Two reference phantoms containing 0.1 M NaCl and 10 mM NaCl in 4% agarose-Na were glued to a plastic animal platform. The 4% agarose-Na phantom was the brightest phantom on the IR image and the NaCl phantom was brightest on the SQ image. For image quality purposes, IR images and SQ images were normalized to the brightest phantom within each slice for comparison using NIH Image J software.

For sensitivity of intracellular sodium MR image intensity, a series of intracellular sodium phantoms was created by dissolving NaCl into four concentrations of agarose (10, 20, 30, and 40% w/v in saline water) for six different NaCl concentrations (5, 20, 50, 100, 120, 150 mM). KCl was added with NaCl to maintain a constant cationic molarity of 150 mM and constant response to radiofrequency coil. The samples were kept at a constant temperature (25°C) and pH (3.54).

Sodium concentrations in tumors were measured by relative sodium concentrations of phantoms placed next to tumor in animals. Mean sodium signal intensity I_k and sodium concentration per kilogram wet weight in region of interest can be represented as:

I_k = A[1 - B exp(-TR/T1)]     (1)

where a and b are constants and depend upon signal intensities I₁ and I₂ and sodium concentrations C₁ and C₂ in both intracellar and extracellular sodium phantoms as following:

where R₁ and R₂ are sensitivity factors. SF₁ and SF₂ are saturation factors 2.

To propagate the tumor in rats, 42 to 50 days old 30 Sprague Dawley rats were obtained from Harlem Sprague-Dawley (Indianapolis, IN) and the 15 rats were injected i.p. with N-methyl-N-Nitrosourea (MNU), at a dose of 50 mg/kg body wt., purchased from (Ash Stevens Inc., Detroit). After injection, the animals were weekly monitored for tumor propagation and tumor development. First palpable tumor was observed and recorded. The animals were fed 4% Purina Chow diet ad libitum and free access to water. The lesions varied in the 0.2 – 1.0 cm range (stage I), presence of cyst, scar, or solid, benign tumor or lesions in the range of 1.0–3.0 cm. (stage II), and malignant or lesions more than 3.0 cm (stage III) 4. For comparison, 2 animals were injected placebo water in similar way instead of MNU at sham site. No injection was given to 3 animals and these served as control.

Taxotere (40 mg/ml) was dissolved in the 20% ethanol diluent to a working solution concentration of 6 mg/ml. Taxotere dose levels were 6.0 mg/kg body weight. For it, Taxotere was dissolved in supplied diluent to prepare working solution 100–150 μL and slowly injected into a femoral vein of rats anesthetized by intradermal injection of 1.25 cc of a mixture (Ketamine 10 mg/ml + Xylazine 0.5 mg/ml) at the rate of 1.25 ml/hr (15 mg/hr Ketamine; 0.75 mg/hr Xylazine). Each animal was injected 6 mg/ml Taxotere (i.e. ~1.5 mg/250 gram body weight) calculated from standard equations which convert weight to surface area for small mammals 5.

When the tumors were developed approximately 2.0 – 2.5 cm in diameter, the tumor bearing, sham and control animals were treated with Taxotere and sodium imaging was done and repeated following 24 hours after injection. For MRI imaging, animals were anesthetized as mentioned above. In units of animal weight, it was a loading dose of 37.5 mg/kg wt for Ketamine and 1.875 mg/kg for Xylazine. This i.p. injecting dose was sufficient to maintain anesthesia in animals for 1 hour.

All imaging experiments were performed on a high field (4.23 Tesla) whole body clinical MRI system at the Columbia University Hatch NMR Center. A small quadrature birdcage radiofrequency coil (50 mm internal diameter, Morris Inc.) and a high strength gradient insert coil (30 mT/m, Bruker model G-33) were employed in this study to achieve 0.5 × 0.5 × 1.0 cubic mm spatial resolution and 1.0 mm in plane resolution.

The anesthetized rats were aligned in a deep groove with the tumor positioned through a small elliptical opening in the plastic holder, thus maintaining the same relative rat eye position to the phantoms from experiment to experiment. 24 slices were acquired for 3D gradient-echo single quantum image (acquisition time 15 min) and subsequently, an inversion recovery image (acquisition time 45 min). Acquisition parameters were: TR = 100 msec, TE = 5.6 ms, FOV = 40 mm, slice thickness = 2.5 mm, inversion time = 25 msec, and flip angle = 90°, acquisition matrix was 64 × 64 × 8. Both inversion and excitation pulses were non-selective pulses 5.

