Morphologic changes in the vastus medialis muscle in patients with osteoarthritis of the knee




To investigate the frequency of structural changes in the vastus medialis muscle in patients with osteoarthritis (OA) of the knee.


Specimens of vastus medialis muscle from 78 patients with end-stage OA of the knee undergoing total joint arthroplasty were examined histopathologically. Morphologic changes were assessed in relation to clinical features that might have contributed to muscle injury.


All muscle specimens exhibited atrophy of type 2 fibers. In 32% of the patients, atrophy of type 1 fibers was also noted. Fiber type grouping of type 1 fiber in 15% of the patients and type 2 fiber in 37%, indicating reinnervation, led to the diagnosis of neurogenic muscular atrophy in 32% of the patients; selective atrophy of type 2 fiber in 68% of the specimens was interpreted as possibly resulting from pain-associated disuse. Signs of muscle degeneration and regeneration were found in 65% and 96% of the samples, respectively. Soft tissue changes indicating long-term disease, such as calcification, fibrosis, and lipomatosis, were frequently observed (in 69%, 71%, and 94% of the patients, respectively). Statistical analysis of clinical and morphologic parameters revealed a significant association between degenerative muscle changes and the presence of a varus deviation of the leg axis.


Patients with OA of the knee frequently exhibit muscle changes, with probable multifactorial etiology. Selective atrophy of type 2 fibers might reflect pain-related immobilization of a limb. Changes such as neurogenic muscular atrophy, muscle fiber degeneration, and regeneration might contribute as cofactors in the development or progression of OA.

The muscles surrounding the knee play an important role as active stabilizers and are crucial in protection of the joint structures against damaging influences (chondroprotection) (1–13). In a longitudinal study of female subjects (14), Slemenda et al showed that adequate quadriceps strength in healthy knees had a protective effect against the development of osteoarthritis (OA). In addition, there is considerable evidence that an imbalance in the activity of the joint-stabilizing muscles due to age-related reduction in sensory joint innervation (1, 2, 4, 15) or proprioception (9–12, 16–20) plays a decisive role in the development of OA of the knee. Slemenda and colleagues (20) and Brandt (21) have suggested that quadriceps weakness and atrophy might not only be the result of OA and resultant pain-related immobilization of the affected limb and muscles, but might also have to be regarded as direct preceding risk factors for the development and progression of OA. It is conceivable that, in addition to imbalance of muscle function, structural muscle changes might interfere with the chondroprotective effect of the musculature and thus contribute to the development of OA.

Normal muscle contains 2 major fiber types, which differ with respect to histochemical staining pattern, twitch contraction time, and resistance to fatigue. The speed of contraction is mainly determined by particular isoforms of the contractile proteins actin and myosin and their associated regulatory proteins. Fast and slow muscle fibers exhibit distinct isoforms of myosin, which vary in their ATPase activity. In myofibrillar ATPase preparations after preincubation at alkaline pH, pale type 1 fibers and dark type 2 fibers can be readily seen. After preincubation at pH 4.6, 3 fiber types can be distinguished: type 1 fibers are dark, type 2A fibers are unstained, and type 2B fibers are intermediate in their staining intensity. In ATPase at pH 4.3, type 1 fibers are dark and both type 2A and 2B fibers are pale (22, 23). The slow-twitch type 1 fibers rely on oxidative mechanisms for the generation of energy, and therefore contain more mitochondria and lipid and have higher activity of oxidative enzymes and a higher degree of vascularization than type 2 fibers, and are highly resistant to fatigue. In contrast, fast-twitch type 2 fibers contain more glycogen and enzymes, such as myophosphorylase and phosphofructokinase, in the glycolytic pathways. Fast glycolytic type 2B fibers use only anaerobic glycolytic pathways and are rapidly fatigued. The combination of oxidative and glycolytic metabolism in type 2A fibers allows more sustained tension (fatigue-resistant fibers). Changes in diameter of fast-twitch type 2 fibers mainly reflect exercise training and detraining (22, 23). As a consequence of pain-related immobilization due to OA, selective atrophy of type 2 muscle fibers may occur (22, 23).

Muscle fiber typing is determined by the motor neuron. A motor unit consists of a single motor neuron and the supplied muscle fibers. Thus, all muscle fibers in each motor unit are of the same fiber type (22, 23).

