|Home | About | Journals | Submit | Contact Us | Français|
Structural changes in water molecules are related to physiological, anatomical and pathological properties of tissues. Near infrared (NIR) optical absorption methods are sensitive to water, however detailed characterization of water in thick tissues is difficult to achieve because subtle spectral shifts can be obscured by multiple light scattering. In the NIR, a water absorption peak is observed around 975nm. The precise NIR peak shape and position is highly sensitive to water molecular disposition. We introduce a Bound Water Index (BWI) that quantifies shifts observed in tissue water absorption spectra measured by broadband Diffuse Optical Spectroscopy (DOS). DOS quantitatively measures light absorption and scattering spectra and therefore reveals bound-water spectral shifts. BWI as a water state index was validated by comparing broadband DOS to Magnetic Resonance Spectroscopy, diffusion-weighted MRI and conductivity in bound water tissue phantoms. Non-invasive DOS measurements of malignant and normal breast tissues performed in 18 subjects showed a significantly higher fraction of free water in malignant tissues (p<0.0001) compared to normal tissues. BWI of breast cancer tissues inversely correlated with Nottingham-Bloom-Richardson histopathology scores. These results highlight broadband DOS sensitivity to molecular disposition of water, and demonstrate the potential of BWI as a non-invasive in-vivo index that correlates with tissue pathology.
Approximately 60–80% of the human body is composed of water (Marieb, 1995). A variety of techniques including MRI, NMR and Diffuse Optical methods have been used to measure tissue water content in brain (Brooks et al., 1989; Kreis et al., 1993), breast (Cerussi et al., 2007; Jakubowski et al., 2004; Pogue et al., 2004), bone (Fantazzini et al., 2001; Wang, 2003) and other tissues (Mathur-De Vre, 1984; Mirrashed and Sharp, 2004)
In addition to the volume fraction of water, water state can provide more detailed cellular and molecular information about tissues. Water mobility measurements using diffusion-weighted MRI (DW-MRI) and diffusion-tensor imaging have received considerable attention for their ability to measure cellularity and fiber orientation in brain and muscle tissues (Basser et al., 1994; Hajnal et al., 1991; Henkelman et al., 1994; Mulkern et al., 1999; Pierpaoli et al., 1996). MRI also has demonstrated that diffusing free water mobility differs between tumor and normal tissues, and correlates inversely with cellularity (Guo et al., 2002; Lyng et al., 2000; Paran et al., 2004). The low mobility of water in tumor tissues may be related to fractional changes in macromolecules, cellular membranes and intracellular components. However, the precise effect of microstructure on DW-MRI parameters is not well-understood (Mardor et al., 2003; Thelwall et al., 2002). Thus accurate measurements of bound water using DW-MRI are difficult to be obtained since macromolecules unbound to water as well as cell membranes bound to water present barriers to water diffusion (Callaghan et al., 1993; Cho et al., 1996; Nusbaum et al., 2000; Putz et al., 1992).
The goal of this research is to develop a new method that measures the molecular disposition of water in thick tissues by analyzing near infrared (NIR) absorption spectra. In previous work, we and others have demonstrated that Diffuse Optical methods can be used to measure increased water concentration in tumors relative to normal tissues (Cerussi et al., 2007; Cerussi et al., 2006; Jakubowski et al., 2004; Paran et al., 2004; Pogue et al., 2004; Spinelli et al., 2004). We have also shown that the optically-measured water concentration correlates with the Bloom-Richardson score of tumor histopathologic grade (Cerussi et al., 2006). The bound water fraction measured by optical techniques can communicate water molecular vibrational states associated with macro molecular complexes in tissues (Brubach et al., 2005; Eisenberg, 1969; Fisher et al., 1970; Libnau et al., 1994; Pimentel and McClellan, 1960; Scherems et al., 1984) and may provide additional insight regarding tissue pathophysiology. Because of increased binding of water to macromolecules such as proteins, the water absorption peak at 975nm undergoes both broadening and red shifting (Bellamy, 1968; Eisenberg, 1969; Pauling, 1960; Pimentel and McClellan, 1960). These spectral changes appear as a consequence of variations in the relative contributions of harmonic overtones from fundamental O-H vibrations at 3.05μm and 2.87μm (Eisenberg, 1969). In this work, we introduce a Bound Water Index (BWI) using these spectral changes obtained from scattering-corrected 650-998nm tissue absorption spectra measured by broadband Diffuse Optical Spectroscopy (DOS).