Analysis of tumor sodium MRI signal intensity change and sodium concentration in all pre- and post- drug treated tumors were measured (in millimoles) relative with corresponding sodium signal intensity in phantom image at its brightest region using NIH J software (see Figure 2). The software puts high quality sodium MRI image sets into common geometrical frame by rotate-translate transform to define region of interest.

Minimum duration of 90 degree Rf pulse with minimum saturation time of preamplifier was used to minimize magnetic inhomogeneity and FID. The image signal intensity depends on both Rf receiver and transmitter coils as:

I_k= R_k.M_0k.|sin (Φ_k)| (5)

where I_k is signal intensity of k^th pixel, R_k is sensitivity at the coil center, M_0k is the equilibrium magnetization. The flip angle Φ_k is function of transmitter gain setting TG as:

Φ_k = (C.R_k).10^TG/20 (6)

(C.R_k) represents relative sensitivity R_k for the k^th pixel. The constant C is independent of spatial position. The receiver sensitivity is directly proportional to transmit field distribution for the coil.

The entire neoplastic explant tumor was removed and sliced into 2 mm thick serial segments starting from the cranial aspect marked by suture and proceeding caudally. A touch preparation of the caudal aspect of each slide was prepared and slides were air dried and fixed with 10% phosphate buffered formalin for Feulgen staining to determine nuclear DNA content. Thereafter slices were fixed by immersion in 10% buffered formalin and embedded in paraffin. For histology, 4 micrometer thick sections were prepared of each segment. The MRI co-registered histology sections were stained for hematoxylin and Eosin (H & E) to determine cytomorphology. Different tumor cytomorphic features were counted as number of micrometer squares per mm² under high power field (HPF) at magnification (× 400) microscope equipped with micrometer squares based on: mitotic rate (number of mitotic figures in micrometer) as mitotic index (MI), amount of apoptosis (20–40 apoptotic nuclei in micrometer square) as apoptotic index (AI), viability (<60% viable cells in micrometer square), cyst space (<100 μm size) and necrosis (<25% necrosis cells in micrometer square) in double-blind manner. The distribution of % extracellular (EC) and intracellular (IC) space was determined with the trichrome stain, where EC appeared light and IC as darker. Malignant cells were characterized according to criteria described elsewhere 4.

The amount of s-DNA content was determined morphometrically using the Feulgen stain to measure the amount of s-DNA per cell nucleus stained using a computer aided image analysis (CAS 200) 11. The S phase was defined as cells with s-DNA content immediate between diploid and tetraploid. On an average, 200 cancer cells were evaluated and their s-DNA content was plotted in histogram form. The histograms were compared to a standard histogram from non-neoplastic cells of the same tumor specimen to determine the standard DNA content for diploid cells (G0/G1) phase of cell cycle and the range of s-DNA content for tetraploid (G2 phase) cells. The proliferation index was expressed as the number of cancer cells in S/G2 phase vs total number of cells (G0/G1 + S/G2) per HPF for each tumor.

The immunoperoxidase enzyme staining technique was used to demonstrate proliferating cells by labeling with KI 67 antigen. Formalin-fixed paraffin-embedded sections were processed for heat retrieval for the KI-67 antigen and then incubated with the primary anti-KI 67 antibody. This antibody was linked to a secondary biotinylated antibody. The binding of the target antigen was demonstrated with avidin-biotin-peroxidase technique using 3, 3' diaminobenzidine as substrate for the peroxidase. It generates a brown reaction product in proliferating nuclei. The non-proliferating nuclei was counterstained green (Methyl-green) and the proliferation index (PI) was evaluated using colorimetric difference on computer aided analysis (CAS 200) as reported earlier 12.

The apoptotic nuclei were visualized on formalin-fixed, paraffin-embedded sections using the antibody MAB3299 single-strand DNA(ss-DNA MAb) using immunogen F7–26 and peroxidase conjugated isotype IgM by Flourescence Activated Cell Sorter (FACS) analysis (Chemicon International Inc. Temecula, CA)13. The formalin-fixed frozen sections were treated with increasing ethanol dilutions. These sections were treated with 0.1 mg/ml saponin-PBS, Protinase K (20 μg/ml PBS) for 20 minutes. Later slices were washed thrice with distilled water and 50% formamide at 55°C. Soon after slices were quenched with endogenous peroxidase (in 3% hydrogen peroxide for 5 minutes) and treated with 100 μl of monoclonal antibody F7–26(1:10) for 15 minutes. After antibody is fixed on slide, these were applied 100 μl peroxidase-conjugated antimouse IgM (1: 200), incubated for 15 minutes and rinsed with PBS and applied DAB chromogen with hematoxylin-counterstaining. These slides were mounted slides for microscopy of MAB 3299 as apoptosis death marker that was independent of internucleosomal DNA fragmentation 13.

