The Shiers hinged knee prosthesis (Fig. 22.2) was first implanted in 1953 [8, 9]. The hinge was made of a molybdenum bearing and stainless steel. The actual hinge consisted of a femoral female surface and a tibial male surface united by a main bearing, which was prevented from unwinding by a reverse-threaded locking screw. In addition, tri-flanged stems of varying lengths, accommodating differing femur and tibia lengths, were screwed onto the hinged surfaces. The design concept allowed uniaxial flexion via the hinge, limited extension to 180°, and preserved lateral stability via the large bearing surfaces [6, 8–12].

The operative technique consisted of a lateral parapatellar approach to expose the knee joint. A patellectomy was performed with care to maintain the continuity of the extensor mechanism. Approximately 0.75 in. of distal femoral condyle was removed with a saw, and the posterior aspects of the condyles were removed with an osteotome. The proximal 0.25 in. of the tibial plateau was resected while preserving the collateral ligaments. The stems of both the female and male components were gently hammered into the medullary cavities of the femur and tibia, respectively. The components were then linked and locked with the main bearing and locking screw. Patients were placed in a cylindrical cast for 10 days, after which full weight-bearing was allowed to impact the hinge. The cast was removed on day 12, and formal physical therapy was initiated [8, 9].

A common complication with this early design of the Shiers hinged knee was stem fracture from metal fatigue [8, 9]. In Shiers’ original series of 17 patients, there were 8 stem fractures in 6 patients. All fractures occurred at the threaded junction of the stem and hinged surface. The fractures occurred at varying intervals ranging from 4 months to 4 years [9]. This complication led to modifications including the elimination of stem modularity. The hinge halves were machined out of a single block of steel, thereby eliminating the easily fatigued stem-hinge modular interface. In addition, the hinges were made a shorter standard length. Finally, the locking screw was made more robust by increasing the diameter [8, 9].

Shiers reported his short-term clinical results in 1961 [10]. After modifications to the hinge, Shiers reported no more prosthetic fatigue fractures and concluded that a short-term successful result was possible in three cases out of four. This report was complicated by multiple cases of skin necrosis, deep infection, loosening, and foot drop. Later reports on the Shiers hinged knee also demonstrated short-term improvements in knee pain. However, many authors noted more severe complications, such as skin necrosis and deep infection necessitating amputation, bolt extrusion, extensor lag and tendon rupture, tibia fracture, hematoma, fat embolism, deep vein thrombosis (DVT), and even four cases of death within 48 h. As with the Walldius hinged knee, the Shiers showed degrading results over time with regard to pain and function [8–12].

The Stanmore hinged knee prosthesis was introduced in 1969 [13]. The early designs were constructed of either titanium 160 with Vitallium bearings or entirely of CoCr. The latest designs were made from CoCr with bushings of ultrahigh molecular weight polyethylene (UHMWPE), in which a stationary metal axle was retained by a titanium 318 clip. The prosthesis had long, oval, tapered medullary stems that were cemented into place and at a fixed angle of 8° of valgus [13, 14].

The clinical results, as with all the first-generation prostheses, were poor in the long term because of the highly constrained design. However, inconsistent short-term results were reported. In 1978, Lettin reported pain relief in 94% of patients at an average follow-up of 2.5 years [13]. On the other hand, Karpinski in 1987 had good results in only 23% of his patients at an average follow-up of 44.7 months [14]. The experience with the Stanmore prosthesis was also associated with an unacceptable rate of major complications [13, 14].

The Guepar prosthesis, introduced in 1969, had several specific design goals and represents the first real attempt to improve on previous design shortcomings [15]. These goals included minimal bone resection, joint stability, valgus alignment, preservation of motion, preservation of patellar tracking, and a dampening effect in extension. The prosthesis had an offset hinge of CoCr that provided 5° of recurvatum and 180° of flexion. There was a choice of either a 7° valgus or modified straight femur, both with 13-cm stems. A trochlear plate provided for patellofemoral articulation. Finally, a silicone rubber bumper was present on the anterior-superior tibia to dampen the femoral-tibial contact by 25% in extension [15].

The clinical results, as with all the highly constrained first-generation prostheses, were poor in the long term. Le Nobel in 1981 reported on 113 knees in 97 patients with an average follow-up of 19 months [16]. Seventy-four of ninety-seven patients reported little or no pain, and 79 patients believed that surgery was worthwhile. Fifty-five results were graded as excellent or good, but 30 were poor [15]. In addition, the complications associated with this prosthesis, like the other first-generation prostheses, were both numerous and severe [7, 15–17].

