Introduction

Meniscal lesion is a common diagnosis among athletes as well as in the regular population [13, 18]. It can be treated by meniscal resection, but this involves an increased risk of osteoarthritis. Therefore, meniscal lesions are increasingly repaired to preserve joint integrity and limit the development of degenerative changes [2, 3, 8, 21, 26].

Meniscal healing after repair is reported in 80–85% of cases [1, 4, 12, 15, 27], which leaves up to one in five patients with an unhealed meniscus, potentially resulting in continuous symptoms and impaired function. When patients return for medical evaluation with a continuation or reoccurrence of symptoms after meniscal repair, there are different modalities for evaluating the meniscal status. Arthroscopy is the gold standard in meniscal healing assessment, but this method is invasive and includes risks of complications [7]. Thus, the goal is achieving good evaluation accuracy with non-invasive methods to identify unhealed menisci after repair.

Specific clinical findings that characterise meniscal lesion have been described in the literature, such as knee locking, joint-line tenderness, swelling and a positive McMurray’s test [5]. Magnetic resonance imaging (MRI) is the primary diagnostic imaging tool for the identification of meniscal lesions [11, 17]. It has been shown that MRI has a good ability to diagnose preoperative primary meniscal tears [17, 22, 23, 33, 34]. A study by Thomas et al. [28] showed that even with the diagnosis of primary meniscal lesion, MRI is not always indicated if the clinical examination strongly suggests pathology. The problem with using MRI for postoperative meniscal healing evaluation is that the rupture interface from the primary injury can be interpreted as a new lesion despite the adaptation from repair and healing at the tissue level [17, 19, 22, 25].

Miao et al. [17] found a low correlation between clinical findings and second-look arthroscopy for patients with meniscal repair, with a clinical healing rate of 70.8%. Steenbrugge [25] and Pujol [22] concluded that there was a high rate of false-positive MRI findings when diagnosing a lack of meniscal healing, reaching 38% in asymptomatic patients [25]. These studies dealt with an assessment of the sensitivity of MRI when diagnosing unhealed menisci in general and regardless of symptoms. However, in a clinical setting, it is important to know how to interpret the combination of imaging and clinical findings to identify meniscal repair failure. Presently, there is a lack of knowledge regarding how combined clinical findings and MRI can improve the diagnostic precision of meniscal repair failure assessment.

The present study investigates the extent to which MRI can identify unhealed meniscal lesions in symptomatic patients with previous meniscal repairs. In addition, the study investigates which clinical findings improve the diagnostic precision of unhealed menisci. The hypothesis is that MRI contributes to better sensitivity and positive predictive values (PPVs) in symptomatic patients with previous meniscal repairs compared to clinical assessment alone.

Therefore, the purpose of this study was to investigate the clinical relevance of whether an MRI is indicated based on the clinical findings of a symptomatic meniscus-repaired patient. The diagnostic value of MRI and clinical assessment based on re-arthroscopy findings has not previously been investigated in this particular patient group. Accordingly, this study endeavours to guide the clinician to decide whether an MRI is indicated in this group of patients.

Materials and methods

The patient cohort was identified retrospectively using the National Health Care Classification System’s surgery code for meniscal repair, whereby we identified all 425 patients who had undergone a meniscal repair at our institution during 2004–2012.

The indication for primary meniscal repair in all patients was a displaced longitudinal meniscus either isolated or in combination with an anterior cruciate ligament (ACL) lesion treated with ACL reconstruction.

Only patients with one or more symptoms of meniscal lesion after primary repair were included. The exclusion criteria included the lack of an MRI after primary meniscal repair, which excluded the majority of patients from our group because they did not return to the clinic after primary meniscal repair with symptoms of an unhealed meniscus. Furthermore, we excluded patients with more than one meniscal lesion in the same knee at the time of primary repair. The final patient cohort consisted of 80 patients.

Clinical signs

The patient records were studied to identify the presenting symptoms and clinical signs after primary arthroscopic meniscal repair. The specific symptoms and clinical signs of the unhealed meniscus were determined as follows: (1) locking, (2) joint-line tenderness, (3) swelling and (4) a positive McMurray’s test [5].

