Journal of Orthopaedlc Reseurc h 10552-561 Raven Press, Ltd , New York 0 1992 OrthopaedlL Research Society
Correlating Magnetic Resonance Imaging with the Biochemical Content of the Normal Human Intervertebral Disc M. Weidenbaum, K. J. Foster, *B. A. Best, F. Saed-Nejad, i-E. Nickoloff, +J. Newhouse, A. Ratcliffe, and V. C. Mow Orthopuedic Research Laboratory, Department qf Orthopuedic Surgery and +Department of Radiology, Columbia University, N e w York, N e w York; and *University oj California-Berkeley, Deportment of Integrative Biology, Berkeley, California, U . S . A .
Summary: Magnetic resonance imaging was used to determine the T2 relaxation times of prepared proteoglycan solutions and of normal human intervertebral disc tissue from the annulus fibrosus (AF) and nucleus pulposus (NP). The collagen, proteoglycan, and water contents of the disc tissue samples were determined by biochemical assays after they were scanned. Correlations among l / T 2 ,collagen, proteoglycan, and water contents of the tissue samples and among l/T2, water, and proteoglycan contents of the proteoglycan solutions were calculated. A moderate negative correlation between 1/T, and water content was noted for the tissue samples, and a very high negative correlation was found beween l / T , and water content for the proteoglycan solutions. The very high positive correlation between l/T, and proteoglycan content of the proteoglycan solutions is probably due to this negative correlation between l/Tz and water content. There was no significant correlation between l/T, and proteoglycan content of the tissues. The moderate positive correlation between 1/T, and collagen content is probably due to the high negative correlation between collagen content and water content. N o significant correlation was found between the collagen and proteoglycan contents of the tissues. Thus it appears that the data confirm previous reports in the literature that the collagen of the disc tissue functions to control its water content. Key Words: Magnetic resonance imaging-Proteoglycan-Collagen-Water content-T, relaxation time-Intervertebral disc.
Magnetic resonance imaging (MRI) is a widely used clinical imaging system. The technique is noninvasive, does not require ionizing radiation, allows direct visualization of soft tissues, and can depict both normal and pathological phenomena (8,11,29,
32,33,36,40,50,51). The intervertebral disc (IVD) is a connective tissue particularly suited to studies using MRI (20,27,35,36,45,47) since it is clinically important and is difficult to access without the use of invasive procedures. The IVD is composed primarily of water (6085%) and a large extracellular matrix of collagen and proteoglycan. The outer annulus fibrosus consists of concentric lamellae of type I collagen fibers (60-70% of the dry weight) and a low concentration of proteoglycan (12-15,42). There is a gradual in-
Received May 30, 1991; accepted February 14, 1992. Address correspondence and reprint requests to Dr. Mark Weidenbaum at Columbia University, Orthopaedic Research Laboratory, Black Building, Rm 1412, 630 W. 168th St., New York, NY 10032, U.S.A.
