SEARCH

SEARCH BY CITATION

Keywords:

  • polyunsaturated fatty acids;
  • PUFA;
  • breast cancer;
  • Sel-MQC;
  • magnetic resonance spectroscopy;
  • MRSI

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS AND FUTURE DIRECTIONS
  7. Acknowledgements
  8. REFERENCES

The spatial distribution of polyunsaturated fatty acids (PUFA) in healthy and cancerous human breast tissues was measured in vivo with a selective multiple-quantum coherence transfer (Sel-MQC) technique. This method selectively detected the olefinic methylene protons (-CH = CH-) of PUFA at 5.3 ppm that were coupled with allylic methylene protons (-CH2-CH2-CH=) of unsaturated acyl chain at 2.8 ppm. Unwanted lipid coherences and tissue water signal were dephased in a single scan. Breast PUFA were mapped at 1 cm3 voxel resolution in sagittal slices of nine breasts in six healthy female volunteers that were compared to invasive ductal carcinoma (IDC) in one breast cancer patient. The healthy breast tissue displayed continuous PUFA distribution. In some individuals, PUFA appeared throughout the breast tissue; in others they were only located in the central breast area. Decreased PUFA levels were detected in the IDC of the breast cancer patient. The magnetic resonance spectroscopic imaging (MRSI) measurement was consistent with the histological findings, ultrasound, and mammography images. PUFA patterns are sensitive to abnormal breast tissue changes including malignant transformations, and thus may serve as a biomarker for early diagnosis and therapeutic monitoring of breast disease. Magn Reson Med, 2007. © 2007 Wiley-Liss, Inc.

Breast examination and X-ray mammography are currently the major clinical screening techniques for breast cancer. Mammography combined with breast biopsy of suspicious lesions is the standard method of diagnosing breast cancer. Development of other modalities for breast cancer detection has been motivated by the low image contrast and broad variation in mammography interpretations. Since tumor microvascularization plays a critical role in tumor progression, the grade of neoplasm by dynamic contrast-enhanced MRI (DCE-MRI) was accessed through signal enhancement by injecting a contrast agent (gadolinium complexes). Malignant tumors have a fast enhancement and a high washout rate (1), as compared to benign tumors that have a slower enhancement and a much lower washout rate (2). This technique detects breast cancer with a sensitivity of 93–99% (3) and a specificity of 37–86%, which can be further improved by utilizing other methods in addition to DCE-MRI (4). For example, using an architectural interpretation model Nunes et al. (5) dramatically improved the negative predictive value (96%) and the diagnostic specificity (80%). Magnetic resonance spectroscopy (MRS) showed an increased choline signal from phosphocholine in malignant breast cancer (6). Localized single-voxel PRESS (point resolved spectroscopy) or STEAM (stimulated echo acquisition mode) experimental protocols for choline detection achieved a modest tumor detection sensitivity of ≈78%, while the tumor detection specificity was improved to ≈86% (7, 8). Neoadjuvant chemotherapy in a 14-patient trial caused a reduction of choline level in responding patients in 24 hrs, while predicting the clinical outcome of breast cancer treatment (9). However, choline has also been detected in healthy lactating breast tissue, and is not always detectable in malignant breast tumors by 1H MRS (10).

Polyunsaturated fatty acids (PUFA) are a potential biomarker for breast cancer diagnosis. PUFA are involved in regulating tumor p53 proapoptotic signal, antiapoptotic Bcl-2, and superoxide dismutase (SOD) levels, telomere shortening, and tumor angiogenesis (11). Subtle differences in PUFA tissue composition may be a critical factor in affecting the biological function of PUFA. For example, omega-6 polyunsaturated fatty acids (n-6 PUFA) promote tumor development, whereas long-chain n-3 polyunsaturated fatty acids (n-3 PUFA) suppress cancer progression (12). The n-3 fatty acid, eicosapentaenoic acid (EPA), inhibits insulin-mediated protein kinase activity (Akt) and enhances the effects of tamoxifen in growth inhibition of the tamoxifen-resistant estrogen receptor α-positive (ERmath image) breast cancer cells (13). Tumors treated with PUFA gamma-linolenic acid regress without any significant side effects (14). In high-resolution magic angle spinning (MAS) experiments the tissue concentration of PUFA in liposarcoma biopsy samples correlated with tumor mitotic activities (15). PUFA were detected in vivo in breast cancer in single-voxel using 2D MRS (16). The MRS-visible fatty acids have been used to distinguish benign lesions from invasive cancer ex vivo in various organs using biopsy specimens (17). In this article we report a successful mapping of the PUFA distribution in healthy and cancerous human breasts with the in vivo Sel-MQC (selective multiple-quantum coherence transfer) method (18). Spatial distribution patterns of PUFA in human breast tissues are highly individual and sensitive to abnormal breast tissue changes. This corroborates cell culture studies that show PUFA changes occurring in the early stage of cancer development (19).

