Complications associated with tibial plateau leveling osteotomy: A retrospective of 1519 procedures

Abstract

This retrospective study identified complications associated with tibial plateau leveling osteotomy (TPLO) and predisposing factors for these complications in a large population of dogs from a metropolitan area with cruciate ligament deficiency. There were 943 dogs that underwent unilateral TPLO and 288 with staged bilateral TPLO for a total of 1519 procedures. There were 47 cases with at least 1 major complication and 126 cases with at least 1 minor complication but no major complications. The total complication rate (major or minor) was 11.4% [95% confidence interval (CI) estimate: 9.8%, 13.2%]; the major complication rate was 3.1% (95% CI: 2.3%, 4.1%); and the minor complication rate was 8.3% (95% CI: 7.0%, 9.8%). Factors associated with development of complications included being a German shepherd dog [odds ratio (OR): 3.2], tibial plateau angle > 30° (OR: 1.6), and heavier weights (for every 4.5 kg increase in body weight the OR increased by 1.10). Tibial plateau leveling osteotomy is a common treatment for dogs with cruciate ligament deficiency and has a low complication rate.

Introduction

Cranial cruciate ligament rupture (CCLR) is the most common cause of lameness in dogs (1). Numerous techniques have been described to address this condition. Among board-certified surgeons, the tibial plateau leveling osteotomy (TPLO) is one of the more popular techniques for treating cranial cruciate ligament (CCL) rupture (2).

Many studies have reviewed the clinical outcome of using TPLO to stabilize CCL deficient stifles. Complication rates in these studies ranged from 14.8% (a single surgeon in a private referral facility) (3) to 28.8% (multiple surgeons in a university teaching hospital) (4). Our purpose was to characterize the nature and the rate of short-term complications associated with the TPLO technique in a large number of cases performed by multiple experienced surgeons and to identify factors that might predispose to complications.

Materials and methods

Medical records for all TPLO procedures performed at Veterinary Specialty Services in St. Louis, Missouri from January 2005 to January 2010 were reviewed. Criteria for inclusion included a complete medical record and TPLO performed by a surgeon who had completed at least 100 TPLO procedures prior to the beginning of the study period.

The TPLO was performed using a full cranio-medial arthrotomy allowing for a traditional open evaluation of the stifle joint. All remnants of completely ruptured or partially torn or stretched CCLs were removed by all surgeons. All torn menisci had the torn portion excised. A jig was applied to the medial portion of the tibia before the osteotomy. All osteotomies were stabilized with 2 anti-rotational Kirschner wires (K-wires; 0.045 mm) placed from the tibial crest proximal to the insertion of the straight patellar ligament directed caudo-distally to exit the caudal cortex of the proximal fragment. All osteotomies were stabilized using plates by Slocum Enterprises (Eugene, Oegon, USA) or New Generation Devices (Glen Rock, New Jersey, USA) and Synthes screws (Synthes Vet, West Chester, Pennsylvania, USA). If the craniocaudal dimension of the proximal tibia would accommodate it, the osteotomy was stabilized with an additional 8-hole, 2.7 mm DCP plate (Synthes Vet) placed caudal to the TPLO plate. The K-wires were not removed after application of the bone plate. If the tibial crest was thin or reduction of the osteotomy left a gap > 2.0 mm (subjective assessment), the K-wires were incorporated into a tension band apparatus.

Pre- and post-operative radiographs were taken for all patients. All operated limbs were bandaged using a modified Robert-Jones bandage after surgery. Bandages were removed before or at the time of discharge (24 to 48 h after surgery). Patients were evaluated a minimum of 2 times: 10 to 14 d after surgery at the time of suture removal and 6 to 8 wk after surgery when radiographs were taken. Patients were evaluated thereafter as necessary.

The following information was recorded and evaluated statistically for all patients: age, breed, gender, weight, prior abnormalities, prior orthopedic surgeries, duration of lameness prior to presentation, unilateral versus bilateral CCLR at time of presentation, side affected, preoperative tibial plateau angle (TPA), post-operative TPA, state of CCL, state of the meniscus and treatment, blade size, plate size, additional implants, antibiotics, anesthesia time, surgery time, rupture of the contralateral CCL, days between the first and second TPLO surgery (for animals with bilateral surgeries), and complications including subsequent meniscal injury. All recorded complications occurring in the first 6 mo following surgery were reported. Major complications were defined as those that required further surgical intervention. Minor complications were defined as those that did not require surgery.