MRI evaluation

A blinded radiologist re-examined the MRIs. The most frequent sequences used were sagittal short TI inversion recovery (Sag STIR), sagittal proton density (Sag PD), sagittal transverse relaxation (Sag T2*), coronal longitudinal relaxation (Cor T1) and axial STIR (Ax STIR). The meniscal lesions were graded on a scale of 0–3 (0: low signal intensity; 1: irregularly marginated intrameniscal signal; 2: linear signal not extending to articular surface; 3: linear signal intensity extending to surface). Only a grade 3 change was considered an MRI-diagnosed meniscal lesion [9].

Re-arthroscopy

Re-arthroscopy recordings were used for collecting information on the meniscal status. Senior surgeons performed all re-arthroscopies, using a probe to evaluate the meniscus. The re-arthroscopy approach was used as the gold standard to which the clinical assessments and MRI findings were compared.

The project was approved by the Central Denmark Region Committees on Health Research Ethics (268/2017).

Statistical analysis

The MRIs were evaluated statistically by calculating the PPV, negative predictive value (NPV), sensitivity and specificity of the MRI in general and for each MRI sequence. The clinical symptoms were evaluated in the same manner. Furthermore, different combinations of symptoms and MRI were evaluated to improve the diagnostic values. Stata 15 software was used for calculating the statistical outcomes.

Results

Patient characteristics

In the period of interest, 425 menisci were repaired. Of these patients, 266 were excluded due to the lack of an MRI after primary meniscal repair, 42 due to no re-arthroscopy and 36 either because of inaccessible record files and/or MRIs, new injuries or multiple meniscal lesions in the same knee. One patient died shortly after primary repair and was thus excluded as well.

The final cohort consisted of 80 patients with a median age of 23.8 (range 14–52) years at the time of primary meniscal repair (see Table 1 for patient details).

Table 1 Patient and injury details
Full size table

In 45% of the cases, the primary meniscal lesions were associated with ACL ruptures, which were reconstructed with the meniscal repair or 6 weeks after the meniscal repair. No patient had additional PCL ruptures. No patient had osteoarthritic changes, as characterised by MRI grade 3–4 cartilage lesions in more than one joint chamber.

The median time from injury or debut of symptoms to primary meniscal repair was 8 (range 1–132) months. The median time from primary meniscal repair to MRI was 12 (range 2–61) months.

Re-arthroscopy

Re-arthroscopy identified 65 patients (81%) with a lack of meniscal healing and 15 (19%) with healed menisci.

MRI

Overall, MRI was able to find 55/65 (85%) of the meniscal lesions (sensitivity). The specificity was found to be 80% (12/15). MRI assessment of the meniscal lesions showed a PPV of 0.95 (55/58) and an NPV of 0.65 (12/22). The same statistical values were calculated for each of the five most frequently used MRI sequences. They all ranged from 0.86 to 1.0 in the PPVs and 0.15–0.3 in the NPVs (Table 2).

Table 2 Statistical values of magnetic resonance imaging (MRI) and different MRI sequences on meniscal healing after meniscal repair
Full size table

Clinical findings

The prognostic value of the clinical findings was as follows: the PPVs for swelling (30/34), joint-line tenderness (52/59) and a positive McMurray’s test (21/24) were 0.88, whereas for locking, the PPV was 0.83 (20/24). Joint-line tenderness had a sensitivity of 0.8 (52/65), whereas locking only had a sensitivity of 0.31 (20/65). The NPV was 0.38 for joint-line tenderness (8/21) but only 0.19 for locking (11/56) (Table 3).

Table 3 Statistic values of different clinical findings in patients with symptoms after meniscal repair
Full size table

The patients had varied clinical findings: 11 patients had none of the above-mentioned clinical findings but presented with other symptoms, 18 patients presented with one, 33 patients presented with two, 16 patients presented with three and 2 patients presented with four of the clinical findings (locking, joint-line tenderness, swelling and positive McMurray’s test).

An increasing number of positive clinical findings was found to improve diagnostic precision. One clinical finding predicted a meniscal lesion with 78% (PPV = 14/18) accuracy, whereas three clinical findings increased the PPV to 94% (PPV = 15/16) (Table 4).

Table 4 Statistical values of increasing clinical findings in patients with symptoms after meniscal repair
Full size table

When combining MRI findings with clinical findings, the following diagnostic precisions were found: The highest PPV was found for a grade 3 MRI in addition to a positive McMurray’s test, with a PPV of 1 (17/17). The combination of grade 3 MRI and joint-line tenderness had a PPV of 0.98 (44/45), while grade 3 MRI in addition to swelling had the highest sensitivity, at 0.87 (26/30) (Table 5).