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS AND FUTURE DIRECTIONS
  7. Acknowledgements
  8. REFERENCES

MRI/MRSI experiments were conducted on a 2.1T Bruker Avance (Billerica, MA) human spectrometer (1 meter bore) operated at the 1H resonant frequency of 89.69 MHz. The spectrometer was equipped with actively shielded magnetic field gradients in three orthogonal directions, with a maximum strength of 6141 Hz/cm (1.44 × 10−2 T/m). The Sel-MQC method (Fig. 1) was used to selectively detect the olefinic methylene protons (-CH = CH-) of the PUFA at 5.3 ppm, which was J-coupled with the allylic methylene protons (-CH2-CH2-CH=) of unsaturated acyl chain at 2.8 ppm. The first 90° slice-selective sinc pulse (4 ms, 3-lobe) was used to excite the protons. During the multiple-quantum (MQ) preparation period (τ = 1/(2J)) between the first 90° pulses, the excited olefinic methylene protons at 5.3 ppm evolved into an antiphase magnetization due to spin–spin coupling with the allylic methylene protons at 2.8 ppm. The second 90° sinc pulse (1-lobe, 10 ms) that was applied at 2.8 ppm selectively converted the antiphase magnetization into the zero-quantum (ZQC) or double-quantum (DQC) coherences. The 180° sinc pulse (1-lobe, 10 ms) that was applied at 5.3 ppm interconverted the ZQC and DQC during the MQ-evolution period t1 between the second and the last 90° pulses. Because MQ coherences (MQC) cannot be measured directly under normal experimental conditions, the DQC was converted to a detectable single-quantum (SQ) antiphase magnetization by a final 90° sinc pulse (1-lobe, 10 ms) at 2.8 ppm. An in-phase coherence transfer echo was formed subsequently at τ'=1/(2J) – t1. The ZQ [RIGHTWARDS ARROW] DQ coherence transfer pathway (CTP) from the excited proton coupling network of PUFA was selected by a gradient combination of g1:g2:g3 = 0:–1:2. This suppressed water and other undesirable lipid signals in a single scan (18). In this experiment, the three highly frequency-selective 10 ms 1-lobe sinc pulses (100 Hz FWHH) were critical for PUFA editing without baseline distortions.

thumbnail image

Figure 1. The Sel-MQC pulse sequence and coherence transfer pathways (CTP) to map PUFA in breast tissue. The olefinic methylene protons (-CH = CH-) in PUFA at 5.3 ppm were selected by multiple quantum filter for detection, mediated by spin J-coupling to the allylic methylene protons (-CH2-CH2-CH=) of unsaturated acyl chain at 2.8 ppm.

Download figure to PowerPoint

A butterfly transmit/receive RF coil (Fig. 2) was constructed in-house to transmit RF power and receive proton signals. Four parallel coil loops were formed into a butterfly shape, with two nested copper wire loops bent and placed symmetrically on each side of the breast, covering the entire organ. The coil opened toward the chest wall for breast access, producing a B1 field perpendicular to the B0 direction. An 8-cm diameter glass ball-shaped flask containing corn oil was prepared to test the Sel-MQC sequence and B1 field homogeneity of the butterfly coil (Fig. 3a). The spatial distribution of PUFA from a 1-cm sagittal slice was mapped in 32 × 32 phase encoding steps for FOV = 16 cm (Fig. 3b). A slice-selective gradient of 4.262 ms was applied simultaneously with a 4 ms sinc (3-lobe) slice-selective pulse followed by a slice-refocusing gradient of 1.3 ms. Immediately after this gradient pulse a phase-encoding gradient was applied with a duration of 3.3 ms. MQ preparation time was τ = 65.242 ms. During the MQ evolution time (t1 = 16 ms) an MQ-labeling gradient g2 with a duration of 5 ms and amplitude of 1.3 × 10−2 T/m (90% of the maximum gradient) was applied. The coherence transfer echo formed at 44.242 ms from the last 90° pulse (18). The MQ coherence refocusing gradient g3 also had the same amplitude, but twice the duration (10 ms). The data acquisition period started at 14.02 ms after the last 90° pulse and lasted 172.5 ms. A 5 ms t1-crusher gradient was applied after the second 90° pulse with an amplitude of 0.5% maximum gradient (7.2 × 10−5 T/m) (18). The 2D sagittal Sel-MQC CSI spectral map of PUFA distribution was processed after 3D Fast Fourier Transformation (FFT) in chemical shift and phase-encoding dimensions (Fig. 3b). The PUFA map was overlaid on a T1-weighted MRI image (512 × 256) that was acquired from the same slice using the Bruker gradient echo sequence (GEFI_TOMO) (data not shown). Relatively uniform PUFA distribution was obtained from the Sel-MQC experiment, which indicates a suitable B1 homogeneity of the butterfly breast coil for PUFA investigation in the human breast. A proton spectrum was recorded in a single-pulse experiment (Fig. 3c) to compare with the edited PUFA spectrum from a voxel of the processed Sel-MQC CSI spectral matrix (Fig. 3d). The Sel-MQC procedure suppressed the undesirable lipid signals with an excellent spectral baseline for PUFA detection.

