The Bennett fracture or fracture-dislocation, first described by Bennett1 in 1882, is the most common intra-articular injury involving the trapeziometacarpal joint10,26. Unlike the more comminuted Rolando fracture, the Bennett fracture consists of a single vertical intra-articular fracture line at the base of the metacarpal of the thumb. The fracture line separates the main fragment of the metacarpal shaft from the smaller fragment of the palmar beak (the volar articular prominence), which is attached to the primary static stabilizer of the trapeziometacarpal joint, the palmar beak ligament. Without this ligamentous support, the metacarpal shaft subluxates in a dorsal and radial direction secondary to the combined force couple of the abductor pollicis longus and the adductor pollicis.
Treatment of this fracture has remained controversial. More than twenty methods, ranging from no reduction and an early range of motion2 to the most recent recommendation for open anatomical reduction and internal fixation of the articular surface11, have been described in the literature9,12,14,27,29. Many studies have shown that, regardless of the method of treatment, few patients have clinically important pain after the fracture has united and meaningful functional impairment is uncommon3,13,16. One study even suggested that there is no relationship between reduction of the fracture, the range of motion, pain, and future osteoarthrosis3. Nonetheless, many patients have some restriction of motion3,13,15,24 as well as abnormal radiographic findings demonstrating degeneration of the trapeziometacarpal joint3,10,11,13,16,21,28. Only rarely, however, has a patient with a remote history of Bennett fracture been included in a study of the results of arthroplasty of the trapeziometacarpal joint3,21.
Despite the wide enthusiasm for anatomical reduction of articular fractures in most joints, there is considerable uncertainty with regard to the need for anatomical repair of the metacarpal articular surface in the treatment of Bennett fractures. Good functional results in historical series in which closed treatment and indirect reduction was used do not suggest a mandate for contemporary techniques of open anatomical restoration of the joint surface. Paradoxically, these joints commonly had a residual articular step-off in the region of the fracture after closed treatment; while this does not guarantee the development of late degenerative changes, persistent incongruity of the joint surface is generally believed to be a primary causative factor in post-traumatic osteoarthrosis.
The purpose of the current study was to investigate the effects of residual articular incongruity, after simulation of a Bennett fracture, on loading of the joint surface and the mechanical implications for the late development of degenerative osteoarthrosis of the trapeziometacarpal joint.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. Funds were received in total or partial support of the research or clinical study presented in this article. The funding sources were The American Society for Surgery of the Hand and National Institutes of Health Biomedical Research Support Grants RR5353 (Stanford University) and R27213 (University of Rochester).
†Read in part at the Annual Meeting of the American Society for Surgery of the Hand, Phoenix, Arizona, November 11, 1992.
‡Prevea Clinic, 1551 Dousman Street, P.O. Box 19070, Green Bay, Wisconsin 54303-9070.
§Departments of Orthopaedics and Rehabilitation (M. A. P. and V. D. P., Jr.) and Biostatistics (V. M. C.), Milton S. Hershey Medical Center, Pennsylvania State University, Box 850, Hershey, Pennsylvania 17033. Please address requests for reprints to Dr. Pellegrini.
A biomechanical study of the trapeziometacarpal joint was carried out with use of fifteen fresh-frozen forearm specimens from the cadavera of individuals who had had no history of connective-tissue disease. The radius and the ulna of each specimen were transected at the proximal level of the forearm and mounted on a Plexiglas stage. Dissection of the specimens and simulated lateral pinch-testing with pressure-sensitive film were conducted as described previously4-6,22.
