Fay A. Marks

General Electric Corporate Research and Development, One Research Circle, Niskayuna, NY 12309

Received 11/25/97 Accepted 12/5/97

3. MATERIALS AND METHODS

For studies of this kind, where an indirect marker of pathology is sought, the ultimate goal would be to use models which mimic the In vivo physiological conditions under investigation. The experimental nature of the hunt in this case precluded human studies, so the samples used in this study are tissue specimens which represent, as nearly as possible, the metabolic function of the tissue under normal physiological conditions. The studies described here move though two phases. The existence of exploitable, functional effects is investigated by first examining the In vitro spectral response of breast biopsies and human breast cancer xenografts (16-17), collected and harvested according to a strict protocol. Then, in a follow-on phase, measurements of human breast cancer tumors are performed In vivo in nude mice to test the In vitro experimental findings.

3.1. In vitro Experimental Method

Thirty-four breast biopsies (11 malignant, 4 fibroadenoma, 3 male breast gynecomastia and 16 fibrocystic & other benign breast processes), twenty-five mouse model tumors stored in growth medium, and 1 gm of cancer cell cultures were studied. The sizes of the biopsies ranged roughly from a few millimeters to 1 cm and the sizes of the mouse model tumors ranged from about 4 mm spheroids to sections roughly 3 cm in diameter.

Sample collection and handling was designed to acquire samples that would mimic metabolic function under In vivo conditions as closely as possible within the scope of the study. The samples chosen to best represent these conditions were: (1) breast biopsies frozen (at -70 °C) within 5 - 10 minutes after removal from the host, (2) mouse model tumors, harvested immediately after sacrifice and stored in growth medium at wet ice temperatures, and (3) approximately 1 gm of a viable culture of cancer cells. The breast biopsies were obtained from a local hospital: Ellis Hospital in Schenectady, NY. The human breast cancer xenografts were grown in nude mice at the UCLA Medical Center in California, harvested and shipped to Schenectady in wet ice, and the cancer cell cultures were prepared at Rensselaer Polytechnic Institute in Troy, New York.

Prior to spectrophotometry, the breast biopsies were maintained in the laboratory at or below -50°C and the mouse model tumors were kept in growth medium either at 0 °C or at -50°C. The cancer cells were washed in phosphate buffer and spun down into a pellet. The pellet was placed in a quartz cuvette with a 1 mm path length. The biopsy samples and mouse model tumors were cut in a Nitrogen atmosphere while still frozen, to the desired thicknesses ranging from approximately 25 microns to 5 mm, and placed between quartz microscope slides with appropriate spacers. Care was taken to drain all samples after mounting and prior to measurement to get rid of any excess fluid and growth medium, which has absorption resonances in the visible that overlap those of tissue. As a control check, a few model tumors were snap frozen with liquid nitrogen (LN₂) after removal from the host, and their spectra compared with those of tumors stored in growth media. For selected samples, pathology was performed on material adjacent to and within 2 mm of the piece under study to eliminate errors of identification due to heterogeneity in large samples. As a final step in minimizing the time between sample collection and measurement, the optical spectra of the tissue samples were recorded typically within two to ten minutes after removal from dry ice storage or growth medium.

A Hitachi 3410 UV-VIS-NIR spectrophotometer with integrating sphere was used to measure the optical response of the tissue samples. The spectrophotometer is a double beam, ratio-recording instrument equipped with a system of detectors and lamps which permits the uninterrupted monitoring of wavelengths from 240 nm to 2500 nm. The use of the integrating sphere allowed us to collect all of the light scattered through 180 degrees in the back plane of the turbid sample.

Transmittance measurements were taken, and two types of scans were recorded for each tissue sample: a spectrum from 240 nm to 2500 nm and a narrow scan from 240 nm to 700 nm. For the narrow scans, improved resolution and signal-to-noise was achieved using a 1 nm bandpass and signal averaging of at least five repeated scans. Quantitative analysis of the spectra was performed on the average of the narrow scans using the Hitachi software package, Spectracalc.

3.2. In vivo Experimental Method

Nineteen nude mice were implanted with four human breast cancer cell lines: Mitchell, UCLA #157, UCLA #645 and UCLA #231. The cells were implanted subcutaneously on the back and intramuscularly on the hind leg. The heterotransplantations were performed at UCLA in the laboratories of D. J. Castro, M.D, Ph. D and R. Saxton, Ph. D. The implantation schedule was staggered in an attempt to produce tumors at various stages (18) of growth: UCLA cell lines #231 and #645

were implanted in two mice each every Monday for four weeks. These tumors were studied twice within a 2-week time period beginning the week of the last implantation. The other two cell lines were implanted and allowed to grow for approximately four weeks.

A remote VIS-NIR model 260 Guided Wave Spectrophotometer was used in conjunction with a 7:7 bundle fiber reflectance probe to perform the In vivo optical studies of the model tumors. Reflectance spectra were recorded over the wavelength range 350 - 1000 nm in 1 nm steps. The reflectance probe delivered ~ 9 mW of white light to the tumors through seven fibers and collected the reflected light from the tumors with seven fibers. This reflected light was then passed through a monochromator with a 0.5mm slit to a silicon detector and recorded. The diameter of the fiber bundle was approximately 2 mm.

Optical reflectance measurements were taken by touching the probe against the tumors with firm, but moderate pressure. Depending on the size of the tumor relative to the probe, measurements were taken at at least two different positions on the tumors. For each measurement, three to five consecutive scans were recorded as a reproducibility check and for off-line averaging. Measurements were taken in two cases: (i) with the skin over the tumor intact, and (ii) with the tumor exposed by raising a flap of skin over the lesion. In the first case, the mouse was awake and allowed to sit quietly during data acquisition; in the second case, the mouse was anesthetized throughout the entire procedure. After data acquisition, the flap of skin was repositioned and sutured and the mouse allowed to recuperate, making it possible for further measurements on that particular tumor at a later date.