thumbnail image

Figure 2. a: The butterfly proton transmit/receive coil constructed for breast MRI/MRSI experiments, and (b) the electrical circuit of the coil. C1 = 15 pF and C2 = 20 pF. Two high-power tunable capacitors C4 (5–25 pF, 10 kV peak working voltage, 7.2A maximum current) (Polyflon MRP-VC25-15) were used to tune and match the L-C circuit to a resonant frequency of 89.69 MHz. A 20 pF capacitor (C3) was used as a matching capacitor to make a balanced circuit. The diameters of the large and small copper loops were 17.5 cm and 14.0 cm, respectively.

Download figure to PowerPoint

thumbnail image

Figure 3. a: The T1-weighted gradient echo image (512 × 256) from a 1-cm sagittal slice of an 8-cm glass-ball phantom containing vegetable oil. TE = 7 ms, TR = 200 ms, and SW = 50,000 Hz. A 2-ms 3-lobe sinc pulse was used for 30° flip angle excitation. Each echo was acquired with 512 sampling points in a single scan. The artifact at the bottom edge of the phantom was due to the nearby copper wire of the coil. b: The 2D sagittal Sel-MQC CSI spectral map of PUFA distribution (32 × 32) acquired from the same slice with 32 × 32 phase-encoding steps (only 16 × 16 spectra were displayed). FOV = 16 cm, TR = 3 sec, SW = 1502.40 Hz. Each MQ-coherence transfer echo was acquired with 512 sampling points. c: A single-scan lipid proton spectrum acquired with 4096 digital points from the entire sample in a one-pulse experiment. SW = 5,000 Hz. d: The MQ-edited PUFA spectrum in voxel (16, 15) of the Sel-MQC CSI spectral matrix displayed in the magnitude mode.

Download figure to PowerPoint

Healthy volunteer subjects and the breast cancer patient were positioned prone (head first) and the tested breast was placed in the sensitive region of the RF coil. Two soft sponges were used to stabilize the breast tissue, minimizing breast movement. The breast tissue was imaged in three orthogonal planes using a Bruker gradient-echo (TRIPLOT) sequence in order to position the breast at the magnet isocenter. The butterfly coil provided excellent signal sensitivity for human breast imaging. All breast PUFA were mapped in the Sel-MQC CSI (chemical shift imaging) experiments at 1 cm3 voxel resolution, with either 32 × 32 or 16 × 16 2D phase encoding procedures for FOV = 32 cm and 16 cm, respectively. A full MQ-coherence transfer echo was acquired in 512 data points immediately after the last refocusing gradient g3. The typical proton spectral width was SW = 1502.40 Hz and TR = 2 sec. The MQ-coherence transfer echo of PUFA was acquired in a single scan without phase cycling. The duration of a 16 × 16 Sel-MQC CSI experiment was ≈10 min. In addition, fat-suppressed, T1-weighted gradient-echo MRI images (512 or 256 × 256) were obtained from the same slice of the breast without contrast agents, typically with TE = 9.6 ms, TR = 300.0 ms, SW = 50,000 Hz, and FOV = 18 cm. In the gradient-echo imaging experiment a 3-lobe sinc frequency-selective RF pulse (excitation bandwidth = 250 Hz) was applied to saturate the lipid signal at 1.3 ppm for fat suppression. In the study of the breast cancer patient the PUFA spatial distributions were mapped from a sagittal slice that contained the invasive ductal carcinoma (IDC) and from a control slice of surrounding tissue in the same breast. The Sel-MQC CSI spectra were processed using the XSOS signal processing software from Columbia University, installed in an IBM Pentium III ThinkPad computer. The MRSI maps of PUFA were compared with the clinical mammography, ultrasound, and pathology reports. The investigation of human subjects was approved by the local Institutional Review Board and informed consent was obtained from each subject.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS AND FUTURE DIRECTIONS
  7. Acknowledgements
  8. REFERENCES