The sensitivity and accuracy of DOS to water state was validated by measuring bound water phantoms with controlled amounts of bound water. BWI, T2 relaxation times, conductivity and apparent diffusion coefficient (ADC) were measured and compared to BWI. Once established, BWI measurements were obtained from 18 patient volunteers with breast cancer. Breast tissue types were differentiated based on a significantly higher fraction of free water in malignant tissues compared to normal tissues (p<0.0001). Our clinical results also show that although there is significantly (~116%) increased water content in breast tumors, there is ~41% greater bound water fraction in normal vs. tumor containing breast. BWI was inversely correlated with Nottingham-Bloom-Richardson histopathology scores (R = −0.96) for all participating breast cancer patients. These findings are discussed in terms of underlying tumor-induced changes in cellularity and extracellular matrix and the potential for BWI to be used as prognostic indicator of tumor grade.
The details of the broadband DOS system have been previously described (Bevilacqua et al., 2000; Cerussi et al., 2006). Broadband DOS combines multi frequency domain photon migration (FDPM) with broadband steady state (SS) spectroscopy. Multi-frequency FDPM separates the effects of absorption and scattering in tissues (Fishkin and Gratton, 1993; Tromberg et al., 1993). Specifically, we have used a P1 approximation to the transport equation in the semi-infinite geometry using an extrapolated boundary condition (Fishkin et al., 1996; Haskell et al., 1994). The SS component enables the acquisition of continuous high resolution absorption spectra of more than 1000 wavelengths from 650 to 998nm, the upper spectral limit of our spectrometer (B&W Tek, Newark, DE, model 611a). Since the spectrometer is auto-calibrated, the absolute wavelength spectrum can be obtained for every measurement. Instrument response is determined using tissue-simulating phantoms for FDPM and spectraflect-coated integrating spheres for SS measurements with known scattering and absorption values as described in Cerussi et al. (Cerussi et al., 2006).
Measurements are performed by placing a specially-designed handpiece onto the surface of the test material or subject. The multi-frequency (50~600MHz) swept FDPM data for 658, 682, 785, 810, 830 and 850nm lasers (<20mW power) and SS data for all wavelengths between 650 and 998nm are acquired in a total of ~ 6 seconds. The source-detector separation is 29mm for breast measurements and the interrogated tissue volume by DOS is approximately 10cm3. Model fits to SS and FDPM data were performed in order to recover absorption (μa) and reduced scattering (μs′) values for all NIR wavelengths (650–998nm) as described in Bevilacqua et al. and Cerussi et al.
The extinction coefficient spectra of major tissue chromophores (oxy- and deoxy-hemoglobin, water and bulk lipid were used to recover tissue concentrations of these components (ctO2Hb (μM), ctHHb (μM), ctH2O (%, a relative ratio to pure water concentration (55.6M)) and ctLipid (%, a relative value to pure lipid density of 0.9 g·ml−1), respectively) from tissue absorption spectra (Cerussi et al., 2002 and 2007, Cubeddu et al., 2000, Pogue et al., 2004, Quaresima et al., 1998)). For oxy- and deoxy-hemoglobin, we employed molar extinction coefficients of Zjilstra et al (Zijlistra, 2000). For water, we obtained extinction coefficients by measuring distilled water in a cuvette using a spectrophotometer (Beckman DU 650) at various temperatures in order to account for possible temperature effects (Merritt, 2005). The employed extinction coefficient values of lipid were obtained from mammalian fat by van Veen et al (van Veen et al., 2005). The tissue absorption spectrum and extinction coefficient spectra provided needed information to solve nonnegative least-squares constraints problems (using ‘lsqnonneg’ in MATLAB®) to recover the concentrations of the four principal NIR absorbers.
