1. Biomechanics Vol. 25, NO. 10. pp. 1185-1194,

1992.

0021.9290/92 $5.00+ .&.I fit 1992 Pergamon Press Ltd

Printed in Great Britain

MECHANICAL PROPERTIES OF THE HUMAN LUMBAR ANTERIOR LONGITUDINAL LIGAMENT P. NEUMANN,* T. S. KELLER,~$ L. EKSTR~M, L. PERRY,~ T. H. HANSSON* and D. M. SPENGLER~ *Department of Orthopaedics, Sahlgren Hospital, Gothenburg, Sweden, tDepartment of Orthopaedics and Rehabilitation, Vanderbilt University, Nashville, Tennessee, U.S.A., IVeterans Administration Medical Center, Nashville, Tennessee, U.S.A. and §Surgical Research Laboratory, Inc., Nashville, Tennessee, U.S.A. Ahstract-A new technique incorporating a motion analysis system and a materials testing machine was used to investigate regional differences in the tensile mechanical properties of the lumbar spine anterior longitudinal ligament (ALL). Bone-ALL-bone specimens were prepared from young human cadaveric motion segments with no disc or bony pathology. Each specimen was distracted until failure at a constant crosshead displacement rate of 2.5 mm s- ’ (approximately 1.0% strain per second). Strains were evaluated from digitized video recordings of markers attached to the ALL at 12 sites along its length and width, including the ligament substance and insertions. The ‘overall’ strain in the ligament was calculated from the outermost pairs of markers along the ligament length. The average tensile strength, the ‘overall’ tensile modulus and the ‘overall’ strain of the ALL at failure were 27.4 MPa (S.D. 5.9), 759 MPa (SD. 336) and 4.95% (S.D. 1.51), respectively. Large and significant variations in the strains were present along the width and length of the ALL. Peak substance strains were over twofold greater than peak strains at the ligament insertion sites, whereas across the ligament width, peak strains in the outer portion of the ligament were over 40% greater than in the central region. Failure consistently occurred in the ligament mid-substance and ultimate strains at the ligament failure site averaged 12.1% (SD. 2.3). These results indicate that the strains are highly nonuniform in the normal ALL.

INTRODUCTION

Knowledge of the mechanical properties of ligaments is a necessary prerequisite to understanding boneligament structural function. This knowledge is important for reconstruction of damaged ligaments as well as clinical assessment of ligament injuries resulting from trauma. Analyses of the mechanical properties (stress-strain relationships) of ligamentous tissues, however, are complicated due to problems associated with the precise determination of strains in the ligament. Special apparatuses with mechanical attachments have been developed to directly measure ligament strains in vitro (Nachemson and Evans, 1968; Edwards et al., 1970; Wang et al., 1973; Waters and Morris, 1973; Arms et al., 1984; Kennedy et al., 1977; Panjabi et al., 1982) and in uiuo (Howe et al., 1990), but these techniques are not well suited for strain evaluation at multiple sites on the ligament, and may also interfere with ligament substance deformation. Woo et al. (1982) addressed these problems by developing a noncontact optical-video technique that utilizes a

Received in final form 4 February 1992.

Author to whom correspondence should bc addressed: Tony S. Keller, University of Vermont, Department of Mechanical Engineering, 209B Votey Building, Burlington, VT 05405-0156, U.S.A. Research supported by the Department of Veterans Affairs Medical Center, Swedish Work Environment Fund, and the Folksam Research Foundation. Presented in part at the 7th Meeting of the European Society of Biomechanics, Aarhus, Denmark, July 8-11, 1990.

video dimensional analyzer (VDA) to measure surface strains of soft tissues. Using this technique to evaluate strain variations in the femoral and tibia1 regions of the knee medial collateral ligament substance, Woo et al. (1983, 1986) later demonstrated that strains were highly nonlinear and nonuniform within the ligament substance region. Few studies have examined the mechanical properties of the bone-ligament interface. Panjabi et al. (1984) described a photographic technique to measure spinal-ligament insertion and substance strains using a 4 x 3 grid of points. They observed that ligament failures occurred at mid-substance in some preparations and at the ligament-bone interface in others, but did not present any stress-strain data related to these observations. Pintar et al. (1986) speculated that regional strain variations exist within ligaments, but to our knowledge there are no comprehensive studies in the literature regarding the distribution of strains and mechanical property variations within ligament insertions and substance. This paper reports a new, optics-based technique for measurement of the strain distribution and tensile mechanical properties of the lumbar spine anterior longitudinal ligament (ALL). The ALL is the largest and strongest of the spinal ligaments and stabilizes the spine during extension, lateral bending and rotation, and also serves to protect the spinal cord by restricting the motions of spinal segments within certain limits. In the lumbar region, the ALL is a well-defined, relatively flat, thin and broad ligament of uniform cross section, attaching firmly to all the anterior surfaces of the vertebrae. These features make this ligament ideally