Healthy Volunteer Subjects

Nine breasts from six healthy human subjects were studied to obtain the baseline distributions of PUFA in breast tissue. Each of the healthy breasts gave a unique continuous distribution pattern of PUFA (Figs. 4, 5). A proton spectrum was acquired using the Bruker ONEPULSE sequence to set water on resonance. An example spectrum presented here (Fig. 4a) was acquired from the left breast of a 68-year-old healthy subject. The tissue water at 4.7 ppm and the methylene protons of lipid at 1.3 ppm dominated the proton spectrum. The olefinic methylene protons (-CH = CH-) at 5.3 ppm (arrow) of the unsaturated lipid resonance were not detected due to the strong water resonance at 4.7 ppm under these experimental conditions. However, our Sel-MQC experiment successfully detected the olefinic methylene protons (-CH = CH-) at 5.3 ppm that were coupled to the allylic methylene protons (-CH2-CH2-CH=) of unsaturated acyl chain at 2.8 ppm (Fig. 4b), while filtering out water and unwanted lipid peaks. Two-dimensional phase encoding was carried out in 32 × 32 steps in the Sel-MQC CSI experiment to define a voxel size of 1 cm3. The PUFA distribution from the Sel-MQC (in blue) was overlaid on the fat-suppressed T1-weighted MRI image from the same slice (Fig. 4c). A PUFA image was obtained by displaying the spectral integration of the PUFA spectrum in each voxel (data not shown). In this subject, PUFA were slightly more concentrated in the center of the breast and were continuously distributed across the tissue slice.

thumbnail image

Figure 4. The PUFA map from the left breast of a 68-year-old woman. a: The proton spectrum from a one-pulse sequence acquired in a single scan in 2048 sampling points. SW = 50,000 Hz. b: The PUFA spectrum (in magnitude mode) from the Sel-MQC CSI voxel (15, 13). Voxel size = 1 cm3. TR = 1 sec, SW = 1,502.40 Hz, NS = 1, and 512 data points were acquired for each MQ coherence transfer echo. c: The Sel-MQC CSI maps (32 × 32, FOV = 32 cm) of PUFA were overlaid on a fat-suppressed T1-weighted gradient-echo image (512 × 256) acquired from the same slice (FOV = 18 cm). Water and unwanted lipid peaks were completely suppressed in a single scan.

Download figure to PowerPoint

thumbnail image

Figure 5. Representative 2D Sel-MQC CSI maps of PUFA obtained from two healthy female volunteers. A Sel-MQC CSI data matrix (32 × 32) was acquired from a 1 cm sagittal tissue slice in the (a) right breast of a 63-year-old woman, (b) left breast of a 40-year-old woman. The PUFA maps were superimposed on the corresponding fat-suppressed T1-weighted gradient echo images from the same slice. c: The 2D Sel-MQC CSI maps of the discontinuous patterns of PUFA from a 37-year-old woman who complained of right breast pain. A 1 cm sagittal slice of the right breast was mapped in 32 × 32 phase-encoding steps. FOV = 32 cm. The PUFA CSI spectra were overlaid on the fat-suppressed T1-weighted gradient echo image, revealing differentially reduced PUFA levels in the breast regions highlighted by the hyperintensive T1-weighted MRI imaging areas. d: A transverse view of the hyperintensive area in the T1-weighted MRI image of the right breast. Only one of the 16 slices acquired in a 3D gradient echo imaging sequence is displayed. Slice thickness = 10 mm / 16 = 0.625 mm, FOV = 18 cm.

Download figure to PowerPoint

There were instances of PUFA distribution throughout the breast, as observed in both breasts of the 63-year-old woman when overlaying the PUFA spectra (in blue) on the fat-suppressed T1-weighted images of the breasts (Fig. 5a). In other instances, however, PUFA were only distributed in the center of the breast area and appeared with high ductal glands, as observed in the left breast of a 40-year-old woman (Fig. 5b) and the right breast of a 66-year-old woman (data not shown). The PUFA patterns measured in human subjects were not due to instrumental artifacts, considering the homogeneous PUFA distribution pattern detected in the phantom of isotropic corn oil (Fig. 3b). Thus, the Sel-MQC CSI experiments can be used to measure the distributions of PUFA in human breast tissue.

