|Home | About | Journals | Submit | Contact Us | Français|
A method is described for quantitating autoradiographs of bone-seeking isotopes in microscopic sections of bone. Autoradiographs of bone sections containing 45Ca and internal calibration standards are automatically scanned with a microdensitometer. The digitized optical density output is stored on magnetic tape and is converted by computer to equivalent activity of 45Ca per gram of bone. The computer determines the total 45Ca uptake in the bone section and, on the basis of optical density and anatomic position, quantitatively divides the uptake into 4 components, each representing a separate physiologic process (bone formation, secondary mineralization, diffuse long-term exchange, and surface short-term exchange). The method is also applicable for quantitative analysis of microradiographs of bone sections for mineral content and density.
Although radiocalcium and related isotopes have been widely used in the study of bone metabolism, several fundamental issues remain incompletely resolved. Accretion of tracer in bone involves several processes, only one of which represents mineralization of collagen.1–3 The relative importance of the different accretionary processes, the change in their distribution as a function of time, and the validity of different kinetic models for radiocalcium metabolism need further study and clarification. The approach to these problems requires quantitative autoradiographic measurements of tracer deposition and distribution in bone. Although quantitative autoradiographic methods have been previously described, none has proved entirely satisfactory in the solution of these problems.
In this report we describe a new autoradiographic method for quantitating total uptake and regional distribution of radiocalcium in microscopic sections of bone. The method, based on automated microdensitometric scanning, is also applicable to quantitative analysis of bone microradiographs for mineral content and density.
Development and testing of the method was accomplished with the use of autoradiographs and microradiographs of biopsy samples from the distal ulna and rib of dogs. Specimens were obtained at various times, but usually 7 days, after injection of 45CaCl2 (100 µCi per kilogram of body weight).
The bone biopsy samples were fixed in 70 per cent ethyl alcohol, embedded in methyl methacrylate, cut with a rotating saw on a milling machine, and hand ground to a final thickness of 100 ± 2 µ. Autoradiographs were exposed by holding Type A Eastman Kodak autoradiographic plates in contact with the sections with spring clips on a brass slide holder.4, 5 The emulsion was developed with Eastman Kodak DK 90 developer solution diluted to 40 per cent. Small indentations in the brass slide holder were filled with known amounts of 45Ca uniformly mixed in plaster of Paris (1.2 ml. of radioactive solution per 2 Gm. of dry plaster) for calibration of radioactivity. There was a sufficient range of radioactivity in the standards to encompass all ranges of emulsion darkening due to 45Ca in the bone section. Because plaster has a mass absorption coefficient similar to that of bone, it is possible to express results in absolute values of 45Ca per gram of bone.4 The autoradiograph itself was the transparency used for scanning by the densitometer for 45Ca uptake and distribution studies.
Microradiographs were prepared by passing soft, filtered x-rays from a copper target through the section placed on a 649-0 Eastman Kodak spectroscopic plate. The plate was developed with Eastman Kodak D19 developer solution. A step-wedge of aluminum foil, 6.25 µ thick, was simultaneously exposed with the section as a calibrating standard; aluminum has been shown to have a mass absorption coefficient similar to that of hydroxyapatite.6 Because the blackened background of the microradiograph negative was too dense for the scanning beam to penetrate, a positive image, enlarged twofold, with a translucent background was made by projecting the negative image of the section and calibrating standards on a medium-contrast Eastman Kodak projector slide plate and developing with D19 developer. This reversed image of the microradiograph was the transparency used for scanning by the densitometer for bone-density studies.
The transparencies were scanned automatically, and the digitized optical density output was continuously recorded on magnetic tape by using 3 instruments operating in a series.* The first instrument, a modified Joyce-Lobel microdensitometer, has a split light beam, one beam passing through the specimen and the other through an optical wedge on a sliding carriage. A servomechanism continuously achieves optical balance by moving the optical wedge until the transmission intensities of the 2 beams are equal.
An attachment to the specimen table of the microdensitometer provides for automatic parallel scanning across the section, and a stepping motor assembly with an interval pre-selected by the second instrument, the Isodensitracer, advances the table the required distance between parallel scans. When the scan of the section is completed, a trip switch turns off the machine. The movement of the optical wedge during scanning constitutes a continuous analog signal which is mechanically digitized into 180 optical density levels by a shaft encoder of gears and wire filaments mechanically linked to the wedge carriage. The digitized information of position and optical density obtained during scanning is transformed by a third instrument, the Laboratory Data Collector, into a binary coded decimal (BCD) format and stored on magnetic tape for subsequent computer analysis. For convenience, multiple recordings from scans are stored on a single tape reel.
The following instrument settings were found to be satisfactory and were used for all scans. An optical wedge with a range of 0 to 2.52 optical density units was used so that each of the 180 density levels was equivalent to an increment of 0.014 optical density unit. For examining autoradiographs, a scanning window of 20 by 20 µ was used with optical density readouts occurring at regular intervals of 100 µ along parallel scan paths, with steps of 100 µ between scan paths. Readings could not be obtained with an aperture window of less than 4 by 4 µ because of noise due to the grain of the emulsion. When line tracings (pen and ink recorder) were made across small “hot spots,” maximal density was about 15 per cent higher with the 4 by 4 than with the 20 by 20 µ window; however, the integrated area of the scan path across the hot spot varied only slightly.