Statistical analysis

The software SAS v9 (SAS Institute, Cary, North Carolina, USA) was used for analysis. Summary statistics such as the mean and standard deviation were found for all data with quantitative outcomes [e.g., body weight (BW), days of lameness prior to surgery, days between first and second TPLO surgery, TPA, blade size, plate size, duration of anesthesia, duration of surgery, days to complication(s), days to subsequent meniscal tear]. The median value is reported for variables with a highly skewed distribution. Frequency tables and cross tabulations were constructed for categorical and ordinal variables (e.g., breed, gender, bilateral CCL at first evaluation, state of CCL at time of surgery, meniscal evaluation and treatment during surgery, post-operative antibiotics used, complications, and other orthopedic procedures).

Statistical evaluation was performed on complications occurring in the 6 mo immediately after TPLO was performed. Estimation of complication rates, confidence interval estimates of those rates, and examination of predisposing factors were complicated by the fact that some dogs had surgery on both sides. To account for the potential dependency of responses from the same animal, logistic regression was done using generalized linear models. Specifically, the GENMOD procedure in SAS was used for the primary analysis. Potential patient factors (e.g., body weight, breed, TPA angle > 30°), and potential surgical factors (e.g., antibiotic use, prior orthopedic procedures, surgical time, and anesthetic time) were examined singly as well as jointly. For statistical comparison, the population was separated into 4 groups: Labrador retrievers, Newfoundland dogs, German shepherd dogs (GSD), and all other dogs (AOD). These groups were made to compare our study with previous ones (3,5,6) and to investigate breeds that have been anecdotally singled out as having higher rates of complication. Due to the large number of tests that were done, a more stringent level of significance of P < 0.01 was used rather than the standard P < 0.05. All statistical tests used two-sided alternatives. Results are expressed as mean +/− standard deviation (SD) for continuous numeric variables.

In examining factors that might be associated with the occurrence of any complication (either major or minor), attention was restricted to a small number of factors which have been previously reported in the literature (3) or factors suggested by the experience of the surgeons in this practice. The factors studied were breed (4 groups as described), weight, duration of surgery, duration of anesthesia, prior orthopedic surgeries (5 categories: 0 = None, 1 = same leg, 2 = opposite pelvic limb, 3 = both pelvic limbs, 4 = forelimb), indicator of TPA > 30, indicator of antibiotic use, and indicator of having a bilateral rupture before the first surgery or at a subsequent surgery. These factors were first considered individually for their relationship to the presence of any complication by using simple logistic regression taking into account that there were 2 observations for some animals by using general linear models. The Wald statistic was used to determine if the coefficient for the factor in the model was significantly different from zero. For factors with more than 2 categories pairwise comparisons were done using a Tukey adjustment for multiple tests.

Next factors that were significant at the 0.05 level when considered singly were used as candidates in a multivariable logistic regression model. A backward elimination procedure was used with factors being excluded from the model if the P-value for the Wald statistic was greater than 0.05. When all remaining variables showed significance, pairwise interaction terms for these factors were added to the model. These were dropped from the final model if they were not significant at the 0.05 level. A 0.05 level was used in this analysis rather than the more conservative 0.01 level since only a modest number of predictor variables was considered in this modeling step. For factors that were continuous, the linearity of their effect was tested by using the Box-Tidwell approach as suggested by Hosmer and Lemeshow (7). The fit of the model was summarized by calculating the c-statistic and doing the Hosmer-Lemeshow goodness of fit test. Collinearity was addressed by looking at the Condition Number (8).

Results

A total of 1579 TPLO procedures were performed. Of these, 60 cases were excluded from the study because of incomplete medical records. Of a total 1231 dogs, unilateral TPLOs were performed in 943, and staged, bilateral TPLOs in 288 patients.

