Notch morphology is a risk factor for ACL injury: a systematic review and meta-analysis
Renato Andrade, Sebastiano Vasta, Nuno Sevivas, Rogério Pereira, Ana Leal, Rocco Papalia, Hélder Pereira, João Espregueira-Mendes
DOI: 10.1136/jisakos-2015-000030 Published 18 April 2016
Importance Several bone morphological parameters have been identified in the scientific literature as risk factors for sustaining an anterior cruciate ligament (ACL) injury; however, a clear consensus on which are the most predisposing factors is still missing.
Aim This systematic review and meta-analysis aims to investigate the association between bone morphological parameters and the risk of sustaining an ACL injury.
Evidence review We conducted a comprehensive search using PubMed, Cochrane Library, Scopus, CINAHL and SPORTDiscus databases from 2005 until 2015. Two authors independently searched for relevant studies that assessed the association between bone morphology and ACL injury. Other search sources were used for hand-searching additional potential studies and the reference list of recent studies was screened. The methodological quality of the included studies was assessed through an adapted scale for radiological studies. A fixed-effects or random-effects model was used accordingly to estimate the mean differences with 95% CIs regarding the association of ACL injury with intercondylar notch (ICN) width, notch width index (NWI) and tibial slopes.
Findings 23 studies were included for analysis comprising a total of 3452 participants, 1681 with an ACL injury and 1763 with an intact ACL. The ACL-injured individuals had narrower ICN width (p<0.001), smaller NWI (p=0.005) and steeper tibial slope (p<0.001).
Conclusions On the basis of the current scientific literature, narrower ICN widths, smaller NWI and increased tibial slopes put the individual at higher risk of injuring the ACL. Future research should focus on developing indexes for different parameters rather than absolute measurements.
What is already known
Several bone morphological parameters have been identified as possible risk factors for anterior cruciate ligament (ACL) injury.
Stenosis of the intercondylar notch (ICN) and smaller notch width index (NWI) have been associated with higher risk of ACL injury.
Increased tibial slopes have been hypothesised as possible risk factors for ACL injury.
What are the new findings
A narrower intercondylar notch width puts the individual at higher risk of anterior cruciate ligament (ACL) injury.
A smaller notch width index increases the risk of sustaining an ACL injury.
Steeper posterior, medial and lateral tibial slopes increase the risk of injuring the ACL.
The anterior cruciate ligament (ACL) plays a key role in knee stability and has, therefore, gained a special interest from the orthopaedic surgical community over time. It is estimated that between 80 000 to more than 250 000 new ACL injuries will occur every year1 with a reconstruction rate of approximately 175 000–200 000 per year.2 Considering the high incidence rate and the subsequent economic burden,3–5 it is crucial to predict the injury risk for sustaining an ACL injury.
It is known that the mechanisms underlying ACL injury are multifactorial;6therefore, several elements must be considered when assessing ACL status in order to better prevent the occurrence of an ACL tear. The scientific literature describes many predisposing factors for ACL injury, including biomechanical, functional, anatomical and morphological factors. The bony anatomical and morphological parameters described in the scientific literature focus primarily on the femoral intercondylar notch (ICN) and femoral condyles’ shape and dimensions,7–20 notch width index (NWI)8–10,12 ,14 ,18 ,19 ,21–23 and tibial slopes.8 ,17 ,19 ,21 ,22 ,24–29 The pathological association between the ICN and the ACL was first described by Palmer,30in 1938, when he noticed, as the knee was flexed and supinated, that the ACL was placed in a vulnerable position by being stretched over the inner margin of the lateral femoral condyle.
Since Palmer’s30 study, dozens of other studies have been conducted to investigate the ICN and other bone morphological parameters’ role on ACL injury. However, unequivocal evidence and a clear consensus on which morphological parameters are the most predisposing to ACL injury have not been reached so far. This systematic review and meta-analysis aims to investigate the most recent scientific evidence regarding the association between bone morphological parameters and the risk of sustaining an ACL injury.