In the present cross-sectional study, we performed histopathologic analysis of vastus medialis muscle from patients with OA of the knee. We analyzed the frequency of structural changes (in addition to disuse atrophy) and assessed associations with clinical parameters, such as joint stability and medication, that might have caused structural muscle changes.


Muscle biopsy samples (1 cm3) were obtained from 78 patients (62 female, 16 male) with end-stage knee OA undergoing total knee arthroplasty. All patients had radiographically advanced OA affecting all 3 compartments of the knee, with a Kellgren/Lawrence score of 4 (24). OA predominantly affected the medial femorotibial joint in 59 patients and the lateral joint in 19. Criteria for exclusion were as follows: OA secondary to rheumatoid arthritis or other inflammatory arthritis, or history of avascular necrosis, injury, joint infection, neuropathic arthropathy, ochronosis, acromegaly, hemochromatosis, Wilson's disease, osteochondromatosis, gout, pseudogout, or osteopetrosis. Patients with underlying peripheral nerve disorders that might affect muscle structure and function, such as lumbar compression syndrome or polyneuropathy, were also excluded.

Laxity of the knees was examined clinically by one experienced examiner (JS), and patients with obvious collateral ligament instability of >5 degrees were excluded from the study, based on a report by Sharma et al (12) of a possible negative effect of muscle strength on the development of OA in unstable knees (with laxity of >6.75 degrees). The cutoff of 5 degrees was used as an exclusion criterion in our study because the reliability of physical examination of laxity is low, with reported interobserver agreement of 0.55 (25). In our study participants the mean ± SD knee laxity measurement was 2.7 ± 3.0 degrees varus. To assess alignment, a single anteroposterior radiograph of the leg was obtained, with the patient in a standing position. For participants who were tall, a 130 × 36–cm graduated grid cassette was used to include the full limb. Alignment was defined as the measure of the angle formed by the intersection of the line connecting the centers of the femoral head and the intercondylar notch, and the proximal extension of the line connecting the centers of the ankle talus and tibial spines. The analyses were performed by one experienced reader (MF). Alignment was classified as varus (1–5 degrees), neutral (0–1 degree), or valgus (1–5 degrees). Patients with malalignment of >5 degrees of varus or valgus were excluded from the study, because Sharma et al have reported a possible negative effect of muscle strength on the development of OA in knees with malalignment of >5 degrees (12), and Bade and colleagues demonstrated histomorphologic changes in the vastus lateralis muscle in severely malaligned knees (26). Mean ± SD alignment in our study patients was 3.8 ± 2.0 degrees of varus. Range of motion was analyzed clinically by one experienced examiner (JS), using a goniometer. The mean ± SD flexion preoperatively was 104.8 ± 15.5 degrees, and patients had flexion contracture of 6.4 ± 6.6 degrees.

At the time of knee arthroplasty, the mean ± SD age of the patients was 68.5 ± 8.2 years (range 51–83). The following long-term medications had been taken in the past or were being taken currently: nonsteroidal antiinflammatory drugs (NSAIDs) (36 patients), statins (6 patients), bisoprolol (1 patient), metoprolol (7 patients), furosemide (3 patients), and corticosteroids (3 patients); 21 patients had received medications of which myalgia may be a side effect (nifedipine, captopril, nizatidine).

Knee arthroplasty was performed using a midvastus approach for purposes of surgical access (27). The site of muscle biopsy was located 3 cm above the median upper patellar pole, and biopsy did not necessitate an enlargement of the surgical approach. The biopsy specimens were mounted in tragacanth perpendicular to the longitudinal orientation of the muscle fibers, immediately shock-frozen in isopentane, and then stored at −80°C. Serial crosscut cryostat sections were stained with hematoxylin and eosin and Gomori's trichrome, and were then stained with ATPase after preincubation at pH 4.6 and 9.4 (for fiber type differentiation), NADH–tetrazolium (for identification of structural changes), oil red O, periodic acid–Schiff, Sudan black, and acid phosphatase (22, 23). In addition, immunohistochemical staining was performed as previously described (28–30), using commercially available monoclonal antibodies against neonatal myosin heavy chain (MHCn) and neuronal cell adhesion molecule (NCAM) for identification of regenerating fibers and HLA class I antigen, in order to detect up-regulation of class I major histocompatibility complex as an indirect parameter indicating an inflammatory process.