1185

1186

P. NEUMANNet al.

suited for the evaluation of regional differences in ligament mechanical properties. The goals of this study were as follows. (1) To describe a new approach in the measurement of the mechanical properties of ligamentous tissue. (2) To measure the regional strain variation along the length and width of the lumbar spine ALL. (3) To evaluate the tensile mechanical properties of the ALL substance, insertions and overall ligament. All the specimens tested were from young individuals with no skeletal or disc pathology. TEST SPECIMRN PREPARATION

Fresh human lumbar motion segments were obtained from cadavers. Frontal and lateral X-rays were obtained to identify any skeletal pathology and the segments were also graded based on their macroscopic degree of disc degeneration. Only segments with normal-appearing discs and no skeletal pathology were selected for this study. A total of six specimens were subsequently obtained from three male cadavers

Table 1. Summary of human ALL anatomical measurements. The parameters are defined in the text. Age (yr) 21

29 43

Preparations 1

2 3 4 5 6

Level T12-Ll L2-L3 LA-L5 Ll-L2 T12-Ll L2-L3

MSH DHA DHP LA (mm) (mm) (mm) (mm2) 77.0 76.0 73.8 78.4 73.0 83.0

6.5 9.8 10.0 8.5 9.5 9.0

5.3 8.4 8.0 7.2 9.0 9.0

36.5 39.8 43.5 33.5 39.5 36.5

Fig. 1. Schematic representation of the fixation device designed for implantation of the Steinmann pins. The motion segment is supported along the superior-inferior axis by two clamps (A)which are tightened using thumb screws (B).Drill guides (C) attached to the clamps above and below the segment are used to insert four Steinmann pins through the lateral aspect of each vertebrae. The drill guides can be positioned along the longitudinal axis of the motion segment in order to accommodate different-size segments and disc spacing. The hole pattern in the drill guides matches a similar pattern in the clamps attached to the mechanical-test apparatus.

aged 21,29 and 43 yr. Four levels between T12 and L.5 were represented (see Table 1). Each segment was sealed in an airtight plastic bag and frozen at - 35°C. Prior to testing, the specimens were thawed overnight at room temperature, after which the anterior motion segment height (MSH) and the anterior and posterior disc heights (DHA and DHP) were determined with a dial caliper. MSH was calculated as the distance from the anterior aspect of the upper endplate of the most superior vertebrae to the lower endplate of the inferior vertebrae. DHA and DHP were computed as the distance from the lower endplate of the superior vertebrae to the upper endplate of the inferior vertebrae. These measurements were repeated several times and the results averaged. Segments were prepared for testing by excising the posterior elements and carefully removing all connective soft tissues except for the ALL. The anterior column of the motion segment was then mounted in a special stainless-steel fixture which isolated the ligament and its bony attachments from damage during the fixation procedure (Fig. 1). The fixture supported each vertebrae along the superior-inferior axis and was used to insert several Steinmann pins into the lateral aspect of each centrum. Segments were carefully aligned in the fixture in order to provide an axial alignment and load on the ligament during mechanical testing. An L-shaped frame (drill guide) with a rectangular pattern of 4 mm diameter holes through each arm of the frame was attached to each end of the fixture and centered over each vertebrae. Four Steinmann pins (4 mm diameter) were driven through the drill guide and into the lateral aspect of each centrum. The rectangular pinhole pattern in the drill-guide frame matched a similar pattern on clamps attached to the tensile-testing apparatus. After the pins were inserted into both vertebrae, the fixture was removed, leaving the pins in place. With the help of a stereomicroscope, the posterior longitudinal ligament and intervertebral disc of each preparation were carefully dissected using a scalpel, leaving the ALL intact. This process was facilitated by flexing the motion segment during dissection, and continued until the demarkation between the red-blue, longitudinally oriented ALL fibers and the whitish, horizontally oriented annulus fibers was reached, at which point the annulus fibers were loosely attached to the ALL and easily separated. Care was taken not to disturb any part of the ligament and ligament-bone insertion sites during the fixation and dissection procedures, and the specimens were frequently irrigated with 0.9% saline in order to minimize drying. Twelve plastic beads (2 mm diameter) were then sewn to the ligament using 5-O nylon suture. Each suture penetrated the entire ligament thickness. A clear plastic template was used to aid the positioning of the beads in a 4 x 3 (column x row) grid with points at approximately 10 mm intervals and centered with respect to the intervertebral disc space (Fig. 2). The

1187

Mechanical properties of spinal ligaments

VIDEO

CCD B’ CAMERA

(MOTION

I

LOAD FRAME li0~D _________________-___--_____-_!

