1.

Introduction

Breast cancer (BC) is a leading oncologic disease of active working age, being also the main malignant tumor mortality cause among women. In accordance with the data of the International Agency for Research on Cancer, BC is the most common cancer in women worldwide representing about 12% of all new cancer cases and 25% of all cancers in women.¹ Since a systemic model of BC pathogenesis had been accepted,^2,3 a substantial enhancement of the role of chemotherapy (CT) and hormonal therapy for its treatment has occurred.⁴ Neoadjuvant CT has been initially considered as a method of “downstaging” of locally advanced tumors prior to surgery;^5,6 however, in the past years, the indications for it were widely extended.⁷ The response of breast tumors to the neoadjuvant CT is considered to be a surrogate biomarker, allowing to predict prognosis and outcome of the disease.⁸ The best marker of therapeutic efficiency having influence of the general and recurrence-free survival is the pathological complete tumor response (pCR), which is defined as absence of viable tumor cells in the breast tissue and regional lymph nodes.^7,9 However, pCR is only achieved in $\sim 13\%$ to 33% of the patients, due to the grade of malignancy, tumor immunophenotype, and the CT scheme.^5,10,11 Moreover, in a certain percentage of the patients, the preoperative CT shows an absence (grade I or II) of pathological tumor response.^12–16 In this case, problems occur connected with a delay of surgery, development of toxicity, and high cost of the actually ineffective therapy.¹⁵ From this point of view, development of methods allowing for early prognosis of the tumor response after the beginning of CT is of great importance.

Methods of structural imaging [ultrasound investigation (USI), x-ray mammography, magnetic resonance imaging (MRI)] that allow detecting dimensional changes of the lesion¹⁶ are not suitable for early assessment of the tumor response to the administered therapy. In the past years, active studies have been conducted to use metabolic imaging methods (MRI and positron emission tomography)^17–19 for such a purpose, as well as studies have been started for the investigation of opportunities provided by optical methods for assessment and monitoring of the oxygen status of biological tissues.^20,21 The latter are based on the calculation of optical parameters of biological tissues (scattering and absorption) using multiple-scattering light interacting with the object.^22–26 The experiments conducted by the method of diffuse optical spectroscopy (DOS) have shown that CT, dependent on its efficiency, leads to a certain change of parameters of neoplasm’s oxygen saturation.²⁷ The results build a basis for use of optical methods in clinical research.^28,29 The papers^30–32 summarize the results of oxygen status monitoring of breast tumors in the course of neoadjuvant CT by means of DOS. The studies conducted enabled offering of criteria of early tumor response. As such predictors, the initial (basic) level of total hemoglobin,³³ the initial level of oxyhemoglobin,³⁴ the trend of changes of the total hemoglobin after the first CT cycle,³⁵ dynamics of tumor saturation,³⁶ and the initial level of deoxyhaemoglobin³⁷ were suggested.

Along with the oxygen saturation of the tumor and its dynamics in the course of the therapy, an important parameter is the tumor vascularity that determines the oxygen delivery to the tissue. Its monitoring during a treatment might provide valuable additional information concerning tumor response to therapy.³⁸ However, tumor hemodynamics evaluation over the course of cancer therapy is limited in a routine clinical practice because of the lack of appropriate and convenient technologies. Several methods exist for the measurement of tumor hemodynamics, such as optoacoustic imaging³⁹ and optical coherence tomography angiography.⁴⁰ The latter’s limitation is the imaging depth (the instrumentation may be used only in case of superficial neoplasms). Nowadays, a prospective technology for microvasculature blood flow measurement, namely, diffuse correlation spectroscopy (DCS), that uses the temporal fluctuations of near-infrared light to measure blood flow has been developed.^41–43 The main advantage of the method is its opportunity to study deep-seated tissues such as brain⁴³ and muscle.⁴⁴ The method was also used for therapeutic monitoring of the (chemo)radiation therapy of head/neck cancer and for prediction of tumor response to treatment,^41,45 demonstrating a predictive ability in revealing the tumors sensitive and nonsensitive to the treatment.

