Near-Infrared Spectrometric Imaging in Stroke Research.

Contact: lodder@pop.uky.edu

Kentucky is part of the southern stroke belt, a geographic zone in the U.S. in which the risk of death from stroke is elevated. The purpose of this research is to develop near-IR spectrometry as an imaging technique and to identify molecules that might contribute to the formation and/or progression of carotid atherosclerotic plaque, ulceration, or stroke.

Carotid artery atherosclerotic plaque is an important cause of cerebral ischemia, including devastating or fatal strokes. Plaque thickness increases with increasing age and also with rising levels of exposure to tobacco smoke. This thickening of carotid plaque can result in flow-significant stenosis or increase the risk of emboli to the cerebral vascular system.

The oxidation of low density lipoprotein (LDL) enhances its incorporation into atherosclerotic plaque. Cholesterol travels freely in the circulatory system in particles composed of LDL. LDL is a complex lipoprotein with a significant component of natural antioxidant present within its structure. It is reasonable to assume that processes that decrease this antioxidant protection can enhance age-related processes in the vessel. Lipoprotein particles have external ligands that enable them to bind to tissues such as those in the liver, which regulate the level of cholesterol in the blood. Oxidation of LDL changes its chemical structure, enabling LDL to enter the cell through any of a number of scavenger receptors that recognize modified forms of LDL. Cholesterol- laden macrophages are involved in the sites of lesions in the endothelium of the blood vessel wall. As more and more macrophages become attached, an atherosclerotic plaque laden with cholesterol is able to develop. It has been hypothesized that greater levels of oxidation of LDL in the blood lead to more rapid growth of some atherosclerotic plaques. However, traditional sonography is incapable of providing chemical analyses of plaque. If such an analysis were available, it might be possible to select patients for different drug interventions, obviating the need for endarterectomy. In order to rapidly measure the oxidation status in the plaque, near-infrared spectroscopy is proposed as a nondestructive assay for oxidized lipoproteins.

Scanning of atherosclerotic plaque with a near-IR video camera occurs in the operating room during carotid endarterectomy, a surgical procedure for removing plaque, usually plaque located near the carotid bifurcation. Endarterectomy is proven to reduce the incidence of stroke by 25-50%. While the ultimate goal is to provide near-IR analyses of plaque transcutaneously in patients, simultaneously with duplex ultrasonography, it is first necessary to prove that near-IR spectra provide an accurate indication of the oxidation status of carotid plaque. This proof requires that excised plaque tissue be analyzed by reference methods in vitro. It should be noted that the reference procedures are destructive and much more tedious and error prone than near-IR spectrometry.

Using the InSb camera During Surgery

The InSb focal plane array video camera is built into a cart powered through an on-board extreme isolation transformer. The cart holds four tungsten light fixtures, and the two used in the operating room provide 300 watts of black body light on the patient. The cart also holds a computer with a 12-bit video digitizer, and a color targeting monitor connected to the camera for the surgeon. Warm filters are used on the camera to select a number of near-IR wavelengths (each with 10 nm bandpass) in the 1000 to 2400 nm range, which is set by the cold filter (located inside the camera under vacuum). The camera and lamps are mounted on a single tripod. The tripod and cart are readied near the operating table beyond the sterile field as the blood vessel is exposed.

A safety clamp and bolts mounted on the tripod allow the camera to be rotated and focused while preventing the camera or its liquid nitrogen contents from falling on the patient through an accident. Interference filters in black holders are placed over the camera lens. The back of the camera has output jacks for the amplified array signals (which go to the 12-bit video digitizer), composite video for the targeting monitor, and selected test points. The focal-plane array data lines must be individually shielded and tied together to prevent cross-talk and loss of video synchronization. The surgeons step back for 8 minutes and the camera is angled down, aimed at the plaque, and the focus is checked. A visible-light image is recorded at this point. Light and dark reference spheres are placed as close as possible to the site of the plaque, and the surgeon marks the plaque location with forceps for the visible and near-IR cameras. Before collecting images, a final video synchronization and aim check on is made on the computer. The computer image matches the image on the camera monitor when all components are working properly. Actual scanning of images is performed with lights on and off at each wavelength to compensate for existing lights and windows as well as black body sample emission. Images are collected for almost 8 seconds at each wavelength. In this time period, 256 frames are collected.