CELL

PROCESSOR

ANALYSIS

PC-AT ,

I

GRIP

VP310

N I COLET

4094(’

Fig. 2. Schematic diagram of the mechanical-test apparatus and data acquisition system. An enlarged view of the boneALL-bone preparation is also shown illustrating the 4 x 3 (column x row) positions of the 12 ligament markers. The central columns of the markers (2 and 3) are positioned at the level of the vertebral endplate, which is the margin between the ligament substance and the insertions regions. Insertion A corresponds to the superior aspect of the ALL and was attached to the fixed end of the tensile test apparatus.

attaches firmly to the edges of the vertebral bodies, which were about 10 mm apart in these segments (see DHA in Table 1); so the central columns of beads (Columns 2 and 3) were located on the margin between the ligament substance and the insertions. The region between the two innermost columns (Columns 2 and 3) of beads along the longitudinal axis of the ligament is, therefore, referred to as the ALL substance region, whereas the two outermost columns (Columns 1 and 2 and Columns 3 and 4) of beads adjacent to the substance region are referred to as the insertion regions (B and A, respectively). Insertions A and B correspond to the superior and inferior aspects of the ALL, respectively. The Steinmann pins served to attach the boneALL-bone preparations to two clamps, which were pushed onto the pins on either side of the centrum of each segment, and were then bolted to the crosshead of a screw-type mechanical testing apparatus. The hole pattern in the clamps precisely matched that of the pins, thereby preserving the original in vitro anatomical orientation of the vertebral bodies and the ALL. Prior to mechanical testing, a normalizing distraction force (approximately 15% of the body weight) was applied to each preparation in order to restore the ALL to its in vivo preloaded condition (Tkaxcuk, 1968). Measurements of the ligament thickness and width at the level of the disc were then obtained using ALL

dial calipers (accuracy of 0.025 mm and reproducibility of about 5%). Preloading of the ligament helped to facilitate these measurements by increasing the tension of the collagen fibers, thereby increasing the stiffness of the ligament substance. The preload also approximated the load required to restore the segments normal DHA. Cross-sectional areas of the ligaments (LA) were calculated from ligament width and thickness by assuming a rectangular cross section. The ligaments were not mechanically preconditioned in any other way. MECHANICAGTESTING PROCEDURE

Each of the bone-ALL-bone specimens exhibited some relaxation following initial distraction at the 100 N preload. In order to minimize these viscous effects, each specimen was allowed to relax until the load stabilized, typically 3-5 min. The bone-ALL-bone preparations were then distracted to failure at a constant crosshead displacement rate of 2.5 mm s-i. Load and crosshead deformation were recorded at 500 Hz on a digital oscilloscope (Nicolet 4094C, Milwaukee, Wisconsin). These data were also digitized at 30 Hz and stored in a PC-AT computer for comparison with the video displacement data. Displacement of the 12 ligament markers were collected from a noninterlaced, 60-Hz CCD camera (NEC TI-23A,

P. NEUMANNet al.

1188

Tokyo, Japan) with a field of view of 60 mm. A second CCD camera with a field of view of 125 mm recorded the clamp-to-clamp deformation from two 5-mm diameter retroreflective markers for direct comparison with the crosshead deformation measured using an LVDT. The ligament markers and clamp markers were thresholded separately and the field of each camera was calibrated during each test (approximately 2 and 4 pixels mm-’ for cameras 1 and 2, respectively). Video data were collected at 30 Hz, resulting in 100-200 frames of video for each test. A trigger pulse was used to initiate the start of each test and to start data collection by the video processor and PC-AT digitizer. Thresholded video data were processed in real time for marker perimeter coordinates using a video processor (Motion Analysis VP310, Santa Rosa, California), and transferred to a Sun 3/l 10 computer. The raw video data were also stored on a videotape.

DATA ANALYSIS

Following testing, each frame of the thresholded video was processed by the Sun 3/110 computer, which computed the x, y coordinates of the centroids from the perimeter coordinates of each of the 12 ligament markers (Ml, . . . , M12) and two crosshead markers (Cl, C2). These data were then transferred to a PC-AT computer for evaluation of the ligament strains. Tensile strains of the ligament substance (three in total) and insertions (six in total) were determined from the longitudinal path of each pair of markers along each row. The strains were computed as the change in displacement divided by the initial distance between the markers. Zero strain was defined at the 100 N preload. In this fashion, nine separate ligament strain histories were computed, three each for the regions enclosed by Columns 1 and 2 (insertion B), Columns 2 and 3 (substance) and Columns 3 and 4 (insertion A). The ‘overall’ strains &l-4 were also computed from the outermost pairs of markers (Columns 1 and 4 in Fig. 2). Hence, a total of 12 ligament strain histories were determined from the 12 marker displacement histories. Assuming a field width: target diameter ratio of 30: 1, an accuracy of 1 part in 6500 has been reported for marker centroid data processed on this system (Walton, 1988). Using the field of view of the ligament marker camera as a reference (60 mm), the predicted marker centroid displacement resolution was 0.01 mm (60/6500), indicating that strains of approximately 0.1% could be measured. Comparisons of LVDT displacement histories with the marker displacement histories obtained from the crosshead indicated that the marker displacement measurements were accurate to within 0.02 mm. Ligament stress u was computed as the load/ligament substance area, and the ultimate tensile strength cult was defined as the stress at failure. The peak strain for each of the 12 ligament stress-strain histories was

