Abstract
The purpose of our study was to determine the most favourable combination of core suture material and peripheral repair technique for Kessler tendon repair. Thirty freshly thawed pig flexor tendons were repaired by a Kessler technique, either with braided polyester or monofilament nylon suture. A peripheral augmentation was done using one of the three techniques – running, cross-stitch and Halsted. All repairs were tested by cyclic loading, followed by load-to-failure. During cyclic loading six of the 15 tendons with a nylon core failed, but none with a braided polyester core. Irrespective of peripheral technique, the monofilament nylon core suture allowed early central cyclic gapping, resulting in failure of the repair. During load-to-failure testing, the running stitch proved weakest and the cross-stitch repair toughest.
INTRODUCTION
Early controlled motion after flexor tendon repair improves tendon nutrition, healing and the remodelling response, and results in decreased adhesions with greater tendon excursion. However, early motion also increases the risk of gapping and failure of the repair (Peck et al., 1998). Gapping increases adhesion formation and subsequent tenolysis rate, resulting in poor outcome following tendon repair (Seradge, 1983). A postoperative active motion protocol thus requires a strong suture repair technique.
Complex techniques such as the multi-strand repairs (4–6 suture strands across the repair site) are strong but technically difficult to perform (Wagner et al., 1994) and have limited practical application compared to two-strand repairs. Kessler’s core with a running peripheral repair is considered representative of two strand repairs by several authors (Greenwald et al., 1994a; Komanduri et al., 1996; Mclarney et al., 1999; Urbaniak et al., 1975) but modified peripheral suture techniques, such as the cross-stitch technique (Silfverskiold and Andersson, 1993) and the horizontal mattress-type Halsted technique (Wade et al., 1989) may further increase the repair strength.
Most authors in the past two decades have favoured braided polyester for core repair (Mclarney et al., 1999; Savage, 1985; Silfverskiold and Andersson, 1993; Trail et al., 1992) though some prefer mono-filament nylon (Greenwald et al., 1994a; Komanduri et al., 1996; Tang et al., 2001; Wagner et al., 1994). The two-strand Kessler repair is widely used clinically but there is currently no consensus on the best choice of core suture material or peripheral repair technique.
MATERIALS AND METHODS
Fresh pig trotters were obtained from a local abattoir and stored at −24°C. On the day of testing, the pig trotters were thawed in normal saline solution at room temperature, and the flexor tendons (flexor digitorum profundus) were dissected out. Each trotter had two long flexor tendons in the pulley zone, which had cross-sectional dimensions similar to human flexor tendon, and had no muscular connections over a length of about 8 cm. The maximum cross-sectional diameter of each tendon at the site of repair was measured using a Vernier calliper.
Surgical technique
The tendon was divided with a surgical scalpel and randomly assigned to one of six repair groups. These six groups were defined on the basis of two suture materials for the core Kessler repair and three techniques for the circumferential epitenon repair (running, cross-stitch and Halsted; Fig 1, Table 1). The two core suture materials were 3-0 braided polyester (ETHIBOND: Ethicon, UK) and 3-0 monofilament nylon (ETHILON: Ethicon, UK). Circumferential repairs were performed with 6-0 monofilament nylon suture. All repairs were done by a single author (VM) to maintain consistency.
The suture technique was perfected during preliminary study and the suture and tendons were handled gently to minimize trauma. All sutures were knotted by surgical knots, first a double followed by three single square throws. We found that cyclic loading potentially causes knot unravelling. Hence all knots were tied meticulously using square throws to prevent unravelling. Tendons were gently preloaded to the point of deformation before tightening the core knot. This removed slack in the suture material and tendon substance, thereby increasing the initial gap resistance of the repair. During the preliminary study we noted that loading the repaired tendons resulted in gapping that started away from the core knot end of the repair. This eccentric gapping eventually resulted in point loading of the knot and failure. Hence it was decided to place the core and peripheral suture knots at opposing ends of the tendon diameter. This encouraged uniform loading of the repair and increased gap resistance on cyclic loading. Five tendons were repaired according to each method, making a total of 30 repairs. Following repair, the maximum cross-sectional diameter at the repair site was measured using a Vernier calliper, and the increase in diameter was calculated.