Histologic samples were evaluated independently by 2 experienced investigators (BF and EN-J). Each biopsy specimen was examined with respect to fiber size, fiber type distribution, selective fiber type involvement, and possible fiber type grouping indicative of reinnervation (Table 1). According to standard protocols, muscle fiber atrophy was defined as a fiber diameter of <30 μm in female subjects and <40 μm in male subjects (22, 31, 32). Fiber type grouping was defined as the occurrence of large clusters of ≥15 fibers of the same fiber type (in normal muscle, the different histochemical fiber types are distributed in a mosaic checkerboard pattern); fiber type grouping is associated with reinnervation by collateral sprouting of terminal axons (33). Neurogenic muscular atrophy was defined as atrophy of both fast-twitch type 2 fibers and slow-twitch type 1 fibers in addition to fiber type grouping, indicating reinnervation of denervated muscle fibers (22, 23). In contrast, selective atrophy of fast-twitch type 2 fibers, mainly reflecting the state of exercise training, is suggestive of pain-related immobilization of the leg (22, 23).

Table 1. Frequency of histologic changes in the muscle biopsy specimens from 78 patients with advanced knee osteoarthritis
Histologic abnormality (definition)No. (%) of patients with abnormality
Atrophy of type 1 fibers (muscle fiber diameter <30 μm [females] or <40 μm [males])25 (32)
Atrophy of type 2 fibers (muscle fiber diameter <30 μm [females] or <40 μm [males])78 (100)
Grouping of type 1 fibers (clusters of ≥15 type 1 fibers)12 (15)
Grouping of type 2 fibers (clusters of ≥15 type 2 fibers)29 (37)
Disuse atrophy (selective atrophy of type 2 fibers with or without fiber type grouping of type 2 fibers53 (68)
Neurogenic muscular atrophy (atrophy of both type 1 and type 2 fibers, with fiber type grouping of type 1 and/or type 2 fibers, with or without regenerative changes)25 (32)
Degenerative changes (necrosis, ghost fibers, centrally placed nuclei)51 (65)
Regenerative changes (basophilic regenerating fibers, neonatal myosin heavy chain–positive fibers, neuronal cell adhesion molecule–positive fibers, target fibers)75 (96)
Architectural changes (motheaten fibers, whorled or coiled fibers, ringbinden, cytoplasmic bodies, rimmed vacuoles)20 (26)
Lipid content (increased compared with controls)34 (44)
Glycogen content (increased compared with controls)2 (3)
Angiopathy (stenosis, thrombosis, microangiopathy, arteriosclerosis, intramural calcification)13 (17)
Calcification (juxtavascular)54 (69)
Fibrosis (endomysial or perimysial)55 (71)
Lipomatosis (increased compared with controls)73 (94)
Changes in nerve endings (reduction in number of myelinated fibers, demyelination, endoneurial fibrosis, endoneurial edema)7 (9)
Steroid myopathy (selective atrophy of type 2 fibers combined with increased lipid content of type 1 fibers, combined with increased glycogen content of type 2 fibers)3 (4)

In addition, possible signs of degeneration, such as necrosis, “ghost fibers,” and centrally placed nuclei, were assessed. Moreover, changes such as basophilic fibers, neonatal MHCn-positive or NCAM-positive fibers or target fibers indicative of regeneration were analyzed. Furthermore, architectural changes within the individual muscle fibers and the content of lipid and glycogen in the muscle fibers were analyzed in order to detect possible drug-induced myotoxic side effects or structural changes due to disturbed lipid metabolism. The distribution of fibrous and connective tissue and the occurrence of calcification, reflecting a long-term course of the disease or possible risk factors, also were determined. Moreover, possible changes in the nerve endings and the vessels were documented. The morphologic parameters assessed are detailed in Table 1.

Clinical features that might have contributed to histologic muscle changes were also evaluated. The features assessed included basic clinical parameters (age, sex, body weight, body size, body mass index), secondary diagnoses (hyperlipidemia, hyperuricemia, diabetes mellitus, chronic obstructive pulmonary disease), previous surgical interventions on the affected knee (arthroscopy, synovectomy, rotational osteotomy), leg axis, ligament instability, flexion contracture (more than or less than 10 degrees), laboratory data (levels of C-reactive protein, high-density lipoprotein, low-density lipoprotein, triglycerides, sodium, potassium, calcium), and medications taken for >6 weeks (NSAIDs, steroids, statins, bisoprolol, metoprolol, furosemide, drugs with myalgia as a potential side effect).