The raw data from each scan were read from the magnetic tape by a digital computer (Model 3200, Control Data Corporation, Minneapolis, Minn.). The computer analysis had 3 functions: (1) to determine the total uptake of 45Ca in the bone section, (2) to divide the uptake into its component parts, and (3) to produce a symbolically coded representation of the original scan so that the optical density readings assigned to the components could be verified by direct comparison with the original autoradiograph.
For autoradiographs made from bone specimens taken more than 24 hours after injection of 45Ca, 4 components of uptake were recognized by the computer. Each component represents entry of 45Ca into bone by a different physiologic process.
Marshall and co-workers3 have shown that diffuse activity occurring throughout previously formed, fully mineralized bone is due to a slow exchange process which contributes to the kinetically determined calcium accretion rate but does not increase bone density.
Rowland7 has produced evidence that surface activity represents rapid exchange of isotope in plasma with bone, which is probably largely measured by kinetic analysis as a part of the exchangeable calcium pool.
The slow completion of mineralization (“secondary mineralization”) of newly formed bone is visualized as focal areas of intermediate activity (usually 3 to 5 times that of the diffuse activity). By this process, mineral but not additional matrix is added to the skeleton.2
Areas of new bone formation are visualized in autoradiographs as focal areas of high-density darkening (hot spots).
For analysis of total 45Ca content of the bone section, the optical density readings obtained during scanning of the autoradiographs were converted by the computer to microcuries of 45Ca per gram of bone by interpolation from the calibration curve of the optical density versus 45Ca concentration (Fig. 1), with background darkening automatically set at 0 activity. The individual 45Ca concentrations representing the converted optical density readings were then summed to give the total activity of 45Ca, of the bone section. Similarly, total mineral content of the section was determined directly by summation after conversion of the densities obtained by scanning the microradiograph to an equivalent thickness of aluminum foil standard.
Distributional analysis of the uptake in the 4 components was performed by the computer on the basis of differences in optical density and anatomic location. For the part of the analysis based on differences of density, the computer separated high, intermediate, and low activities on the basis of a frequency array (Fig. 2) of 45Ca equivalent activity held in computer memory. Low activity, representing the diffuse uptake component, was automatically recognized as a sharp, homogeneous peak. The boundary between high and intermediate activities was taken, as a first approximation, to be 5 times the 45Ca activity of the diffuse activity peak because experience has shown that it is never less than this. The computer then printed out a numerical representation of the original scan composed of the actual optical density readings as stored on magnetic tape and sequentially printed out those individual optical density readings tentatively assigned to the high, intermediate, and low activity ranges. From direct comparison of the print-out representation of the scan with the original autoradiograph (examined by microscopy),* final assignments of the optical densities for high, intermediate, and low activities were made.
Two components of autoradiographic darkening must be separated by the computer on the basis of anatomic position because their optical densities overlap with those of other components. The first of these components is surface activity. For autoradiographs obtained about 7 days after injection of 45Ca, surface activity falls within the less-dense part of the intermediate range. The second is the halo of intermediate activity surrounding hot spots. The halo is a result of high-energy beta rays originating within the hot spot or of the scanning window including adjacent diffuse activity with the edge of a hot spot. To recognize these components, a computer search program compares each density reading with all immediately adjacent readings. Surface activity is recognized as low intermediate activity that is immediately adjacent to background activity.† Halo activity is recognized as intermediate activity that is immediately adjacent to hot-spot activity; the halo activity is subtracted from the intermediate activity total and added to the high activity total. The computer then totals the component activities and prints out a representation of the original scan, assigning a symbolic code to each component (Fig. 3), for final verification by comparison with the original autoradiograph.
During analysis, data are entered into the computer twice and comparison of the autoradiograph with a coded print-out of the scan is made twice. An additional computer program, which requires manual card punching of input data, is available for removal of artifacts.
Preparation of the autoradiograph or microradiograph requires about 10 minutes of technician’s time per section. Setting up for automatic scanning requires about 5 minutes per slide. Data analysis (including key-punching, tape transfer, and verification by comparison of the autoradiograph and data print-out) require from 30 to 60 minutes per scan.
The processing of each scan requires a total of about 5 minutes of computer time. Automatic scanning of an autoradiograph of a dog rib, a total area of 30 mm.2, requires about 15 minutes; for that of a dog midfemur or ulna, a total area of 120 mm.2, the time is about one hour.
The microdensitometer recordings were quite reproducible. For repeated scans over a transparency of uniform darkness, such as the background fog of autoradiographs or microradiographs, variations of density readings were always less than one per cent.
To determine the cumulative error of the method, duplicate and triplicate transparencies of the same bone section were developed and scanned. For 27 replicate autoradiographs, the coefficient of variation for determination of total 45Ca activity in the sections was 6.4 per cent. The coefficient of variation for the mineral content of 22 replicate microradiographs was 2.1 per cent.