The tissue absorption spectra (figure 1(a) solid line) were post-processed to measure bound water state using a 3 step process. First, a tissue water spectrum (μa, tissue water) was obtained by eliminating non-water tissue absorbers’ contributions by subtracting the extinction coefficient spectra of oxy- and deoxy- hemoglobin and bulk lipid multiplied by the concentration of each tissue chromophore (obtained using the method described above) from the overall tissue absorption spectrum. Second, the residual between the tissue (μ a, tissue water) and pure water spectra (μ a, pure water) was calculated by subtracting the pure water spectrum (at physiological temperature, 36°C for breast tissues (Jeffrey et al., 1999)) from a normalized tissue water spectrum in the wavelength range from 935nm to 998nm. Finally, in order to represent the residual using an index, the absolute values of the differences were combined and divided by the number of points in the sum to form the Bound Water Index (BWI) (figure 1, equation (1)).
where λi is ith wavelength (935nm ≤λi ≤ 998nm), ctH2O is the water concentration described above and N is the number of wavelength points in the sum. 935nm was chosen because above 935nm the tissue water absorption spectrum is linearly shifted to higher wavelengths relative to the pure water spectrum (Merritt, 2005).
We have investigated the effect of the temperature of the pure water spectrum used in the second step above. It is possible that the temperature could vary depending on the status of the tissue, personal difference, room temperature etc. We measured the temperature dependent NIR absorption spectra of pure water between 15 and 65 °C. We then simulated the potential impact of physiological temperature fluctuations by using water reference spectra obtained from 34°C to 38°C. Our results showed that BWI varied by up to 4% with temperature fluctuations in this range (data not shown). Since the simulation demonstrates that the effect on BWI would be small, we chose one temperature (36°C), previously measured for in vivo breast tissue (Jeffrey et al., 1999), for all in-vivo BWI calculations
The accuracy of DOS bound water measurements was tested using bound water phantoms with varying bound water fractions. Measurements were compared to 2 gold standard methods: Magnetic Resonance Spectroscopy and conductivity. Diffusion-weighted MRI was also used to characterize the apparent diffusion coefficient of each phantom.
Different concentrations of bound water phantoms were made by mixing varying amounts of gelatin powder (0, 8, 16 and 20g for figure 2 and 4, and 0, 6, 9,15 and 21g for figure 3) from porcine skin (G2500-1KG, Sigma-Aldrich) with intralipid (Liposyn II, 20%, Abbott Laboratories), nigrosin (Sigma-Aldrich) and distilled water. The gelatin powders, 30ml intralipid, 1.5ml nigrosin (511mg/1L stock solution) were dissolved in 300ml water at 37°C and incubated at 37°C while stirred with a magnetic stir bar for 22–24 hours. Then the fully dissolved phantoms were cooled and hardened in a cold room at 5–7°C for 22–24 hours (te Nijenhuis, 1997).
All phantoms were measured in semi-infinite geometry three times using broadband DOS with a source-detector separation of 21.5mm. BWI for phantoms was calculated in the same manner as for tissue (described in section 2.2). However, unlike tissue, phantoms do not contain hemoglobin and the main phantom spectral components are a composite of water, lipid, and nigrosin contributions. Consequently, a phantom water spectrum was obtained by subtracting the complementary phantom spectral components, (i.e. lipid and nigrosin) from the composite phantom spectrum. The employed extinction coefficient spectrum of nigrosin was measured in a cuvette using a spectrophotometer. The resulting phantom water spectra were compared to pure water spectra (at the corresponding temperature) in order to obtain phantom BWI. The temperature of the phantoms was measured before and after each measurement using a thermistor. This allowed us to correct for possible temperature effects in phantoms and measure water spectral changes only due to the bound water.