Fisher's exact test was used to assess associations between morphologic features, or between morphologic and clinical features. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated. P values less than 0.05 were considered significant. Because this investigation was explorative, we did not control for global Type I error.

All patients had been informed preoperatively of the purpose of the muscle biopsy and the study, and all provided consent. Institutional approval of the study protocol was obtained.


Twenty-two of the 78 patients (28%) had comorbidities (hyperlipidemia in 9, type 2 diabetes mellitus in 6, hyperuricemia in 2, and chronic obstructive pulmonary disease in 5). Two of these 22 patients had 2 comorbidities (type 2 diabetes mellitus and hyperuricemia in 1, and hyperuricemia and hyperlipidemia in the other). Twenty-six patients had undergone arthroscopy and 5 had undergone rotational osteotomy prior to the knee arthroplasty; in 1 patient both procedures had been performed.

The results of the histologic analyses are summarized in Table 1. Reliability of the histologic evaluations was high, with an intrarater intraclass correlation coefficient (ICC) of 0.99 and an interrater ICC of 0.97.

All of the muscle specimens exhibited atrophy of fast-twitch type 2 fibers (Figure 1a). In 25 patients (32%), atrophy of slow-twitch type 1 fibers was also noted (Figure 1b). Samples from 32 patients showed additional fiber type grouping (Figure 1c), affecting type 1 fibers in 12 cases (15%), type 2 fibers in 29 cases (37%), and both fiber types in 9 cases. The combination of atrophy of types 1 and 2 fiber with fiber type grouping, indicating reinnervation and/or regenerative muscle changes, led to the diagnosis of neurogenic muscular atrophy in 25 patients (32%) (Figure 1d). In 53 patients (68%), selective atrophy of type 2 fibers with or without fiber type grouping restricted to type 2 fibers, which mainly reflects the degree of exercise training (Figure 1a), was interpreted as possibly due to pain-associated disuse (22, 23). Signs of muscle degeneration, including ghost fibers (Figures 2a and b), necrotic muscle fiber undergoing phagocytosis (Figure 2a), or centrally placed nuclei (Figure 2a), were observed in 65% of the patients. Basophilic fibers, neonatal MHCn- or NCAM-positive fibers, or target fibers indicative of regeneration were found in 96% of the cases (Figures 1d and 2b–d). Architectural changes, such as motheaten fibers (Figure 2c), were encountered in 25%. Soft tissue changes indicating long-term disease were frequently observed (calcification in 69% of the patients, fibrosis in 71%, and lipomatosis in 94%) (Figures 1c and 2a, b, and e).

Figure 1.

Fiber type differentiation identified using combined staining with NADH–tetrazolium (blue staining of type 1 fibers) and ATPase at pH 9.4 (brown staining of type 2 fibers), showing variation in muscle fiber size and fiber type grouping in vastus medialis muscle specimens from patients with end-stage osteoarthritis of the knee. a, Selective atrophy of type 2 fibers, with fiber type grouping restricted to type 2 fibers. b, Single fiber atrophy of both type 1 and type 2 fibers. c, Atrophy as well as fiber type grouping of both fiber types, indicating reinnervation. Marked lipomatous changes are also evident. d, Group atrophy and hypertrophy of type 2 fibers more than type 1 fibers, marked fiber type grouping of both fiber types, and target formations, consistent with chronic neurogenic muscular atrophy. (Original magnification × 100.)

Figure 2.

Degenerative and regenerative changes in vastus medialis muscle specimens from patients with end-stage osteoarthritis of the knee. a, Hematoxylin and eosin (H&E)–stained section, showing degenerative changes, including “ghost fibers” (arrows), necrotic muscle fibers undergoing phagocytosis (arrowheads), and centrally placed nuclei. Endomysial fibrosis is also evident. b, Gomori's trichrome–stained section, showing group atrophy and hypertrophy, many target fibers, and endomysial fibrosis. Note the single ghost fiber (arrow). c, NADH–tetrazolium-stained section, showing architectural changes, including motheaten fibers and targets. d, Neonatal myosin heavy chain monoclonal antibody–stained section analyzed by immunohistochemistry, showing regenerating fibers. e, H&E-stained section, showing marked calcification of the perimysial tissue, as well as lipomatosis. (Original magnification × 200 in ad; × 100 in e.)