Different patterns of noncontinuous PUFA distribution were observed in the breasts of a 37-year-old woman complaining of pain in her right breast. Prior to the start of this study, this subject had three breast cancer screening tests for possible abnormality, as requested by her physician. However, neither the clinical ultrasound, mammography, nor clinical breast exams detected any malignancy or benign breast tissue changes. In the sagittal view of the Sel-MQC spectra of PUFA superimposed on the fat-suppressed T1-weighted MRI image from the same slice (Fig. 5c), PUFA spectra were missing in some voxels (arrow), which can be recognized by counting PUFA resonances (red spectra) along the columns of the Sel-MQC CSI spectral matrix. Reduction of PUFA occurred in the hyperintensive fat-suppressed T1-weighted image area, as verified by the coronal view of the hyperintensive fat-suppressed T1-weighted breast MRI region (Fig. 5d). The hyperintensive T1-weighted areas may reflect possible fibrostic changes in the breast tissue; however, tissue specimens were not available to examine the corresponding histopathological changes. Breast compression between two sponges gave a rectangular-shaped coronal breast image. MRSI examination of this subject's left breast displayed a PUFA pattern different from the right breast (data not shown). It appears that PUFA distribution is sensitive to detecting nonmalignant human breast tissue changes. The PUFA in the right breast originated mainly from the adipose tissue areas, which became more evident with the corresponding T1-weighted sagittal image of the right breast tissue slice without fat-suppression (data not shown). Later, a similar case of discontinuous PUFA distribution pattern was observed in the right breast of a 66-year-old woman, whose left healthy breast gave a different continuous PUFA distribution pattern (data not shown). Voxels with a reduced PUFA level were found in the Sel-MQC map in the area of hyperintensive T1-weighting, as observed when the Sel-MQC PUFA spectra were overlaid on the fat-suppressed T1-weighted MRI sagittal image of the right breast. Five years prior to this study, the physician of this subject had a suspicion of a possible abnormality in the right breast. The subsequent breast exams did not find any evidence of malignancy.

Breast Cancer Patient Study

The PUFA in an IDC of the subject's left breast were investigated and compared with the above baseline PUFA distribution in healthy breast tissue. The breast PUFA were mapped by Sel-MQC CSI procedures from a 1-cm sagittal slice in 16 × 16 phase-encoding steps (Fig. 6a,b). In the fat-suppressed T1-weighted MRI image (256 × 256), the tumor appeared to be bright in the same location of the spectral void in the Sel-MQC map of PUFA (Fig. 6a). In the IDC lesion area of ≈2 cm cross-sectional diameter, there were no detectable PUFA peaks (arrow). For comparison, a control slice (Fig. 6b) in the same breast without the tumor mass was scanned. The control displayed a continuous distribution pattern of PUFA covering the entire breast, similar to that observed in healthy volunteers (Figs. 4, 5). IDC was also detected by clinical mammography, ultrasound, and contrast-enhanced clinical MRI. Mammogram (Fig. 6c) revealed a 2-cm irregular mass (arrow) with an indistinct margin in the left breast at 1 o'clock in the middle depth with pleomorphic calcifications, suggesting malignancy. Real-time ultrasound (Fig. 6d) found a 2.5-cm irregular-shaped lobulated mass (arrow) at 2 o'clock in the anterior depth. In clinical breast MRI exams on a 1.5T GE scanner, fat-suppressed T1-weighted MRI images (data not shown) acquired before and after intravenous injection of 0.1 ml/kg gadolinium chelate also found a tumor of 2.7 × 3.2 × 2.1 cm3.

thumbnail image

Figure 6. a: Abnormal PUFA distribution pattern in the breast cancer patient with an invasive ductal carcinoma. FOV = 16 cm, TR = 2 sec, NS = 1, SW = 2003.21 Hz. Total scan time was ≈10 min. The MRSI of PUFA was scanned from a 1-cm sagittal slice in 16 × 16 phase-encoding steps. FOV = 16 cm. b: A control tissue slice without tumor in the same breast displayed a continuous distribution pattern of PUFA. The Sel-MQC CSI maps of PUFA were superimposed onto the corresponding fat-suppressed T1-weighted gradient echo images (256 × 256). The 3D gradient echo MRI parameters were: FOV 16 cm, SW = 50,000 Hz, slice thickness = 1 cm, number of slices = 16, TE = 7 ms, TR = 50.15 ms. The tumor was also detected by (c) clinical mammography and (d) ultrasound imaging. e: The hematoxylin and eosin tissue staining of the invasive ductal carcinoma in the breast cancer patient from core biopsy. The tumor was (f) estrogen receptor-positive (ER+) and (g) progesterone receptor-negative (PR) from immunohistochemical staining.

Download figure to PowerPoint

A core biopsy tissue sample of the above IDC tumor was obtained prior to the MRSI study. The pathology report diagnosed the tumor as an IDC (Fig. 6e). The immunohistochemical staining of estrogen receptor (ER) and progesterone receptor (PR) using antibodies against these proteins (DAKO, Carpinteria, CA) showed the tumor to be ER+ (Fig. 6f) and PR (Fig. 6g). The immunohistochemical staining with herceptin antibody (DAKO) showed HER2/Neu amplification in the tumor. After the core biopsy the patient had a partial mastectomy and sentinel lymph node excision. The tumor was a poorly differentiated IDC (histologic grade 2) with focal colloid carcinoma (10%), signet ring cells (<5%), and apocrine features. This invasive ductal carcinoma was 1.8 cm in the largest dimension, with three positive lymph nodes and a central acellar area of fibrosis in the primary tumor. It was within 1 mm of inferior margin. The tumor size identified by histological staining was slightly smaller than the mass measured in contrast-enhanced MRI and ultrasound images, but still consistent with the tumor size detected by MRSI when considering the partial volume effect.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS AND FUTURE DIRECTIONS
  7. Acknowledgements
  8. REFERENCES

