Michael J. Farquharson

Breast tissue classification using x-ray scattering measurements and multivariate data analysis

This study utilized two radiation scatter interactions in order to differentiate malignant from non-malignant breast tissue. These two interactions were Compton scatter, used to measure the electron density of the tissues, and coherent scatter to obtain a measure of structure. Measurements of these parameters were made using a laboratory experimental set-up comprising an x-ray tube and HPGe detector. The breast tissue samples investigated comprise five different tissue classifications: adipose, malignancy, fibroadenoma, normal fibrous tissue and tissue that had undergone fibrocystic change. The coherent scatter spectra were analysed using a peak fitting routine, and a technique involving multivariate analysis was used to combine the peak fitted scatter profile spectra and the electron density values into a tissue classification model. The number of variables used in the model was refined by finding the sensitivity and specificity of each model and concentrating on differentiating between two tissues at a time. The best model that was formulated had a sensitivity of 54% and a specificity of 100%.

Efficient and Rapid Labeling of Transplanted Cell Populations With Superparamagnetic Iron Oxide Nanoparticles Using Cell Surface Chemical Biotinylation for In Vivo Monitoring by MRI

Determination of the dynamics of specific cell populations in vivo is essential for the development of cell-based therapies. For cell tracking by magnetic resonance imaging (MRI), cells need to internalize, or be surface labeled with a MRI contrast agent, such as superparamagnetic iron oxide nanoparticles (SPIOs): SPIOs give rise to signal loss by gradient-echo and T2-weighted MRI techniques. In this study, cancer cells were chemically tagged with biotin and then magnetically labeled with anti-biotin SPIOs. No significant detrimental effects on cell viability or death were observed following cell biotinylation. SPIO-labeled cells exhibited signal loss compared to non-SPIO-labeled cells by MRI in vitro. Consistent with the in vitro MRI data, signal attenuation was observed in vivo from SPIO-labeled cells injected into the muscle of the hind legs, or implanted subcutaneously into the flanks of mice, correlating with iron detection by histochemical and X-ray fluorescence (XRF) methods. To further validate this approach, human mesenchymal stem cells (hMSCs) were also employed. Chemical biotinylation and SPIO labeling of hMSCs were confirmed by fluorescence microscopy and flow cytometry. The procedure did not affect proliferation and multipotentiality, or lead to increased cell death. The SPIO-labeled hMSCs were shown to exhibit MRI signal reduction in vitro and was detectable in an in vivo model. In this study, we demonstrate a rapid, robust, and generic methodology that may be a useful and practical adjuvant to existing methods of cell labeling for in vivo monitoring by MRI. Further, we have shown the first application of XRF to provide iron maps to validate MRI data in SPIO-labeled cell tracking studies.

The use of Compton scattering to differentiate between classifications of normal and diseased breast tissue.

This study describes a technique for measuring the electron density of breast tissue utilizing Compton scattered photons. The Kalpha2 line from a tungsten target industrial x-ray tube (57.97 keV) was used and the scattered x-rays collected at an angle of 30 degrees . At this angle the Compton and coherent photon peaks can be resolved using an energy dispersive detector and a peak fitting algorithm. The system was calibrated using solutions of known electron density. The results obtained from a pilot study of 22 tissues are presented. The tissue samples investigated comprise four different tissue classifications: adipose, malignancy, fibroadenoma and fibrocystic change (FCC). It is shown that there is a difference between adipose and malignant tissue, to a value of 9.0%, and between adipose and FCC, to a value of 12.7%. These figures are found to be significant by statistical analysis. The differences between adipose and fibroadenoma tissues (2.2%) and between malignancy and FCC (3.4%) are not significant. It is hypothesized that the alteration in glucose uptake within malignant cells may cause these tissues to have an elevated electron density. The fibrotic nature of tissue that has undergone FCC gives the highest measure of all tissue types.

Characterization of the depth distribution of Ca, Fe and Zn in skin samples, using synchrotron micro-x-ray fluorescence (SμXRF) to help quantify in-vivo measurements of elements in the skin.

