2.1.

Near-Infrared Time-Resolved Spectroscopy

Optical properties of normal breast tissue and cancer were measured using a single-channel near-infrared TRS system (TRS-20SH, Hamamatsu Photonics K.K., Hamamatsu, Japan). This system uses a time-correlated single-photon counting method for acquiring the temporal response profiles from the tissue against optical impulse inputs and enables quantitative analysis of light absorption and scattering in tissue by the photon diffusion theory.⁷ An ultrashort pulsed laser light source (PLP, Hamamatsu Photonics K.K.) emits light pulses at three different wavelengths (758, 795, and 833 nm) with an average power of $200\mu \mathrm{W}$, full width at half maximum (FWHM) of $<100\text{picoseconds}$ (ps), and repetition frequency of 5 MHz. The detector section of this system consisted of a photomultiplier tube (H7422P-50MOD, Hamamatsu Photonics K.K.) followed by constant fraction discriminators, time-to-amplitude converters, A/D converters, and histogram memories. The instrumental response function of the system was approximately $320\text{-}\mathrm{ps}$ FWHM. The three light sources emitted a pulse on a time series, and the three-wavelength optical pulses were guided into one optical fiber via a fiber coupler (NTT Advanced Technology Corp., Kawasaki, Japan). Each single optical fiber (GC200/250L, Fujikura Ltd., Tokyo, Japan) used for light irradiation had a numerical aperture (NA) of 0.25 and a core diameter of $200\mu \mathrm{m}$. The optical bundle fiber (SCHOTT MORITEX Corp., Asaka, Japan) used to collect the light had an NA of 0.26 and a bundle diameter of 3 mm. The nonlinear least-squares method was used to fit the solution of the time-domain photon diffusion equation with a semi-infinite homogeneous model to the observed temporal profiles. The zero boundary condition of the reflectance mode was employed. The time resolution of the temporal profile was 10 ps, and the data acquisition time was 3 s. The distance from the light source to the detector was 3 cm. The absorption coefficient and reduced scattering coefficient were obtained at each wavelength. The data over the time range from $-560$ to 5440 ps were selected for fitting. The reduced chi-square value was used to evaluate fitting accuracy. Our fitting process was successfully implemented if this value was within the range of 0.8 to 1.2.⁸ The concentrations of oxygenated and deoxygenated hemoglobin were calculated from the absorption coefficient at each wavelength by the least-squares method after subtracting water and lipid absorption in breast tissue. We assumed that water content was 18.7% and lipid content was 66.1%.⁹ The concentration of tHb was the sum of the concentrations of oxygenated hemoglobin and deoxygenated hemoglobin.

2.2.

Ultrasonography

An ultrasonography (US) system (EUB-7500, Hitachi Medical Corporation, Tokyo, Japan) with a linear probe (EUP-L65, Hitachi Medical Corporation) was used. The probe had a frequency range of 6 to 14 MHz. A spectroscopic probe was attached to the US probe.³

2.3.

Patient Characteristics

We examined 86 patients with breast lesions between June 2013 and April 2016. We excluded 32 patients who had undergone chemotherapy or hormonal therapy. Three patients in whom the final diagnosis was not cancer were excluded. We excluded two patients just after biopsy and two patients just after subcutaneous injection of a radiopharmaceutical around the tumor for sentinel lymph node biopsy. We also excluded two patients whose tumors were undetectable on US and one patient in whom ultrasound images were not recorded. Finally, 44 patients were included in this study. All patients were female and their median age was 52.3 years, with a range of 33 to 78 years. Twenty patients were premenopausal and 24 patients were postmenopausal. The patients underwent TRS and US scan at least 12 days after core needle biopsy. Six patients had multiple lesions, and a total of 53 lesions were evaluated. Forty tumors were found to be invasive ductal cancer, 12 tumors were ductal carcinoma in situ, and one tumor was invasive lobular carcinoma. The mean maximum transverse diameter of the tumors was 19.5 mm, with a range of 7 to 61 mm. The study protocol was approved by the Ethical Review Committee of Hamamatsu University School of Medicine. All patients signed a written informed consent form.