552
MRI AND BIOCHEMISTRY OF THE INTERVERTEBRA L DISC crease in the ratio of type I1 to type I collagen toward the inner annulus fibrosus. The nucleus pulposus has a high concentration of proteoglycan (>50% of the dry weight) with a lower concentration of collagen, primarily type 11 (9,12,13,15,29). The collagen network provides much of the basic structure of the IVD and accounts for the tensile properties of the tissue (7,13,14,19). The IVD proteoglycans are similar to the proteoglycans of articular cartilage (2,41,56). They consist of an extended protein core to which are attached many chondroitin sulfate and keratan sulfate glycosaminoglycan chains. They have the ability to form large aggregates with hyaluronate, and this binding is stabilized chemically and mechanically by a separate link protein (10,21,37,43,58). The large number of carboxyl and wlfate groups on the glycosaminoglycan chains possess a high fixed charge density, and thus the proteoglycans provide the IVD with a high Donnan osmotic swelling pressure (62-64). This swelling pressure tends to expand the IVD, an expansion that is resisted by the tensile 3tiffness of the collagen network within the IVD tissue and by axial loading on the spine. This balance of force determines tissue hydration. The compressive, tensile, and shear stiffnesses of the IVD are provided by the intrinsic mechanical properties of the collagenproteoglycan solid matrix. Finally, the effects of the swelling pressure on the state of mechanical stress acting in the collagen-proteoglycan solid matrix has been described by a mechanoelectromechanical theory for charged, hydrated soft tissucs. such as the IVD (28). MRI has been used to distinguish between normal and pathological tissues of the IVD (8,32,34). Degenerative changes and disc herniation in the IVD have previously been shown to be accompanied by changes in the water and proteoglycan content of the tissue (30,42). However, it is not clear which components of the IVD contribute to the MRI signal (34). It should be noted that the MRI signal is also probably influenced by the environment of the molecules (interfaces between phases, local viscosity, etc.) as well as by their composition (17). Most previous studies have dealt with weighted MR images that contain a combination of both T , and T2, and not actual relaxation times. Such weighted images are protocol dependent and therefore do not lend themselves to direct comparisons. It has recently been suggested that the T2 relaxation time reflects the proteoglycan content of the intervertebra1 disc (15,44,60). However, there is a lack of
553
information in the literature concerning the statistical correlations of proteoglycan, collagen, and water contents of the intervertebral disc tissue with actual (not weighted) T2 relaxation times. The objective of this study was therefore to determine which components could be correlated with the T2 relaxation time. To achieve this objective, samples of purified proteoglycan solutions and samples of normal IVD tissue were analyzed using MRT to determine their T2 relaxation times. Correlation coefficients for relaxation time versus biochemical composition of the samples were then calculated and compared.
MATERIALS AND METHODS Preparation of Tissue Samples Four intervertebral discs (two L1-2, two L2-3) were dissected from three fresh frozen human spines (34-45 years old) that were without visual signs of degeneration [Thompson grade I1 (61)l. Sixty-seven cylindrical tissue samples (4 mm diameter by 8 mm length) were cored from the IVDs; 41 samples were taken from the annulus fibrosus and 26 samples from the nucleus pulposus. The tissue samples were inserted into plastic vials (4 mm diameter by 3 cm length) with the bottom portions filled with 2 cm of paraffin (Fig. 1). This method kept all specimens positioned within the top 1 cm of the vials through which the MR imaging plane was centered. Preparation of Proteoglycan Solutions Proteoglycans were extracted from bovine nasal cartilage in 4 M guanidine HCV0.05 M sodium aceAnterior
, ANNULUS FIBROSUS
FIG. 1. Tissue sample collection and placement in vials
J Orthup Res,
V O ~10, . N O . 4 , 1992
M . WEIDENBA UM ET AL.
554
tate buffer, pH 5.8, in the presence of protease inhibitors (21). The proteoglycan components were separated by equilibrium-density-gradient centrifugation, as described previously (48), to produce proteoglycan monomer (A 1D1). Four AlDl solutions at concentrations of 25, SO, 75, and 100 mg/g (w/w) were prepared in 0.15 M NaCl. This range spans the approximate physiological range of proteoglycan concentrations found in disc tissue (42). These solutions were placed in the sample vials without the use of paraffin, since the entire vial could be filled. MRI Scanning of the Samples
All vials were placed randomly in a sample holder inserted into a custom-made Plexiglas "phantom" filled with 9 L of 0.7 g/L CuSO, solution (Fig. 2). This CuS0,-filled phantom was used to simulate the body tissue normally present with the scanning configuration (46). The phantom and samples were allowed to equilibrate to room temperature, which was maintained at 21.1 t I.1"C. The samples were scanned with a Philips Gyroscan with a field strength of 0.5 Tesla at a frequency of 21.3 MHz. A Carr-Purcell/Meiboom-Gill (CPMG) spin-echo se-
I
24cm
- 1
4
32 c m--
13
i
I
V I A L HOLDER
Correlating Magnetic Resonance Imaging with the Biochemical Content of the Normal Human Intervertebral Disc M. Weidenbaum, K. J. Foster, *B. A. Best, F. Saed-Nejad, i-E. Nickoloff, +J. Newhouse, A. Ratcliffe, and V. C. Mow Orthopuedic Research Laboratory, Department qf Orthopuedic Surgery and +Department of Radiology, Columbia University, N e w York, N e w York; and *University oj California-Berkeley, Deportment of Integrative Biology, Berkeley, California, U . S . A .