In the early 1970s, it became apparent that midterm results with the first-generation hinged prostheses were poor and that early results with unlinked prostheses were promising. As such, designers began attempting to meld the concepts of linked and unlinked knee arthroplasty [18]. The successive design changes throughout the second generation document a newer, more scientific approach to prosthesis design, outcomes analysis, and knee arthroplasty. The newer prostheses were a clear attempt by investigators to decrease joint constraint, decrease bone cement-prosthesis stress, and improve longevity. As a whole, the design modifications associated with the second generation of hinged knee implants may be summarized as the inclusion of varus/valgus motion and modest axial rotation to a linked design [17, 19–32].

The Sheehan hinged knee was introduced in 1971 [19]. This design was both constrained and unconstrained depending on the degree of flexion and extension of the knee. The prosthesis was made up of femoral and tibial components with intramedullary stems, which were mirror images for the left and right knees. The external surface of the femoral component was designed to have a curvature simulating a normal knee, thus allowing for a constantly changing instant center of rotation. The tibial component had a high-density polyethylene surface mounted on an intramedullary stem. The tibial polyethylene had an expanded intracondylar stud shaped like a rugby football. This polyethylene stud interlocked between the femoral bearing surfaces and engaged the inner radius of the femoral component. When the knee was fully extended, the tibial stud engaged the notch of the femoral component and prevented axial rotation and allowed 2–3° of side-to-side motion. With 30° of flexion, the gradual widening of the femoral notch allowed approximately 20° of rotation and 6–7° of side-to-side motion. Beyond 90° of flexion, there was no direct linkage between the tibial stud and the femoral component. This allowed femoral rollback and reduced tensile and distraction forces on the components. The prosthesis did not have an accommodating patellar surface; nevertheless, the patella made contact with the prosthesis after 50° of flexion [19].

Sheehan reported his short-term results in 1978 with 157 knees and an average follow-up of 34 months. He reported good results with regard to pain relief and had no cases of clinical or radiological loosening. However, there were four cases of the plastic-metal interface detaching on the tibial component and two cases of fracturing of the tibial stud [19]. Furthermore, long-term results deteriorated, like the rest of the first- and second-generation hinged knees. Rickhuss et al. reported in 1994 the 5- to 10-year follow-up for the Sheehan hinged knee [20]. Using the Hospital for Special Surgery Scoring System, only 15.6% had good results, while 40% had poor results. At review, 31% of the patients had undergone revision surgery or were awaiting such surgery. Therefore, the authors thought that the Sheehan knee replacement should be considered obsolete [20].

One of the earliest second-generation prostheses was described in 1973 by Herbert [21] (Fig. 22.3). The ball-in-socket Herbert design consisted of a polyethylene femoral socket and a CoCr tibial sphere on a shank. While providing unrestrained flexion and extension, the ball-in-socket also allowed 10° of varus and valgus and some limited rotation. The surgical technique called for a limited notch resection, posterior femoral condylar resection, and cementing of left or right fixed valgus femoral stems [21].

Original laboratory testing showed significant shank wear from metal-on-metal gliding between the femoral housing and the tibial shank (Fig. 22.4). Shank wear created increased varus/valgus motion at 500,000 flexion/extension cycles. It was assumed that 1 million cycles represented 1 year of expected in vivo use. Medial condylar and shank fractures were also observed [21] (Fig. 22.3). Clinical experience with 23 prostheses in 22 patients implanted at the Cleveland Clinic from 1973 to 1974 was disastrous. Three dislocations and four medial housing fractures occurred between 5 and 23 months postoperatively. [21].

The prosthesis was modified in late 1974 to add metal to the femoral housing and narrow the notch. The ultimate strength of the prosthesis was increased while decreasing varus/valgus and rotatory motion. Laboratory testing showed significant shank wear at 2 million cycles. Medial housing fracture was noted at 2.8 million cycles. Clinically, one medial housing fracture occurred at 13 months postoperatively in 12 knees. In total, the Herbert prosthesis was found to have a 15% failure by prosthetic fracture within 2 years. The prosthesis was discontinued in April 1976 [21]. Although a clinical failure, the Herbert prosthesis experience emphasized the relevance of laboratory assessment in new prosthetic designs .