defined as the strain corresponding to e,it. The maximum strain value recorded from the 12 ligament strain histories was defined as E,,,. Stress-strain curves for the ligament substance, for insertions A and B, and for the overall ligament were used to evaluate the ALL mechanical properties. For these calculations the strains were expressed as column averages. Tensile moduli for insertion A (EA),insertion B (En), substance (Es) and ‘overall’ ligament (El+) were computed as the ratio of stress/strain from the straight-line portion describing the linear region (LZ) of the stress-strain curves prior to yield (White and Panjabi, 1990). The same stress range was used, therefore, to calculate E,, Es, Es and E,, for each specimen, but varied between specimens since the stress-strain values in the LZ region were not consistent for all the specimens. Load-crosshead-displacement data were also used to evaluate the ALL mechanical properties. Crosshead displacement data were calculated from both the LVDT signal and the video markers located on the crosshead (markers Cl and C2). LVDT displacement values were corrected for the testing-machine compliance. Strains were then estimated by dividing the crosshead displacement by the motion segment height. MSH was used to normalize the crosshead displacement, since the exact length of the ALL could not be precisely determined prior to testing without damaging the ligament. The tensile modulus (E,,,) and ultimate strain (si,,sH)were then computed from the stress-strain curves in the same manner as described previously. RESULTS

The average motion-segment height, anterior disc height, posterior disc height and ligament crosssectional area were 76.9f 3.6 mm, 8.9f 1.3 mm, 7.8 + 1.4 mm and 38.2 + 3.5 mm2, respectively (Table 1). All the motion segments had normal-appearing discs (grade zero on a scale of three) and had no skeletal pathology when examined radiographically. Crosshead displacement

The constant deformation test conditions employed in this study resulted in an average ‘overall’ ligamentous strain rate of 1.07% s-r (S.D. 0.44). The load-crosshead-displacement curves were nonlinear in the initial region of the curve (NZ) and in the region just prior to failure (PZ). The NZ region of the curves accounted for approximately 10% of the total deformation. The region (LZ) between the NZ and PZ was relatively linear (less than 7% absolute deviation of the stress values about the tensile line) and was the region with the greatest stiffness. The PZ region of the curves accounted for approximately 30-70% of the total deformation. Recordings of analog deformation (LVDT) and video crosshead deformation (markers Cl and C2) agreed closely (less than 2% average error).

Mechanical properties of spinal ligaments

of the ligament length. Notable exceptions were markers in the two outermost rows of Columns 2 and 3. These markers tended to follow a diagonal path toward the center of the ligament, resulting in a transverse motion equal to about 20-25% of the total longitudinal motion. Load-marker-displacement response curves (Fig. 3) exhibited the characteristic nonlinear regions (NZ and PZ) and a linear region (LZ), which were observed in the load-crossheaddisplacement curves. The relative displacement between markers along each row was greatest in the regions between Columns 2 and 3, resulting in higher peak strains in the mid-substance in comparison to the insertion sites (Fig. 4). During the early phase of loading, strains were generally higher in the ligament insertion sites in comparison to strains in the ligament substance. At the point of yielding, a slight decrease in strain occurred at one or both insertions. Following yield, all six specimens exhibited increased strains in

The average load, deformation and stress at failure were 1055 N (S.D. 308), 12.3 mm (S.D. 4.9) and 27.4 MPa (SD. 5.9), respectively (Table 2). Loads and stresses at failure were highest in the bone-ALL-bone preparations from the 21 yr old. Crosshead deformations at failure ranged from 8 to 20 mm. Examination of the raw video data indicated that a substantial portion of this deformation (approximately 4 mm or about 3@40% of the total deformation) occurred at the bone-pin fixation sites in two of the preparations (Preparations 2 and 5). Consequently, the computed strain values eDISHvaried considerably (mean= 19.7 f7.0%, range= 14.0-32.0%). The tensile modulus E MSHcomputed from the linear portion of these curves ranged from 103 to 237 MPa (mean 169 _t 53 MPa). Marker

1189

displacement

Marker displacement paths in the bone-ALL-bone preparations generally followed the longitudinal axis

Table 2. Summary of human ALL structural and mechanical properties. The parameters are defined in the text Tensile moduli (MPa)

Peak strain (%) Age (vr) 21

Preparations

1

Level

Load (N)

Difference (mm)

Stress W’a)

Ti2-Ll L2-L3 L4-L5

1060 1163 1607

8.0 20.1 11.1

29.0 29.2 36.9

h

cl4

14.0 32.0 18.3

G,,,

ES

Q

%

4,

4

4

Ee

29

4

Ll-L2

890

10.5

26.6

17.1

4.24 11.4 7.30 4.56 1.61 863 615 811 894 2.65 12.1 1.72 5.04 6.02 1367 640 871 447 6.94 12.5 10.1 3.15 5.62 505 379 1200 469 5.70 14.7 10.3 4.74 4.12 713 724 565 502

43

5 6

T12-Ll L2-L3

885 756

14.2 9.8

21.7 20.7

23.2 13.3

4.35 8.0 5.79 13.8

2 3

COLUMN I

COLUMN 2

LIGAMENT

COLUMN 3

DEFORMATION

5.01 3.69 5.24 8.72 3.46 6.75

680 835 424 286

481 481 551 349

COLUMN 4

(m)

Fig. 3. Typical recordings of the 12 ALL load-marker-displacement load histories (Preparation 4). Each marker displacement history represents the change in marker displacement relative to its initial location prior to loading. The maximum displacement value (mm) for each marker (Ml, . ., M12) is indicated in the lower right-hand corner of each curve. Individual marker displacements are scaled to the maximum displacement recorded among the 12 markers (i.e. M9 for this specimen). The largest deformation between markers and the ultimate site of failure occurred in the outer portion of the ligament substance (markers M2, M3 and MlO, Mll). Similar findings were obtained for the other five preparations.