Table 5 Statistical values of combined positive MRI finding and clinical findings in patients with symptoms after meniscal repair
Full size table

Discussion

The most important finding of the present study regarding symptomatic meniscal-repaired patients was that the MRI of knees with previously repaired menisci had a PPV of 0.95 for the detection of non-healed menisci. This means that a grade 3 meniscal change seen on MRI predicts an arthroscopically verified meniscal lesion in 95% of cases. This is a high precision for MRI in diagnosing non-healed meniscal repair. It is important to keep in mind that the investigated cases consisted of patients who returned to the hospital with knee symptoms after meniscal repair. Therefore, the high PPV is only valid for symptomatic patients.

The low NPV of 0.55 is not very informative to the clinicians dealing with a symptomatic patient after meniscal repair. Therefore, it may be necessary to perform explorative re-arthroscopy, especially at multiple clinical signs of meniscal lesion.

The MRI sensitivity in this study was found to be 0.85. This signifies that MRI is capable of finding 85% of meniscal lesions in this specific group of patients. When looking at the different MRI sequences, the Sag STIR sequence had the highest sensitivity (0.68).

Sixty-five out of the cohort of 80 symptomatic patients in this study had a lack of meniscal healing verified by re-arthroscopy. This means that approximately 80% of the patients who presented postoperative symptoms of a non-healed meniscal repair actually did have a non-healed meniscal lesion.

Each clinical finding was associated with a high PPV (0.83–0.88). These calculations are independent of the number of clinical findings and do not represent isolated clinical findings, as most of the patients had more than one positive clinical finding.

The PPV increased with the number of clinical findings for the patient. If the patient had three clinical findings, the clinical assessment achieved a similar PPV (0.94) to that of MRI (0.95). However, if the patient only presented one (PPV = 0.78) or two (PPV = 0.85) clinical findings, an improvement of diagnostic accuracy could still be achieved with the addition of MRI. As seen in Table 3, the sensitivity decreases with an increasing amount of clinical findings. This simply reflects that the number of patients decreases with the increase of clinical findings.

The combination of MRI and different clinical findings was also investigated. In particular, the combination of a grade 3 MRI and a positive McMurray’s test or joint-line tenderness resulted in a PPV of 1. However, this study population consisted of patients with symptoms and, therefore, the MRI PPV (0.95) already included at least one symptom or clinical finding.

The study reflects the clinical reality, and the results can be considered when a clinician must decide how to manage a patient with a previous meniscal repair who presents with on-going symptoms. An MRI does not improve diagnostic accuracy if three or more clinical findings are present. If only one or two clinical findings are present, an MRI contributes to further diagnostic information and an increase in the PPV.

Miao et al. [17] studied 81 patients with a total of 89 meniscal lesions. All the patients underwent primary repair followed by clinical assessment, MRI and second-look arthroscopy, regardless of symptoms. They found a total healing rate of 86.5% after arthroscopy. The healing rate based on clinical assessment was 70.8%. The MRI sequences presented PPVs ranging from 0.18 (Sag PD) to 0.83 (Cor T2) and sensitivities ranging from 0.42 (Cor T2) to 0.92 (Sag T1). These results differed from the results of the present study, which could be explained by the different study populations. Miao et al. [17] included patients regardless of their symptoms, whereas only symptomatic patients were included in the present study.

Steenbrugge et al. [25] reported clinical outcomes compared with MRI outcomes in 13 patients after meniscal repair in a 13-year follow-up study. The patients were asymptomatic and had fair-to-excellent clinical outcomes. MRI demonstrated hyperintense abnormal signals in 5 of 13 patients. They did not use second-look arthroscopy for evaluation. The abnormal signals could represent new asymptomatic meniscal tears or oedematous scar tissue [25]. Assuming the latter, we can state that the number of false-positive results is decreased when dealing with a symptomatic group of patients.