In vivo monitoring of trace and biometals in skin is normally quantified using phantoms that assume a constant elemental distribution within the skin. Layered calibration skin phantoms could potentially improve the reliability of in vivo calibration skin phantoms by better representing the actual in vivo distribution. This work investigates the micro-distribution of iron, calcium and zinc in prepared human skin samples taken from a number of locations on the body. Slices (orientation running from the skin surface into the dermis) were extracted from 18 formalin-fixed necropsy samples and scanned using the micro-XRF setup at the VESPERS beamline (Canadian Light Source). Elemental surface maps were produced using a 6×6 μm(2) beam in steps of 10 μm. Microscope images of histology slides were obtained for comparison. Statistically significant differences (p<0.01) were noted between the epidermal and dermal layers of skin for the elements examined (Ca, Fe and Zn), demonstrating the ability to clearly distinguish elemental content in each layer. Iron was consistently noted at the epidermal/dermal boundary. These results would indicate that when using phantoms to quantify elemental levels measured in the skin, note should be taken of the appropriate depth distribution.

The use of skin Fe levels as a surrogate marker for organ Fe levels, to monitor treatment in cases of iron overload

A system based on the detection of K-shell x-ray fluorescence (XRF) has been used to investigate whether a correlation exists between the concentration of iron in the skin and the concentration of iron in the liver, as the degree of iron loading increases. The motivation behind this work is to develop a non-invasive method of determining the extent of the bodys iron stores via measurements on the skin, in order to monitor the efficacy of chelation therapy administered to patients with beta-thalassaemia. Sprague-Dawley rats were iron loaded via injections of iron dextran and subsequently treated with the iron chelator CP94. The non-haem iron concentrations of the liver, heart and spleen were determined using bathophenanthroline sulphonate as the chromogen reagent. Samples of abdominal skin were taken and the iron concentrations determined using XRF. A strong correlation between the skin iron concentration and the liver iron concentration has been demonstrated (R2 = 0.86). Similar correlations exist for the heart and the spleen. These results show that this method holds great potential as a tool in the diagnosis and treatment of hereditary haemochromatosis and beta-thalassaemia.

Zinc presence in invasive ductal carcinoma of the breast and its correlation with oestrogen receptor status

Zinc is known to play an important role in many cellular processes, and the levels of zinc are controlled by specific transporters from the ZIP (SLC39A) influx transporter group and the ZnT (SLC30A) efflux transporter group. The distribution of zinc was measured in 59 samples of invasive ductal carcinoma of breast using synchrotron radiation micro probe x-ray fluorescence facilities. The samples were formalin fixed paraffin embedded tissue micro arrays (TMAs) enabling a high throughput of samples and allowing us to correlate the distribution of trace metals with tumour cell distribution and, for the first time, important biological variables. The samples were divided into two classes, 34 oestrogen receptor positive (ER+ve) and 25 oestrogen receptor negative (ER-ve) based on quantitative immunohistochemistry assessment. The overall levels of zinc (i.e. in tumour and surrounding tissue) in the ER+ve samples were on average 60% higher than those in the ER-ve samples. The zinc levels were higher in the ER+ve tumour areas compared to the ER-ve tumour areas with the mean levels in the ER+ve samples being approximately 80% higher than the mean ER-ve levels. However, the non-tumour tissue regions of the samples contained on average the same levels of zinc in both types of breast cancers. The relative levels of zinc in tumour areas of the tissue were compared with levels in areas of non-tumour surrounding tissue. There was a significant increase in zinc in the tumour regions of the ER+ve samples compared to the surrounding regions (P < 0.001) and a non-significant increase in the ER-ve samples. When comparing the increase in zinc in the tumour regions expressed as a percentage of the surrounding non-tumour tissue zinc level in the same sample, a significant difference between the ER+ve and ER-ve samples was found (P < 0.01).

X-ray fluorescence measurements of arsenic micro-distribution in human nail clippings using synchrotron radiation