The tendons of the extensor pollicis longus, flexor pollicis longus, and abductor pollicis longus were identified and isolated in the forearm proximal to their natural retinacular pulleys. The origins of the abductor pollicis brevis and flexor pollicis brevis muscles radial to the tendon of the flexor pollicis longus were dissected together and then elevated to expose the underlying capsule of the trapeziometacarpal joint. Similarly, the origins of the adductor pollicis and the ulnar portion of the flexor pollicis brevis were dissected jointly from the palmar aspect of the third metacarpal. The tendons were then extended with monofilament nylon and were routed through holes in the Plexiglas stage to recreate their physiological directions of muscle force. The abductor pollicis brevis muscle was routed tangential to the plane of the palm toward the pisiform. The adductor pollicis was extended with a wire directed through the third intermetacarpal space to the dorsum of the hand. The extrinsic tendons were held in position by the retinacular pulleys and were kept in line with the axis of the forearm.
To simplify the model and to prevent collapse deformity, the wrist and the metacarpophalangeal joint of the thumb were stabilized in 30 degrees of extension and 20 degrees of flexion, respectively, with multiple Kirschner wires. To provide a stable platform for lateral pinch, the metacarpophalangeal and proximal interphalangeal joints of the index finger were similarly stabilized in 90 and 30 degrees of flexion, respectively.
Ultra-low-pressure-sensitive film (Fuji Medical Systems U.S.A., Stamford, Connecticut) was cut from a template of each joint surface. The film was inserted into the joint through an incision in the membranous thenar capsular recess, between the expansion of the abductor pollicis longus and the palmar beak ligament, which did not compromise the stability of the trapeziometacarpal complex. The orientation of the film was consistently maintained in the joint by the remaining three sides of the capsule, and the film was affixed to the trapezial surface with double-faced tape.
Lateral pinch-testing was performed in two modes. Static pinch was simulated by loading the extensor pollicis longus (at 0.83 newton), the adductor pollicis (at 4.18 newtons), the abductor pollicis brevis (at 5.01 newtons), the flexor pollicis longus (at 8.36 newtons), and the abductor pollicis longus (at 13.37 newtons) in a 1:5:6:10:16 ratio to create a pulp pressure of five to ten kilopascals4,22. Dynamic pinch-testing involved ten cycles of simultaneous loading of the motor units just mentioned, alternating with extension produced by the removal of load from the abductor pollicis brevis, adductor pollicis, and flexor pollicis longus motor units22.
After standard contact-area patterns had been measured in each intact joint specimen, a Bennett fracture was simulated by performing an osteotomy of the metacarpal base, halfway between the palmar beak and the dome of the dorsopalmar concavity of the metacarpal articular surface. The metacarpal beak fragment, with the palmar beak ligament attached, was then recessed two millimeters and secured in close apposition to the metacarpal shaft with a transverse Kirschner wire (Fig. 1). These conditions, with osseous apposition of the fracture fragments and a residual step-off in the articular surface, reproduced the result that is commonly obtained after use of the popular method of closed reduction and percutaneous pinning of the metacarpal shaft for the treatment of a Bennett fracture28,30. Contact-area patterns were generated again during static and dynamic lateral pinch-testing.
After all measurements of pressure had been made, the condition of the joint surface and the supporting structures was inspected and recorded to allow relationships between gross abnormalities and contact-area-pattern data to be determined. The specimens were then divided retrospectively, on the basis of the condition of the articular cartilage, into three groups: those that were non-osteoarthrotic, those that were moderately osteoarthrotic, and those that had end-stage osteoarthrosis. Non-osteoarthrotic specimens were defined as those that had articular cartilage throughout the articular surfaces of the metacarpal and the trapezium; moderately osteoarthrotic specimens, as those that had eburnation of bone limited to the palmar portions of the metacarpal and trapezial articular surfaces; and specimens with end-stage osteoarthrosis, as those that had more extensive articular damage with nearly complete absence of both the metacarpal and the trapezial cartilage.
The contact pressures were quantified with use of a standard curve (Fuji Medical Systems U.S.A.), corrected for relative humidity and temperature, which allowed conversion of color density to numerical values for peak contact pressure with use of the NIH Image 1.37 graphics program (National Institutes of Health, Bethesda, Maryland). Although the ultra-low-pressure-sensitive film has a useful pressure range of fifty to 200 kilopascals, the spectrum used for computer analysis of joint contact pressures was eighty to 200 kilopascals as this range was effective in eliminating artefact caused by handling and cutting of the film. Pressure-sensitive film records the summation of all loading events to which it is exposed on the surface in question. For a single loading event, the recorded pressure represents the peak or maximum pressure at any given point on the film. The mean peak pressure reflects the arithmetic mean of the peak pressures for all specimens.