This systematic review was conducted according the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement, which aims to improve the standard of reporting of systematic reviews and meta-analyses.31
A comprehensive search was conducted using PubMed, Cochrane Library, Scopus, CINAHL and SPORTDiscus databases, in order to search for relevant studies that assessed the association between bone morphology parameters and ACL injury. Additionally, other search sources were used for hand-searching additional potential studies and the reference list of recent studies was scanned in order to achieve the greatest number of available studies in the scientific literature. Two independent assessors (RA and SV) conducted the searches, and results from both were combined and compared to check if they matched; any disagreement was resolved by consensus. All searches were conducted including papers published from 1 January 2005 to 1 December 2015. The following key words were selected by the authors to build the search strategy: ‘bone morphology’; ‘bony morphology’; ‘knee morphology’; ‘femur morphology’; ‘tibia morphology’; ‘tibial morphology’; ‘intercondylar notch’; ‘transcondylar axis’; ‘condyle width’; ‘notch width’; ‘tibial slope’; morphology; morphometrics; ‘anterior cruciate ligament’; ACL; injury; risk factor; tear; rupture; lesion.
Initially, all titles and abstracts from the selected databases were reviewed and potential studies identified to be included in the systematic review. To achieve this, we retrieved and read the full text, and assessed the eligibility according the following criteria: (1) English language studies; (2) diagnostic and prognostic studies and (3) assessment of at least one bone morphology parameter as potential risk factor for ACL injury. Studies were excluded if they were: (1) reviews or meta-analyses; (2) clinical commentaries or expert opinions; (3) single study cases; (4) animal studies; (5) of a skeletally immature population and (6) not clearly defining the study’s inclusion/exclusion criteria.
The main outcome of interest was the association between bone morphology parameters and ACL injury. Thus, after applying the eligibility criteria, the included studies were analysed based on their methodological design and level of evidence, their sample characteristics (gender and age), inclusion and exclusion criteria, determination of ACL injury, measurement method for bone morphology, bone morphological measurement parameters assessed, most relevant results and conclusions to the risk for ACL injury.
A methodological quality scale for radiological studies, proposed by Arrivé et al,32 was used without the standards 3, 5, 10, 11 and 12, since they did not apply to the purpose of this study, because a gold reference standard of knee morphological parameters that can predict an ACL injury has not yet been established. The standards were classified as full, partially met and not met, which we scored as 1, 0.5 and 0 points, respectively. Additionally, the level of evidence was assessed. Two authors (RA and SV) independently assessed the methodological quality of all included articles and all disagreements were discussed until consensus was reached.
The data quantitative analysis was performed using RevMan software, V.5.3 (Cochrane Collaboration, Oxford, England). The mean difference was used to analyse continuous variables and it is reported with the correspondent 95% CIs. A p value of 0.05 was set for statistical significance. The statistical heterogeneity between the included studies was evaluated by the χ2 and I2tests (percentage of total variation across studies resulting in heterogeneity rather than chance), with significance set at p<0.10.33 In this sense, the Cochrane Collaboration33 established thresholds to interpret the I2percentages: 0–40%: might not be important; 30–60%: may represent moderate heterogeneity; 50–90%: may represent substantial heterogeneity; 75–100%: represents considerable heterogeneity. Therefore, when the I2statistic was lower than 60%, the fixed-effects method was used. For other comparisons the random-effects method was used.
The database and hand searches yielded 145 titles, which were reduced to 118 articles after removal of duplicates. Afterwards, 53 studies were removed based on their title and abstract, resulting in 65 full-text articles that were fully screened for eligibility. After screening, another 42 studies were excluded, leaving 23 studies eligible for inclusion within the qualitative synthesis and 15 for the quantitative synthesis. Reasons for exclusion are highlighted in the PRISMA flow chart (figure 1).