The Spherocentric knee was first introduced in 1973 near the same time as the Herbert prosthesis [22]. As in the design of the Herbert knee, the Spherocentric knee was designed to address specific problems experienced with earlier designs. The designers identified three main problems with earlier designs: (1) metal-on-metal contact generates extensive wear and fatigue of the implant; (2) uniaxial rotation creates high-torsional loads that are transferred from the prosthesis linkage to the prosthesis-cement or bone interfaces, thus producing early loosening; and (3) mechanical extension stops produce high-impact loads that are also transferred to the prosthesis-bone or cement interfaces, creating early loosening [22]. The design of the Spherocentric knee included free motion in all rotational axes through a ball-in-socket articulation but provided for load sharing with condylar outriggers and tracks. A cam mechanism that provided controlled deceleration reduced end loading in extension. All metal-on-metal contact was eliminated by incorporating replaceable polyethylene-bearing surfaces. All prostheses were cemented, and all polyethylene surfaces were loaded in compression. These design features provided for multiaxial motion with decreased prosthesis-cement interface stress, thereby theoretically improving longevity [22, 23].

Before clinical experience with the Spherocentric knee , mechanical testing was performed in extension, flexion, varus and valgus, and compression on implants in cadaveric knees. The investigators maintain that the testing documented not only the stability and strength of the assembly but also of the prosthesis-cement-bone interfaces. The tests also demonstrated satisfactory range of motion, kinematics, and deceleration cam mechanism function [22]. The early failure of the Herbert prosthesis led the investigators to perform fatigue investigations of the linkage and housing. Early results identified several areas of considerable surface strain where fatigue failure could occur. Multiple design revisions resulted in the thickening of all prosthetic surface intersections as well as reinforcing of the anterior notch housing. At the conclusion of these mechanical investigations and subsequent design modifications, the institutional review board at the University of Michigan approved clinical use of the Spherocentric knee in 1973 [22].

Matthews et al. reported a midterm result of 58 of the first 81 Spherocentric knees in 1982 [22]. The specific indications for using the Spherocentric knee were fixed varus or valgus greater than 20°, flexion contracture greater than 30°, instability greater than a 20° arc, and severe metaphyseal bone loss. Duration of follow-up averaged 48 months, with a range of 24–73 months. All implants were cemented in the first-generation technique. No patellae were resurfaced. The majority of patients experienced markedly improved range of motion, stability, ambulatory capacity, and pain. In comparison with reports with other devices, the complication rate was quite low. The deep infection rate was 3.5% (3 of 84 knees). Only 7 knees (8.3%) required reoperation for infection, instability, or pain [22]. Early clinical enthusiasm was dampened when it was reported that 52% of radiographically followed patients displayed some radiolucency at the prosthesis-cement or cement-bone interface. Longer-term follow-up displayed only modest deterioration of the results but was limited to only 21 patients [23]. Nevertheless, the basic principles for the design of modern linked prostheses and the methodology for investigating the devices, both in the laboratory and in the clinical setting, are grounded in the Spherocentric experience .

The Attenborough hinged knee was introduced in 1974 and was one of the first prostheses to compromise between the highly constrained first-generation hinged knees and the unconstrained condylar prosthesis [24]. The Attenborough hinged knee was comprised of a polyethylene tibial component, which was cemented in place. The metal femoral component consisted of the femoral articular surface and a short stem, which was also cemented in place. The original knee prosthesis had a stabilizing rod, which was contained in the femoral component. This rod fits inside the tibial component and allowed some lateral and rotational laxity. In the newer modified models, the stabilizing rod is separated from the femoral component and is locked into the femoral component with a polyethylene circlip. This separation of the rod from the femoral component allowed for greater ease in insertion of the prosthesis and facilitated the removal of cement. The femoral-tibial articulation of this prosthesis is similar to the knees used today. The difference lies with the stabilizing rod. The stabilizing rod provides the linkage of the prosthesis but allows for some lateral and rotational laxity. When the lateral and rotational movements occur, the joint opens and tightens the soft tissues, which produce a gradual deceleration of movements instead of a sudden block to movement [24]. This was the conceptual advantage over other second-generation hinged knees that limited movement with a hard block, which may lead to early loosening.

Early clinical results were promising . Attenborough short-term results of 245 knees showed only two cases of tibial loosening [24]. Vanhegan also presented his short-term results with 100 knees at 2.5 years of follow-up. He found 85% good results with only two knees having loosened [25]. However, as with the early generation hinged knees, long-term results deteriorated. Kershaw et al. in 1988 reported on 132 arthroplasties with a 77-month average follow-up (49–120 months). He found a 30% loosening rate and a 19% wound-healing complication rate. The survivorship analysis using revision as the end point showed survivorship to be 77% at both 6 and 10 years. However, if pain and radiographic loosening were used, then survivorship declined to 65% at 6 years and 52% at 10 years [26].

