Introduction

Bone tunnel enlargement in anterior cruciate ligament reconstruction has been initially reported with allograft of the patellar tendon, sterilized by ethylene oxide [18] as well as with patellar or hamstring autograft tendons [2, 6, 11,27]. For a long time widening has not modified the clinical results when considering a follow-up of 2 to 3 years, but this is not the case with a follow up of more than 5 years [29]. The phenomenon of widening can have mechanical or biological reasons. The persistence of micro mobility of the graft within the tunnel has been largely and extensively described and attributed to a non-anatomic fixation of the graft far from the tunnel outlet [16, 17, 23]. Such a graft conserves longitudinal and sagittal mobility leading to erosion of the bone walls. After 3 or 4 months, the widening has reached its maximum diameter and can be characterized by sclerotic margins on X-rays [6, 29]. The surgical goal should be a solid fixation [22] as close as possible to the tunnel entry, as has already been proved by animal experiments [37]. Radiological reports concerning screw fixation more often mention widening with hamstrings than with patellar tendons [23, 24, 28, 36]. A recent study reported more frequent widening with non-anatomical fixation than with anatomical fixation [7]. Biological factors may act differently: immunologic response in the case of allograft [11, 14, 35], inflammatory reaction due to the rupture of the ACL [3], to the necrosis of the graft [6], to the resorption of the interference screws [9], to several components of the metallic screws [25], or to the synovial inflow within the tunnel [12]. In order to improve the biological conditions around the graft within the tunnel, we have developed a surgical technique using a periosteal flap. Michael Strobel has recently published a similar technique in his book [34]. The goal of our study is to evaluate tunnel widening in patients with or without a periosteal flap.

Material and methods

Forty-one patients (six females, 35 males; age range 15–41; mean age: 26.5 years) took part in a prospective and randomized study conducted in 2000. Inclusion criteria were isolated rupture of the ACL with a normal contralateral knee and with differential laxity inferior to 10 mm as measured with the KT-1000. Patients with greater laxity or with radiological abnormalities were excluded. One senior surgeon carried out all the operations using an endoscopic technique. A four-strand gracilis and semitendinosus autograft was harvested, and a blind femoral tunnel created through the tibial socket. The femoral tunnel was drilled to a 7 mm diameter and a depth of 3 cm then progressively dilated to the exact size of the graft.. The femoral fixation was secured by a non-absorbable cross-pin (Transfix, Arthrex, France) [5] in addition to either a resorbable screw or a periosteal flap. Tibial fixation was always secured by a screw or a staple. The femoral poly-L-lactic acid bio-absorbable interference screw (28 mm long) corresponded exactly to the diameter of the tunnel entry. The periosteal flap was harvested at the superior and medial metaphysis of the tibia by a sharp periosteal elevator prior to drilling. The flap was 2 cm long and 1 cm large; it was wrapped around the proximal part of the new ligament at a maximum of 25 mm from the upper end, its osseous surface faces externally (Fig. 1 and Fig. 2). When the periosteal flap was used, the femoral tunnel was dilated by another 0.5 mm. In the postoperative course, the patients wore a splint limiting flexion (0°/0°/70°) for 3 weeks but full weight bearing was allowed within the first 2 weeks. Patients were discharged from the hospital on day 2 or 3 and attended a non-aggressive rehabilitation program 3 times a week for a period of 2 months. Close chain exercises were encouraged immediately postoperatively and open exercises were allowed at 3 months. They resumed work 4 to 6 weeks postoperatively. Jogging, cycling and swimming were allowed after 3 months, but for twisting activities they had to wait for 8 to 10 months.

Fig. 1a, b.
figure 1

The periosteal flap is harvested at the superior and medial metaphysis of the tibia