Edge detection co-registration method was used to delineate boundaries and locate center point of tumor on MRI images and histologic sections with accuracy of 0.5 mm pixel resolution 14. Semi-automated segmentation was used to distinguish location of different IC-Na signal intensities and corresponding tumor features 14. Different cytomorphology features as histology end points were compared with the total sodium by SQ MRI and [Na]_i by IR MRI signal intensities as following : 1) Ratio of intracellular and extracellular space by trichrome stain; 2) Tumor viability (intracellular space) using proliferation and apoptosis index expressed as number of proliferating and/or apoptotic cells per total number of cancer cells; 3) Cell cycle analysis based on DNA content (Feulgen stain); and active necrosis.

The correlation was analyzed to co-register digitized histology maps with matched sodium MR images at different regions in the tissue to test null hypothesis. The hypothesis was that altered sodium signal intensities do correlate with immunohistology and cytopathologic biomarkers indicating different tumor features. Step down multivariate statistical analysis was performed to determine the order of immunohistolgical and pathologic parameters that correlate with the intracellular sodium (IC-Na) MRI signal 15.

Out of total 30 animals, only 9 MNU induced animals (out of 15 animals) showed total 16 breast cancer tumors. Six animals (out of 15 animals) treated with Taxotere only and three animals served as untreated control group. These control untreated animals did not show any tumor. Other animals served as 3 sham and 3 water placebo treated second control animals. These six pre- and post-treated tumors were analyzed for IR intracellular sodium MRI and SQ total sodium MRI imaging. MRI data were correlated with histopathologic features and immunostaining parameters. Tumors were grouped for quantitative observations as control, obtained from MNU induced animals and compared with 16 tumor tissues obtained from Taxotere treated MNU induced animals.

A series of sodium agarose phantoms, showed distinct sodium MRI signal intensities at varying sodium concentrations as shown in Figure 2 top panel on left. T2 values were measured between the phantom and the extracellular fluid as shown in Table 1 and Figure 2. Figure 2 (panel at bottom on left) represents a plots of visibility as M_o^SQ vs. sodium concentration for all concentrations of agarose used. The linear relationship between visibility as M_o^SQ and M_o^IR using relaxation constants and sodium concentration was observed with change in agarose concentration. The visibility of M_o^IR increases more with % agarose concentration. Sodium MRI signal intensity showed linear relationship with sodium concentration as shown in Figure 2 on right panels.

The inversion recovery (IR) pulse sequence suppressed the signal from Na nuclei with long T1 relaxation constants (free EC Na). For it, the inversion time (the time between the 180° and 90° pulses in the IR pulse sequence) was calculated as (ln 2)(T₁^ex), where T₁^ex was composite longitudinal relaxation time. It generated an intracellular (IC) weighted sodium signal and MRI image. The inversion time selectively suppressed the signal from either long T1 (EC-Na) or short T1 (bound IC-Na) in agarose phantoms at lesser or more than the optimal TI choice as shown in Figure 3. At inversion time TI = 30 msec, the long T1 signal from free EC-Na in NaCl phantom was suppressed completely. Two 1 M NaCl phantoms (with 4% agarose shown as ○ open circles; and, without 4% agarose shown as ● closed circles) were examined during both single quantum MRI acquisition and inversion recovery MRI acquisition for nine different inversion recovery times. Data points represented image intensities while the lines showed theoretical relationships and two null points assuming TI = 27.4 and 43.3 msec for sodium images 5.

Single quantum (SQ) sodium images (Figure 4, panels in rows 1 and 2) showed distinct bright tumor on the SQ images. IR pulse sequence perhaps suppressed the extracellular sodium at optimal inversion time (TI) applied on rat model of breast cancer. Optimum inversion time (TI_opt) was chosen 25–30 msec based on quantitative difference in spin-lattice relaxation constants (T1). At this inversion time, the signal was totally nulled for cyst regions e.g. a tumor sac that was predominantly fluid, showed up darker shown as blue arrows in Figure 4, right panels D, E in rows 3 and 4. It enhanced the visibility of semi-solid or proliferative tumor regions that appeared relatively brighter in comparison to the rest of rat body shown as red arrows in Figure 4, panels F, H, I. It may be presumably due to the higher intracellular (IC) sodium in proliferative regions in tumor vs normal breast tissue. IR sodium images (Figure 4, image panels in rows 3 and 4) showed distinct tumor heterogeneity i.e. cyst, solid tumors but poor tumor boundaries.