Biomechanical testing
The ends of the repaired tendon were positioned in non-slipping, non-crushing self-tightening jaws, which were mounted in a servo hydraulic testing machine (ESH Testing, Brockley, West Midlands, UK; Fig 2). The tendon was kept moist by regular spraying with normal saline solution. The tendon was subjected to 3000 cycles of tensile load, which varied between 2 and 20 N at a frequency of 1 Hz. The chosen load mimicked the physiological load on the human long flexor tendon during active unresisted flexion movement (Schuind et al., 1992). A modified Belfast post-operative rehabilitation regime after flexor tendon repair surgery involves about ten active, unresisted flexion movements of the repaired tendon each hour, for 10 h every day, starting from week 2 following surgery. The number of cycles and the magnitude of loading in our experiment mimic the first 4 weeks of our active motion rehabilitation protocol.
A personal computer was used to generate the loading pattern and record the applied load and the resulting displacement. During testing, the repair site was monitored with a video camera (FP22, Hitachi, Tokyo, Japan) to determine gap formation. Gap size was quantified with a measuring scale adjacent to the repaired tendon (Fig 2). A digital display positioned behind the tendon showed the applied load, such that gap formation could be quantified as function of applied load. At the end of cyclic loading, gap size during the last cycle was noted for the surviving tendons.
Tendons that survived the cyclic loading phase were tested for tensile load-to-failure. Tendons were pulled at a constant displacement rate of 4 cm/min until failure occurred. Ultimate strength of the tendon repair and the corresponding gap size were noted.
Failure mechanism of the repair, including failure of suture material, was determined by experimental observation and confirmed by running the videotape. The load displacement curves for the three repair techniques were drawn for gapping up to 10 mm, and the area under the curve (energy to failure/toughness of repair) was calculated (Fig 3).
Statistical analysis
The mean initial tendon diameter and mean increase in diameter due to the repair were calculated and compared using 2 × 3 analysis of variance (ANOVA). The cyclic loading phase was analysed by comparing the number of tendons in each group that survived this phase using Fisher’s exact test. After testing for normality, ultimate strength for the two core suture materials was compared using 95% confidence limits and t-tests. Ultimate load and energy to failure of the three peripheral repair techniques with the braided polyester core were compared using one-way ANOVA.
RESULTS
Tendon diameter
The average tendon diameter before repair varied from 5.5 to 6 mm, without any significant difference between the repair groups (P>0.2, 2 ×3 ANOVA). Average diameter increase due to repair was less than 1 mm for the running and cross-stitches, and more than 1 mm for the Halsted method (Table 2). The Halsted repair caused a significantly larger increase in thickness compared to the other techniques (P=0.01, 2 × 3 ANOVA), but no significant difference was found between the two core suture materials (P = 0.95, 2 × 3 ANOVA).
Cyclic loading
None of the 15 tendons repaired with a braided polyester core suture failed, but six of the 15 tendons repaired with a monofilament nylon core suture failed. This difference between the two materials was highly significant (Fisher’s exact test, P = 0.01). Failures occurred with each peripheral repair technique with the nylon core suture. Gapping after cyclic loading varied from 0 to 4 mm (Table 3). No significant difference in gap size between the groups was found (P>0.75, 2 × 3 ANOVA).