The pressure-sensitive-film images were digitized with a high-resolution scanner. The trapezial joint surface was divided into five areas, from palmar (areas 1 and 2) to central (area 3) to dorsal (areas 4 and 5), to allow regional analysis of the contact areas. The NIH Image 1.37 graphics program was used to measure the contact areas and to calculate the relative shift in the primary contact-area zones. The articular contact area, as defined in this study, represents the area of the joint in which peak pressures were eighty kilopascals or more. The total contact area includes all areas with peak pressures of at least eighty kilopascals within any part of the joint. The regional contact area is defined as the fraction of the total contact area located within one of the five (palmar-to-dorsal) zones of interest. The mean contact area represents the arithmetic mean of the contact areas for all specimens.
Statistical Analysis
Contact area and contact pressure were measured in each of the five zones before and after static and dynamic lateral pinch-testing. A Student t test was used to compare the individual measurements of area and pressure between specific compartments within each group of specimens. Additionally, difference scores were constructed as the measurement after the fracture minus the measurement before the fracture for each compartment in each group. Thus, we conducted four sets of statistical analyses: pressure under static pinch, pressure under dynamic pinch, area under static pinch, and area under dynamic pinch. Within each set, a multivariate analysis of covariance was applied with use of the five measurements corresponding to the five locations as the response variables. The multivariate analysis of covariance model included intercepts and effects for the osteoarthrotic condition (non-osteoarthrotic, moderately osteoarthrotic, and severely osteoarthrotic). The objective of the multivariate analysis of covariance was to determine whether the set of five intercepts differed significantly from zero while adjusting for the osteoarthrotic condition. In addition, univariate analysis of covariance was applied to each of the difference scores at each location. All calculations were performed with use of PROC GLM software (SAS Institute, Cary, North Carolina).
Of the fifteen specimens in this study, six were non-osteoarthrotic, five were moderately osteoarthrotic, and four showed end-stage osteoarthrosis. Analysis of covariance for the osteoarthrotic condition of the joints demonstrated no significant differences, with the numbers available, among the three groups with regard to the regional parameters measured in this investigation. Accordingly, the regional areas of contact and the mean peak pressures are first considered separately for each group of specimens and then as an aggregate analysis of all specimens.
Non-Osteoarthrotic Specimens
Gross examination of the specimens in the non-osteoarthrotic group revealed some modest wear of the articular cartilage and chondromalacia; however, cartilage was present throughout the articular surface. This group therefore included specimens with both normal articular cartilage and cartilage showing changes of physiological aging. The surrounding ligamentous structures, including the palmar beak ligament, were intact and functional.
Static pinch-testing before the fracture revealed a contact-area pattern that was predominantly palmar (Figs. 2-A and 5-A), with peak pressures being highest in the palmar compartments (Fig. 6-A). A significant dorsal shift in both contact area and peak pressure (p < 0.05 for each) was observed after simulation of the Bennett fracture (Figs. 2-B, 5-A, and 6-A). With the numbers available, the mean total contact area showed a non-significant (40 per cent) increase (p = 0.16), from 18.2 to 25.4 square millimeters.
Dynamic pinch-testing produced similar data for both regional contact area and peak pressure before simulation of the Bennett fracture, with both peak pressures and contact areas concentrated in the palmar-central compartment. After simulation of the Bennett fracture, dynamic pinch-testing also resulted in a significant dorsal shift in contact area and pressure (p = 0.02). With the numbers available, the mean total contact area showed a non-significant (14 per cent) increase (p = 0.16), from 55.4 to 63.0 square millimeters.