Description of studies
Table 1 shows the features of the 23 included studies,7–29 comprising a total of 3452 participants (2347 male and 1102 female participants), of whom 1681 had an ACL injury and 1763 had an intact ACL. The majority of the studies used MRI7–17 ,19–21 ,23 ,24 or radiography7 ,18 ,21–23 ,25–29 for measurement of bone morphology parameters. Additionally, Vrooijink et al12compared their measurements performed by MRI or intraoperatively, which did not find any correlation. Most of the studies were retrospective case–control studies and their methods are described in table 2. One study15although retrospective, did not included a control (non-injured) group and three other studies followed a prospective fashion.14 ,18 ,26 In addition, one study performed a case-series of ACL-reconstructed and ACL-deficient patients.25 Overall, the studies provided a moderate description of their inclusion and exclusion criteria: 11 studies included isolated unilateral ACL tears,7 ,9 ,11 ,13 ,16 ,20–26 7 studies included ACL reconstruction/surgery patients,8 ,10 ,12 ,15 ,19 ,27 ,28 one study explored ACL-deficient knees,29 one study investigated the contralateral knee of ACL-injured patients17 and one study also included bilateral ACL tears.23 Nonetheless, 12 of 23 studies (52%)7–10 ,12 ,13 ,15–17 ,20 ,23 ,25 did not report how did they determined the status of ACL.
Table 3 displays the score of each standard on Arrivé’s et al‘s32methodological quality scale for radiological studies. Overall, the studies had a good methodological quality, scoring a mean of 7.3±1.2 of 10 points; 5 studies scored below 7 points.11 ,17 ,21 ,23 ,29 The most common issue was the intraclass coefficient correlation values, that is, the interobserver and intraobserver, which were not reported in 114 ,9 ,11 ,14–23 ,25 ,26 ,29 and 8 studies,9 ,11 ,16–18 ,20–23 ,25–27 ,29 respectively. Additionally, other common issues were exclusion and inclusion criteria; 4 studies11 ,20–23 did not report their exclusion criteria and 1 study24 mentioned them but did not give a clear definition; 3 studies15 ,23 ,24 did not provide enough details on inclusion criteria in order to replicate the study; 8 studies12 ,13 ,17 ,19 ,23 ,27 ,28 ,29defined and developed several purposes. The final score for each individual study can be seen in table 2.
A total of 12 studies reported data on 1822 patients (762 ACL injured and 1060 non-injured) regarding ICN width.7–14 ,16–20 The statistical analysis indicated a statistical significant mean difference in the ICN width of −1.67 mm in favour of the ACL-injured patients (95% CI −2.57 to −0.76 mm; p<0.001). The statistical test for heterogeneity showed substantial heterogeneity (I2=80%; p<0.001) (figure 2).
Seven studies10 ,11 ,14 ,16 ,17 ,20 ,22 showed that a narrower ICN correlated with a higher risk of ACL injury (table 4). Conversely, Alentorn-Geli et al,19Stijak et al7 and Lombardo et al18 found no significant difference on ICN width between ACL-injured and non-ACL-injured participants. Similarly, Vrooijink et al12 found no significant difference on the ICN width between ACL-injured and non-injured participants; however, they found smaller bicondylar width in injured male participants.
Notch width index
Overall, 10 studies8–10 ,12 ,14 ,18 ,19 ,21–23 reported the NWI on 1775 patients (731 ACL-injured and 1044 non-injured). The random-effects model showed a statistical significant mean difference of −0.02 on the NWI for the ACL-injured patients (95% CI −0.04 to −0.01; p=0.005). The tests for statistical heterogeneity indicated considerable heterogeneity (I2=93%; p<0.001) (figure 3).
A smaller NWI was reported to predispose a higher risk of ACL injury,14 ,21–23 especially for values under or equal to 0.25.23 However, there are studies showing no correlation between NWI and ACL injury.8 ,12 ,19 Al-Saeed et al9stated that the NWI did not seem to influence the risk of sustaining an ACL injury and Lombardo et al18 were not able to predict ACL injury rates through the NWI in professional male basketball players. In addition, Alentorn-Geli et al19 found no significant differences between ACL-ruptured and non-injured male patients, while Park et al10 found significant differences only for the female cohort.
Shape of femoral notch
Three studies reported on the femoral notch shape, comprising a total of 714 patients (357 ACL injured and 357 non-injured).9 ,13 ,16 An ‘A-shaped’ femoral notch was reported as the most frequent type among the patients with an ACL tear (73% of participants with type A notch had an ACL tear).9Moreover, van Eck et al13 found a trend to larger notch volumes in patients with unilateral acute ACL tear when compared with non-injured participants. In addition, Everhart et al16 found thicker notch bone ridges in the anteromedial aspect of the ICN in participants who previously sustained a non-contact ACL injury, reaching a significantly strong association (p=0.0014) with the occurrence of this injury.