Summary: Magnetic resonance imaging was used to determine the T2 relaxation times of prepared proteoglycan solutions and of normal human intervertebral disc tissue from the annulus fibrosus (AF) and nucleus pulposus (NP). The collagen, proteoglycan, and water contents of the disc tissue samples were determined by biochemical assays after they were scanned. Correlations among l / T 2 ,collagen, proteoglycan, and water contents of the tissue samples and among l/T2, water, and proteoglycan contents of the proteoglycan solutions were calculated. A moderate negative correlation between 1/T, and water content was noted for the tissue samples, and a very high negative correlation was found beween l / T , and water content for the proteoglycan solutions. The very high positive correlation between l/T, and proteoglycan content of the proteoglycan solutions is probably due to this negative correlation between l/Tz and water content. There was no significant correlation between l/T, and proteoglycan content of the tissues. The moderate positive correlation between 1/T, and collagen content is probably due to the high negative correlation between collagen content and water content. N o significant correlation was found between the collagen and proteoglycan contents of the tissues. Thus it appears that the data confirm previous reports in the literature that the collagen of the disc tissue functions to control its water content. Key Words: Magnetic resonance imaging-Proteoglycan-Collagen-Water content-T, relaxation time-Intervertebral disc.
Magnetic resonance imaging (MRI) is a widely used clinical imaging system. The technique is noninvasive, does not require ionizing radiation, allows direct visualization of soft tissues, and can depict both normal and pathological phenomena (8,11,29,
32,33,36,40,50,51). The intervertebral disc (IVD) is a connective tissue particularly suited to studies using MRI (20,27,35,36,45,47) since it is clinically important and is difficult to access without the use of invasive procedures. The IVD is composed primarily of water (6085%) and a large extracellular matrix of collagen and proteoglycan. The outer annulus fibrosus consists of concentric lamellae of type I collagen fibers (60-70% of the dry weight) and a low concentration of proteoglycan (12-15,42). There is a gradual in-
Received May 30, 1991; accepted February 14, 1992. Address correspondence and reprint requests to Dr. Mark Weidenbaum at Columbia University, Orthopaedic Research Laboratory, Black Building, Rm 1412, 630 W. 168th St., New York, NY 10032, U.S.A.