The analysis of possible associations between histologic parameters revealed that juxtavascular calcification was frequently accompanied by generalized fibrosis. In specimens from 46 of the 78 patients, both features were present, and in 15 specimens, neither was present. This association was statistically significant (OR 10, 95% CI 3–30, P < 0.001 by Fisher's exact test). Juxtavascular calcification was also found to be significantly associated with perimysial fibrosis (OR 6, 95% CI 2–20, P = 0.001) and endomysial fibrosis (OR 5, 95% CI 1.7–14, P = 0.003). In addition, perimysial and endomysial fibrosis were associated with one another (OR 3, 95% CI 1.2–9, P = 0.022).

Furthermore, a perfect relationship between atrophy of type 1 fibers and neurogenic muscular atrophy was observed. In all 25 patients with atrophy of type 1 fibers, neurogenic muscular atrophy was also diagnosed, whereas among the remaining 53 patients without evidence of type 1 fiber atrophy, there were no cases of neurogenic atrophy (P = 0). There were no further statistically significant associations between neurogenic muscular atrophy and other histologic parameters.

Table 2 shows the results of selected clinical assessments. Statistical analysis of the relationship between clinical and morphologic parameters revealed a significant association between degenerative muscle changes and the presence of varus deviation of the leg axis (OR 3, 95% CI 1.1–9, P = 0.032). There were no other statistically significant bivariate associations between histologic and clinical parameters.

Table 2. Distribution of findings and frequency of abnormal results for selected clinical parameters assessed in the 78 patients with advanced knee osteoarthritis
Parameter (definition of abnormal result)Distribution (% with abnormal result)
Body mass index, kg/m2 (≥25)<20 n = 3; 20–24 n = 12; 25–29 n = 31; >29 n = 32 (81)
Axis (varus or valgus)Neutral n = 22; varus n = 37; valgus n = 19 (72)
Extension, degrees (deficit >10)No deficit >10 degrees n = 55; deficit >10 degrees n = 23 (29)
Flexion, degrees (<120)<90 n = 8; 90–99 n = 15; 100–109 n = 18; 110–119 n = 13; 120–129 n = 19; >130 n = 5 (69)
Cholesterol, mg/dl (>200)≤200 n = 28; >200 n = 50 (64)
High-density lipoprotein, mg/dl (<45)<30 n = 5; 30–49 n = 36; 50–69 n = 29; 70–89 n = 10; >90 n = 2 (35)
Low-density lipoprotein, mg/dl (>155)<70 n = 1; 70–119 n = 16; 120–169 n = 45; 170–219 n = 14; >220 n = 2 (42)
Triglycerides, mg/dl (>200)≤200 n = 68; >200 n = 10 (13)


A key result of the present study was the finding of atrophy of type 2 fibers in all biopsy specimens from the patients with advanced OA. Since the diameter of fast-twitch type 2 fibers mainly parallels the state of exercise training, atrophy of type 2 fibers might reflect pain-associated immobilization of the affected leg, directly due to knee OA. However, 32% of the patients also exhibited atrophy of slow-twitch type 1 fibers and additional fiber type grouping, suggesting reinnervation; this was interpreted as being indicative of neurogenic muscular atrophy in these patients (23, 33). The occurrence of neurogenic muscular atrophy in the knee-stabilizing vastus medialis muscle in patients with OA of the knee has not, to our knowledge, been reported previously, although histopathologic studies of the vastus lateralis muscle in patients with knee OA have been performed. Glasberg et al (34) reported fiber type grouping in 9 of 12 patients studied, and muscular atrophy in all 12. Nakamura and Suzuki (35) described atrophy and fiber type grouping of type 2 fibers, indicative of motor neuron dysfunction, in 73% of 26 patients studied. In these earlier studies, however, the authors did not differentiate between cases in which there was selective atrophy and grouping of type 2 fibers, and cases with atrophy of both fiber types with additional fiber type grouping and/or signs of regeneration. Thus, the higher frequency of neurogenic muscular atrophy reported by those authors compared with the present study reflects a less precise definition of neurogenic atrophy, rather than a higher incidence of motor neuron dysfunction.