A visible image and near-IR image (shown inset in visible image at a lipid absorbance wavelength) are collected from cameras located side by side. The two spherical reflectance standards are evident left of center in the image, just above the carotid bifurcation. The brightness and contrast of the images at each wavelength are adjusted to make the references identical. The specular reflectance of the source lamps from the references can be used to calculate the positions of the lamps and camera. The software enables users to zoom in on the references, for example, and re-scale the colors on them to cover just the absorbance values representing the references. The white sphere is visible in front of the black one, and the two purple zones on it represent reflections from the source lamps. The yellow spot to the right of the white sphere is the reflection of one source lamp from the black sphere. This close-up image of the references shows the spatial resolution attainable, as well as the S/N, using the InSb camera. The color scheme goes from white, which represents the lowest absorbances, through purple, blue, green, yellow orange red, and finally black. Black represents the highest aborbance values. A single computer program corrects images for bad pixels (both on and off), black body sample emission, and variable sample lighting, and calculates absorbance values from the images for storage and later manipulation.

The near-IR images from each patient are stored in a three-dimensional matrix. Examination of the matrix at any wavelength generally shows the spherical references, tissue retractors, surgical drape, and other major features. Minor spectral features are usually visible only after more sophisticated computer processing of the data.

InSb Camera Modifications.

After a number of uses (and perhaps thermal cycles, in which the InSb focal plane array (FPA) is cooled to liquid nitrogen temperature and allowed to reheat to room temperature), some pixels in the array lose their response. Loss of pixels is most notable in the corners of the FPA. InSb FPAs are "bump bonded" to a multiplexer chip, which in a practical sense means the InSb elements are actually in contact with only a small portion of the underlying multiplexer, and are not chemically bonded. This bonding method leaves the arrays subject to slow failure by delamination of the InSb from the silicon multiplexer. This picture of the InSb array was taken after removal of the Ge-coated lens, variable leaf aperture, sapphire window, double baffles, and cold filter. The cold filter was changed to one with a bandpass from 1 to 2.2 micrometers. This portion of the camera is under vacuum in normal operation. The bottom of the dewar is opened to access the InSb array leads. To create an electronic service manual for future reference, the signals on all array leads were recorded on videotape from an oscilloscope. The use of a clean room is recommended when opening the camera for extended servicing. Prudent practictioners seal openings not in use with polyethylene during repair and modification operations. The sapphire window retaining ring should be screwed on in the lens position to hold the window in when vacuum is broken. It is not necessary to remove the electronics from the chassis to access the dewar. However, the boards were removed to permit video taping of the signals at all of the boards' test points. Gain adjustments may be made on these boards as necessary after changing the baffles, warm filters and cold filters for a particular experiment.

Some of the InSb cameras had bad power supply connectors, in which supply wires of a large gauge were connected to pins in the plug by butt joints covered with silicone rubber. Over time, the butt joints broke apart, but the breaks were concealed by the silicone rubber. The supply pins need to be resoldered with silver solder and reinforced with heat shrinkable tubing, and re-assembled without more silicone rubber. The number of dead pixels in the FPA also increases over time. Dead pixels can be reactivated with pressure on the InSb array. The pressure should be applied with a nonconductive material directly on the array. In this picture, the pixels in the upper right of the display are stuck on. The application of controlled pressure to the array restores normal response to the pixels in the upper right of the display.

