Clinical versus MRI grading of the medial collateral ligament in acute knee injury

ABSTRACT

Sensitivity, specificity, and agreement between clinical and magnetic resonance imaging (MRI) gradings of the medial collateral ligament (MCL) after acute knee injury were evaluated in 362 patients. Ninety-seven per cent were injured during sports/recreation. Sensitivity and specificity of MRI for grade II or III MCL injury was 68% (95% CI 58–77%) and 90% (95% CI 86–93%), respectively. Weighted Kappa analysis showed moderate agreement between clinical and MRI grading (0.56 [95% CI 0.48–0.65]). Findings were similar for patients with and without concomitant cruciate ligament rupture (0.57 [95% CI 0.48–0.66] and 0.55 [95% CI 0.35–0.75], respectively) and for specialists in orthopaedics and knee sub-specialists (0.55 [95% CI 0.39–0.70] and 0.57 [95% CI 0.47–0.67], respectively). Agreement between clinical and MRI grading of MCL injuries by orthopaedic specialists in a general hospital is at least moderate regardless of the presence of cruciate ligament injury.

KEYWORDS:

• Medial collateral ligament injury
• magnetic resonance imaging
• clinical examination

Introduction

The medial collateral ligament (MCL) is one of the most frequently injured knee ligaments (Olsson et al., 2016). Grade I and II injuries are generally managed non-surgically, whereas both surgical and non-surgical management have been documented with successful outcome in grade III injuries (Marchant et al., 2011; Smyth & Koh, 2015). A range of motion brace may be recommended particularly in higher grade injuries (Laprade & Wijdicks, 2012; Lundblad et al., 2019). Although examination by an experienced clinician has traditionally been the accepted standard for grading, the possibility of concomitant injury, most commonly the anterior cruciate ligament (ACL; Olsson et al., 2016), means MRI is often utilized after the initial clinical examination as part of the evaluation. The relationship between clinical and MRI findings with regard to the MCL is not well defined ,and there is a possibility that discrepancies in grading may lead to uncertainty about the degree of ligament injury present. Several previous studies have suggested questionable agreement between clinical and magnetic resonance imaging (MRI) grading (Mirowitz & Shu, 1994; Schweitzer et al., 1995; Yao et al., 1994). The studies by Mirowitz & Shu (Mirowitz & Shu, 1994) and Yao (Yao et al., 1994) are, however, both limited by relatively small sample sizes, and in addition, the latter studied a mix of both acute and chronic collateral ligament injuries. In a recent study of 130 elite professional athletes, investigators reported excellent agreement between clinical and MRI grading of MCL injuries (Lundblad et al., 2019). It is, though, uncertain whether findings from a group of high-level athletes and highly specialized team doctors are applicable outside of elite sport. Rasenberg et al. (1995) also demonstrated excellent agreement between MRI and clinical examination; however, their use of a varus/valgus testing device in a group of only 21 patients means their results may not be generalizable to a wider population. As such, uncertainty remains regarding the relationship between clinical and MRI grading of the MCL. The purpose of the present study was to investigate the agreement between clinical grading by an orthopaedic specialist and MRI grading of the MCL in a large cohort of patients presenting to a district general hospital with acute knee injury.

Methods

Experimental design

Cohort study

Participants

Consecutive patients with acute knee injury presenting to either the emergency department or the acute knee injury clinic at a district general hospital in southern Sweden. The catchment area has an urban dominance mixed with a rural population. Ethical approval was obtained from Lund University Ethics committee (2009/237 and 2015/205). The full clinical protocol has previously been described in detail (Olsson et al., 2016). In brief, all patients with acute knee trauma between 2002 and 2008 were referred for subacute MRI after an initial clinical examination. Patients who had their examination performed by an examiner other than a specialist in orthopaedics (OSO) or a knee sub-specialist (OSK), suffered an acute patella dislocation, did not have their valgus stability assessed at initial examination or were given an indeterminate injury grading (where the examiner was unable to decide between two gradings and documented I–II or II–III) were excluded.

A total of 1145 knees were evaluated of which 469 knees were examined by an OSO or OSK. Sixty-two cases of acute patella dislocation, 40 cases where valgus stability was not tested and 5 cases of indeterminate injury grading were excluded, leaving 362 knees for analysis. Of 362 knees, 258 were examined by an OSK and 104 by an OSO. Two hundred and seventy-five (76%) had a concomitant cruciate ligament injury on MRI examination.