Condyle width was reported in five studies8 ,10 ,12 ,18 ,19 regarding 807 patients (270 ACL injured and 537 non-injured). The medial condyle width means across the included studies ranged from 23.8 to 29 mm for ACL-injured patients and 22.5–28 mm for non-injured patients.8 ,10 ,12 ,19 In turn, the lateral condyle width means ranged from 26.1 to 31 mm for ACL-injured patients and 24.8 to 31.5 mm for non-injured patients.8 ,10 ,12 ,19 In this sense, van Diek et al8 found a significantly larger lateral condyle width in female patients who had suffered an ACL injury. Additionally, Vrooijink et al12found a significantly larger lateral condyle width in individuals who had undergone ACL reconstruction, in both genders. In contrast, Park et al10found significantly larger medial condyles in both genders among ACL-injured patients, while Alentorn-Geli et al19 and Lombardo et al18 found no significant differences in condyle widths among ACL-injured and non-injured players.
The ratio between the medial and lateral condyle size was assessed in the studies from Vrooijink et al12 and Park et al10 by dividing the medial condyle width by the lateral condyle width. While Vrooijink et al12 found no statistically significant differences between ACL-injured and non-injured participants, Park et al10 found increased ratios in ACL-injured patients.
Two studies15 ,24 investigated the tibial morphology on 98 ACL-injured and 55 non-injured patients. Smaller medial-to-lateral tibial plateau diameters produced higher grades in the pivot shift test.15 In addition, Hashemi et al24reported significantly shallower medial tibial depths in ACL-injured patients, that is, shallower medial tibial concavity, resulting in an odds ratio of 3.03 per each 1 mm decreased in its total value.
Angulation of the tibial plateaus and femoral condyles
The angulation of the tibial plateaus and femoral condyles was reported in six studies8 ,11 ,17 ,19–22–29 including 1863 patients (960 ACL-injured and 903 non-injured). We pooled the results regarding the measurements of the tibial slope,8 ,17 ,21 ,22 ,24 ,26–28 subgrouping them into posterior, medial and lateral tibial slope. The test for heterogeneity showed that there was moderate heterogeneity for the overall effect (I2=54%; p=0.03) but no heterogeneity for the subgroup differences (I2=0%; p=0.87). A statistical significant mean difference of 1.55° in the fixed-effects model was found favouring the ACL-injured pool of patients (95% CI 0.73 to 2.64; p<0.001). When analysing the subgroups individually, statistical significance was found in all the three subgroups, also favouring the ACL-injured group: posterior tibial slope (mean difference, 1.57; 95% CI 1.14 to 2.00; p<0.001); medial tibial slope (mean difference, 1.32; 95% CI 0.31 to 2.33; p=0.01); lateral tibial slope (mean difference, 1.68; 95% CI 0.73 to 2.64; p<0.001) (figure 4).
Sonnery-Cottet et al21 and Hashemi et al24 found that an increased posterior tibial slope was a risk factor for ACL injury and that this measure was negatively correlated with the NWI (r=−0.28, p=0.0052).21 Along this line, other authors also found that patients with non-contact ACL injuries had steeper posterior tibial slopes than ACL-intact patients.19 ,26–28 Furthermore, Şenişik et al26 added that participants had an increased posterior tibial slope in their ACL-injured leg when compared to the contralateral one, and that participants with a posterior tibial slope >9.57° would have a nearly sixfold increased risk of ACL injury. Conversely, Chung et al22 reported no significant differences on the posterior tibial slope between ACL-deficient and ACL-intact knees.
Brandon et al29 studied the posterior-inferior tibial slope and found that ACL-deficient patients had increased slopes and that higher slope values were associated with higher grades of the pivot-shift test. In addition, Hohmann et al29 reported that, when the posterior tibial slope was divided into three subgroups (0–4°, 5–9° and >10°), there was a greater correlation between the posterior tibial slope and knee functionality in ACL-deficient and ACL-reconstructed patients.