2.4.

TRS and US Scan Protocol

After the spectroscopic probe that was attached to the US probe was placed on the skin just above the tumor, the absorption coefficient and reduced scattering coefficient at three wavelengths were measured simultaneously. Tumor thickness, depth, and skin-to-chest wall distance were measured by US. The distance from the skin to the anterior surface of the tumor was defined as the depth. The distance between the anterior and posterior surface of the tumor was defined as the thickness. The chest wall was defined as the surface of the major pectoral muscle or anterior serratus muscle. The default setting of the maximal depth in our US system was 36 mm, and we were not able to measure the skin-to-chest wall distance in two normal breasts and four tumor-containing breasts. They were excluded from the analysis using the correlation coefficient between optical parameters and the skin-to-chest wall distance.

The optical parameters of normal breast tissue were measured in the contralateral breast at the corresponding area. Although the number of tumors examined was 53, the number from normal breasts was 46. The number of the measurement points in the normal side was less than the number of tumors in the patients who had multiple lesions.

Net tHb (${\mathrm{tHb}}_{\text{net}}$) was defined as the difference between tumor tHb and tHb in normal breast tissue (Fig. 1). The reference curve of the tHb concentration in normal breast tissue as a function of the skin-to-chest wall distance for pre- and postmenopausal patients was derived from the previous study.³

Fig. 1

${\mathrm{tHb}}_{\text{net}}$ was defined as the difference between tumor tHb and tHb in normal breast tissue. The reference curve of the tHb concentration in normal breast tissue as a function of the skin-to-chest wall distance for pre- and postmenopausal patients was derived from the previous study.³

2.5.

Statistical Analysis

Statistical examinations were performed using Microsoft Excel 2013 (Microsoft Corporation, Redmond, Washington) and StatFlex version 6.0 (Artech Co., Ltd., Osaka, Japan).

The Pearson’s correlation coefficient was calculated to compare the relationship between tHb and tumor thickness, ${\mathrm{tHb}}_{\text{net}}$ and tumor thickness, tHb and tumor depth, ${\mathrm{tHb}}_{\text{net}}$ and tumor depth, tumor depth and skin-to-chest wall distance, ${\mathrm{SO}}_2$ and the skin-to-chest wall distance, ${\mu }_{\mathrm{s}}^{\prime }$ and body mass index (BMI), and ${\mu }_{\mathrm{s}}^{\prime }$ and the skin-to-chest wall distance. Linear regression analyses were carried out for each pairing. The differences in ${\mathrm{SO}}_2$ and ${\mu }_{\mathrm{s}}^{\prime }$ between tumor and normal breast tissues were determined using nonparametric Mann–Whitney U tests. Differences with $P<0.05$ were considered significant.

3.1.

Total Hemoglobin Concentration

tHb in normal breast tissue was plotted near the reference curve in both pre- and postmenopausal patients. tHb in breast cancer was plotted above the curve in both pre- and postmenopausal patients [Figs. 2(a) and 2(b)].

Fig. 2

(a) In premenopausal patients, tHb in normal breast tissue was plotted near the reference curve ($n=24$) and tHb in all 27 tumors was plotted above the curve. (b) In postmenopausal patients, tHb in normal breast tissue was plotted near the reference curve ($n=20$) and tHb in all 22 tumors was plotted above the curve.

Although no correlation was found between tHb and tumor thickness for all the tumors, a positive correlation was found for tumors with a skin-to-chest wall distance larger than 21 mm [Figs. 3(a) and 3(b)]. There was also a positive correlation between ${\mathrm{tHb}}_{\text{net}}$ and tumor thickness [Fig. 3(c)]. There was a negative correlation between tHb and tumor depth and between ${\mathrm{tHb}}_{\text{net}}$ and tumor depth [Figs. 4(a) and 4(b)]. The correlation coefficient for tHb was stronger than for ${\mathrm{tHb}}_{\text{net}}$. The tumor depth was small in tumors with a small skin-to-chest wall distance [Fig. 4(c)].