Full size image
Fig. 2.
figure 2

The periosteal flap is wrapped around the proximal part of the four strands

Full size image

Forty-one patients were included: the first group of 20 patients had femoral fixation by Transfix and resorbable screw, the second group of 21 patients had femoral fixation by Transfix and periosteal flap. The patients were clinically monitored and tested with the KT-1000 arthrometer (133-N and manual maximal anteroposterior traction) at 1, 2, 3, 5, 8, 12 and 16 months postoperatively, and underwent X-rays at 10 weeks and 11 months postoperatively on average (anteroposterior and lateral weight bearing in extension). The anteroposterior and lateral diameters of the tunnel were measured between the sclerotic margins at the tunnel entrance and 1 cm above and adjusted for magnification (Fig. 3a, b). The result was compared to the peroperative drill size. The percentage change in diameter was calculated as: (tunnel diameter−drill size)/drill size. Tunnel widening was defined as any increase in comparison to the diameter of the last drill. The shape of the tunnel was classified as either "line type", "cavity type" or "cone type" (27). A senior examiner performed all the measurements. A subset of 20 X-rays was randomly selected and reviewed by a second observer to determine the interobserver variability. Femoral tunnel placement was documented by measuring the posterior aspect of Blumensaat's line to the anterior aspect of the femoral tunnel [1]. Statistical analysis was performed using the Student t test and the Mann-Whitney test. Statistical significance was accepted at the 95% confidence level.

Fig. 3a, b.
figure 3

Measurements of the enlargement at the tunnel aperture and 1 cm above. a Anteroposterior view. b Lateral view

Full size image

Results

The two groups of patients were comparable as to gender, side, age, KT-1000 side-to-side difference, femoral tunnel diameter and follow up (Table 1).

Table 1. Demographics of the study population
Full size table

Interobserver variability (p=0.71) and intraobserver variability (p=0.92) for measurement of tunnel enlargement was not significant. At 2.5 months postoperatively on average (1–4 months), tunnel walls were not visible in five cases of the first group, no widening was seen in two cases and the average widening at the outlet of the tunnel was 29.8% in the frontal view and 27.2% in the sagittal view. In the second group (with periosteal flap), tunnel walls were not visible in two cases, no widening was seen in four cases and the average widening at the outlet of the tunnel was 10.3% in the frontal view and 13.7% in the sagittal view. These differences were all significant. Furthermore, differences between the two groups were significant at 1 cm from the outlet of the tunnel (Table 2).

Table 2. Tunnel enlargement on anteroposterior and lateral views before 3 months. The differences are significant
Full size table

At 11 months on average (8 to 16 months), tunnel walls were always visible, and widening was constant in both groups. The average enlargement slightly increased in group 1 (37.4% on AP [anteroposterior] view and 31.8% on lateral view) and in group 2 (19% on AP view and 17.7% on lateral view). These differences are significant (Table 3). The "line type" tunnel was dominant in both groups (Table 4). The femoral tunnel was more anteriorly located in group 1 (58±8%) than in group 2 (65±6%). There was no significant difference in laxity between the two groups (Table 5).

Table 3. Tunnel enlargement on anteroposterior and lateral views after 6 months. The differences are significant
Full size table
Table 4. Types of enlargement according to Peyrache's classification
Full size table
Table 5. Results of knee stability with KT-1000 arthrometer at 133 N and manual maximal anteroposterior traction
Full size table

Discussion

There is a wide variability in the amount of tunnel enlargement in the literature depending on the different methods of measurement employed and the minimum thresholds defined. Computed tomography or MRI are reliable methods for detecting this phenomenon and seem superior to plain radiographs in the early postoperative period [12, 19]. Due to the difficulties in orientation, there is a poor correlation between sagittal MRI and lateral X-rays [19]. We suggest that radiographic measurements are more suitable in evaluating bone tunnel dimensions in clinical practice. In the absence of a threshold, an average widening of 33% was present in all cases of Jansson's series. With a threshold of 25%, the incidence of widening was 94% (average widening is 48%), and if the threshold was 50% the incidence was 84%. In our series, without the periosteal flap, the incidence at the tunnel entry was 87.5% at 3 months and 100% after 6 months. We only referred to the enlargement at the outlet of the tunnel, as the interference screw can lead to local widening by penetration of the spongiosa. Our first group demonstrated greater widening within the tunnel than at the tunnel entry. The widening appeared within the first 3 months and then slightly increased in group 1 (7%) and in group 2 (10%); the efficacy of the periosteal flap seemed to continue.

The periosteal flap can reduce widening in two different ways: either it assists tendinous integration by contributing cells and growth factors, or it prevents synovial fluid inflow. Numerous experimental studies have dealt with the tendon–bone interface.