Median age was 25 (25^th, 75^th percentile 19, 34) years. The median time from injury to examination was 4 (25^th, 75^th percentile 2, 9) days. The median time from injury to MRI was 11 (25^th, 75^th percentile 6, 18) days. Of 362 patients 351 (97%) sustained their injury during sports or recreational activities, of which football was the most common (n = 167).

Procedures

The MCL was graded clinically as 0 (no injury), I, II or III (Fetto & Marshall, 1978). MRI was performed with one of the two scanners: a 1.5 T imager (Gyro-scan, Intera, Philips, Eindhoven, the Netherlands) and a 1.0 T imager (Impact, Siemens, Erlangen, Germany), both using a circular polarized surface coil. Patients were examined with a T2-weighted turbo spin-echo sequence (tSEPdT2) and a T2-weighted turbo short tau inversion recovery sequence (tSTIRT2) in the coronal and in the sagittal views (Olsson et al., 2016). MRIs were evaluated by several experienced musculoskeletal radiologists in a standardized fashion according to Khanna et al. (2001). MCL injuries were graded as none, grade I (oedema without signs of discontinuity), II (discontinuity visualized but without displacement) and III (discontinuity with displacement).

Statistical analysis

Sensitivity, specificity, and positive and negative predictive values were calculated for MRI as a test for grade II or III MCL injury. As there is little difference in the management of the MCL without injury and with grade I injury, these grades were combined for a separate analysis resulting in three categories for both clinical and MRI gradings (Table 1) and agreement calculated using a weighted kappa with linear weights. Statistical analyses were carried out in SPSS (version 27) and STATA (version 16).

Table 1. The distribution of clinical and MRI grade of injury.

		MRI grade (n)			Total
		0 or I	II	III
Clinical grade (n)	0 or I	241	19	8	268
	II	30	23	14	67
	III	0	8	19	27
Total		271	50	41	362

Results

The sensitivity and specificity of MRI as a test for grade II or III MCL injury compared to clinical examination was 68% (95% CI 58–77%) and 90% (95% CI 86–93%), respectively. The positive predictive values (PPV) and negative predictive values (NPV) were 70% (95% CI 62–78%) and 89% (95% CI 86–92%), respectively. The prevalence of clinical grade II or III MCL injury was 26% (94/362).

Weighted kappa analysis yielded an agreement of 0.56 (95% CI 0.48 to 0.65) between clinical and MRI grading corresponding to moderate agreement (Landis & Koch, 1977). Agreement for MCL grading was similar in patients with and without concomitant cruciate ligament rupture (weighted kappa values: 0.57 [95% CI 0.48–0.66] and 0.55 [95% CI 0.35–0.75], respectively) and for OSO and OSK (weighted kappa values 0.55 [95% CI 0.39–0.70] and 0.57 [95% CI 0.47–0.67], respectively).

Discussion

Although it appears that the clinical and MRI gradings are not entirely equivalent, our findings as interpreted according to Landis and Koch (Landis & Koch, 1977) indicate that the agreement is at least moderate. While the sensitivity and specificity were relatively high, there was a degree of uncertainty indicated by the confidence intervals. The prevalence of 26% for grade II or III injury ought to be taken into account when interpreting the PPV and NPV.

While this is a large cohort study of consecutive patients, only around a third were assessed by an orthopaedic specialist. Many of the participants in the study presented initially to the emergency department, which is generally not staffed by orthopaedic specialists and subspecialists. Of those assessed, the majority had either no injury or a grade I injury to the MCL, leading to a relatively low prevalence of grade II and III injuries. Although the cohort was examined in the early to mid-2000s, the clinical assessment and management of MCL injuries has not changed markedly in the last two decades (Smyth & Koh, 2015). Despite the fact that MRI technology has improved, the advantages of high-field MRI appear to be most apparent in the assessment of cartilage lesions rather than ligament injury (Cheng & Zhao, 2018; Nouri et al., 2017; Van Dyck et al., 2013). Both clinical and MRI evaluations were performed by multiple orthopaedic surgeons and radiologists. Despite the possible consequences for scientific validity, this is reflective of routine clinical practice in a general hospital setting. As such, the findings ought to be generalizable to similar clinical settings.

Conclusion

Agreement between clinical and MRI grading of MCL injuries by orthopaedic specialists and subspecialists in a general hospital setting is at least moderate regardless of the presence of concomitant cruciate ligament injury.

Declaration of interest: none.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Stig and Ragna Gorthons Foundation for medical research and the Swedish research council.