Fig. 3

(a) The tHb in tumors was plotted against the tumor thickness. No correlation was found between tHb and tumor thickness ($n=49$, $R^2=0.01$). (b) In tumors with a skin-to-chest wall distance larger than 21 mm, a positive correlation was found between tHb and tumor thickness ($n=27$, $R^2=0.36$, $P<0.001$). (c) There was a positive correlation between ${\mathrm{tHb}}_{\text{net}}$ and tumor thickness ($n=49$, $R^2=0.17$, $P=0.003$).

Fig. 4

(a) There was a negative correlation between tHb and tumor depth ($n=49$, $R^2=0.47$, $P<0.001$). (b) A negative correlation between ${\mathrm{tHb}}_{\text{net}}$ and tumor depth remained after calculating ${\mathrm{tHb}}_{\text{net}}$ ($n=49$, $R^2=0.16$, $P=0.005$). (c) There was a positive correlation between tumor depth and the skin-to-chest wall distance ($n=49$, $R^2=0.34$, $P<0.001$).

3.2.

Tissue Oxygen Saturation

${\mathrm{SO}}_2$ in normal breast tissue decreased when the skin-to-chest wall distance decreased [Fig. 5(a)]. The mean ${\mathrm{SO}}_2$ in normal breast tissue was $78.1\%\pm 2.9\%$ ($\text{mean}\pm \mathrm{SD}$). To minimize the effects of the chest wall, ${\mathrm{SO}}_2$ in breast with a skin-to-chest wall distance larger than 21 mm was analyzed. The tumor ${\mathrm{SO}}_2$ was lower than normal breast tissue ${\mathrm{SO}}_2$, although the difference was not significant [Fig. 5(b)].

Fig. 5

(a) In normal breast tissue, there was a positive correlation between ${\mathrm{SO}}_2$ and the skin-to-chest wall distance ($n=44$, $R^2=0.14$, $P=0.01$). (b) In breast tissue with a skin-to-chest wall distance larger than 21 mm, the tumor ${\mathrm{SO}}_2$ ($n=27$) was lower than breast tissue ${\mathrm{SO}}_2$ ($n=14$), although the difference was not significant ($P=0.221$). Bars and boxes represent the median and interquartile range, respectively.

3.3.

Reduced Scattering Coefficient

We reported results on ${\mu }_{\mathrm{s}}^{\prime }$ at a wavelength of 795 nm (${\mu }_{\mathrm{s}795}^{\prime }$). In normal breast tissue, a negative correlation was found between ${\mu }_{\mathrm{s}795}^{\prime }$ and BMI [Fig. 6(a)]. A negative correlation was also found between ${\mu }_{\mathrm{s}795}^{\prime }$ and age (data not shown). There was a negative correlation between ${\mu }_{\mathrm{s}795}^{\prime }$ and the skin-to-chest wall distance [Fig. 6(b)]. The ${\mu }_{\mathrm{s}795}^{\prime }$ in tumors was higher than that in normal breast tissue when the skin-to-chest wall distance was larger than 21 mm [Fig. 6(c)].

Fig. 6

(a) The ${\mu }_{\mathrm{s}795}^{\prime }$ in normal breast tissue decreased when the BMI increased ($n=46$, $R^2=0.30$, $P<0.001$). (b) In normal breast tissue, a negative correlation was found between ${\mu }_{\mathrm{s}795}^{\prime }$ and the skin-to-chest wall distance ($n=44$, $R^2=0.46$, $P<0.001$). (c) In breast tissue with a skin-to-chest wall distance larger than 21 mm, the ${\mu }_{\mathrm{s}795}^{\prime }$ in tumors ($n=27$) was higher than that in normal breast tissue ($n=14$) ($P=0.005$). Bars and boxes represent the median and interquartile range, respectively.