A standard method of blood flow detection, color Doppler USI, allows evaluating relatively large vessels. A combination of USI and DOS has been used for studies of nonhomogeneous oxygen saturation and vascularization of large breast tumors.⁴⁶ The paper⁴⁷ presents the dynamics of tumor blood flow in patients to whom neoadjuvant CT for BC was administered. Divergent changes of this parameter were detected, dependent on the degree of tumor response, as well as a considerable “delay” of the dimensional and the vasculature changes in the tumor. Color Doppler ultrasonography was used to monitor the response to therapy and to identify possible correlations between survival and various Doppler indices (resistivity index, pulsatility index, and maximum flow velocity) in Refs. 48 and 49. The pre- and post-CT indices were compared with each other in order to assess and evaluate the response to CT.

The study objective was investigation of a predicting ability of early changes of breast tumor oxygenation (by DOS) and tumor vascularity (by Doppler ultrasonic imaging) in the course of neoadjuvant polychemotherapy.

2.

Materials and Methods

2.2.

Study Design

Mammography, ultrasonic breast and peripheral lymph node imaging, tumor biopsy with determining of receptor status, HER2neu status, and proliferation index of all patients were made before the therapy. The assessment of the oxygen saturation level of the tumor by DOS and of the blood flow of the tumor area by Doppler scanning were conducted prior to CT start and prior to the second CT cycle. Study design is presented in Fig. 1.

Fig. 1

Study design.

After CT, surgery was carried out with subsequent pathological study of the tumor and assessment of the pathological tumor response to the therapy. Based on the pathological results, the patients were divided into groups of complete and partial response, as well as those with absence of tumor response to the therapy. The degree of pathological tumor response was determined in accordance with Ogston et al.⁵⁰

2.3.

Diffuse Optical Spectroscopy

Experiments on DOS were performed on the experimental setup with parallel plane geometry developed at the Institute of Applied Physics, RAS (Nizhny Novgorod, Russia).²⁶ Three laser fibers coupled in a bundle illuminated the studied volume at three wavelengths: 684 nm corresponding to the maximum absorption of deoxygenated hemoglobin, 850 nm corresponding to maximum absorption of oxyhemoglobin, and 794 nm isosbestic point. In this device, a high-frequency amplitude modulation (140 MHz) was used, allowing determination of absorption coefficients more accurately due to separate determination of scattering and absorption coefficients. The diffuse light was detected by the Hamamatsu photomultiplier tube with an automatic gain control. The images were obtained by simultaneous scanning of the source and the detector located along the sagittal axis from the opposite sides of the studied subject with a step of 1 to 2 mm synchronously. In each position, data were obtained from all three sources (Fig. 2).

Fig. 2

Investigation methods: (a) arrangement of the source and the detector; (b) example of an x-ray mammogram and DOS image of BC (the red oval line marks the tumor area). The dimension of DOS image is $6\times 8\mathrm{cm}$. (c) A sample of US image and breast skin marking in the tumor projection for ultrasonic scanning by means of a mask.

For scanning, a patient was lying in a prone position with a breast between two sliding transparent plates preventing its contact with the source and the detector, slightly pressed for equal thickness of the breast in the scanning area. The distance between the plates was fixed. The scanning area selection was based on the results of the clinical examination and the mammographic data. The local signal change allowed for detection of optical nonuniformity in the scanned area for subsequent visual check of the scanning area.

3.

Results

Breast tumors have demonstrated diverging dynamics of the oxygen saturation level dependent on pathological tumor response to CT [Fig. 3(a)]. In case of degree IV and V of pathological tumor response (massive extinction of invasive cells, $>90\%$ of cell loss), an enhancement of the oxygen saturation of the tumor referring to the initial value after the first cycle of CT was observed in eight patients of nine. The lowering of the level of oxygen saturation by 8% was detected in one patient with IV degree of pathological tumor response [Fig. 3(a) (yellow and green columns)].

Fig. 3

Oxygen saturation and blood flow rates of the tumor after the first cycle of CT, dependent upon the degree of pathological tumor response (25 patients). (a) DOS data and (b) USI data.

In case of the first and the second degrees of pathological tumor response (hard-to-notice separate tumor cells, but without their number going back), a lowering (in six patients) or absence of dynamics (in two patients) of the oxygen saturation level of the tumor in comparison with the initial one was observed in eight of nine patients [Fig. 3(a) (blue and orange columns)]. In one patient with the II degree of pathological tumor response to therapy, enhanced level of oxygen saturation by 12% as compared with the initial level was observed.

In case of the third pathological tumor response degree (reduction in tumor cells of between 30% and 90%), both enhancement and decrease of the oxygen saturation level of the neoplasm was observed. In three patients, enhancements of the saturation were detected, also one significant enhancement value of 57%. In two cases, a decrease was observed and in the other two no oxygen saturation level changes were detected after the first CT cycle [Fig. 3(a) (gray columns)].