Breast tumors overall showed around 2 – 3 times higher intracellular(IC) sodium MRI signal over normal body IC sodium MRI signal intensity in baseline control animals shown in Table 2 and Figure 4 (panels in rows 5 and 6).

Comparison between pre- and post-treatment single quantum (SQ) sodium images (Figure 4, panels in rows 1 vs 2) showed increased size as tumor boundaries but poor tumor heterogeneity. However, the tumor showed up as bright on the post-treatment SQ image. Viable, active proliferative necrosis and apoptosis regions showed up hyperintense on post-treated IR images but unchanged or hypointense on post-treated SQ images. Cysts, extracellular space showed hypointense regions on post-treatment IR images but hyperintense on IR images as shown with arrows in Figure 4 (rows 5 and 6). These enhanced IR signal intensities after Taxotere 24 hour post-treatment were associated with both enhanced apoptotic index (9%) and reduced proliferation index (8.6 vs 2.2; 4 fold decrease over pre-treated tumors) in these regions based on NIH Image J pixel reader counts at different colored locations in post-segmented tumor as shown on Figure 4 (rightmost insert A in row 6). Using calibration graph, the high sodium signal intensities in active regions were measured relative to phantom (P) as shown in Figure 4 and Table 3 showing distribution of sodium in different tumor regions.

Histology and the pentachrome staining measured the ratio of intracellular vs. extracellular space (EC) as shown (red arrow) in the panel on third row with polymorph cells (P) rich with mitotic figures (M) in Figure 6. Pentachrome staining further demonstrated the distribution of extracellular space (EC) shown as lighter and intracellular space (IC) as darker in 54/58 (in 90%) histology slides. The % IC/EC space ratio indicated viable cells and cells at risk or dying cells. Viable cells showed the average ratio '% IC/EC' ratio as 60–80%. It represented the number of viable cells and higher mitotic Index. The dying cells showed higher '% EC/IC ratio' 70% or above with higher Apoptotic Index (AI) of cells possibly in late apoptosis phase. The extracellular space showed cell debris, dying cells with fluids. It contributed minimum to intracellular space. Gross morphology further determined the intracellular contents including collagen, extracellular matrix, cyst fluid (color ranging lighter for cyst to darker for collagen) as shown with arrows in IR sodium MRI image. The comparison of IC/EC ratio with the MRI images is shown in Table 2. Areas comprising less than 30% IC space showed up as dark to hypointense regions whereas areas comprising 30–50% intracellular space showed up hypointense to isointense on IC-Na MRI images. More than 50% IC space (i.e. less than 50% EC) showed up brightest or hyperintense on IR-Na MRI images. Intracellular sodium images (IR-Na) showed medium to bright MRI areas in (36/54 tumor image slices) representing more than 50% IC space. Low intracellular sodium signal intensity in MRI images represented less than 40% IC space in the tumor regions. Control tumors showed the more viable cells and more intracellular space than the post-Taxotere treated tumors by H&E staining.

The apoptotic index was increased marginally in post-treated breast tumors (160 per mm²) vs. control tumors (150 per mm²) that indicated the negative correlation between intracellular sodium signal intensity by MRI and cell viability as shown in Tables 2 and 4. Single cell necrosis showed enhanced IC-Na sodium signal intensity depending on the degree of necrosis while extensive late necrosis showed low SQ Na MRI signal intensity as shown in Table 4 and Figure 6, perhaps due to chemosensitivity effect, ischemia or increased tumor vascular supply. Large areas of late necrosis (low signal intensity) were not observed within 24 hours after initiation of Taxotere chemotherapy in breast cancer.

A "touch slide preparation" of 4 micron thick serial sections by 10% Fuelgen staining (41/48 tumor sites) determined the s-DNA content (in 3 control tumor tissues and 16 Texotere treated tumor tissues) as shown in rightmost panel on third row in Figure 6. The 4 micron thick histology section staining by H&E showed histology features of necrosis, apoptosis, proliferation, mitotic rate (number of mitotic figures per mm²) and amount of apoptosis (number of apoptotic nuclei per mm²) in 36/48 (75%) tumor sites.