In Vitro Near-IR Spectrometry of Excised Plaques

The carotid plaques are frozen in liquid nitrogen in the operating room after removal by the surgeon and examination by a pathologist. The reference procedures for plaque analysis are destructive as well as much more tedious and error prone than near-IR spectrometry. To preserve the plaques during in vitro near-IR scanning, the scans are collected from the plaques in an environment chamber filled with cold nitrogen gas. Both near-IR scanning and biochemical plaque processing occur in the chamber under a cold nitrogen atmosphere. The box on the left side of the chamber has a 9x12 inch beam port in the top. Plaques placed on this beam port can be scanned in the near-IR using a spectrometer or camera located safely beneath the port. (The inside of the box beath the port is actually black, but flash reflections obscure it.) A new plaque chopper built into the center floor of the chamber cuts the plaques into small pieces for extraction while maintaining the plaques under a cold extraction buffer containing antioxidants. The chamber protects plaques from the atmosphere and workers from potential biohazards in the plaques. After cutting, the plaques are sealed in vials with extraction buffer in the chamber.

Ultracentrifugation and SDS-PAGE of Plaque Extracts.

The plaques shake gently overnight in a cold room on an orbital shaker under the extraction buffer with nitrogen headspace. For a close-up view of the extracts in tubes, click here. Some tubes contain visible blood, lipids, and different amounts of solid material. The amount of extraction buffer is adjusted in each tube according to initial excised plaque mass, regulating total solution volume for the ultracentrifuge tubes.

After overnight extraction, a Sorvall centrifuge pellets the solid material in a few minutes spin at 5000 rpm. A nitrogen tank and bubbler are used to replace the headspace gas for solutions in re-opened vials. Centricon tubes with 10 kD membrane cut-offs are used to concentrate lipoproteins in samples where the Lowry assay indicates the lipoprotein concentration may be low. Sample pre-concentration in these tubes requires up to 60 minute spins. A potassium bromide density gradient from about 1 to 1.2 g/ml is prepared and loaded with samples into the ultracentrifuge tubes, which are then sealed into the rotor. The samples spin at 100,000 rpm for 4 hrs. After the spin is complete, lipoprotein layers in the tube are detected and identified in two ways: by low angle laser light scatter and by comparing tube to LDL and oxLDL standards in tubes pre-stained with Coomassie brilliant blue. The oxLDL layer is removed from the tube and placed in dialysis tubing to remove the potassium bromide. During dialysis the potassium bromide diffuses out of the tubing and into a large volume aqueous solution.

The remaining proteins in solution in the tubing are placed on SDS-PAGE gels and separated. The molecular weight markers on the gels run from 14 to 200 kD. The LDL and oxLDL standards from the ultracentrifuge are also typically run on this gel with the patient samples.

In Vivo Near-IR Catheter

The purpose of these studies is to demonstrate that near-IR catheters are capable of detecting the growth and chemical composition of atherosclerotic plaque in vivo. Test specimens are fed either normal food or high cholesterol food for a period of weeks before their blood vessels are examined with the catheter. High cholesterol intake through food causes serum cholesterol to rise and fatty streaks and plaques to form in blood vessel walls. In ordinary use, the near-IR catheter is supplied with light by a Nd:YAG-pumped potassium titanyl phosphate optical parametric oscillator (KTP/OPO). In the figure, however, the catheter is illuminated with a HeNe laser. Placing the catheter against the aluminum surface of the optical table in the figure illuminates one of the light-receiving fibers from the catheter tip.

The KTP/OPO system is all solid-state and produces tunable near-IR light from 1400 to 4100 nm under computer control, with an effective maximum power of 3.3 megawatts. The Nd:YAG pump beam (1064 nm) is doubled (532 nm) and the doubled YAG split away on entering the OPO cabinet. The 532 nm line (105 mJ) pumps the OPO located in a nonresonant oscillator (NRO). The OPO outputs a signal beam and an idler beam, whose photon energies sum to the energy of the pump beam (532 nm). The 1064 nm line (375 mJ/pulse) pumps the optical parametric amplifier (OPA), which also produces a signal and an idler whose wavelength varies from 1440 to 4100 nm, and whose energies sum to the energy of the pump beam at 1064 nm. The mirrors on the output of the OPO cabinet show the pump and output beams when viewed through a CCD camera with the laser firing. (From left to right - 532 nm mirror, 1064 nm mirror, and the output coupler with [1064 nm plus near-IR signal and idler] and a little residual 532 nm.)