The analysis of the dynamics of the blood flow through the tumor could not show a significant dependence between the changes of this parameter after the first CT cycle and the degree of pathological tumor response. In most cases, a decrease of the blood flow rate was observed, sometimes significant and sometimes negligible [Fig. 3(b)]. In four patients with different degree of pathological tumor response, increased blood flow ratings of the neoplasm were observed, in one case, more than a fourfold increase [Fig. 5(c)]. Obviously, the blood flow rating within the tumor area might be influenced by processes not directly connected with the death of the neoplasm’s cells or with the inhibition of the tumor neoangiogenesis under the influence of CT. It might be associated with the development of the inflammation in the peritumoral area. To illustrate the obtained results, see the following clinical examples.

A comparison of the indicators of tumor oxygenation and tumor response revealed a moderate but significant positive correlation between the analyzed parameters [Fig. 4(a)]. Linear fitting of the experimental data provided a satisfactory value of approximation reliability ($R^2>0.34$) and moderate Spearman’s rank correlation coefficient ($r=0.628$) with a significance level $p=0.0003$. We failed finding a statistically significant correlation between the number (percentage) of dead cells and blood flow changes ($r=-0.394$, $p=0.051$), BC stage ($r=-0.199$, $p=0.340$), and immunophenotype ($r=0.267$, $p=0.198$).

Fig. 4

A correlation between pathological tumor response (the number of dead cells after CT completion) and (a) $\mathrm{\Delta }{\mathrm{StO}}_2$ and (b) $\mathrm{\Delta }$ vascularization index (%) after the first cycle of CT.

Case report 1. Patient A, age of 53, was administered neoadjuvant CT (schedule AC) due to triple-negative left BC of stage 2a (T2N0M0). During the ultrasonic investigation prior to neoadjuvant CT, a tumor node with uneven and vague contour $41\times 41\mathrm{mm}$ was detected, with necrosis in the center and poor peripheral blood flow [Figs. 5(b) and 5(c), top image]. The pathologic investigation showed vast layers of necrotic tissue with a small layer of invasive BC with crash-syndrome phenomenon [Fig. 5(d) (top image)].

Fig. 5

Dynamics of tumor oxygen saturation and blood flow (patient A) (a) DOS imaging (${\mathrm{StO}}_2$) with tumor area marked by red oval, area of necrosis marked by white oval. (b) USI imaging (B-mode), area of necrosis marked by white oval. (c) US imaging (power Doppler mode). (d) Histologic samples (after core-biopsy, top image; postoperative material, bottom image). Top row, results prior to CT; bottom row, after (a–c) the first CT cycle or (d) on accomplishment of CT. DOS image dimensions $6\times 8\mathrm{cm}$.

In DOS images after the first CT cycle, an enhancement of level of oxygen saturation was detected by 14% [Fig. 5(a)] as compared with the initial value, allowing to suppose a good response of the tumor to the administered therapy. In accordance with the ultrasonic investigation, an increase of the tumor size (up to $49\times 48\mathrm{mm}$) was detected, along with a considerable growth of the central necrosis area and occurrence of a distinct edema of perinodular tissue, which is equal to the development of an inflammation process around the tumor node [Fig. 5(b), bottom image]. In addition, a significant (by 430% as compared with the initial level) vascularization growth in the tested area was detected. The blood vessels appeared all-over the peripheral area of the tumor tissue (around the necrotic area) and in the perinodular area [Figs. 5(c) and 5(d)]. After four CT cycles, the patient underwent mastectomy. During the pathological investigation, the III degree of pathological tumor response after Ogston et al. was detected. The central part of the tumor node was occupied by a vast necrosis area [Fig. 5(d) (bottom image)]. Having analyzed the results of the pathologic investigation after core-biopsy and of the removed tumor, we suppose association of the dimensional growth of the tumor blood vessels with the vast necrosis area in the center of the tumor node. Thus, in this case, the vascularization has grown due to the presence of an inflammation process in the peritumoral area stipulated by the vast necrosis area, however, not for the reason of the absence of response to the administered therapy.