Proton MR spectroscopy (MRS) was performed on a 3.0T (e.g., 127 MHz) scanner (Achieva; Philips Medical System, Cleveland, Ohio) with a birdcage head coil. Axial images using a T1-weighted spin echo sequence (repetition time (TR)/echo time (TE) = 1100/12 ms) was used for placing the volume of interest (VOI) in the phantoms before MRS. The voxel size was 2 × 2 × 2 cm3 in the phantom. For localization, MRS spectra were obtained using a point-resolved spin-echo sequence (PRESS) (Bottomley, 1987). Shimming was performed automatically on the water resonance for the optimization of homogeneity in each VOI. A fully relaxed, unsuppressed spectrum (e.g., 32 averages and 1024 data points) was acquired in order to measure the amplitude of the water peak in the localized volume. For data analysis, no line broadening was applied before Fourier transformation. Manual zero and first-order phase corrections without baseline correction were applied and water peak integration was performed on processed spectra using jMRUI (Naressi et al., 2001) version 3.0 package.
For the T2 relaxation time measurement, TR of 2000 ms and TE of 60, 120, 240, 480 and 960 ms were used. The T2 values of water were calculated using the following equation
where Ms denotes the signal intensity of the measurement at a given TE and M0 denotes the signal where TE = 0.
The T2 relaxation time with increasing gelatin concentration was fit to a mono-exponential curve. A mono-exponential decay with increasing protein amount is expected from a two-state model with rapid exchange between bound and bulk water in phantoms with protein (Fukuzaki et al., 1995; Lambelet et al., 1988; Moser et al., 1996; Oakes, 1976a). The relationship between T2 relaxation rate (R2 = 1/T2) and BWI was also investigated since a linear relationship between the proton relaxation rate and protein concentration is predicted in this model.
A four electrode conductivity measurement cell was employed to measure bound water phantom conductivity. Gelatin was hardened inside the conductivity cell for its complete contact to each electrode. After the current was measured at both longitudinal ends of the conductivity cell, the current was used to calculate electrical resistance of the gelatin using two electrodes in the middle of the cell. Then, the conductivity of the phantom was calculated by dividing the length between the two electrodes in the middle by the electrical resistance times cross sectional area of the cell. (σ=l/(R·A), σ: conductivity, l: length between the two electrodes in the middle, R: electrical resistance of the gelatin, A: area of the cell). Each phantom was measured two times.
The ADC of water is obtained from diffusion weighted MRI, an imaging technique that measures the mobility of water in tissues and is sensitive to such parameters as cell organization and microstructure. Details of the technical approach for diffusion weighted MRI used in this research are reported in the paper of Patridge et al (Partridge et al., 2001). Images were acquired using a 1.5T GE Signa scanner (GE Medical Systems, Milwaukee, WI) with a bilateral phased-array breast coil (MRI Devices Corp., Waukesha, WI). Employed imaging parameters were TE 49.5 ms, TR 5348 ms, flip angle 90°, 0.5 NEX (number of excitations), 6-mm slice thickness, pixel size 1.56 × 1.56 × 6mm, and a 128 × 128 acquisition matrix. Diffusion weighting factor (b value) used for the data in figure 4 was b= 400 seconds/mm2. Diffusion weighted images were fit on a pixel-by-pixel basis to equation (3).
Equation (3) relates the decrease in signal intensity observed with increasing b-factor or ADC where S and S0 are the signal intensities in each pixel with and without diffusion weighting.