When comparing medial and lateral tibial slopes, Simon et al17 and Alentorn-Geli et al19 showed that ACL-injured patients had an increased lateral tibial slope and that this measure was negatively correlated to the ICN width at the inlet, considering this increased slope a potential factor for injuring the ACL on the contralateral knee.17 The lateral and medial tibial slopes were also measured by van Diek et al,8 however, they could not find any differences between ACL-injured and non-injured participants.
When measuring the inner angle of the lateral femoral condyle, Miljko et al11reported that this angle was increased in athletes with an ACL tear when compared to those who were non-injured. Moreover, it was reported that ACL-injured patients had higher α angles (formed by the Blummensaat line and the long axis of the femur)20 and decreased ICN angles (formed by the Blumensaat line and anterior tibial slope), 19 which were considered risk factors for ACL injury. Furthermore, Alentorn-Geli et al19 added that the ICN angle measurement seems to have superior clinical relevance than ICN width for evaluating the risk of sustaining an ACL injury.
Most of the studies included both genders in their cohort, with the exception of four studies.11 ,18 ,19 ,26 Nonetheless, only four studies found statistically significant differences between genders in the bone morphological parameters measured.8 ,10 ,12 ,20 When comparing male and female participants, it was found that female participants had statistically significant smaller bicondylar width,8 ,10 ,12 smaller notch width10 ,20 and smaller medial8 ,10 ,12 and lateral condyle widths.10 ,12
The scientific literature reports several bone morphology and morphometric parameters that have been correlated with the risk of sustaining an ACL injury. The main findings of the current systematic review and meta-analysis were that narrower ICN widths, smaller NWI and steeper posterior, medial and lateral tibial slopes, put individuals at higher risk of sustaining an ACL injury.
A narrower ICN has been associated with an increased risk of sustaining an ACL injury.10 ,11 ,14 ,16 ,17 ,20 ,22 ,34–37 The meta-analysis performed confirms this association, showing that ACL-injured patients had significantly narrower ICN widths when compared to those who were non-injured. In contrast, three studies were not able to show the same results.7 ,18 ,19Lombardo et al‘s18 was the only study in which the authors measured ICN width at 45° of knee flexion on X-rays instead of, as all the other studies did, on MRIs. Moreover, they only included a very specific population—all the participants were professional basketball players from the National Basketball Association. Previous studies have shown that bone morphology is influenced by the participant’s height and weight.38 ,39 ,40
The scientific literature has been identifying a smaller NWI as a predisposing factor to ACL injury.14 ,21–23 ,34 ,41 ,42 Along the same lines, our results indicated that ACL-injured patients have significantly smaller NWIs, which might have predisposed them to a higher risk of sustaining an ACL injury. In particular, Hoteya et al23 showed that an NWI ≤0.25 would lead to a higher risk of injury. Notwithstanding, some authors found no significant differences in NWI between ACL-injured and non-injured participants.8 ,9 ,12 ,18 ,19 Also, notch shape has been investigated as a possible risk factor, and an ‘A-shaped’ notch, the most stenotic type of femoral notch, has been related to ACL injury.9 ,43 Moreover, larger notch volumes or thicker notch bone ridges were also found in participants who had sustained an ACL injury.13 ,16
Concerning femoral condyles, it has been suggested that a decreased lateral femoral condyle width compared to a medial femoral condyle would predispose the individual to knee valgus collapse, increasing the risk of ACL injury.44–46 In this sense, two studies reported that larger lateral condyle width is related to ACL injury,8 ,12 while others found no differences between ACL-injured and non-injured participants,18 ,19 and one study did not find that larger medial condyles are predisposing to ACL injury.10 Maybe the disproportion between medial and lateral condyle size is more important than the condyle size itself. In fact, a positive association was shown between increased medial and lateral condyles ratio and ACL injury in female patients,10 although this is not a univocal finding.12 Additionally, Pereira et al47 suggested that a high ratio between anteroposterior length of the lateral condyle and the anteroposterior length of tibial plateau significantly increases the risk of sustaining an ACL tear.