Lipid unsaturation is sensitive to cell proliferation and growth arrest. These two processes can be differentiated using lipid metabolism related to phosphocholine (PC), glycerophosphorylcholine (GPC), and total choline (Cho). For example, the proliferating MCF-7 and HeLa tumor cells in early-log phase produced increased mobile lipid with high PC and low GPC or Cho, whereas the postconfluence cells gave low PC and high GPC or Cho (20). Proliferating cells growing into a confluence state were accompanied by a gradual decrease of unsaturated fatty acid chains. In cultured tumor cells, rapid tumor growth with a low rate of lipid peroxidation correlated with reduced PUFA. Malignant tumor cells with a low rate of lipid peroxidation were highly susceptible to free radical-induced toxicity (11). This process seemed to occur at the premalignant stage in tumorogenesis (19).

In early NMR experiments of tumor cells and spheroids in culture medium, tumor cells increased mobile lipid signals (21) as a response to the environmental changes, including cell media pH, cell–cell interactions upon cell confluency, and cell exposure to chemical agents (20). When treated by ganciclovir (GCV), the animal gliomas tumor model accumulated mobile lipid, including PUFA, when cells were arrested in the late S- or G2-phase during apoptotic processes (22). These NMR-detectable PUFA signals originated mostly from intracellular lipid bodies, or as argued by some investigators, from the plasma-membrane microdomains with altered membrane fluidity (23). Recent research presents evidence that lipid bodies may have important biological functions in mammalian cells. For example, cyclooxygenase and prostaglandin synthase activities have been observed in cytoplasmic lipid bodies (24). A relatively high level of fatty acid synthase (FASN) expression is often observed in human breast cancer and has been involved in the early stage of tumor development (25).

Although lipid bodies may contribute to the PUFA signal that was detected in our case study of human breast cancer, the decreased PUFA level in the IDC is likely due to lack of cells in the central region of the tumor as well as high cellular density and low adipose tissue content in the other tumor regions. This is consistent with the high water-to-fat ratio often observed in breast cancer (26). Interestingly, clinical MRSI of brain tumors have frequently displayed a high level of saturated lipid signals in the necrotic tumor areas (27). Tumors in other extracranial organs present changed PUFA levels, as already observed in the liposarcoma biopsy samples (15). In this study each subject appears to have a distinct PUFA distribution pattern, which may serve as a reference to characterize the PUFA distribution changes in breast tissues. A clinical trial has been designed to find out whether the detailed changes in PUFA distribution patterns mapped by MRSI can differentiate between benign, malignant, and healthy breast tissues. Using 2D COSY, Thomas et al. (16) detected decreased unsaturated lipids and increased choline in a single voxel of human breast tumors. This is consistent with our multivoxel observation of diminished PUFA in one IDC (Fig. 6). In the tissue biopsy study of human liposarcoma using magic angle spinning (MAS) techniques, Singer et al. (15) reported semiquantitative correlations of the increased level of PUFA with tumor mitotic activities. In most ex vivo biopsy studies and the single-voxel in vivo MRS measurements, however, individual PUFA distribution in a healthy breast was not available as a baseline control. Thus, the in vivo Sel-MQC measurement of PUFA distribution in the human breast reported here provides a simple method of identifying breast tissue changes. We have recently developed the T1- and T2-Sel-MQC sequences that can be applied to measure T1 and T2 spin relaxation times of PUFA for their absolute quantification in vivo (28). Although quantification of PUFA is difficult in 2D MRS, 2D MRS methods provide valuable information of spin coupling topologies of metabolites or drugs.

Further investigations are needed to determine how well the PUFA pattern changes in human breast tissues will correlate with the histopathological conditions of benign and malignant breast diseases. The literature reports a reduction of the delta-6-desaturase in human breast cancer (29). Lack of this enzyme inhibits conversion of the essential fatty acids linoleic acid and α-linolenic acid series into important PUFA such as γ-linoleic acid (GLA), dihomo-γ-linoleic acid (DGLA), eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA) that inhibit cancer cell growth. In an NMR study of an epigenetic rat model fed a choline-deficient diet, a lower degree of lipid double bonds in fatty acyl containing compounds was found to be associated with early-stage hepatocarcinogenesis (30). An abnormal breast PUFA pattern may serve as a potential marker for identifying premalignant and malignant breast lesions and for monitoring therapeutic processes.