Kernwein et al. [21] studied this anchorage in a tunnel of long bone in rabbits and dogs using either fascia lata femoris or the extensor carpi radialis longus. The graft was gradually invaded by osteoblasts, providing osseous stability. Studies by Whiston and Walmsley [38] demonstrated the early existence of a sleeve of connective tissue around the tendon followed by infiltration of chondral cells generating an enchondral ossification. Rodeo et al. [31] studied the anchorage of an extensor tendon in the trans-tibial tunnel of dogs. After a few weeks, a fibrovascular interface developed out of the pluripotential cells of the bone marrow. Some cells underwent osteoblastic metaplasia, others developed fibroblastic metaplasia. The new collagen fibers were traction oriented, securing the tendon to bone fixation, consequently resembling Sharpey fibers. Immature bone invaded the interface and was mineralized. The quantity of collagen fibers and osteoid tissue was greater at the aperture than within the tunnel. Grana et al. [13] demonstrated an identical anchorage by fibrocellular interface in an intra-articular tunnel at 12 weeks. Weiler et al. [37] studied the Achilles tendon-bone fixation in the goat by intra-articular resorbable interference screw. This fixation depends on the different strengths of the tendon and the location of the tendon. At 6 weeks, they found a mixture of direct insertion and Sharpey fibers in contact with woven bone, within the tunnel. At 12 weeks, insertion was continuous, with alternately Sharpey fibers, fibrocartilage and bone. From 6 to 24 weeks there is an intense osteoblastic activity at the tunnel aperture. After 12 weeks, the chondral zone of the articular surface and the graft are continuous and direct fixation in four layers is complete at 24 weeks. The periosteal flap might accelerate the development by providing pluripotential cells and growth factors. This hypothesis has been confirmed by experimental work in rabbits [4]. Extratendinous implantation using a periosteal flap has been compared to a sham group: the periosteal flap accelerated the osseous integration of the fibrous interface. Moreover, several authors [15, 32] have added growth factors to the tendon bone interface to accelerate the osteoid reaction. According to Yamazaki, the periosteal flap has to be fixed at the entry of the femoral tunnel where the fixation is strongest [39].

There are five reports in the literature concerning tendon–bone biopsies of anatomical screw fixation using hamstring tendons. In Scranton's series [33], Sharpey fibers are present at 6 weeks postoperatively, whereas Pinszewsky [28] described them only at 12 weeks after operation. Eriksson et al. described the presence of Sharpey fibers between tendon and bone and a fibrochondral interface at 1 year after operation [10].

The interface between tendon and bone has been demonstrated with MRI as intermediate signal intensity on T1-weighted images [6, 19, 23]. Two reports have studied the anchorage in non-anatomic fixation: according to Petersen and Laprell [26] (femoral fixation by Endobutton and tibial fixation by cramp) indirect fixation was observed at 6 months postoperatively within the tunnel, whereas Robert et al. [30] noted fixation at the top of the socket (femoral fixation with Transfix) at 10 months. It may be possible that biopsies far from the tunnel outlet do not show the exact maturation of the interface. Anchorage at the tunnel entry seems faster and is of the direct type as has been shown by Colombet [8]. The periosteal flap appears to accelerate fixation at this site but cannot avoid widening of the tunnel.

In a second hypothesis, the periosteal flap can act as a barrier to synovial fluid inflow. Fluid leakage around the tendons in the tunnel is documented by CT arthrograms [12] or MRI [6, 11, 23] during the first weeks after surgery. Cytokines affecting bone resorption (IL-1, IL-6, IL-8, TNFa, PGE-2) have been identified in synovial fluid following ACL disruption [20] or graft necrosis [3]. In a similar way to articular prostheses, these uncontrolled cytokines (IRAPS) lead to bone resorption. Very low concentrations of particles released from the metallic screws can act on the synovial cells and macrophages to stimulate cytokine secretion [25]. Stimulation of these cells could equally be generated by thermal bone necrosis [11] or by micro mobility of the graft [6]. Individual variability in the concentration of these cytokines could account for the variable frequency of widening [3].

The periosteal flap may use either or both these mechanisms for accelerating tendinous integration at the tunnel entry to reproduce direct anchorage in four layers. It seems that its biological role is of considerable importance since we rarely find a circular periosteal cover on the graft.