Case report 2. Here is the example of changes in the tumor oxygen saturation and blood flow of patient E, age of 45. Diagnosis: right BC, stage IIa (T2N0M0), HER2neu-positive cancer (estrogen receptors 0 points, progesterone receptors 0 points, Ki67 40%). During ultrasonic investigation, prior to administering of neoadjuvant therapy, a tumor node was detected in the breast with uneven and vague contour with dimensions $28\times 14\mathrm{mm}$ [Fig. 6(b)]. The pathomorphology showed invasive ductal carcinoma [Fig. 6(d) (top image)]. The initial level of vascularization was 5.3%. The patient was assigned neoadjuvant polychemotherapy (AC + trustuzumab schedule). After the first cycle of CT, a tumor shrinked to $18\times 10\mathrm{mm}$ [Fig. 5(b)], oxygen saturation enhancement by up to 7% after the results of DOS [Fig. 6(a)] and reduction of the tumor blood flow by 40%, as compared with the initial values [Fig. 6(c)]. A pathologic investigation after six CT cycles and surgery failed to find vital tumor cells, equivalent to the V degree of pathological tumor response [Fig. 6(d) (bottom image)].

Case report 3. Here is the example of changes in the oxygen saturation and blood flow of breast tumor of patient C, age of 50. Diagnosis: left BC, stage IIIa (T2N2M0), luminal B HER2neu-negative cancer (estrogenic receptors eight points, progesterone receptors six points, Ki67 30%). Prior to neoadjuvant therapy, a node with an uneven and vague contour with dimensions of $30\times 25\mathrm{mm}$ was detected in the breast [Fig. 7(b)]. The pathologic test showed invasive ductal carcinoma [Fig. 7(d) (top image)]. The initial level of vascularization was 6.9%. The patient was assigned neoadjuvant polychemotherapy (AC schedule). After the first cycle of CT, the size of the tumor node remained unchanged [Fig. 7(b)]. Based on the results of DOS, the oxygenation level was unchanged [Fig. 7(a)], the tumor blood flow increased by 5% [Fig. 7(c)]. A pathologic investigation after six CT cycles and a mastectomy detected the I degree of pathological tumor response, equal to therapy failure [Fig. 7(d) (bottom image)].

4.

Discussion

Since the mid-2000s, there were attempts to use various modifications of optical diffuse spectroscopy as a method to predict the sensitivity of BC to CT. At the base of all suggested predictive criteria were indicators of the oxygen status of the breast tumor and/or their changes in the course of treatment.^32,51 When planning the study, we expected that the oxygenation level of any tissue reflects the ratio of oxygen delivery and oxygen consumption by tissues, i.e., tumor tissue.^53,54 The disproportion between the escalating demands of the tumor tissue in oxygen and its inadequate delivery of the pathological vascular bed leads to the phenomenon of tumor hypoxia that is known to be a characteristic feature of many solid tumors.⁵⁵ In our study, the difference between the tumor oxygenation before and after the first course of treatment was chosen as the main criterion, which reflects the tumor sensitivity to CT. When analyzing the dependence of changes in this parameter on the pathologic tumor response, a moderate but statistically significant correlation was revealed ($r=0.628$, $p=0.0003$). The trend of changes in tumor oxygenation was different depending on the effect of the treatment. Most tumors that appeared to be sensitive to CT (IV and V degrees of pathologic response that corresponds to 90% up to 100% of dead tumor cells) have demonstrated improved oxygenation in 3 weeks after treatment beginning compared to baseline. The changes of the oxygen saturation level of tumor tissue may depict changes of oxygen consumption by the tumor as a result of mortality of a certain part of the tumor cells.⁵⁶ In this case, it is possible to draw an analogy with the process of tumor reoxygenation after fractionated radiation therapy (4R of radiotherapy).⁵⁷ Tumor tissue reoxygenation appears because of decrease of the number of viable tumor cells after irradiation, and, accordingly, the decrease of tumor oxygen demands.⁵⁸ In case of ineffective CT, in most cases there was a decrease in the level of tumor oxygenation or lack of its dynamics. From our point of view, this indicates the preservation or increase of oxygen consumption by a tumor tissue, i.e., the preservation of the number of viable tumor cells.

Fig. 6

Dynamics of tumor oxygen saturation and blood flow (patient E) (a) DOS imaging (${\mathrm{StO}}_2$) with tumor area marked by red oval. (b) USI imaging (B-mode). (c) US imaging (power Doppler mode). (d) histologic samples (after core-biopsy, top image; postoperative material, bottom image). Top row, results prior to CT; bottom row, after (a–c) the first CT cycle or (d) on accomplishment of CT. DOS image dimensions $6\times 8\mathrm{cm}$.