The mean age was 5.4 +/− 2.6 y (range: 0.4 to 14.6 y). There were 392 mixed breed dogs in the study accounting for 31.8% of the population. The remaining 839 dogs were represented by 61 breeds. The most common breeds were: Labrador retriever (29.7%, n = 366), golden retriever (7.0%, n = 86), and Rottweiler (5.3%, n = 65). The mean weight was 37.3 +/− 11.0 kg (range: 8.0 to 106.8 kg).

Before TPLO surgery, prior orthopedic problems and attempted corrections were noted in the ipsilateral limb in 42 cases (2.8%) and in the contralateral limb in 347 cases (22.9%).

The median duration of lameness prior to presentation was 56 d (range: 1 to 2190 d). At initial presentation 1060 dogs (86.1%) had unilateral CCL rupture and 171 (13.9%) had bilateral rupture. Tibial plateau leveling osteotomies were performed 794 times on the left (52.3%) and 725 times on the right (47.7%) stifle.

The median pre-operative tibial plateau angle was 27° (range: 18° to 45°). The median post-operative TPA was 6° (range: 3° to 15°). The CCLs were completely torn in 1305 stifles (85.9%), and partially torn in 195 stifles (12.8%). The remaining 19 stifles (1.3%) had CCLs that were stretched with marked instability or had interstitial tears. Medial menisci were torn in 611 stifles (40.2%) and intact in 908 stifles (59.8%). Meniscal releases were performed in 452 of the stifles (29.8%). Newfoundland dogs had fewer meniscal tears (20.8%) (P = 0.004) than other breeds.

One hundred seventy-five plates were produced by Slocum Enterprises and 1344 plates were produced by New Generation Devices. Additional implants were used in 178 cases (11.7%). Implants included a tension band apparatus in the tibial crest in 158 stifles, an 8-hole, 2.7-mm DCP plate placed caudal to the TPLO plate in 18 stifles (the average weight of these dogs was 60.9 kg). When cerclage wires were used 1.9% had pin migration (for all dogs 2.2% had migration) and 4.4% developed a tibial crest fracture (for all dogs 3.4% had a tibial crest fracture). Lateral fabellotibial sutures were placed in 2 cases because continued marked instability was noted following the TPLO procedure by means of continued cranial tibial thrust and cranial drawer. One case of instability was noted during the initial surgery after placement of the plate and was corrected at that time. This dog had a pre-operative TPA of 26° and a post-operative TPA of 4°. The second dog had a pre-operative TPA of 25° and a post-operative TPA of 6°. This dog was lame several weeks after surgery with a meniscal click and continued instability. A lateral suture was placed at the time of meniscectomy.

In all cases peri-operative antibiotics were given every 8 h during the first 24 h. Post-operative antibiotics were administered in 1426 cases (93.9%).

The duration of anesthesia was 133.6 +/− 24.3 min (range: 60 to 240 min). Duration of surgery was 65.4 +/− 14.7 min (range: 30 to 135 min).

Three hundred thirty-nine dogs (27.5%) developed CCL rupture in the contralateral limb after the first surgery, a median of 280 d after surgery (range: 19 to 1822 d). Not all dogs had the second ruptured CCL repaired within the time frame of this study.

Complications

Complications were classified as major or minor (Tables 1 and 2). There was a total complication rate (major or minor) of 11.4% (95% CI: 9.8%, 13.2%) with a major complication rate of 3.1% (95% CI: 2.3%, 4.1%) and a minor complication rate of 8.3% (95% CI: 7.0%, 9.8%). Complications were diagnosed by means of physical examination, radiographs, ultrasound, and joint culture.

Table 1.

Major complications among 1519 TPLO procedures in 943 dogs

Major complications	Total	Percent of 47
MPL/LPL (surgery required)	10	21.3%
Joint infection	9	19.2%
Tibial fracture/implant failure	9	19.2%
Plate removal	5	10.6%
Fracture to fibula and subsidence of TPA	3	6.4%
Bone graft for slow healing osteotomy	3	6.4%
Persistent valgus	3	6.4%
Lameness resolves after negative exploratory	1	2.1%
Medial collateral ligament rupture	1	2.1%
Continued instability and placement of lateral suture	1	2.1%
Esophageal stricture	1	2.1%
Renal failure	1	2.1%
Total	47	100%

Table 2.

