Impact of type of meniscal tear on radiographic and symptomatic knee osteoarthritis: A sixteen-year followup of meniscectomy with matched controls




To investigate long-term radiographic and patient-relevant outcome of isolated limited meniscectomy with regard to type of meniscal tear and extent of surgical resection.


We studied 155 patients with intact cruciate ligaments (mean ± SD age 54 ± 12 years) who had undergone meniscectomy an average of 16 ± 1 years earlier. The patients were examined using standardized radiography and validated self-administered questionnaires. The Knee Injury and Osteoarthritis Outcome Score (KOOS) was used to quantify knee-related symptoms, and the definition of a symptomatic knee was determined. We used 68 control subjects matched for age, sex, and body mass index to calculate the relative risks (RRs).


Radiographic tibiofemoral osteoarthritis (OA) (Kellgren/Lawrence grade ≥2) was present in 66 index knees (43%), of which 39 (59%) were considered to be symptomatic according to the KOOS. In total, 77 patients (50%) had a symptomatic index knee. In a multivariate model, degenerative meniscal tears were associated with both radiographic OA (P = 0.030) and combined radiographic and symptomatic OA (P ≤ 0.015). The RRs for combined radiographic and symptomatic OA after degenerative and traumatic types of meniscal tear were 7.0 (95% confidence interval [95% CI] 2.1–23.5) and 2.7 (95% CI 0.9–7.7), respectively, compared with matched controls.


An isolated meniscal tear treated by limited meniscectomy is associated with a high risk of radiographic and symptomatic tibiofemoral OA at 16-year followup. Factors associated with worse outcome were degenerative meniscal lesions and extensive resections. We suggest that degenerative meniscal tears may be associated with incipient OA, and that the meniscal tear signals the first symptom of the disease.

Meniscal tear (and the resulting meniscal surgery) is a well-recognized risk factor for osteoarthritis (OA) of the knee. Most studies are based on followup of patients treated with total meniscectomy (1–5). The fibrocartilaginous menisci of the knee improve joint stability, load distribution, shock absorption, and cartilage lubrication (6). Over the past several decades, surgical methods have shifted to use of arthroscopic technique and minimal resections of damaged meniscal tissue, with the intent to reduce the risk of subsequent OA by preserving as much of the meniscal function as possible. Short-term benefits such as shortened hospital stay and recovery time have been proven for limited meniscus resection, as opposed to more extensive surgery, but so far improved long-term outcome has not been convincingly documented (7–13). The radiologic and clinical outcomes reported in previous studies vary considerably, probably due to heterogeneous groups of patients with respect to extent of injury, ligament status, range of followup times, and to high dropout rates, small samples, or lack of appropriate controls (13). The contralateral knee has been proven inappropriate as a control, because this knee also frequently shows pathologic changes (3, 14).

Knee trauma in itself has been reported to be a significant risk factor for subsequent knee OA (13, 15–17). For meniscal tears, however, a history of knee trauma frequently is not present. Smillie observed that longitudinal tears usually occur in association with a definite knee trauma, while horizontal tears occur in middle-aged or older patients, apparently resulting from degenerative changes in the meniscal tissue (18). Results of necropsy studies and a large epidemiologic study have supported these suggestions (19–21). Additionally, meniscal degenerative pathology has been reported to correlate with degenerative cartilage changes (22). In previous studies, no significant difference in outcome with regard to OA attributable to different meniscal tears has been reported. However, varying definitions of such tears, small patient numbers, use of total meniscectomy, or associated ligament injuries may have confounded interpretation of the results (3, 11, 13, 23–25).

The objectives of this controlled retrospective cohort study were to investigate the long-term outcome of isolated meniscal tears in stable knees treated with limited meniscectomy; to evaluate radiographic outcome alone and in combination with knee symptomatology as measured by the Knee Injury and Osteoarthritis Outcome Score (KOOS), and to analyze the influence of the type of meniscal tear and size of resection on these outcomes.



The ethics committee of the medical faculty of Lund University approved the study, and written informed consent was obtained from all subjects. The patients were identified by a manual search of the surgical records at the Department of Orthopedics, Lund University Hospital. All patients who underwent meniscectomy between 1983 and 1985 were identified. The medical records of the patients until the time of followup were reviewed. Based on the exclusion criteria (as detailed in Figure 1), 254 individuals, whose current addresses were obtained through the National Population Records, were invited to enroll in the study. These patients received a written invitation to participate in the followup, which included radiographic and clinical knee examination and a self-administered questionnaire. The final study cohort included 155 patients, representing 61% of the available cohort (Figure 1). In 1998, the majority of these patients also participated in a 14-year followup study, which included mailed patient-relevant questionnaires (26).

Figure 1.

Flow chart detailing inclusion and exclusion criteria. OA = osteoarthritis.

Twenty orthopedic surgeons performed the operations. The surgical records used for retrospective collection of data were not standardized, but the surgeon's findings were usually satisfactorily described. Two investigators (ME and one experienced orthopedic surgeon) performed the data extraction. In cases in which interpretation of the records was uncertain, efforts were made to reach a consensus. When uncertainty remained, the data were recorded as nonclassified. In cases in which the patient first had a diagnostic arthroscopy, and the following meniscectomy was performed by arthrotomy, the operation was classified as open.