CONCLUSIONS AND FUTURE DIRECTIONS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS AND FUTURE DIRECTIONS
  7. Acknowledgements
  8. REFERENCES

Molecular markers are continuously sought after to achieve early diagnosis of cancer. Success in the genetic characterization of cancer permits cancer diagnosis by oligonucleotide probes of the body fluid and stool of cancer patients. However, locating the tumors diagnosed by these molecular signatures, predicting treatment outcome, and monitoring cancer recurrence remain a challenge. In this preliminary investigation we found that the PUFA pattern is unique to each individual and sensitive to breast tissue changes, including cancerous alterations. The healthy breast tissue displayed continuous PUFA distribution. In some individuals, PUFA appeared throughout the breast tissue; in others they were only located in the central breast area. Decreased PUFA levels were detected in the IDC of one breast cancer patient. To define abnormal PUFA changes in disease states, it is important to measure the baseline PUFA distributions in healthy breast tissue. The PUFA distribution in human breasts and breast cancer measured in vivo merits a further investigation as an index of abnormal tumor fatty acid metabolism for cancer detection. Altered PUFA patterns in high-risk women may be monitored for premalignant breast tumor development. PUFA and other potential biomarkers will be investigated for possible early interventions, especially before benign breast tumors and ductal carcinoma in situ (DCIS) convert to malignant tumors or metastasis.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS AND FUTURE DIRECTIONS
  7. Acknowledgements
  8. REFERENCES