Minor complications among 1519 TPLO procedures in 943 dogs

Minor complications	Total	Percent of 126
Incisional	47	37.3%
Fracture to tibial crest/patella	44	34.9%
Pin removal/migration	27	21.4%
Patellar tendon thickening	5	4.0%
MPL/LPL (resolution without surgery)	3	2.4%
Total	126	100%

MPL — medial patellar luxation, LPL — lateral patellar luxation.

Subsequent meniscal tears were observed in 12 of 1519 TPLOs. Of 457 menisci that were intact and not released, 8 developed a subsequent meniscal tear. Of 452 cases that had undergone a meniscal release, 3 dogs developed a subsequent meniscal tear. There was 1 meniscus of all 1519 cases that was diagnosed with a tear, the torn portion was removed and another tear developed after surgery. The number of post-operative days until diagnosis of a complication was 42.8 +/− 45.3 d (median = 26 d, range: 0 to 180 d). The time until a subsequent meniscal tear (SMT) was diagnosed was 142.3 +/− 24.5 d (median = 147 d, range: 99 to 179 d).

The results of looking at factors that might be associated with any complication are shown in Table 3. Breed and weight are significant at the 0.01 level. Duration of surgery and TPA > 30° would be significant at the 0.05 level but not at the 0.01 level. Pairwise comparisons among the breeds of Newfoundland, GSD, Labrador retriever, and AOD showed significant differences between GSD and AOD as well as between GSD and Labrador retrievers. Adjusted P-values were < 0.05 for these pairs. The total complication rates for Newfoundlands, GSD, Labrador retrievers, and AOD were 24.0%, 31.0%, 8.4%, and 11.8%, respectively.

Table 3.

Association of individual factors with any complication

Factor	P-value
Breed	0.0019*
Bilateral	0.4515
Prior orthopedic procedures	0.4991
TPA > 30°	0.0338*
Antibiotic use	0.3843
Weight	0.0016*
Duration anesthesia	0.6118
Duration surgery	0.0331*

^*Significant; TPA — tibial plateau angle.

These factors were evaluated jointly for total complications using a multiple logistic regression. For breed, indicator variables for GSD and Labrador retrievers were included. Following a backwards elimination procedure, TPA > 30, GSD, and weight were significant at the 0.01 level. Duration of surgery was not significant when the other variables were in the model. Pairwise interaction terms for the factors were considered for the model but none of the coefficients were significantly different from zero, so interaction terms were not included in the model. The linearity of the effect of the continuous variable weight was tested by introducing a factor of weight*ln(weight) in the model as per the Box-Tidwell approach. The P-value for this factor was 0.9909 indicating that the linearity assumption is reasonable. The Condition Number was 7.4 indicating that collinearity was not of concern. The c-statistic for the final model was a modest 0.596. The Hosmer-Lemeshow goodness-of-fit test had a P-value of 0.0838 indicating an acceptable fit. The estimated coefficients, standard errors, odds ratios (ORs), and 95% CI for the odds ratios are shown in Table 4.

Table 4.

Parameter estimates for multivariable model

Parameter	Estimate	Standard error	Wald Chi-square	P-value	Odds ratio	Lower 95% limit	Upper 95% limit
Intercept	−2.9645	0.2864	107.1734	< .0001	*
TPA ≥ 30°	0.4662	0.2095	4.9543	0.0260	1.5940	1.0570	2.4030
Weight	0.0095	0.0031	9.2518	0.0024	1.0997	1.0336	1.1700
GSD	1.1561	0.4138	7.8047	0.0052	3.1770	1.4120	7.1500

^*Odds ratio for weight is for a 4.5 kg change; TPA — tibial plateau angle; GSD — German shepherd dog.

Based on a model using these factors, a German shepherd dog had an odds of developing a complication 3.2 (95% CI: 1.4, 7.2) times greater than that for a non-GSD. Dogs with a TPA > 30° had an odds of developing a complication that was 1.6 (95% CI: 1.1, 2.4) times greater than that for dogs with a TPA ≤ 30°. For every 4.5 kg increase in body weight, the odds of developing a complication increases by 1.1 (95% CI: 1.0, 1.2).