Correspondingly, the tibial morphometrics also showed association with a higher risk of ACL injury. A shallower concavity of the medial tibial plateau seems to predispose to a threefold higher risk for ACL injury per each decreased millimetre.24 Moreover, smaller medial to lateral tibial plateau diameters were associated with higher grades in a pivot shift test,15 which could lead to mechanical events predisposing to ACL tears.48 Nevertheless, these results should be interpreted with caution since the pivot shift test is performed manually and subject to some intra-rater and inter-rater variability.
Contrasting data have been reported about tibial plateau angulation being a possible risk factor for ACL injury.8 ,17 ,19–22 ,24–29 ,49 Moreover, caution should be taken when comparing the results from different studies due to the different methods for measuring the tibial slope reported in the scientific literature,28 ,50 including the longitudinal axis,24 anterior tibial cortex axis51and posterior tibial cortex axis.25 In this sense, Zeng et al28 compared the reliability and consistency of these three different measuring methods on radiograph, concluding that the anterior tibial cortex method results in the highest values and the posterior tibial cortex method in the smallest values, suggesting that the longitudinal axis method might be the most reliable method for measuring the posterior tibial slope. The meta-analysis regarding the tibial slopes showed that ACL-injured patients had significantly steeper tibial slopes than those who were non-injured. When individually analysing the posterior, medial and lateral tibial slopes, it was also found that each of these slopes was significantly increased in the ACL-injured patients (1.57°, 1.32° and 1.68°, respectively). In this sense, it has been previously shown that an increase of 10° in the posterior tibial slope can lead to a 6 mm increase in the anterior tibial translation, creating a threefold higher ACL load.52 Along this line, it has been believed that a steeper tibial slope creates an anterior shear force and increased quadriceps muscle force, resulting in greater sagittal knee laxity.24 ,25 ,51–53 Furthermore, it has also been shown that ACL-reconstructed knees with increased posterior tibial slope have higher re-injury risk.53 In contrast, Hohmann et al25 found that ACL-deficient and ACL-reconstructed patients with steeper posterior tibial slopes had more functional knees; however, unequivocal evidence on causality for these results was not established. Similarly, a steeper lateral tibial slope was pointed out as a possible risk factor,17 though again with conflicting evidence.8 Although van Diek et al8 found no differences in the medial tibial slope between ACL-injured and non-injured patients, an increased medial tibial slope has been indicated as an ACL injury risk factor.19 ,24 Differences in the lateral and medial tibial slopes would lead to a higher torsional load, which in turn increases the risk of ACL injury.54 ,55Along this line, several authors showed that an increased lateral slope would result in external rotation of the femur, which has been shown to put the ACL into more stress.56–58
It has been demonstrated that female athletes are at higher risk of ACL injury compared to male athletes.59–61 This is in accordance with studies investigating bone morphological parameters, in which females’ morphological parameters were significantly associated with higher risk of ACL injury.8 ,10 ,12 ,19 ,20 ,25
Several limitations should be considered in the present study. Since no reference gold-standard values have been yet established in the literature, the results could only be compared between ACL-injured and non-injured participants, without any reference to established standardised values. Furthermore, the population included in the control groups also differed through the studies, ranging from healthy subjects and medical staff to patients with knees affected by pathologies other than ACL injury, such as meniscal injuries or patellofemoral pain (although none of the studies included ligamentous injury in the control group). The gold standard for confirming an ACL tear is through arthroscopy, however, most of the studies did not report how they defined the presence of an ACL injury/rupture, and even those that did showed heterogeneity in the methods used for determining ACL tear, which may lead to misclassification bias. The lack of access to non-reported data (ie, means and/or SDs) precluded meta-analysis of medial and lateral femoral condyle width and between male and female participants.
Our meta-analysis showed that ACL-injured individuals had smaller ICN widths, decreased NWI and increased posterior, medial and lateral tibial slopes. These findings suggest that individuals with narrower ICN widths, smaller NWI and steeper tibial slopes are at greater risk of sustaining ACL injuries. Furthermore, since bone morphology is variable among populations, being influenced by body size, future research should focus on developing indexes that put different parameters into relation rather than absolute measurements.
Competing interests None declared.
Provenance and peer review Commissioned; externally peer reviewed.