Neurogenic muscular atrophy is caused by damage of the central or peripheral nervous system. Under normal conditions, the different histochemical muscle fiber types are distributed in a mosaic checkerboard pattern. Degeneration of motor neurons or nerve fibers leads to group atrophy. Collateral sprouting of terminal axons during reinnervation results in larger clusters of the same fiber type (i.e., fiber type grouping) and is paralleled by larger motor units seen on electromyography (22, 23).

Limitations of our study include the fact that varus–valgus laxity was assessed by physical examination, which is known to be associated with poor reliability in OA knees (25). We used a cutoff of 5 degrees for exclusion of patients, which is lower compared with the definition proposed by Sharma et al (12). Moreover, reliability of the clinical examinations (laxity, alignment, range of motion) performed in the present study was not calculated. However, reliability of the procedures used in the present study for measuring alignment and range of motion has been reported to be high (12, 36–40).

This was an explorative study, and only bivariate analyses were carried out. A multivariate analysis of the clinical and histologic parameters was not performed due to the relatively small sample size. Because this was a cross-sectional analysis, no conclusions can be drawn regarding causality of structural muscle changes in knee OA: whether pathology of the vastus medialis muscle antedates the development of OA and therefore is a predisposing factor or is a phenomenon secondary to the development of arthritis is unknown. However, disuse-associated muscle atrophy can be explained by pain-related reduced muscle force in the affected leg. It is unlikely that neurogenic muscular atrophy is directly due to OA of the knee, since there are no reports in the literature of structural changes of the peripheral nerves as direct consequence of OA. It does seem conceivable, however, that denervation or reinnervation processes had already started before or soon after the initial clinical manifestation of OA, as evidenced by findings reported by Salo et al (41). In their study, experimentally induced loss of joint afferents in rats led on average to a 54% higher score of histologic severity of OA compared with control knees. Salo and colleagues thus concluded that a loss of afferents, as occurs in the aging process, might predispose a joint to development of OA. This would imply that neurogenic muscular atrophy might contribute to the development or acceleration of OA, by reducing the chondroprotective function of the musculature surrounding the knee. This hypothesis is supported by the findings of Slemenda and colleagues (14, 20) and Brandt (21), who concluded that quadriceps weakness and atrophy might be a direct risk factor for development of OA of the knee.

Neurogenic muscular atrophy may have a variety of causes, e.g., peripheral nerve lesions due to lumbar compression syndrome or polyneuropathy. In our study, however, underlying peripheral nerve disorders had been excluded in the patients, and none of the patients exhibited any clinical evidence of peripheral nerve dysfunction. We found no statistically significant association between neurogenic muscular atrophy and any of the clinical or histologic parameters investigated. This suggests a multifactorial etiology, which might have differed among the individual patients.

Degenerative and regenerative muscle changes, as well as soft tissue changes such as fibrosis, lipomatosis, and calcification, were observed in the vast majority of the muscle samples examined. It might be speculated that the frequent occurrence of these features in patients with OA was not a chance finding and that they may impair the function of muscle as an active stabilizer and chondroprotector, leading to the development or progression of OA. The statistically significant bivariate association of the histologic parameter degenerative changes with the clinical parameter varus deformity of the leg might be regarded as evidence in support of this hypothesis.

In conclusion, the results of this study highlight the important role of muscle function in the pathogenesis of knee OA. Selective atrophy of fast-twitch type 2 fibers might reflect pain-related immobilization of the affected limb, whereas changes such as neurogenic muscular atrophy, muscle fiber degeneration, and regeneration might contribute as cofactors in the development or progression of OA. The imbalance of muscle function due to structural muscle changes could result in a decrease or loss of the muscle's ability to provide active knee stabilization and chondroprotection. Strengthening of the quadriceps muscles as a component of a conservative treatment strategy during the initial stages of OA has been reported to have an important physiotherapeutic effect (42–45), corroborating the results of the present morphologic study.


Dr. Fink had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Fink, Neuen-Jacob.

Acquisition of data. Fink, Egl, Singer, Fuerst.

Analysis and interpretation of data. Fink, Bubenheim, Neuen-Jacob.

Manuscript preparation. Fink, Neuen-Jacob.

Statistical analysis. Bubenheim.


The authors are grateful to Dr. Martin S. Schulz (Orthopaedic Clinic Markgröningen) for valuable suggestions made during the manuscript revision process.