Fig. 7

Dynamics tumor oxygen saturation and blood flow (patient C) (a) DOS imaging (${\mathrm{StO}}_2$) with tumor area marked by red oval. (b) USI imaging (B-mode). (c) US imaging (power Doppler mode). (d) histologic samples (after core-biopsy, top image; postoperative material, bottom image); top row, results prior to CT; bottom row, after the (a, b, c) first CT cycle or (d) on accomplishment of CT. DOS image dimensions $6\times 8\mathrm{cm}$.

In most works devoted to the predictive criteria for the response of breast tumors to CT, the tumors are divided to the subgroups of complete (pCR) or noncomplete (non-pCR) response. Subgroup of non-pCR aggregates tumors with grade 1 (the absence of cell’s loss), grade 2 (minimal cell’s loss), grade 3 (cell’s loss 30% to 90%), and grade 4 (more than 90% of dead cells), i.e., is very heterogeneous in respect of biological characteristics.⁵¹ The authors underscore that a more refined clinical assessment of the non-pCR group using a continuously variable endpoint, such as the residual cancer burden or the Miller–Payer system, could lead to a better understanding of the non-pCR changes, and to a better separation between response groups.

The same approach was used in Ref. 59, when evaluating a predictive value of an initial level of total hemoglobin of tumor tissue before treatment and its changes after three cycles of CT. Tumors with grades 1 to 3 and 4 to 5 were compared. We assume that the evaluation of tumor oxygenation in case of grade 3 is the most difficult problem because this subgroup includes very heterogeneous neoplasms that demand an individual analysis from the point of view of cell loss.

A separate discussion is necessary when a tumor node contains extended necrotic areas which occupy a significant part of the volume. In this case, tumor oxygenation will be highly influenced by the anoxic area characterizing a necrotic tissue. The selection of the region of interest is, in this situation, a difficult matter. The used method of manual contouring may lead to an incorrect selection of the area of interest overlapping with the necrotic area. That may drastically distort the results demonstrating reduced oxygen saturation, as compared with the real value. In addition, around the necrosis area inflammation may occur, which provokes vascular reaction and increase of the blood flow being an additional distortion factor of the data on the oxygen status of the tumor. It is exactly the case represented in the clinical example. It is possible that necrotic changes of the tumor node could be a contraindication for DOS use on the present state of the method development.

In its turn, the oxygen saturation of any tissue is directly associated with the features of its blood flow, that is, dependent on the state of the vascular bed. In our investigation, an attempt was made to assess the change of the breast tumor blood flow dependent on the effect of the neoadjuvant CT. Available literature shows that early decrease or complete disappearance of the tumor blood flow evaluated by Doppler sonography is the indicator of the efficiency of CT. On the contrary, the increase of the vascularization of the tumor during therapy may be an indicator of the tumor’s progression.^38,60,61 In our investigation, 3 weeks after therapy, a certain decrease of tumor vascularization was observed in the overwhelming majority of the patients. In four tumors with the first, second, and third degrees of tumor response a growth of the blood flow was detected, rather significant in two cases (Fig. 3). Those changes were observed in patients with triple-negative BC for which occurrence of necrotic areas and development of a peritumoral inflammation is distinguishing. We may assume that the multiple increase of the blood flow may be explained in these patients by exactly that fact. The absence of the correlation of the blood flow change and the response of the tumor may be explained as follows. The limit of resolution of contemporary ultrasonic devices is confined by blood vessels of minimum 0.1 mm diameter,⁶² changes in which do not show the processes in the capillary vessel network due to tumor’s neoangiogenesis. It is evident that for the assessment of processes occurring in the tumor’s vascular network methods of higher resolution are required which enable imaging of vessels with diameters of 10 to $100\mu \mathrm{m}$.

5.

Conclusion

Oxygen saturation level of BC under short-term chemotherapeutic impacts may allow for a prognosis of the tumor response to the therapy. Increased oxygen saturation is a predictive factor of the fourth and fifth degrees of pathological tumor response. A reduction or absence of the oxygen saturation dynamics allows for assumption of the first or the second degree of pathological tumor response. The most complicated is the prognosis of the third degree of pathomorphism, since in such cases both increasing and decreasing of the oxygen saturation level can be observed.