P. NEUMANN et al.

1190 30.5 - INSERTION

A

E&S 4.7%

-SUBSTANCE

r OVERALL

STRAIN

(%I

16.0 P

645.7%

STRAIN

(96)

I6

Fig. 4. Stress-strain relationships of human ALL substance and insertions (Preparation 4). The curves shown represent row averages of the ALL strain response. The ‘overall’ stress-strain curve is also shown in the lower right-hand corner. Initial positions of the 12 ligament markers are illustrated in the inset of each curve. The larger dots in each inset indicate the marker pairs which correspond to the region analyzed. The yield point is indicated by a ‘+’ symbol on each curve. Substance strains were over twofold greater than strains at the bone-ligament insertion sites. A slight decrease in strain occurred at both insertion sites, neither of which exhibited any appreciable yielding. Yielding and ultimate failure was clearly evident in the ligament substance. region.

the ligament

substance and each failed within this region. The stress-strain behavior of the ‘overall ligament lay between that of the insertions and the substance. The peak strain values and modulus values computed from the marker displacement data were relatively consistent among the six bon&ALL-bone preparations tested (Table 2). No particular trends for any of the peak strain values were observed on the basis of specimen age or segment level. Of note, however, was the finding that the ‘overall’ peak strain Q_., (mean 4.95 & 1.51%) was significantly lower (ANOVA, P ~0.01) than the peak crosshead strain by a factor of about 2-12. In accordance with the ultimate stress data, tensile modulus values for the ‘overall’ ligament and ligament insertions tended to be higher for the younger preparations (Preparations l-3). The mean values for El,, E,, E, and E, were 759 MPa (SD. 336), 747 MPa (S.D. 271), 524 MPa (S.D. 189) and 580 MPa (S.D. 209), respectively. These modulus values are about three to five times greater than the modulus calculated from the load-crosshead-displacement data. The slightly higher average Ei+ in comparison to the individual component tensile moduli (EA, EB and Es) reflects, in part, the contribution of values from Preparation 2 (see Table 2). In this specimen there was nearly a twofold greater El+ in comparison to the average of the component moduli, which was attributed primarily to transverse marker misalignment in Columns 3 and 4. Variations in ligament strains were significant across the ligament width and also along the ligament

and peak substance strains in the outer portions of the ligament (Rows 1 and 3) were significantly greater (over 40%, ANOVA, P~0.05) than the central region (Row 2). Highest strains consistently occurred in the outer portions of the ligament substance (Row 1, Column 2-3 and Row 3, Column 2-3), and maximum strains E,,, for these regions averaged 12.1+ 2.3%. The average peak strain in the ligament substance es (8.18 + 1.96%) was significantly greater (ANOVA, P < 0.01) than the average peak strain in the ligament insertions (sA=4.11f0.77, ~,=4.89+_1.83%). There was some asymmetry in the strain distribution across the width of the ligament, particularly at the ligament insertion sites, as was evident by the approximately 50% higher strains in Row 3, Column l-2 and Row 1, Column 3-4 in comparison to Row 1, Column l-2 and Row 3, Column 3-4 (see Fig. 5). length (Fig. 5). Peak ‘overall’ strains

DISCUSSION

The measurement and analysis of the ligament strain is a difficult problem. This study has described an optics-based strain measurement technique and an analysis of boneligament-bone preparations of the ALL which provides a direct measure of ligamentous mechanical properties, eliminates the complications associated with the testing of isolated ligament structures, and allows one to determine regional variations in the mechanical properties of bone-ligament-bone preparations. Therefore, the data obtained should be

Mechanical properties of spinal ligaments

Peak Strain (Wl

IXlRowl

ORow2

-Row3

Fig. 5. Histogram showing the mean variation in peak strain along the length (columns) and width (rows) of the human ALL. Substance strains were significantly greater than insertion strains, and peak strains were significantly higher in the outer portion of the ALL in comparison to the central portion. Comparison of the strain distribution in Columns 1 and 3 indicates that some asymmetry in the peak strain pattern was present, which was attributed to a local deformation at the bone-pin fixation sites.

more representative of the actual ligament mechanical behavior. We have used this technique to evaluate regional strain variations in the normal lumbar ALL, and we provide the first detailed description of the mechanical properties at the ALL substance and at the insertion regions. Assumptions and limitations In this study we have examined the in vitro biomechanical behavior of the anterior longitudinal ligament by applying loads along the longitudinal axis of the ligament after removing the intervertebral disc. In addition to disturbing the normal functional environment of the ALL, several assumptions were required in order to determine the mechanical properties of the ALL. First, the ligament stresses were based on measurements of ALL cross-sectional area at a single site corresponding to the level of the intervertebral disc. In our analysis of the ALL mechanical properties we have assumed that the cross-sectional area of the ALL can be approximated by that of a rectangular cross section, which is constant along its length and remains constant during testing, whereas in reality the width and thickness of the ALL vary along its length and may decrease in thickness and width during loading. Magnetic resonance imaging and/or nondestructive techniques for measurements of connective tissue structure (Shrive et al., 1988) could be used to more GM 25:10-G