The Noiles total knee (Fig. 22.5) was introduced in the late 1970s by Joint Medical Products (Stamford, CT) [27]. The prosthesis consisted of a modified constrained hinge that allowed 20° of varus/valgus as well as axial rotation. The cemented femur and uncemented tibial components were linked via a cemented poly sleeve and a hinge pin. Knee simulator data showed torque similar to that of an unconstrained design and less than that of a semiconstrained design total knee prosthesis [27].

The clinical adaptation of the Noiles design was intended for patients with anticipated heavy use and severe varus/valgus instability as well as revision surgery [27, 28]. The late 1970s and early 1980s clinical experience was very positive. However, poor results were reported by Shindell in 1986 [28]. Twenty-three knees in nineteen patients with an average age of 61 years were followed for up to 75 months. HSS scores improved from 41.3 to 76.8 at 6 months, but 10 knees failed at an average of 32 months. The majority of failures were in heavy patients (>200 lb) and in patients with large tibial metaphyses. A significant rate of subsidence of the tibial prostheses occurred (5.1 mm) even in well-functioning knees. Subsidence of greater than 10 mm was reported in rheumatoid patients [28]. Despite the clinical failure of the original Noiles hinge design, the device further advanced hinge technology by coupling decreased constraint with decreased mechanical failure of the link .

In 1978, the Kinematic rotating hinge device was introduced for clinical use [17]. Like the Noiles and Spherocentric prostheses, the Kinematic rotating hinge prosthesis was designed to decrease the clinical and mechanical failure mechanisms of earlier designs. Several fundamental principles required for a well-functioning linked prosthesis were identified, and extensive mechanical and wear testing of the design was performed before clinical release.

The design team proposed five primary questions: (1) How is hyperextension limited and what is the range of flexion before impingement? (2) Is the prosthesis unrestricted in axial rotation? (3) How is varus/valgus alignment restricted? (4) Is there provision for patellar replacement/resurfacing? (5) How much bone is resected from the intercondylar area? [17] The resultant design was a cast cobalt chrome femoral component with condylar replacement and intramedullary stems for use with cement. Removable, condylar, polyethylene bushings prevent metal-on-metal contact between the femoral component and a snap-in axle that provides flexion and extension. A cobalt chrome tibial bearing component articulates between the femoral snap-in axle and an all-polyethylene tibial component. The all-polyethylene tibial component is cemented to the tibia and has a central cylinder to receive the rotational axle of the cobalt chrome tibial bearing component [17].

The prosthetic linkage controls 2 of 3° of linear freedom, while the soft tissue sleeve limits distraction [17, 29]. The prosthesis also controls varus/valgus motion while allowing flexion-extension and axial rotation. The limits of flexion are related more to soft tissue restraints than to prosthetic design. Extension is limited by the posterior soft tissues and also by a polyethylene bumper on the tibial bearing component that engages the femoral axle at 3° of hyperextension. Posterior placement of the axle in the condyles helps facilitate unlimited flexion and lockout in hyperextension. Axial rotation of the prosthesis is limited to 12° internal and external rotation by the incongruent curvatures of the all-polyethylene tibia and the base plate of the cobalt chrome tibial bearing component.

Wear analysis of the polyethylene bushings was performed through a 30° arc of motion in a simulator loaded to approximately three times standard body weight, at 37° C, with distilled water as a lubricant, for up to 5 million cycles [29]. Most flexion-extension rotation occurred between the axle and the polyethylene bushings, creating a maximum wear of 0.23 mm at 5 million cycles. No significant changes were noted in any other component. However, when an off-center load was applied in a similar experiment, permanent deformation of both the bushing and polyethylene tibial component were noted [29]. Despite the authors’ claim that the deformation was mild in both components, the results indicate that reconstruction of a neutral mechanical axis of the lower extremity is crucial to the longevity of this design.

Finite element analysis of the relationship between the cobalt chrome tibial bearing component and both the condylar portion of the femoral component and the all-polyethylene tibial component concluded that the majority of weight-bearing force in a normally aligned knee reconstruction passes from the tibia to the femur via the condyles [29]. The risk of fatigue fracture of the rotational axle is extremely low. Mechanical testing of the rotational axle confirmed the fatigue limit of the metal to be slightly higher than the expectant forces as calculated by finite element analysis [29]. These results also indicate that neutral axis reconstruction with the Kinematic rotating hinge prosthesis is critical to the longevity of the prosthesis. Excessive varus or valgus produces moments greater than those predicted and could result in fatigue failure of the polyethylene bushings, the all-polyethylene tibial component, or the rotational axis of the cobalt chrome tibial bearing component. Five of the first 200 devices implanted suffered fatigue fracture of the rotational axle at its junction with the base plate. Subsequently, the design was modified to thicken the rotational axle and improve the tolerance between the femoral condyles and the cobalt chrome tibial bearing component [29]. The Kinematic rotating hinge experience furthered the scientific approach to introducing a new prosthesis, and many of the design principles are preserved in newer design-linked prostheses .