Arsenic (As) distribution in nail clippings from three healthy human subjects was investigated using the microbeam experimental setup of the hard x-ray micro-analysis (HXMA) beamline from the Canadian Light Source (CLS) synchrotron. A pair of toenail and fingernail clippings was collected from each of three subjects (one contributed two fingernail clippings). The fingernail and toenail clippings were embedded in polyester resin and cut in cross-sectional slices with an average thickness of 270 µm. Nine nail clipping cross sections were analyzed from the three subjects. The same method was used to produce five cross sections of nail phantom clippings with concentrations of As ranging from 0 to 20 µg g−1, in increments of 5 µg g−1. These samples were used to produce a calibration line for the As Kα peak. The energy of the x-ray beam was set at 13 keV for optimal excitation of As and the beam size was 28 × 10 µm2. Each sample was analyzed using a point-by-point scanning technique in a 45° beam-sample and 90° beam-detector geometry. The dwelling time was set at 30 s for the human nail clippings and 20 s for the nail phantom clippings, using a step size of 50 µm in both the horizontal and vertical directions for all samples. As concentration for each point was calculated based on the calibration line parameters and the fitted amplitude of the observed As Kα peak. As concentration maps were produced for each nail clipping cross section. The maps show that small regions (<0.1 mm2) with higher As concentrations (>1 µg g−1) are located predominantly in the ventral and dorsal layers of the nail. The results are in agreement with findings reported in a recent study and can be linked to nail histology and keratin structure.

The optimization of an energy-dispersive X-ray diffraction system for potential clinical application

In the past decade, energy-dispersive X-ray diffraction (EDXRD) has been used to identify the nature of tissues. However, these systems have limited clinical use because of problems such as the long measurement times. In this study, the relation between various setup parameters and some performance specifications such as sensitivity, spatial resolution and momentum transfer resolution were assessed using both geometrical calculations and modeling. Accuracy of the derived relations was also confirmed by means of experimental measurements. As an example, the optimum parameters were determined for obtaining diffraction patterns of breast tissue for an efficient acquisition time. Accordingly, the results of this study could introduce a useful tool for EDXRD optimization in clinical application.

The classification of secondary colorectal liver cancer in human biopsy samples using angular dispersive x-ray diffraction and multivariate analysis.

The motivation behind this study is to assess whether angular dispersive x-ray diffraction (ADXRD) data, processed using multivariate analysis techniques, can be used for classifying secondary colorectal liver cancer tissue and normal surrounding liver tissue in human liver biopsy samples. The ADXRD profiles from a total of 60 samples of normal liver tissue and colorectal liver metastases were measured using a synchrotron radiation source. The data were analysed for 56 samples using nonlinear peak-fitting software. Four peaks were fitted to all of the ADXRD profiles, and the amplitude, area, amplitude and area ratios for three of the four peaks were calculated and used for the statistical and multivariate analysis. The statistical analysis showed that there are significant differences between all the peak-fitting parameters and ratios between the normal and the diseased tissue groups. The technique of soft independent modelling of class analogy (SIMCA) was used to classify normal liver tissue and colorectal liver metastases resulting in 67% of the normal tissue samples and 60% of the secondary colorectal liver tissue samples being classified correctly. This study has shown that the ADXRD data of normal and secondary colorectal liver cancer are statistically different and x-ray diffraction data analysed using multivariate analysis have the potential to be used as a method of tissue classification.

An evaluation study of trace element content in colorectal liver metastases and surrounding normal livers by X-ray fluorescence

Background Trace elements are involved in many key pathways involving cell cycle control. The levels of trace metals such as iron, copper, and zinc in colorectal liver metastases have not previously been assessed. Methods The trace element content in snap-frozen cancerous liver tissue from patients who underwent liver resection for colorectal liver metastases was compared with the normal surrounding liver (distant from the cancer) using X-ray fluorescence (XRF). Results X-ray fluorescence was performed on a total of 60 samples from 30 patients. Of these 29 matched pairs (of cancer and normal liver distant from cancer from the same patient) were eligible for univariate analysis. Iron (0.00598 vs. 0.02306), copper (0.00541 vs. 0.00786) and zinc (0.01790 vs. 0.04873) were statistically significantly lower in the cancer tissue than the normal liver. Iron, copper, and zinc were lower in the cancer tissue than in the normal liver in 24/29 (82.8%), 23/29 (79.3%), and 28/29 (96.6%) of cases respectively. Multivariate analysis of the 60 samples revealed that zinc was the only trace element decreased in the cancer tissue after adjusting for the other elements. Zinc levels were not affected by any of the histopathological variables. Conclusion Iron, copper, and zinc are lower in colorectal liver metastases than normal liver. An investigation into the pathways underlying these differences may provide a new understanding of cancer development and possible novel therapeutic targets.