Broadband DOS measurements were acquired from 18 subjects who had infiltrating ductal carcinoma (IDC, N=16) and infiltrating lobular carcinoma (N=2). The mean subject age was 51 ± 9.6 with a range of 37 to 65. The average tumor size, which was taken to be the maximum tumor dimension, was 35±24 mm, with a range of 13 to 100 mm. Linear reflectance linescans were performed in all subjects both over the region of the tumor and in the same location on the contralateral normal side. The details of the linescan have been described in Cerussi et al (Cerussi et al., 2006). The linescan location was chosen based upon a priori knowledge of the tumor location from standard X-ray mammography or ultrasound. We assumed that the contralateral side of the lesion breast was normal tissue unless otherwise indicated in the final pathology report. Linescans were performed with 10-mm increments using a source–detector separation of 29 mm. We kept the handheld probe contact gentle and constant as much as possible. In related studies we examined the impact of probe contact pressure on our measurements. BWI varied 1.2±0.8% (N=3 subjects) when the normal contact pressure was increased by 50% (data not shown). This pressure difference required substantial force and is significantly higher than the pressure applied during patient measurements. Consequently, under normal use conditions, the impact of probe pressure variation on BWI is expected to be negligible.
In order to compare normal and malignant tissues, we calculated an average value of a measured parameter from all locations on a line on the normal breast and an average value of three continuous peak points of BWI on the malignant tumor on the contralateral lesion breast.
The BWI of bound water phantoms with various amounts of gelatin were measured (figure 2). BWI increased linearly with gelatin concentration (R=0.98). Each phantom was measured three times on slightly different locations, and the measurement variation in BWI is also plotted. The offset of BWI at 0g gelatin concentration is likely due to water bound to intralipid components.
Phantoms made by the same fabrication method were measured with MRS and conductivity cell in order to validate the DOS-measured bound water dependence in the bound water phantoms. The T2 relaxation times are shown in figure 3 (a). The water peak (127.8 MHz) was broader in bound water phantoms than a liquid intralipid phantom without gelatin (not shown). As the gelatin concentration increased, the T2 relaxation times decreased with R2=0.996. Each T2 relaxation time was measured once per phantom. Similar trends have been observed by Fukuzaki et al (Fukuzaki et al., 1995) who showed exponential decreases of T2 relaxation times as the amount of albumin increased. The relationship between T2 relaxation rate and BWI is shown on figure 3 (c). R2 had high correlation with BWI (R=0.96).
The conductivity of each bound water phantom was measured twice using a four electrode conductivity measurement cell as described in 2.3.4. The conductivity increased linearly as the gelatin concentration increased (R=0.95) as shown in figure 3(b). The BWI and conductivity of the phantoms had a high linear correlation (R=0.99) (figure 3(d)).
The ADC of water from the bound water phantoms was compared to BWI (figure 4). The phantoms in figure 2 were measured after 10 days using diffusion-weighted MRI as described in 2.3.5. An inverse linear correlation between the ADC and BWI (R=−0.97) was obtained. We confirmed that the BWI of the phantoms did not change significantly after 10 days (2.25±1.3%). This result suggests that the degree of water molecule diffusion decreased with increasing protein-water binding.
In figure 5, TOI and BWI of both lesion and normal breasts of one of the IDC patients are depicted. This patient (57 years old, post-menopausal) had a 4 × 3.5cm tumor on her left breast according to ultrasound. On the lesion side, the TOI values were higher and the BWI values were lower than the normal side. The higher TOI is due to elevated concentrations of water and deoxy-hemoglobin and reduced lipid in tumor tissues. This trend was found in all 18 subjects. The average value of two (for tumor less than 2cm) or three (for tumor bigger than 2cm) continuous BWI peak points in the window in figure 5 was used to calculate BWI of tumor tissues, and it coincides with the location of the lesion. In general, the TOI and BWI values represent an average measurement of tumor and normal tissues of the measured tissue volume (Cerussi et al., 2006).
BWI values from malignant (1.96±0.3) and normal tissues (2.77±0.47) in 18 subjects compared using the Wilcoxon ranked-sum test (figure 6) were significantly different (p<0.0001). Malignant tumors had lower BWI than normal tissues, suggesting that tumors appear to have more free than bound water. The relationship between bulk water concentration and BWI of normal and malignant tissues was also studied (figure 7). In general, BWI is inversely proportional to tissue water content. In normal tissues, the water content range is relatively narrow (10~28%) and the BWI spans between 2.3 and 4.3. However, in cancer tissues, the bulk water content values are high and widely spread (22~73%) while the BWI is clustered between 1.4 and 2.4.