1191

precisely quantitate regional variations in the ALL substance cross-sectional area prior to mechanical testing and would, therefore, provide an improved characterization of stresses in these regions. Second, the video strain measurement technique required markers (targets) to be attached directly to the ligament using fine suture. A direct examination of the ligamentous tissue and a frame-by-frame review of the raw video did not reveal any artifacts such as premature failure at the marker attachment sites. In analyzing these data we have also assumed that the ALL is a planar structure, whereas in reality the anterior surface of the motion segment is curved--the central row of markers were approximately l-2 mm closer to the camera than the outer rows. However, this has very little effect on the video-based marker displacement measurements since a 25 mm height differential produces less than 1% error in the measured centroid distances between two markers located 10 mm apart. Third, the mechanical testing protocol involved pre-tensioning of the bone-ALL-bone preparations to a load of 100 N or approximately 10% of the failure load. This protocol provided a reproducible method for obtaining accurate estimates of the crosssectional geometry and is consistent with the notion that the spinal ligaments are normally in a pre-tension physiological state (Nachemson and Evans, 1968; Tkazcuk, 1968; Chazal et al., 1985). The 1OON preload, however, is higher than the physiologic preloads (about 20 N) estimated for the spinal ligaments (Nachemson and Evans, 1968). We recognize, therefore, that by assuming zero strain at the preload, we have underestimated the magnitude of the peak strains. The data presented by Chazal et al. (1985) suggest that our peak strain values were underestimated by about 10% of the total strain. Pre-tensioning of the ALL would not be expected to alter the peak strain patterns within the ALL, nor would it affect the ultimate stress and tensile modulus calculations. Finally, the spatial resolution of the video processor (0.01402 mm) limited the accuracy of strain measurements on the ALL to about 0.1-0.2% strain. This was judged to be adequate to characterize the mechanical behavior of the ALL, since peak strains were greater than about 3%. Structural properties The mean deformation at failure for the boneALL-bone specimens in this study was 12.3 mm. This agrees well with the ultimate crosshead deformation of the bone-ALL-bone preparations and with the values reported in the literature. Pintar et al. (1986) reported deformations at failure for the lumbosacral ALL ranging from 3 to 36 mm (mean 16) for fresh human cadavers with an average age of 69 yr. In a survey of the literature, White and Panjabi (1990) reported a mean ALL deformation at failure of 15.2 mm (range 7-20). Our strength values (loads and stresses) for the ALL, however, were two to threefold greater than the

1192

P. NEUMANN et al.

average strength data obtained in previous studies (450 N, 11.6 MPa; see review by White and Panjabi, 1990). In one of the studies reviewed by White and Panjabi (Chazal et al., 1985), the forces required to fail isolated ALL preparations obtained from five 63-80 yr old human lumbar spines ranged from 440 to 520 N or roughly 3&40% lower than the weakest specimen tested in the present study. Pintar et al. (1986) reported even lower mean failure loads for the aged lumbar ALL (391 N). We believe that these marked differences are due to the fact that we tested tissues obtained from relatively young and normal spines. Tkaczuk (1968) found that ultimate loads, deformations and strength (stress at failure) of isolated preparations of the ALL decreased with age and were lower for specimens associated with degenerated discs. Nachemson and Evans (1968) noted a fivefold decrease in the mechanical properties of isolated preparations of the lumbar spine ligamenturn flavum between the ages of 20 and 80 yr. Similar strength disparities have been reported for other ligamentous tissues (Noyes et al., 1984). In this study we noted a slight decrease in the ALL strength between the ages of 21 and 43 yr. Strain measurements

We found that ALL strains measured directly using the marker displacement data were much lower than ‘specific deformation’ values calculated using traditional crosshead deformation. The crosshead measurement technique significantly overestimated the overall and regional strains by a factor of two or more. This finding is consistent with previous studies of connective tissues that have compared directly measured surface strains to strains based on crosshead measurements (Woo et al., 1983; Noyes et al., 1984). These authors attributed the differences between specific deformation and surface strains to increased strains at the ligament insertion sites. In this study, differences found between marker strains and crosshead strains reflect, in part, deformations occurring at the pin-bone grip interface, which were estimated to be as much as 40% of the total crosshead deformation as opposed to increased strains at the insertion sites. These differences also reflect the difficulty of precisely determining the initial ligament length. Here we have assumed that the initial ligament length could be approximated by the segment height. Had we used lumbar ALL lengths based on insertion-to-insertion measurements (approximately 10-15 mm), our crosshead-based measurements of ALL strains would have been much greater. Few reports of ultimate strain values for the ALL can be found in the literature and data for other spinal ligaments have been highly scattered. Chazal et al. (1985) reported specific deformations for five isolated lumbar ALL preparations (ages 63-80 yr) ranging from 43 to 59% and, in a survey of the literature, White and Panjabi (1990) tabulated a range of strain

