NAC treatment. The AUC of oxygen content percentage change was similar to that of tumor size. Thus, change in oxygen content percentage could also be used as an alternative method to analyze tumor response. Therefore, DOS image (ES-DM-X) including blood-oxygen content could provide an alternative approach to assess breast tumor response, which represented a supplementary measurement besides tumor size. According to our analysis, blood content was more precise and earlier than oxygen content to predict tumor response. Compared to other imaging modalities, AUC of blood and oxygen content percentage change in the response and nonresponse group were similar with those obtained by quantitative ultrasound (QUS) (AUC of 81%) [
18], or even higher than PET/CT (AUC of 63.6%) or MRI (AUC of 75.7%) [
3]. Although we could not demonstrate that blood and oxygen content by DOS (ES-DM-X) was more precise than QUS, PET/CT and MRI, it could be speculated that they all have the same power to predict tumor response after NAC treatment. A typical blood-oxygen images and MRI of a 41-year-old responder (Fig.
4) directly show that the change in blood and oxygen content was synchronous with that of tumor size in MRI.
However, our study still had some limitations. It included patients with many different clinical characteristics and different NAC treatment protocols, although this aspect is common in oncology research. In addition, this work lacks standard criteria to perform the blood-oxygen content test or analyze the result, for example to evaluate how long or how often the blood-oxygen content should be tested, and which blood-oxygen content is the best threshold to evaluate the positive or negative result. Thus, a larger prospective study is needed to improve the present result.
In conclusion, our data indicated that blood content could detect the breast tumor response as early as after the first treatment, and both blood and oxygen content could be used to discriminate between responder and non-responder after NAC treatment. Although blood content was better than oxygen content to predict tumor response, they all could be supplementary choices besides measuring tumor size.