552
MRI AND BIOCHEMISTRY OF THE INTERVERTEBRA L DISC crease in the ratio of type I1 to type I collagen toward the inner annulus fibrosus. The nucleus pulposus has a high concentration of proteoglycan (>50% of the dry weight) with a lower concentration of collagen, primarily type 11 (9,12,13,15,29). The collagen network provides much of the basic structure of the IVD and accounts for the tensile properties of the tissue (7,13,14,19). The IVD proteoglycans are similar to the proteoglycans of articular cartilage (2,41,56). They consist of an extended protein core to which are attached many chondroitin sulfate and keratan sulfate glycosaminoglycan chains. They have the ability to form large aggregates with hyaluronate, and this binding is stabilized chemically and mechanically by a separate link protein (10,21,37,43,58). The large number of carboxyl and wlfate groups on the glycosaminoglycan chains possess a high fixed charge density, and thus the proteoglycans provide the IVD with a high Donnan osmotic swelling pressure (62-64). This swelling pressure tends to expand the IVD, an expansion that is resisted by the tensile 3tiffness of the collagen network within the IVD tissue and by axial loading on the spine. This balance of force determines tissue hydration. The compressive, tensile, and shear stiffnesses of the IVD are provided by the intrinsic mechanical properties of the collagenproteoglycan solid matrix. Finally, the effects of the swelling pressure on the state of mechanical stress acting in the collagen-proteoglycan solid matrix has been described by a mechanoelectromechanical theory for charged, hydrated soft tissucs. such as the IVD (28). MRI has been used to distinguish between normal and pathological tissues of the IVD (8,32,34). Degenerative changes and disc herniation in the IVD have previously been shown to be accompanied by changes in the water and proteoglycan content of the tissue (30,42). However, it is not clear which components of the IVD contribute to the MRI signal (34). It should be noted that the MRI signal is also probably influenced by the environment of the molecules (interfaces between phases, local viscosity, etc.) as well as by their composition (17). Most previous studies have dealt with weighted MR images that contain a combination of both T , and T2, and not actual relaxation times. Such weighted images are protocol dependent and therefore do not lend themselves to direct comparisons. It has recently been suggested that the T2 relaxation time reflects the proteoglycan content of the intervertebra1 disc (15,44,60). However, there is a lack of
553
information in the literature concerning the statistical correlations of proteoglycan, collagen, and water contents of the intervertebral disc tissue with actual (not weighted) T2 relaxation times. The objective of this study was therefore to determine which components could be correlated with the T2 relaxation time. To achieve this objective, samples of purified proteoglycan solutions and samples of normal IVD tissue were analyzed using MRT to determine their T2 relaxation times. Correlation coefficients for relaxation time versus biochemical composition of the samples were then calculated and compared.
MATERIALS AND METHODS Preparation of Tissue Samples Four intervertebral discs (two L1-2, two L2-3) were dissected from three fresh frozen human spines (34-45 years old) that were without visual signs of degeneration [Thompson grade I1 (61)l. Sixty-seven cylindrical tissue samples (4 mm diameter by 8 mm length) were cored from the IVDs; 41 samples were taken from the annulus fibrosus and 26 samples from the nucleus pulposus. The tissue samples were inserted into plastic vials (4 mm diameter by 3 cm length) with the bottom portions filled with 2 cm of paraffin (Fig. 1). This method kept all specimens positioned within the top 1 cm of the vials through which the MR imaging plane was centered. Preparation of Proteoglycan Solutions Proteoglycans were extracted from bovine nasal cartilage in 4 M guanidine HCV0.05 M sodium aceAnterior
, ANNULUS FIBROSUS
FIG. 1. Tissue sample collection and placement in vials
J Orthup Res,
V O ~10, . N O . 4 , 1992
M . WEIDENBA UM ET AL.
554
tate buffer, pH 5.8, in the presence of protease inhibitors (21). The proteoglycan components were separated by equilibrium-density-gradient centrifugation, as described previously (48), to produce proteoglycan monomer (A 1D1). Four AlDl solutions at concentrations of 25, SO, 75, and 100 mg/g (w/w) were prepared in 0.15 M NaCl. This range spans the approximate physiological range of proteoglycan concentrations found in disc tissue (42). These solutions were placed in the sample vials without the use of paraffin, since the entire vial could be filled. MRI Scanning of the Samples
All vials were placed randomly in a sample holder inserted into a custom-made Plexiglas "phantom" filled with 9 L of 0.7 g/L CuSO, solution (Fig. 2). This CuS0,-filled phantom was used to simulate the body tissue normally present with the scanning configuration (46). The phantom and samples were allowed to equilibrate to room temperature, which was maintained at 21.1 t I.1"C. The samples were scanned with a Philips Gyroscan with a field strength of 0.5 Tesla at a frequency of 21.3 MHz. A Carr-Purcell/Meiboom-Gill (CPMG) spin-echo se-
I
24cm
- 1
4
32 c m--
13
i
I
V I A L HOLDER