The first clinical results with the Kinematic rotating hinge were published in 1982 [29]. Twenty-two knees were followed for an average of 12 months with a range of 5–24 months. The indications for a constrained prosthesis were a combination of marked collateral ligament deficiency and bone loss, in which a condylar-type replacement was considered unsuitable. All but one case was a revision procedure. All patients without prior patellectomy underwent patella resurfacing. Half of the extensor mechanisms required lateral release for patellar stabilization. The short duration of follow-up in this series prevented the authors from reporting radiographic or clinical results of mechanical failure. However, 17 of 22 knees reported trivial or no pain, 16 of 22 patients had the same or improved range of motion, there were no cases of postoperative sepsis, and no re-revisions were performed [17].

Good early clinical results using the Kinematic rotating hinge prosthesis were also reported independently by Shaw and Rand [17, 30]. Follow-up periods ranged from 25 to 79 months and averaged approximately 4.0 years. Satisfaction rates in the primary setting range from 80 to 90%. Satisfaction rates after revision surgery to the Kinematic rotating hinge were worse, however, ranging from 74 to 83%. Patellar instability, the most frequently reported complication by both investigators, was reported as high as 36%. More serious complications reported by Rand included sepsis in three cases and implant breakage in one. Of greatest concern was the report by both authors that, despite the short-term follow-up period, progressive radiolucent lines were present in as many as 25% of cases [17, 30, 31].

Unlike the experience with many hinged devices, midterm follow-up with the Kinematic rotating hinge has recently been reported by Springer et al. [32] Sixty-nine knees were followed for an average of 75 months with a range of 24–199 months. The indications for implanting a linked device were (1) severe bone loss combined with ligamentous instability, (2) periprosthetic fracture, (3) severe collateral ligament instability, (4) congenital dislocation of the knee, and (5) reimplantation after sepsis. The average range of motion was from 1° shy of full extension to 94° of flexion. At final follow-up, Knee Society Scores had improved an average of nearly 40 points. However, complications were frequent and often severe. Thirty-two percent of patients experienced at least one complication. Postoperative infection was greater than 14%, and component fatigue failure was 10%. Patellar pain was reported as severe in 13% of patients, the majority of whom had an unresurfaced patella. Radiographic analysis of the surviving components revealed that 13% of patients had definite loosening of either or both the femoral and tibial components [32]. Although unreported, failure for any reason can be interpreted as high as 40% at an average of approximately 6 years in this patient population. The authors concluded that linked prosthetic reconstruction with the Kinematic rotating hinge should be reserved for salvage situations [32].

In 1986, the Swedish Orthopaedic Society published the survivorship analysis of over 8000 knee arthroplasties enrolled in the Swedish Knee Arthroplasty Project between 1975 and 1983 [33]. Included in the report, subdivided by primary diagnosis of osteoarthritis and rheumatoid arthritis, was an independent survivorship analysis of four first-generation and three second-generation hinged knee arthroplasty designs. Arthroplasties were designated as failures if one or more prosthetic components had been added, removed, or replaced during the observation period. At 6 years, 140 first-generation hinges implanted for primary diagnosis of osteoarthritis had a survivorship of only 65%. One hundred two second-generation hinges implanted for the same diagnosis had a survivorship of 83% at the same follow-up duration. The majority of failures in both first- and second-generation designs were secondary to infection and mechanical loosening [33]. The Swedish experience clearly linked improved prosthetic design and surgical technique to improved prosthetic longevity. Nevertheless, the fundamental problems with linked prosthetic designs were also highlighted. Despite improved survivorship, unacceptable rates of loosening and major complications such as deep infection persisted. Survivorship of both first- and second-generation hinges was notably inferior to that of both unicompartmental and tricompartmental unlinked designs. In rheumatoid arthritis, survivorship was similar in all hinged designs but inferior to the survivorship of unlinked tricompartmental arthroplasty [33]. This report accurately encapsulated the unsatisfactory clinical performance of hinge knee arthroplasty designs up to that point; however, it also provided promise that continued design evolution could improve longevity.