Ten explanted Taxotere-treated tumors showed enhanced IR-Na MRI signal intensities by microimaging of tumors at 0 hour and after 24 hours. Central necrosis appeared dark in center and bright peripherypresentation on the IR image as shown in second row on Figure 6. These tumors were uniformly bright on the IR image. Based on histology, these post-treatment tumors showed neoplastic features rich with cyst or absence of cell nuclei, apoptotic cells and proliferation rich necrosis zones as shown in bottom row in Figure 6. An outermost peripheral zone appeared viable tumor tissue. Apoptosis cells were distributed in unspecific manner but apoptosis rich zones were common between viable (V) zones and necrosis (N) rich zones seen in HPF shown in Figure 6 bottom row (see panel a) and Figure 7 (see panels 13).

Under HPFs, the untreated tumors showed high number of mitotic figures (see Figure 6, panel b and Figure 7 panel 1). The mitotic figures measured up to >20 per HPF in the viable zones of tumor. Mitotic figures were altered significantly (p < 0.0001) in the Taxotere-treated 9 tumors (8.5 ± 0.5 per field; n = 160 HPFs) as shown in Figure 7 (see panel 9) and Table 2. All post-treatment tumors showed a significant inverse correlation (p < 0.02) between the reduced mitotic figures and the relative increase in IR sodium signal intensity on IR images.

Pre-Taxotere and post-Taxotere treated tumors showed enhanced SQ- and IR-MRI signal intensities (panels A and B) over the untreated matched tissue (panel A) and sham-control tissue (panel B) on their sodium MRI images as shown in Figure 6. In tumors with small or absent non-viable centers, the IR-MRI images showed uniformly bright regions. The non-viable centers were comparatively smaller than necrosis rich regions. The chemosensitivity effect of Taxotere (enhanced sodium MR image intensity) was evident at histology matched locations of apoptosis (A), neoplastic (B) tumor features rich with cyst or absence of cell nuclei (C), and proliferation rich (D) and necrosis (E) zones as shown in Figures 5 and 6. Taxotere 24 hour post-treatment showed the increased bright area on the IR MRI image as shown in Figure 6 (panels A and B on second row). The IR MRI image in panel C showed different tumor features as dark center indicating late necrosis region (A), hyperintense region indicating apoptosis (B) and lower hypointense region indicating viable cells (C) as shown with red arrows in corresponding post-segmented IR MRI image (A) and its histology matched section. The 24 hours post Taxotere treated tumors showed unchanged IR MRI signal intensities at histology matched necrosis but increased IR MRI signal intensities at histology matched apoptosis showed as yellow arrows on panels A and B (see pre Taxotere and post taxotere images) following the Taxotere drug administration as shown in Figure 6. However, neoplasia and apoptosis were identified by CAS 200 and fluorescent ss-DNA monoclonal antibody staining showed apoptosis cells (next to the viable region) as brightest and the cells closer to the smaller nonviable center were less bright as shown in Figure 6 (red arrows on rightmost panel in third row). The delineation reflected measurement accuracy on histology and matched IR MRI image is shown with red arrows in Figure 6.

Sodium MRI signal intensities showed heterogeneous regions of pre-treated and post-treated tumors corresponding with characteristic pre-malignant and malignant different stages by cytomorphology as shown in Table 4 and Figure 7. However, sodium distinct IR MRI signal intensities and relative IC sodium concentrations using standard sodium phantom data of different tumor stages were evident but not conclusive. Main evident cytomorphic features were apoptosis, necrosis and cysts with characteristic sodium MRI signal intensities. Typical tumor histology features in both pre- and post Taxotere treatment appeared as intraductal proliferation, intraductal carcinoma with cribriform or comedo patterns as shown in Figure 7 (see panels 1 – 8 for pre-treatment tumors and panels 9–22 for post-treatment tumors). Some tumors showed papillary carcinoma, adenoma and neoplasia as shown in Figure 7 and Table 4. IC sodium MRI signal intensities and IC sodium concentrations showed marginal insignificant differences for tumor premalignant and malignant locations confirmed by histology as shown in Table 5 and Figures 7 and 8. However, relationship of drug treatment and tumor histology pre-malignant features was evident but non-specific and limited to one time point.

In co-registered histological digital images and IR-Na sodium images showed % difference ~5% for tumor delineated area as shown in Table 6 and Figure 9 at the top row. Bar diagrams visualized the quick comparison of tumor areas. MRI imaging methods showed correlation with histology (r² = 0.34; n = 6) and s-DNA maps (r² = 0.39; n = 6) as shown in Figure 8 at bottom on right.

Statistical analysis suggested main determinants in the following order: tumor area by histology > tumor area s-DNA map > apoptosis index (immunostaining) > cell viability parameters associated with increased sodium MR signal intensity.