We thank Dr. Douglas L. Rothman in the Magnetic Resonance Center (MRC) of Yale University for providing the Bruker 2.1T spectrometer and Carol H. Lee for recruitment assistance. We thank Dr. Dikoma C. Shungu of Columbia University for the generous gift of the XSOS software package and Ms. Xiangling Mao for XSOS technical support. We thank Drs. Edwin M. Nemoto, Robert Andrews, and Kaung-Ti Yung in the University of Pittsburgh for editorial suggestions.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSIONS AND FUTURE DIRECTIONS
  7. Acknowledgements
  8. REFERENCES
  • 1
    Knopp MV, Weiss E, Sinn HP, Mattern J, Junkermann H, Radeleff J, Magener A, Brix G, Delorme S, Zuna I, van Kaick G. Pathophysiologic basis of contrast enhancement in breast tumors. J Magn Reson Imaging 1999; 10: 260266.
  • 2
    Choyke PL, Knopp V, Libutti SK. Special techniques for imaging blood flow to tumors. Cancer J 2002; 8: 109118.
  • 3
    Kuhl CK, Mielcareck P, Klaschik S, Leutner C, Wardelmann E, Gieseke J, Schild HH. Dynamic breast MR imaging: are signal intensity time course data useful for differential diagnosis of enhancing lesions? Radiology 1999; 211: 101110.
  • 4
    Schnall MD, Rosten S, Englander S, Orel SG, Nunes LW. A combined architectural and kinetic interpretation model for breast MR images. Acad Radiol 2001; 8: 591597.
  • 5
    Nunes LW, Schnall MD, Orel SG. Update of breast MR imaging architectural interpretation model. Radiology 2001; 219: 484494.
  • 6
    Katz-Brull R, Margalit R, Bendel P, Degani H. Choline metabolism in breast cancer; 1H-, 13C-, 31P-NMR studies of cells and tumors. Magn Reson Mater Phys Biol Med 1998; 6: 4452.
  • 7
    Roebuck JR, Cecil KM, Schnall MD, Lenkinski RE. Human breast lesions: characterization with proton MR spectroscopy. Radiology 1998; 209: 269275.
  • 8
    Jagannathan N, Kumar M, Seenu V, Coshic O, Dwivedi SN, Julka PK, Srivastava A, Rath GK. Evaluation of total choline from in vivo volume localized proton MR spectroscopy and its response to neoadjuvant chemotherapy in locally advanced breast cancer. Br J Cancer 2001; 84: 10161022.
  • 9
    Meisamy S, Bolan PJ, Baker EH, Bliss RL, Gulbahce E, Everson LI, Nelson MT, Emory TH, Tuttle TM, Yee D, Garwood M. Neoadjuvant chemotherapy of locally advanced breast cancer: predicting response with in vivo 1H MR spectroscopy—a pilot study at 4 T. Radiology 2004; 233: 424431.
  • 10
    Stanwell P, Gluch L, Clark D, Tomanek B, Baker L, Giuffrè B, Lean C, Malycha P, Mountford CE. Specificity of choline metabolites for in vivo diagnosis of breast cancer using 1H MRS at 1.5 T. Eur Radiol 2005; 15: 10371043.
  • 11
    Das UN. A radical approach to cancer. Med Sci Monit 2002; 8: RA7992.
  • 12
    Ge Y, Chen Z, Kang ZB, Cluette-Brown J, Laposata M, Kang JX. Effects of adenoviral gene transfer of C. elegans n-3 fatty acid desaturase on the lipid profile and growth of human breast cancer cells. Anticancer Res 2002; 22: 537543.
  • 13
    De Graffenried LA, Friedrichs WE, Fulcher L, Fernandes G, Silva JM, Peralba J-M, Hidalgo M. Eicosapentaenoic acid restores tamoxifen sensitivity in breast cancer cells with high Akt activity. Ann Oncol 2003; 14: 10511056.
  • 14
    Das UN, Prasad VSSV, Reddy DR. Local application of gamma-linolenic acid in the treatment of human gliomas. Cancer Lett 1995; 94: 147155.
  • 15
    Singer S, Sivaraja M, Souza K, Millis K, Corson JM. 1H-NMR detectable fatty acyl chain unsaturation in excised leiomyosarcoma correlate with grade and mitotic activity. J Clin Invest 1996; 98: 244250.
  • 16
    Thomas MA, Binesh N, Yue K, DeBruhl N. Volume-localized two-dimensional correlated magnetic resonance spectroscopy of human breast cancer. J Magn Reson Imaging 2001; 14: 181186.
  • 17
    Mountford CE, Mackinnon WB, Russell P, Rutter A, Delikatny EJ. Human cancers detected by proton MRS and chemical shift imaging ex vivo. Anticancer Res 1996; 16: 15211532.
  • 18
    He Q, Shungu DC, van Zijl PCM, Bhujwalla ZM, Glickson JD. Single scan in vivo lactate editing with complete lipid and water suppression by selective multiple quantum coherance transfer (Sel-MQC) with application to tumors. J Magn Reson B 1995; 106: 203211.
  • 19
    Benedetti A, Malvaldi G, Fulceri R, Comporti M. Loss of lipid peroxidation as a histochemical marker for preneoplastic hepatocellular foci of rats. Cancer Res 1984; 44: 57125717.
  • 20
    Delikatny EJ, Roman SK, Hancock R, Jeitner TM, Lander CM, Rideout DC, Mountford CE. Tetraphenylphosphonium chloride induced MR-visible lipid accumulation in a malignant human breast cell line. Int J Cancer 1996; 67: 7279.
  • 21
    Rosi A, Grande S, Luciani AM, Barone P, Mlynarik V, Viti V, Guidoni L. 1H MRS studies of signals from mobile lipids and from lipid metabolites: comparison of the behavior in cultured tumor cells and in spheroids. NMR Biomed 2004; 17: 7691.
  • 22
    Hakumäki JM, Poptani H, Sandmair A-M, Ylä-Herttuala S, Kauppinen RA. 1H MRS detects polyunsaturated fatty acid accumulation during gene therapy of glioma: implications for the in vivo detection of apoptosis. Nat Med 1999; 5: 13231327.
  • 23
    Mountford CE, Wright LC. Organization of lipids in the plasma membranes of malignant and stimulated cells: a new model. Trends Biochem Sci 1988; 13: 172177.
  • 24
    Dvorak AM, Morgan E, Schleimer RP, Ryeom SW, Lichtenstein LM, Weller PF. Ultrastructural immunogold localization of prostaglandin endoperoxide synthase (cyclooxygenase) to non-membrane-bound cytoplasmic lipid bodies in human lung mast cells, alveolar macrophages, type II pneumocytes, and neutrophils. J Histochem Cytochem 1992; 40: 759769.
  • 25
    Zhang D, Tai LK, Wong LL, Chiu L-L, Sethi SK, Koay ESC. Proteomic study reveals that proteins involved in metabolic and detoxification pathways are highly expressed in HER-2/neu-positive breast cancer. Mol Cell Proteom 2005; 4: 16861696.
  • 26
    Jagannathan NR, Singh M, Govindaraju V, Raghunathan P, Coshic O, Julka PK, Rath GK. Volume localized in vivo proton MR spectroscopy of breast carcinoma: variation of water-fat ratio in patients receiving chemotherapy. NMR Biomed 1998; 11: 414422.
  • 27
    Tzika AA, Zarifi MK, Goumnerova L, Astrakas LG, Zurakowski D, Young-Poussaint T, Anthony DC, Scott RM, Black PM. Neuroimaging in pediatric brain tumors: Gd-DTPA-enhanced, hemodynamic, and diffusion MR imaging compared with MR spectroscopic imaging. Am J Neuroradiol 2002; 23: 322333.
  • 28
    Muruganandham M, Koutcher JA, Pizzorno G, He Q. In vivo tumor lactate relaxation measurements by selective multiple-quantum-coherence (Sel-MQC) transfer. Magn Reson Med 2004; 52: 902906.
  • 29
    Lane J, Mansel RE, Jiang WG. Expression of human delta-6-desaturase is associated with aggressiveness of human breast cancer. Int J Mol Med 2003; 12: 253257.
  • 30
    Tesiram YA, Saunders D, Towner RA. Application of proton NMR spectroscopy in the study of lipid metabolites in a rat hepatocarcinogenesis model. Biochim Biophys Acta 2005; 1737: 6168.