values at failure of g-57% for the anterior and posterior spinal ligaments. Our specific deformation values for the bone-ALL-bone (sMsH= 1432%) were comparable to these ranges. The peak strains measured directly from the ligament substance (and insertions) using the motion analysis system (E,,, = 5-15% and ss = 5-lo%), however, indicate that the actual strain values in the ligament were much lower. Although there are no reports of ultimate strains at failure for the bone-ALL-bone in the literature, our results are consistent with those for peak strains of bone-ligament-bone preparations of the posterior spinal ligaments (transverse, capsular, supraspinous, ligamenturn flavum) obtained by Goel and Njus (1986) using the VDA technique. They noted that the total strain at failure ranged from 10 to 15% in the four specimens studied. Our findings also compare well with the maximum physiologic strains computed for the human ALL (13%) using a mathematical model (Panjabi et al., 1982), and are consistent with peak surface strains obtained for other connective tissues using optical strain-measurement techniques (Woo et al., 1983; Noyes et al., 1984; MacKenna et al., 1989; Newton et al., 1989). Mechanical properties

Data concerning the stress-strain characteristics of the ALL are lacking in the literature and we are not aware of any previous publications on the ALL modulus values. Tensile modulus values for the ‘overall’ ligament E,, averaged 759 MPa (range 4241367 MPa). These values are much higher than those reported by Goel and Njus (1986) for human posterior spinal ligaments of unspecified age and pathology (E = 12-24 MPa). The precise reason for this great disparity is unclear, but we speculate that the tissues tested by Goel and Njus must have been from aged subjects, since tensile modulus values obtained from studies of isolated human posterior spinal ligaments (ligamenturn flavum, interspinous ligaments) are much higher, ranging from 17 to 169 MPa for ages between 79 and 13 yr (Nachemson and Evans, 1968; Waters and Morris, 1973). Nevertheless, it would appear that the stiffness of the ALL is substantially greater than that of its posterior spinal ligament counterparts. This is not surprising considering the fact that the ALL has a large percentage of longitudinally oriented and close-packed collagen fibers in contrast to posterior spinal ligaments, such as the ligamenturn flavum, which has a very large percentage of elastin fibers and, hence, greater extensibility (Ramsay, 1966). Regional variations in ALL mechanical properties

The ALL tensile strains were evaluated at nine sites in the six ALL specimens examined. These measurements provided a topographical description of the strain variations along the length and width of the ALL. Along the length of the ligament, the strains at

Mechanical properties of spinal ligaments failure were found to be greatest in the ligament substance. The average substance strains at failure were significantly greater (about twofold) than the insertion strains, and across the width of the ligament the peak strains were significantly greater in the outer portion. At mid-substance, the average strains at failure in the outer portion of the ligament were roughly twofold greater than the central portion. Failure consistently occurred in the outer portion of the ALL substance region. This finding agrees with those of other studies which have determined that the mid-substance region is typically the site of failure in bone-ligament-bone preparations from adult animals (Woo et al., 1986) and adult humans (Noyes and Grood, 1976). Variations in the mechanical behavior of the ALL in the linear portion of the stress-strain curves were also noted along the length of the ALL. The tensile modulus tended to be highest at the superior insertion site, and insertion moduli averaged about 20% higher than the substance modulus values. The functional significance of this is not known. Some asymmetry in the ‘average’ regional peak strain patterns was observed in insertion regions of the six bone-ALL-bone preparations. Based on an examination of individual strain patterns, this asymmetry did not appear to be physiological as there was no consistent pattern from one ligament to another. Rather, the peak strain asymmetry in Fig. 5 reflects a local asymmetrical deformation at the bone-pin interfaces since a significant amount of deformation occurred at the bone-pin interfaces of several of the specimens examined. Preliminary work in our laboratory indicates that deformation of the bone-pin interface appears to be related to the quality of bone at the pin sites, and reflects the fact that the ALL strength is similar to the strength of the cancellous bone comprising the vertebral centrum (Neumann et al., 1990). Improvements in fixation techniques which minimize local bone deformations are a prerequisite for future studies of the ALL mechanical properties, and are the focus of current investigations in our laboratories. Future studies of the age-related stress-strain behavior of spinal ligaments may provide insight into the structure--function relationships associated with ageing and disease. In this paper we have been concerned with the anterior longitudinal ligament (ALL), but the technique could be applied to other ligaments or soft tissues.

Acknowledgements-The authors thank Mr Wan Suwito, Mrs Mary Margaret Peel and Mr Erik Huh for assisting in

preparation of the figures.

REFERENCES Arms, S. W., Pope, M. H., Johnson, R. R., Fischer, R. A., Arvidsson, I. and Eriksson, E. (1984) The biomechanics of anterior cruciate ligament rehabilitation and reconstruction. Am. J. Sports Med. 12, 8-18.