Load-to-failure
All 24 tendons (15 braided polyester and nine mono-filament nylon repairs) that survived the cyclic loading phase were loaded to failure. The average gapping at the point of maximum loading ranged from 1 to 3 mm (Table 4). No significant difference in gap width was found between suture materials or techniques (ANOVA, P>0.3). The average strength of the nylon repairs was 50 N (95% CL 44–56), and that of the polyester repair was 57 N (95% CL 51–64). This difference is not significant (P = 0.11). Strength and energy to failure for the three peripheral suture techniques were compared for the polyester-core-repaired tendons only, because too many nylon-repaired specimens failed during cyclic loading. The average ultimate strength ranged from 47 to 65 N (Table 4). The three techniques were not equally strong (one-way ANOVA, P = 0.03), with the running stitch producing the weakest repair. The average energy to failure (toughness) ranged from 0.3 to 0.4 J, with the cross-stitch repair being significantly tougher than the other two techniques (one-way ANOVA, P = 0.02; Table 4, Fig 3).
Failure mechanism
The failure mechanism was consistent for all repairs. The epitenon repair was the first to fail followed by rupture of the core suture. Failure was initiated by gapping at the repair site. Tendons repaired with a monofilament core suture showed excessive central gapping at the repair site during each loading cycle. Repeated central gapping initiated the failure of the weak peripheral repair resulting in subsequent failure of the monofilament core repair. This central cyclic gapping was minimal with stiff braided polyester core suture. The epitenon repair failed initially by cutting through, but eventually usually ruptured at the knot (Table 5). The core rupture always corresponded to the ultimate strength of repair. Several Halsted and cross-stitch repairs partially held, even after the rupture of core and peripheral repair, showing better tendon grasp.
DISCUSSION
Our study showed that significant numbers of Kessler repairs performed with monofilament nylon core suture failed during the cyclic loading phase, independent of peripheral repair technique. To the author’s knowledge, this is the first study reporting central cyclic gapping with monofilament core suture material and its association with failure of flexor tendon repair.
The mechanical properties of different types of suture material samples have been compared (Greenwald et al., 1994b; Trail et al., 1989; Urbaniak et al., 1975) but very few studies have actually compared the mechanical properties of repaired tendons. Trail et al. (1989) tested the mechanical properties of suture materials and suggested that there was very little to choose between braided polyester and monofilament propylene. They found that monofilament sutures were more stretchable than braided polyester but had comparative strength in their knotted form. Greenwald et al. (1994b) also concluded that braided sutures had a higher elastic modulus than monofilament and hence were less stretchable. Trail et al. (1989) predicted that a mono-filament suture material with lower elastic modulus and hence greater stretchability, when loaded could gap without actually failing mechanically. The result of our study clearly disagrees with this assumption, as a significant proportion of Kessler core repairs using a monofilament nylon suture failed during the cyclic loading phase, independent of the peripheral augmentation technique. This was an unexpected but consistent finding. Most studies (Greenwald et al., 1994a; Komanduri et al., 1996; Tang et al., 2001; Wagner et al., 1994), that have used nylon monofilament as core suture have relied on load to failure (pull out) and not cyclic loading to determine the tensile strength of the repair. In recent years the importance of cyclic loading to test the tendon repair strength has been recognized (Pruitt et al., 1991; Sanders et al., 1997). Cyclic testing causes gap formation at lower loads than with load to failure testing. During cyclic loading, tendons repaired with monofilament core suture showed central gapping at the repair site during each loading cycle. Central cyclic gapping was minimal with braided polyester core suture which suggests it occurred due to the stretchable nature of monofilament suture material. Central cyclic gapping led to increased loading of the peripheral repair, which started to cut through the tendon tissue and ultimately failed by knot rupture. Failure of the peripheral repair in turn reduced the overall tensile strength of the repair that subsequently failed by rupture of the core suture. This led to failure of six of the 15 monofilament nylon core repairs during cyclic loading. Therefore Kessler repair with a stiff core material (braided polyester) minimizes the central cyclic gapping during cyclic loading and prevents early failure of the weaker peripheral repair.