Factors evaluated and found not to have a statistical significance for developing a complication include: age, gender, prior orthopedic surgeries, duration of lameness prior to presentation, unilateral versus bilateral CCL rupture at time of presentation, side affected, preoperative TPA, post-operative TPA, state of CCL, state of the meniscus and treatment, blade size, plate size, additional implants, antibiotics, anesthetic duration, and surgical duration.

Discussion

This study evaluated the complications associated with TPLO and determined predisposing factors for these complications. The experience of the veterinary surgeon as related to complications rate has previously been evaluated. Recently, Fitzpatrick and Solano (3) reported a complication rate of 14.8% with 6.6% being major complications in a large case series from a single surgeon performing 1000 consecutive TPLOs in dogs. Increased body weight and complete versus partial CCL rupture were predictive variables for complications. Being a Labrador retriever and using post-operative antibiotics were positive predictive factors for preventing infection. The difference in results between this study and ours may be due to the variables evaluated and the difference in the study populations.

Plates were removed because of discharge from the surgical site, loosening of implants, exposure of the implants and radiographic lucency around screws in the plate or chronic lameness on the operated limb. In all cases, plate removal resolved the patient’s clinical signs. Most plate removals occurred outside of the 6-month time period. Five plates were removed in the first 6 mo after surgery (0.3% of all complications). An additional 22 plates were removed (381.5 +/− 260.6 days) after the 6-month period. The infections were managed medically and plate removal was delayed to allow union of the osteotomy.

Tibial fracture/implant failure occurred in 9 cases; these occurred shortly after surgery with collapse of the tibial plateau and implant failure. The average weight of these dogs (44.6 kg) was higher compared with the average weight of all dogs in the study (37.3 kg) but the difference was not statistically significant (P = 0.160). There was a significant difference in complications associated with an increase in body weight. Although none were recorded for these 9 cases, technical errors (i.e., fracture through a distal screw hole, improperly located osteotomy, improperly prepared screw holes, improperly contoured plates) may predispose the proximal tibia to failure around the implants. Prior to 2006, there were no broad or oversized plates specifically manufactured for stabilization of the TPLO. In dogs with a broad proximal tibia, a straight plate was often placed caudal to the TPLO plate. In obese dogs, this was often not possible due to their proportionally smaller proximal tibia. This may contribute to the increased complication rate associated with increasing weight. With the advent of numerous TPLO plate designs and sizes, patient weight may be less of a predisposing factor for complications in the future.

Patellar luxation was the most common major complication (10 cases) and a minor complication in 3 cases. These results are similar to previous reports (9) in which 84% of patellar luxations resolved following surgery and 16% with medical management. Potential causes for post-TPLO patellar luxation include muscle atrophy, closure of the medial retinaculum under increased tension, suture failure or incomplete closure of the medial retinaculum, severe joint effusion after surgery, concurrent undiagnosed patellar luxation, or mal-alignment of the proximal tibia following plate application (3,9).

Medial meniscal injury is a common sequella to CCL injury (10–13). Concurrent medial meniscal pathology is present in 58% of dogs with CCL injury (14). The rate of SMT for various cranial cruciate deficient stifle stabilization surgeries in the dog varies from 1.9% (15) to up to 28% (3,10,16,17). In our study, the SMT rate was 0.8%. With this low rate of occurrence, we propose that this injury may not be a complication of TPLO surgery, but a continuation of the abnormal stifle dynamics in a CCL deficient joint (18). Arthroscopy with probing of the medial meniscus is the most sensitive method of diagnosing meniscal pathology (19). We speculate that the lower SMT rate in our study was related to surgeon experience and use of the full cranio-medial arthrotomy with probing of the meniscus. Such practices allowed a more thorough evaluation of the medial meniscus than a caudal arthrotomy or not probing the meniscus.

Incisional problems were the most frequently reported minor complication (47 cases); they included seroma formation, inflammation, bruising, premature suture removal, and excessive skin trauma caused by licking. Incisional complications are common with any surgical procedure in dogs (20,21). Most incisional complications are resolved with medical management and optimal postoperative care supplied by the owner at home during the first 2 wk of recovery.