For the distributional analysis of the components of 45Ca activity, variability was similar for each component and was independent of the magnitude of the quantities measured except in the case of the diffuse component (low activity). The pooled coefficient of variation for high activity, intermediate activity, and surface activity was 3.1 per cent. For the diffuse component the coefficient of variation was 6.4 per cent, and there was a trend toward greater variability as the magnitude of the measurements increased.
The following experimental results illustrate the change, with time, in uptake and distribution of radiocalcium in the bone of 4 mongrel dogs after intravenous injection of 45CaCl2 (100 µCi per kilogram of body weight). Biopsy specimens of rib were taken at 10 minutes, 4 hours, 24 hours, and 7 days after injection and of distal ulna, at 7 days. The results (Fig. 4) are expressed as fraction of dose per gram of calcium in the bone section.
At 10 minutes and at 4 hours, uptake was entirely at bone surfaces. During the first 24 hours, the serum specific activity decreased rapidly as radiocalcium diffused throughout the exchanging compartments of the body; mixing within the exchangeable calcium pool was virtually completed by 24 hours, and, thereafter, the serum specific activity decreased mono-exponentially (when plotted semilogarithmically against time) as radiocalcium was excreted from the body and accreted into the skeleton. Between 24 hours and 7 days, surface activity decreased in proportion to the decrease in serum specific activity, indicating that this uptake is due to a rapid exchange with the serum. Hot-spot activity in the areas of bone formation was apparent by 24 hours; by 7 days it was quantitatively the most important accretionary process. In the distal ulna, which is composed entirely of compact bone, at 7 days the diffuse activity due to long-term exchange accounted for between ⅓ and ⅔ of the total activity. In the rib, which is a mixed compact-trabecular bone with a higher remodeling rate, diffuse activity was less important at 7 days. Intermediate activity due to secondary mineralization was quantitatively a relatively minor process.
Earlier attempts at quantitating autoradiographic darkening consisted of taking the average of short line scans with a microdensitometer and relating this to the calibration curve of the standard.4, 5, 10–13 Such methods are useful only for materials of uniform density (such as the diffuse component) and cannot be used for evaluation of nonuniform darkening such as hot spots of 45Ca activity in bone autoradiographs.
Recently, 2 methodologic advances were published. Jowsey5 described a method in which a positive print of the original autoradiograph or microradiograph with calibrating standards is made by precipitating a blue dye rather than a silver emulsion. Areas of interest in the print are cut out, and the blue dye, the amount of which is proportional to the darkening on the autoradiograph or microradiograph, is extracted chemically and measured spectrophotometrically.
Lloyd and co-workers14 published a preliminary report of a method for autoradiograph and microradiograph densitometry utilizing automatic scanning with a beam of light from a cathode ray tube under computer control. Readings from the scan are calibrated against a wedge with 15 density steps and are read out as the integrated area of optical densities corresponding to the interval between steps. Data regarding errors have not yet been given nor is it known whether areas of focal density such as hot spots can be quantitated accurately.
The reproducibility of our method is comparable to or better than that of other techniques. Although our method has the disadvantage of requiring specialized equipment for scanning and the availability of a digital computer for data analysis, it offers a number of major advantages. Because the specimen is scanned automatically and the data are analyzed by computer, a large number of specimens can be batch-processed with a minimum of operator time. Quantitative analysis of all major components of 45Ca on the autoradiograph, including surface activity, is possible. The optical densities assigned to each of these components are verified by direct visual comparison of the scan output with the original autoradiograph. In contrast to previously described methods, both linear and nonlinear portions of the calibration curve (standard activity versus optical density) can be utilized, which permits a longer exposure of the transparency and a more favorable signal-noise ratio for low-activity darkening. Finally, because the original scan data are permanently stored on tape and are not destroyed by processing, many analyses of the data can be conveniently performed.
This method has been used by us only for analysis of bone autoradiographs and microradiographs. However, it is clearly applicable to any tissue section transparency from which quantitative density information would be useful.
The authors are grateful to Mr. Carl Brandt and Mr. Robert Frey for their technical assistance and to Mr. Martin T. Storma, Mr. William Dunnette, and Dr. Eugene Ackerman, Section of Biomathematics, University of Minnesota, for their many helpful suggestions.
This investigation was supported in part by Research Grants AM-8665 and FR-7 from the National Institutes of Health, United States Public Health Service.
*Manufactured as the Isodensitracer with Laboratory Data Collector (Model 1100) by Beckman and Whitley, Inc., Mountain View, Calif.
*In the developmental stage, the association of hot spots on the autoradiograph with bone formation was confirmed by other histologic criteria (direct comparison of the autoradiograph with the original section with the use of tetracycline fluorescence8 or with low-density surfaces as described by Jowsey and associates9).
†A small, systematic error exists for surface activity on trabeculae which traverse the bone section at an angle, because calibration of activity with standards assumes a full thickness of bone while, in fact, a trabecula in a 100 µ bone section may be less than full thickness. The error is probably of the order ± 10 per cent of the obscured surface activity.