Moderately Osteoarthrotic Specimens
Each specimen in this group had eburnation of bone limited to the palmar aspect of the trapezial and metacarpal surfaces. Although the beak ligaments were functional, they showed mild attenuation and early detachment from the metacarpal insertion as compared with those in the specimens in the non-osteoarthrotic group.
Static pinch-testing showed a contact-area pattern that was predominantly palmar-central, with little contact in the most dorsal compartment (Figs. 3-A and 5-B). Peak contact pressures were also highest in the palmar and central compartments, with a significantly lower peak pressure (p < 0.05) in the most dorsal compartment (Fig. 6-B). A significant dorsal shift in both contact area and peak pressure (p < 0.05 for each) occurred after simulation of the Bennett fracture (Figs. 3-B, 5-B, and 6-B). With the numbers available, the mean total contact area showed a non-significant (75 per cent) increase (p = 0.16), from 16.6 to 29.0 square millimeters, after simulation of the Bennett fracture.
Dynamic pinch-testing revealed a central contact-area pattern, with peak contact pressures that were highest in the central compartment. A dorsal shift in both contact area and peak pressure occurred after simulation of the Bennett fracture. With the numbers available, the mean total contact area had a non-significant (29 per cent) increase (p = 0.16), from 51.9 to 67.2 square millimeters, after simulation of the Bennett fracture.
End-Stage Osteoarthrotic Specimens
The end-stage specimens showed extensive degeneration of articular cartilage with eburnation of subchondral bone and peripheral formation of osteophytes. Articular cartilage was present only on the most dorsal aspect of the trapezial surface. The beak ligament was detached from the apex of the metacarpal, leading to marked dorsal subluxation of the metacarpal base.
The end-stage specimens had a contact-area pattern that was predominantly dorsal during static pinch-testing (Figs. 4-A and 5-C), and the highest pressures were also seen dorsally (Fig. 6-C). Unlike the non-osteoarthrotic and moderately osteoarthrotic specimens, the end-stage specimens showed no meaningful regional shift in contact area or peak pressure after simulation of the Bennett fracture (Figs. 4-B, 5-C, and 6-C). However, the total contact area doubled (p = 0.04), from 11.3 to 22.3 square millimeters, after simulation of the Bennett fracture.
Dynamic pinch-testing also showed a contact-area pattern that was predominantly dorsal, and the highest pressures were also seen dorsally. As with static pinch-testing, no significant shift in regional contact area or peak pressure was seen after simulation of the Bennett fracture. With the numbers available, the mean total contact area showed a non-significant (21 per cent) increase (p = 0.35), from 29.4 to 35.5 square millimeters, after simulation of the Bennett fracture.
Aggregate Analysis of All Specimens
Multivariate analysis of covariance demonstrated no meaningful difference in the response to testing among the three groups of specimens on the basis of the extent of the osteoarthrosis. Aggregate assessment of the total contact area in all fifteen specimens during static pinch revealed a significant (63 per cent) increase (p = 0.02), from 15.8 to 25.8 square millimeters, after simulation of the Bennett fracture. This increase was seen in all three groups of specimens. In the palmar region of the joint surface, the mean contact area decreased, from 58 to 25 per cent of the total area (p = 0.04); in the central region, it increased, from 28 to 52 per cent (p = 0.05); and in the dorsal region, it increased, although not significantly, with the numbers available, from 14 to 24 per cent (p = 0.18). On dynamic pinch-testing, with the numbers available, the mean total contact area showed a non-significant (7 per cent) increase (p = 0.29), from 50.3 to 54.0 square millimeters, after simulation of the Bennett fracture.
Aggregate multivariate analysis of each of the five regional compartments of the joint surface demonstrated a significant increase, on static pinch-testing, in both contact pressure (p = 0.006) and contact area (p = 0.001) in the central compartment of all specimens. During dynamic pinch-testing, after simulation of the Bennett fracture, there was a significant redistribution of pressure from palmar to dorsal across all compartments, and the regional contact areas decreased significantly in both the palmar and palmar-central compartments (p = 0.001).