Based on previous findings that correlate water content with histopathological grade (Cerussi et al., 2006), figure 7 suggests that BWI may provide additional insight into tumor invasiveness. In order to explore this relationship further, we examined the Nottingham-Bloom-Richardson scores (Nottingham scores) of breast cancer tissues. The Nottingham-Bloom-Richardson grading system uses objective criteria to assess three main features; tubule formation, nuclear pleomorphism and mitotic count. A histopathologic grade is given on a relative scale of 3–9, where 9 corresponds to the highest grade tumor.
Figure 8 shows that the BWI decreased with increasing Nottingham score (R=−0.96, p=0.002). The black squares are the average values of BWI for each score. There was only one subject for Nottingham score 3. The solid line represents a linear fit to the data points.
We developed an optical bound water index (BWI) from quantitative tissue absorption spectra in the NIR. In order to validate the accuracy and sensitivity of BWI for water state measurements, bound water phantoms were fabricated and tested using MRS and a conductivity measurement cell.
Figure 2 demonstrates that BWI increases linearly with phantom gelatin content (R=0.98). Since gelatin fraction was the only variable in all phantoms, we conclude that BWI measures water disposition changes caused by gelatin.
In order to confirm that the phantoms indeed have protein bound water, each was measured using MRS and a conductivity measurement cell. T2 relaxation time is strongly affected by the number of water binding sites in tissue (Fullerton and Dornbluth, 1982; Moser et al., 1996). Many studies have shown the ability of T2 for measuring water mobility in biological tissues such as breast, brain and lung (Assaf et al., 1997; Brauer, 2003; Shioya et al., 1997; Tan et al., 2008). The T2 value increases as freely mobile water increases. In figure 3(a), T2 relaxation time decreased mono-exponentially as protein concentration increased. This result is due to reduced excitation of nuclei in bound states and consequent fast recovery. Also the mono-exponential decrease of T2 values satisfies a two-state model with a rapid exchange between the bound and bulk water (Fukuzaki et al., 1995; Lambelet et al., 1988; Moser et al., 1996; Oakes, 1976a). T2 values smaller than T1 relaxation times (less than half of T2 values, not shown) signify the prevalence of heterogeneous water-proton environments in our phantoms, which is associated with bound water molecules (Odajima, 1959; Sasaki et al., 1960). High correlation between T2 relaxation rates (R2) and protein concentration has been reported in other studies (Lambelet et al., 1988; Moser et al., 1996; Oakes, 1976a) which is also from the two-state model. Thus, the high correlation between R2 and BWI (figure 3(c)) suggests that BWI measures macromolecular bound water accurately.
Careri et al. have observed increasing conductivity with increasing bound water fraction as shown in figure 3(b). By analyzing the relationship between dielectric constant and conductivity, they found that the protons involved in the conductivity at 100kHz are not the free hydrogen ions of the solvent, but the ions bound to ionizable groups of proteins (Careri et al., 1985). The high correlation between BWI and conductivity shown in figure 3(d) further demonstrates the reliability of BWI as an index for bound water fraction.
DW-MRI has been used to measure water state in in-vivo tissues using apparent diffusion coefficients (ADC) of water. In figure 4, ADC and BWI from homogeneous phantoms demonstrated an inverse correlation suggesting that as more water is bound to macromolecules the degree of water diffusion decreases. Based on our clinical data where the BWI decreased in tumor tissues (figure 5 and and6),6), this would suggest that ADC and BWI are complementary and ADC should increase in tumors. Interestingly, previous MRI studies have demonstrated lower ADC in malignant tumors compared to both normal tissue and benign tumors (Englander et al., 1997; Guo et al., 2002; Partridge et al., 2001). However, in these studies the ADC is typically measured on regions of interest (ROI) carefully drawn to exclude necrotic and cystic areas. If any of the necrotic, cystic or fat regions are included, ADC values change significantly. Paran et al.(Paran et al., 2004) acquired heterogeneous diffusion parameters on large, progressing tumors due to partial and full necrotic loci spread throughout the tumor tissues. It is challenging to select an ROI that only represents the cellular part of a tumor due to the low resolution of DW-MRI images. In addition, the diffusion of free water can be obstructed by the presence of macromolecules and fibrous structures unbound to water (Callaghan et al., 1993; Cho et al., 1996; Nusbaum et al., 2000; Putz et al., 1992). Thus, the interpretation of bound water state in tissues using ADC will vary substantially depending on tissue micro-structure and the precise ROI selected. In contrast, BWI represents an average bulk property of the tissue and is a relatively simple measurement to acquire and interpret.