At the end of the cyclic loading phase mean gapping amongst the surviving tendons was less than 1 mm (range 0–4 mm), without any significant difference between the six groups. This is in agreement with the observation of Sanders et al. (1997) who found no difference between running, cross-stitch and Halsted repairs with respect to number of cycles to 2 mm gap formation using a staircase cyclic loading testing model. However Sanders et al. (1997) noted an initial gapping of less than 1 mm in several cases after the first loading cycle. This early gapping perhaps represents slack in the suture and tendon substance at the time of repair. We noticed this early gapping in our preliminary study but this was minimized by modifying the repair technique as mentioned in the Methods section.
Load-to-failure testing showed that the mean ultimate strength of the repair with a running peripheral suture was significantly less than with the other techniques. There was no statistically significant difference between the ultimate strength of the cross-stitch and the Halsted repair. The actual values could not be compared with the available literature because of differences in tendon source, suture material, suture calibre and testing models. However, both the cross-stitch and the Halsted peripheral repairs have been shown to be stronger than running peripheral stitch (Silfverskiold and Andersson, 1993; Wade et al., 1989).
As found in other studies (Barrie et al., 2001; Urbaniak et al., 1975), all repairs failed first by peripheral failure and then by core Kessler knot rupture, indicating the greater grasping power of the Kessler repair than the peripheral repair. The occurrence of core suture rupture confirms that the grasping power of the Kessler repair is greater than the strength of 3-0 suture material (both monofilament and braided polyester). Increase in suture calibre from 4-0 to 3-0 has been shown to increase the fatigue strength (Barrie et al., 2001) of tendon repair significantly, independent of peripheral technique. We believe, where possible, 3-0 sutures should be preferred for two strand grasping Kessler repairs in a clinical situation.
The failure of the peripheral repair was initiated by pull through, possibly reflecting weak grasp, followed by rupture of the material. This study agrees with previous studies in that Halsted and cross-stitch repairs had greater grasping power than the running suture (Sanders et al., 1997; Silfverskiold and Andersson, 1993). Cross-stitch repair had the highest energy of failure (toughness) reflecting its superior tendon grasping strength.
The increased thickness of the repaired tendon (Sanders et al., 1997) and amount of suture material at the repair site (Aoki et al., 1995) is thought to have an impact on tendon gliding and adhesions. The Halsted repair caused the largest increase in thickness but the cross-stitch repair appears to have most suture material on its surface. The cross-stitch technique, surprisingly, did not significantly increase the thickness at the repair site as compared to the running repair. This may be due to tightening of its Chinese finger trap-like configuration.
The results of this study support the finding that cyclic loading provides a more realistic assessment of a flexor tendon repair’s response to the rehabilitation protocols requiring early motion (Barrie et al., 2001; Choueka et al., 2000; Pruitt et al., 1991; Sanders et al., 1997). Models that follow cyclic load testing with load to failure, such as used by Choeuka et al. (2000) and the present study, probably best mimic the post-operative rehabilitation regimen. Effectively, they test the residual strength of the repair after a known number of cycles. Such models are used in the engineering industry to determine the remaining strength of a structure after a known service time (Paton et al., 2001). Curvilinear models (Choeuka et al., 2000; Komanduri et al., 1996) may be more physiological but are difficult to set up and control. In the current study, porcine instead of human tendon was used as it is satisfactory for biomechanical testing (Savage, 1985) and was freely available.
We are aware that the result of our study is contrary to the established practice of using a monofilament suture as the core suture for flexor tendon repair. However, this result should be treated with caution as our experimental model suffers from some weaknesses. Firstly, we have used a pig tendon that may have some differences from a human tendon. Also the mechanism of failure in a soft, healing, live tendon in vivo may be very different to our experimental model. The impact of central cyclic gapping on a weak peripheral repair would only be obvious in the clinical situation when an aggressive active motion rehabilitation regime is followed.
Footnotes
Acknowledgements
We acknowledge the help and financial support of the Institute of Orthopaedics, Robert Jones and Agnes Hunt Hospital, Oswestry.