A study by Mustonen et al. [20] examined the MRI signals of 44 patients who had undergone meniscal repair with bioabsorbable arrows. Postoperative MRI showed that 26% of the patients had a grade 3 signal. These patients were not evaluated by second-look arthroscopy. Thus, it is not known whether these signals were due to lesions or scar tissue. However, it is known that there is a general healing rate after repair of 80–85%. Therefore, it can be assumed that a part of these grade 3 MRI signals is due to scar tissue and not meniscal lesions. The above-mentioned studies support the theory that scar tissue from a meniscal repair can give abnormal signals on an MRI that can be interpreted as a new or unhealed lesion.

Walz [32] substantiated this further by describing the challenges of MRI interpretation for a repaired meniscus. Ideally, the bright intrameniscal signal on an MRI indicating a meniscal lesion should resolve after repair. However, in reality, a complete resolution is rare. Especially in the first 2 years after the repair, fibrovascular granulation or scar tissue can cause the persistence of the bright signal, and this can be mistaken as indicating a non-healed lesion.

In a study by Pujol et al. [22], MRIs were performed on 23 patients 10 years after meniscal repair, and the researchers concluded that the scar tissue persisted and could not be differentiated from new lesions. This may indicate that the high diagnostic value of the MRI in our study occurred because the patients were all symptomatic, whereas the patients in Mustonen et al.’s [20] and Steenbrugge et al.’s [25] studies included symptomatic and asymptomatic patients. Our study indicated that the risk of being incorrect on the MRI interpretation of a failed meniscal repair is markedly decreased in symptomatic patients.

The time between injury and primary repair varied among the patients, but a study by Espejo-Reina et al. [10] demonstrated that there is a good clinical healing rate of 83% for meniscal repairs 2–60 months after injury. van der Wal et al. [30] found no significant difference in clinical outcomes in meniscal repairs < 2 weeks or > 12 weeks after injury. Other studies [14, 31] have shown better clinical outcomes in early repairs within 3–6 months after injury.

The time from primary repair to MRI also differed. There was a correlation between the postoperative grade 3 MRI signal and the time between the repair and MRI, indicating that the grade 3 signal decreases with time as a sign of healing [20]. Hence, an even higher MRI PPV could be expected if the MRIs in the present study were performed later.

Two different devices have been used in the cohort, but studies have shown that the healing rate does not differ significantly using sutures compared to arrows [6, 24], and even the MRI signal pattern is quite similar after using the two techniques [20].

A number of patients underwent ACL reconstruction in the same procedure as the meniscal repair or 6 weeks after. Therefore, it is possible that the knee symptoms after surgery originated from the ACL reconstruction rather than the meniscal repair, which blurs the symptoms and the actual meniscal status. However, studies have shown that there is no increased MRI signal or decreased healing rate of repaired meniscal lesions in patients with operated ACLs compared to patients with intact ACLs [10, 16, 20, 21, 29].

This study has limitations. The data retrieval involved difficulties because of the retrospective design. This resulted in a number of patients with meniscal repairs being excluded, which may have caused a selection bias. In addition, no distinction was made between excluded asymptomatic patients and the few symptomatic patients who had no MRI performed after primary repair. Some of these patients had a re-arthroscopy directly due to strong clinical symptoms of meniscal lesion, and others had neither procedure performed. Both groups could have caused a selection bias. Another weakness is that MRI and re-arthroscopy were only performed on symptomatic patients and, therefore, no asymptomatic patients were included, which, combined with the small number of patients with healed menisci, made it difficult to determine the NPV and specificity.

This study reflects the clinical reality, where patients often present with associated diagnoses, such as ACL rupture, and this could explain some of the false-positive cases in which the clinical assessment and/or MRI indicated a lack of meniscal healing but re-arthroscopy invalidated this diagnosis. Despite the possible effect of ACL rupture and reconstruction, a high PPV of MRI in diagnosing non-healed menisci in symptomatic patients was found.

To provide results that can be used in a clinical situation, it is important to investigate a cohort characteristic of the typical scenarios of meniscal repair patients (isolated and with ACL reconstruction).

Conclusion

In this study, MRI of previously repaired menisci in patients with symptoms of non-healed meniscus had a PPV of 0.95 for detecting a lack of meniscal healing. When examining a cohort of symptomatic meniscal repair patients, the PPV of the clinical assessment increased, with more clinical findings evident. A supplementary MRI would increase diagnostic accuracy when fewer than three of the clinical findings specified in the study are present. If the patient exhibits three or four clinical findings, there is no further increase in diagnostic accuracy to be achieved, and thus there is no need for an MRI.