In normal well-differentiated tissues, we expect a relatively narrow range of BWI and water content values due to the absence of significant disorder and the expectation that normal tissues have an intrinsic maximum water content and binding capacity (figure 7). With the appearance of cancer and loss of differentiation, water content increases dramatically and the tissue structural diversity increases. This phenomenon is communicated by the wide range of water content values and reductions in BWI seen in figure 7.
Overall, significantly lower BWI was observed in tumor vs. normal tissues (figure 5–7). This indicates that the water associated with tumor tissues is more like free water. This could be from necrotic regions in the tumors where free water can fill. The measured free water increase might also be related to alterations in extracellular matrix (ECM). Hyaluronan (or, hyaluronic acid, HA), one of the components of the ECM, is a large, negatively charged polysaccharide that participates in transducing signals in proliferating and migrating cells, and it is closely related to tumor growth (Vignal et al., 2002), metastasis (Hayen et al., 1999), increased drug resistance (Baumgartner et al., 1998), cellular invasiveness (Toole, 2001) and angiogenesis in malignant tumors.(West et al., 1985) In breast cancer patients, high levels of HA in the stroma are associated with low survival rate (Auvinen et al., 2000; Vignal et al., 2002). HA has the ability to retain water and the meshwork structure exerts swelling pressure because of increased mutual repulsion between and within HA molecules (Toole, 2004). Therefore, the figure 8 finding that water binding state correlates inversely with tumor histopathological grade maybe related to the increased presence of HA and necrotic regions in malignant tumors. This inverse correlation between BWI and tumor grade complements our previous observation that overall water content positively correlates with histopathological scores (Cerussi et al., 2006). BWI appears to convey additional information about tissue pathological changes specifically related to the molecular disposition of water, possibly due to alterations in cellularity and extracellular matrix. Although preliminary, these results suggest that both water content and disposition as measured by DOS may provide clinical prognostic information related to metastatic potential and therapeutic outcome
In conclusion, we provide evidence that water state is communicated effectively by the optical measurement of BWI. The accuracy of BWI as a water state index has been validated in bound water phantoms by comparing broadband DOS to MRS and a conductivity measurement cell. Breast cancers were found to possess significantly more free than bound water and BWI correlated inversely with tumor histopathological grade. These results highlight broadband DOS sensitivity to tissue water content and state, and underscore the potential for BWI to be used as a complementary index to MRI relaxation time and ADC measurements.
This work was supported by the National Institutes of Health under Grants P41-RR01192 (Laser Microbeam and Medical Program: LAMMP) and U54-CA105480 (Network for Translational Research in Optical Imaging: NTROI); and the Chao Family Comprehensive Cancer Center (P30-CA62203). This work was also supported by programmatic support by Beckman Foundation and Medical Free Electron Laser program (MFEL). Early part of this work was also supported by International Research Internship program by the Korea Science and Engineering Foundation (KOSEF) and the Korea Research Foundation (M07-2004-000-10161-0). The authors thank the patients who cheerfully volunteered to participate in this study. We also thank Montana Compton for her assistance with organizing the clinical trial. So Hyun Chung also thanks Chris Raub for the help for finding the bound water phantom recipe. We specially thank Dr. Enrico Gratton for the discussion related to this research.