Tibial crest and patellar fractures were seen in 44 dogs. This was the second most common minor complication, similar to results from a study by Bergh (22). In most cases, tibial crest and patellar fractures were radiographic diagnoses. Most dogs were not symptomatic for this injury. When radiographics were reviewed, it was noted that most fractures occurred through the holes of the anti-rotational pins placed after the osteotomy was performed. Distal patellar fractures were seen in 2 cases involving spayed female dogs with no evidence of external trauma to the patella after surgery. We speculate that the change in forces at the stifle after TPLO may cause excess stress on the distal portion of the patella leading to fracture. Most traumatic patellar fractures are treated surgically. However, both of these injuries were found on radiographs taken 6 to 8 wk after surgery and both resolved without surgical intervention. A similar finding was reported by Rutherford et al (23).

Migration and removal of the anti-rotational pins was reported in 27 cases. All anti-rotational pins were smooth (0.045 mm) K-wires and 2 pins were used. We left the anti-rotational pins in place after the surgery. It is our belief that doing so prevented stress risers from developing through the pin tracts as well as adding another source of stability as a pin/plate construct. Pin migration was observed because of the smooth nature of the pins used in our patients. A smooth pin placed in bone has a higher tendency for implant loosening compared to a threaded pin. Although a common minor complication, most smooth anti-rotational pins probably do not migrate because of the stability of the plate construct which reduces micro motion. Partially threaded anti-rotational pins should be considered if it is intended to not remove these pins.

Cerclage wires were used in 158 stifles. Their use was determined at the time of surgery if there was a subjective greater than 2-mm gap in the osteotomy or if the tibial crest was thin. An additional 8-hole, 2.7 mm DCP was used on 18 tibiae. The surgeon evaluated each tibia for its overall width taking into account the patient’s body condition and weight compared to the plate being used. These additional plates were used to prevent a perceived overload of the single TPLO plate. There was no difference in post-operative TPA in these dogs compared with the total population.

Multiple logistic regressions showed that TPA > 30°, being a GSD, and increasing body weight were significant predisposing factors contributing to the occurrence of a complication. A larger TPA requires a greater degree of proximal tibial plateau rotation to achieve the desired minimum TPA of 6.5° (24). This increased rotation may cause extra trauma to the fibula and surrounding soft tissues of the proximal tibia, which may lead to bony and soft tissue injury. Also, with greater degrees of rotation, an improperly centered osteotomy can lead to increasingly poor reduction of the osteotomy and an increased level of implant failure. While rotating the proximal tibial plateau less may achieve a higher post-operative TPA, it may also decrease complications in such patients. Further study is warranted.

Increasing body weight leads to an increase in complication rate (3,25). This can be seen in both lateral fabellotibial correction (15) and TPLO (3). Our study confirms this finding. In addition, our study quantified the potential increased complications associated with increased weight. For every 4.5 kg increase in body weight, the odds increase by 1.10. This means that a 45 kg dog would increase his odds of having a complication by 1.30 over a dog that weighs 31.5 kg.

Our study was retrospective and therefore limited by the data entered for each patient. Another limitation was that statistical evaluation was performed on complications occurring in the 6 mo immediately after TPLO was performed. This was done to allow all animals to have the same length of post-operative evaluation time. However, we did compile all complications that occurred during the entire study period. An additional 24 complications (22 of which were plate removals) occurred after the 6-month period. As a result, 88% of all complications occurred in the study period. Other limitations to our study were the lack of control groups and outcome measures for the study population and the absence of follow-up phone calls to check on animal recovery. It was thought that the long time frame (January 2005 to December 2009 and recording of the data starting in July 2010) of this study would cause inaccurate accounts by the owners. We have a close working relationship with referring veterinarians and owners are instructed (orally and in writing) to call us if any lameness or complication arises. We have an emergency clinic located in our facility that is open 24 hours a day.

We conclude that in the hands of experienced surgeons, TPLO is a safe technique. The complication rate increased as the weight of the patient increased and being a GSD, and having a TPA > 30° were also predictive of an increased complication rate. Our study confirms a low complication rate at 11.4%. The rate of SMT (0.8%) was very low whether a meniscal release was performed or not. Most complications were minor and resolved without surgical intervention. Complication rates in this study were not higher than with other techniques used to treat CCL rupture (26–31). Indeed, complications requiring further treatment following a novel extracapsular stabilization procedure (TightRope) was reported to be 12.5% (27). In contrast, in our study, only 7.4% of cases needed further treatment.