Blood samples are obtained before catheterization. The catheter is inserted through a small incision into the femoral artery, and advanced to the aortic arch. On reaching this point, a small suture is tied to the base of the catheter to mark the insertion depth. The catheter is then withdrawn a centimeter at a time, scanning spectra along the way. After the catheter has been completely withdrawn, the heart and aorta are removed and measured to mark the maximum insertion depth of the catheter. The aorta is cut into sections and lipoproteins extracted as described earlier. The spectra of aortas of specimens fed control food show normal spectra and cholesterol content. The spectra of specimens fed high cholesterol food show much more cholesterol in their blood vessels, and no apparent oxidation of the cholesterol.

Near-IR Results to Date

There are correlations between near-IR spectra of atherosclerotic plaque and the concentration of several proteins in LDL fraction extracted from human carotid plaques. The correlation between the 18 kD fragment and near-IR spectra is particularly high, and the 93, 114, 79, 65 kD proteins also appear detectable in near-IR spectra.

There are also correlations between near-IR spectra and medical histories of patients. Darkened boxes are correlations significant at p less than 0.05. Note that orthogonal spectral changes are observed for aging and previous stroke, a result that was not observed in brain tissue previously published (J.M. Carney, W. Landrum, L. Mayes, Y. Zou, and R. A. Lodder, Anal. Chem., 1993, 65, 1305-1313). The only medical history items that are indistinguishable by near-IR spectra are sex and smoking, which are correlated in the medical histories. There are also correlations between gel analyses of plaque proteins and medical histories. In these analyses, sex and smoking show significant correlations to plaque protein composition, and an interaction effect with age.

Near-IR spectra correlate with pathology reports on the excised carotid plaques. Necrosis and hemorrhage appear on the same spectral factor, as do the presence of fibrous cap and ulceration (which may be an earlier form of necrosis and hemorrhage). Neither pair of factors use the same factors as age, although necrosis and hemorrhage are on the same spectral factor as previous stroke.

Cell uptake of LDL particles through LDL receptors appears to be amenable to monitoring by near-IR spectrometry. Some LDL and cholesterol remains in cells for awhile after uptake. Gel electrophoresis results suggest that some oxLDL remains in plaques after uptake through scavenger receptors, too, but some plaques also show very little oxLDL. While near-IR imaging spectrometry using cameras in the operating room show lipoprotein distribution macroscopically, testing hypotheses of atherogenesis on a cellular level, and monitoring uptake of oxLDL, requires better tracking of these chemicals inside cells. Near-IR microscopy of cells is possible through a video lens relay system mounted on a conventional optical microscope. A near-IR near-field scanning optical microscope (NSOM) is now under construction in our laboratories and should make spectrometric monitoring of cell parts, such as receptors and their binding, possible at high resolution. The NSOM is based upon a single mode optical fiber, which is heated and drawn apart until it breaks, forming an aperture of subwavelength dimension. In an NSOM, only one photon at a time can squeeze out of the fiber tip. One optical fiber under a visible microscope shows the original fiber size and the newly drawn size near the tip. In this picture, the outside diameter of the larger fiber is 50 micrometers, and the fiber is stretched to 2 micrometers in the lower half of the image. It is a simple matter to make tips too small to see in light microscope, so characterization of the tips is done by scanning electron microscopy (SEM). Characterizing fiber tips by SEM requires that the optical fibers be gold-coated. At the same time, the gold coating acts to dissipate evanescent waves from the optical fiber and improve the spatial resolution of the NSOM. In this figure, a 400 nm O.D. tip, about 33% of the size of the smallest wavelength we will scan. The newest tips pulled are probably even smaller, but the SEM is down and additional measurements must be delayed until the SEM is repaired.

The purpose of this research is to develop near-IR spectrometry as an imaging technique and to identify molecules that might contribute to the formation and/or progression of carotid atherosclerotic plaque, ulceration, or stroke. Once these molecules ar identified, sequencing and calculation of tertiary structure can be undertaken to establish mechanisms and role in atherosclerosis.