1193

Chazal, J., Tanguy, A., Gaurel, B. G., Escande, G., Guillot, M. and Vanneuville, G. (1985) Biomechanical properties of spinal ligaments and a histological study of the supraspinal ligament in traction. f. Biomechanics 18, 167-176. Edwards, R. G., Lafferty, J. F. and Lange, K. 0. (1970) Ligament strain in the human knee joint. J. basic Engr. (ASME Trans.) 131-136. Goel, V. K. and Njus, G. 0. (1986) Stress-strain characteristics of spinal ligaments. Trans. 32nd Ann. Orthop. Res. sot. 11, 370.

Howe, J., Wertheimer, C., Johnson, R., Nichols, C. and Pope, M. (1990) Arthroscopic strain gauge measurement of the normal anterior cruciate ligament. Arthroscopy 6, 198. Kennedy, J. C., Hawkins, R. J. and Willis, R. B. (1977) Strain gauge analysis of knee ligaments. C2in. Orthop. 129, 225-229. MacKenna, D. A., Newton, P. O., Lyon, R. M., Akeson, W. H. and Woo, S. L.-Y. (1989) The effect of hemarthrosis and synovectomy on the mechanical properties of the anterior cruciate ligament (ACL). ASME Publ. AMD-95, 6568. Nachemson, A. and Evans, J. (1968) Some mechanical properties of the third lumbar interlaminar ligament (ligamentum flavum). J. Biomechanics 1, 211-220. Neumann, P., Hansson, T., Ekstriim, L. and Keller, T. (1990) Strength properties of the anterior longitudinal ligament in relation to vertebral bone mineral content. In Proc. First World Congress of Biomechanics, Vol. ii, p. 68. La Jolla, California. Newton, P. O., MacKenna, D. A., Lyon, R. M., Akeson, W. H. and Woo, S. L.-Y. (1989) Comparison of the mechanical properties of the medial collateral and anterior cruciate ligaments of the rabbit knee. ASME Publ. AMD-95,53-56. Noyes, F. R., Butler, D. L., Grood, E. S., Zernicke, R. F. and Hefty, M. S. (1984) Biomechanical analysis of human ligament grafts used in knee-ligament repairs and reconstructions. J. Bone Jt Surg. 66A, 344-352. Noyes, F. R. and Grood, E. S. (1976) The strength of the anterior cruciate ligament in humans and rhesus monkeys. J. Bone Jt Surg. 58A, 10741082. Panjabi, M. M., Gael, V. K. and Takata, K. (1982) Physiologic strains in the lumbar spinal ligaments. An in vitro biomechanical study. Spine 77 192-203. Paniabi. M. M.. Jorneus. L. and Greenstein. G. 11984) Physical properties of lumbar spine ligaments.’ Trans. 30ti Ann. Orthop. Rex Sot. 9, 112. Pintar, F. A., Myklebust, J. B., Yoganandan, N., Maiman, D. J. and Sances, A. (1986) Biomechanics of human spinal ligaments. In Mechanisms of Head and Spine Trauma (Edited by Sances, A., Thomas, D. J., Ewing, C. L., Larson, S. J. and Unterharnscheidt, F.), pp. 505-527. Aloray Publishers, Goshen, NY. Ramsay, R. H. (1966) The anatomy of the ligamenta flava. Clin. Orthop. 44, 129-140. Shrive, N. G., Lam, T. C., Damson, E. and Frank, C. B. (1988) A new method of measuring the cross-sectional area of connective tissue structures. J. biomech. Engng 110, 104109. Tkaczuk, H. (1968) Tensile properties of human lumbar longitudinal ligaments. Acta Orthop. Stand. 115 (Suppl). Walton, J. S. (1988) The accuracy and precision of a video motion analysis system. In High Speed Photography, Videography, and Photonics IV: Proc. SPIE (Edited by Ponseggi, B. G.), Vol. 693. The International Society for Optical Engineering. Wang, C. J., Walker, P. S. and Wolf, B. (1973) The effects of flexion and rotation on the length patterns of the ligaments of the knee. J. Biomechanics 6, 587-596. Waters, R. L. and Morris, J. M. (1973) An in oitro study of normal and scoliotic interspinous ligaments. J. Biomechanits 6, 343-348.

White, A. A. and Panjabi, M. M. (1990) Clinical Biomechanics ofthe Spine (2nd Edn). J. B. Lippincott Company, Philadelphia, PA.

1194

P. NEUMANNet al.

Woo, S. L.-Y., Gomez, M. A., Seguchi, Y., Endo, C. M. and Akeson, W. H. (1983) Measurement of mechanical properties of ligament substance from a bon~ligament-bone preparation. .I. orthop. Res. 1, 22-29. Woo, S. L.-Y., Gomez, M. A., Woo, Y. K. and Akeson, W. H. (1982) Mechanical properties of tendons and ligaments. Biorheology 19, 385408.

Woo, S. L.-Y., Orlando, C. A., Gomez, M. A., Frank, C. B. and Akeson, W. H. (1986) Tensile properties of the medial collateral ligament as a function of age. J. Orthop. Res. 4, 133-141.