We recorded the type of meniscal tear, its localization within the meniscus, the compartment affected, and the cartilage and ligament status. Classification of the tears was modified according to the criteria described by Newman et al (27) (Figure 2). We classified longitudinal tears as traumatic, and flap tears, horizontal tears, and tears in a meniscus with degenerative changes as degenerative. In epidemiologic (21) and necropsy studies (19, 20), complex, horizontal, and flap tears have been described as degenerative. There is no consensus on the origin of radial tears, and we therefore chose not to include them when comparing traumatic and degenerative tears. Joint cartilage changes recorded at the time of surgery were graded as no change, lesion without visible bone, or exposed bone. If any part of the meniscus was removed, leaving a minimum of two-thirds of the meniscal surface intact, we classified the resection as partial. We considered a meniscectomy of more than one-third of the meniscal surface as subtotal. Demographic data of the cohort at followup and at the time of the index surgery are shown in Table 1. The corresponding main parameters for nonresponders (n = 88) and patients without radiographic examination at followup (n = 10) were as follows: for female sex, 23%; for mean age at index surgery, 32.5 years (P < 0.001); for open surgery, 64%; for medial meniscectomy, 82%; for partial resection, 46%; and for degenerative tear, 46%.

Figure 2.

Classification of meniscal tears. Longitudinal tear = vertical tear in the meniscus with a longitudinal direction; usually located in the periphery of the meniscus. The longer the tear, the more unstable it is, and the end result is the dislocated central part of the meniscus (bucket-handle tear). Horizontal tear = horizontal cleavage in the meniscal tissue. Radial tear = vertical tear starting in the free (central) margin of the meniscal tissue. Flap tear = oblique vertical cleavage causing a flap tear (parrot beak); a flap tear can also be caused by a horizontal tear. Tear in degenerative meniscus = a tear or multiple tears in a degeneratively changed meniscal tissue (not shown).

Table 1. Characteristics of patients and control subjects
CharacteristicsPatient group (n = 155)Control group (n = 68)
  • *

    Eleven patients, who had undergone total meniscectomy, were counted as having undergone subtotal meniscectomy.

Demographic data at followup  
 Sex, no. (%) female27 (17)18 (26)
 Age, mean ± SD years54.3 ± 11.956.3 ± 11.5
 Followup time, mean ± SD years16.1 ± 0.9
 Body mass index, mean ± SD kg/m226.2 ± 3.525.8 ± 3.6
 Median occupational loadLight laborClerical work
 Median physical activity level, spare-timeModerateModerate
Demographic data at index surgery  
 Age, mean ± SD years38.2 ± 12.0
  No. (%) open/no. arthroscopic107 (69)/48
  No. (%) medial/no. lateral125 (81)/30
  No. (%) partial/no. subtotal*62 (40)/93
Type of meniscal tear  
 Traumatic, longitudinal, no. (%)63 (41)
  Flap, no. (%)50 (32)
  Horizontal, no. (%)8 (5)
  Tear in degenerative meniscus, no. (%)13 (8)
  Radial, no. (%)14 (9)
  Nonclassified, no. (%)7 (5)
Joint cartilage changes noted at index surgery  
 Index compartment, no. (%)23 (15)
 Contralateral compartment, no. (%)10 (6)
 Patellofemoral compartment, no. (%)18 (12)

During the followup period, 23 patients (15%) underwent reoperation of the meniscus in the index knee, and 23 patients (15%) underwent meniscectomy in the contralateral knee. Five patients who had a previous partial meniscectomy in the index knee now had a subtotal resection. All five of these patients underwent reoperation within 3 years of their original meniscectomy, and the resections were referred to as subtotal in the data analysis. During the followup period, 5 patients underwent high tibial osteotomy for OA in the index knee. One of these subjects and 1 additional patient also underwent high tibial osteotomy of the contralateral knee. Two other patients were treated with knee arthroplasty in the index knee, and 1 patient (with osteotomy in the index knee) received arthroplasty in the contralateral knee. Information on subsequent surgery was based on the medical records at Lund University Hospital and self-reported information from the patients.

Control subjects.

The control group comprised 68 individuals with no previous knee surgery and no meniscal or cruciate ligament injury, as described in a previous study (4). The size of the control group was based on the assumption that the relative risk (RR) of developing OA after meniscectomy was at least 4.5. Thus, a control group of 60 individuals was needed to provide statistical power of 83% at the 5% significance level. Control subjects were identified using National Population Records, matching for sex, birth year, and zip code. In addition, because age at the time of examination, the sex ratio, and the general geographic living area of controls were similar to those of patients in the present study, we regarded the controls as appropriate for study (Table 1).

Knee Injury and Osteoarthritis Outcome Score.

We used the KOOS, Swedish version LK 1.0 (28, 29). Patients completed the questionnaire on their own during the intervals between examinations. The KOOS is a 42-item, self-administered, knee-specific questionnaire based on the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) (30). WOMAC scores can be calculated from the KOOS data. KOOS comprises 5 subscales: pain, symptoms, activities of daily living (ADL), sports and recreation function (sport/rec), and knee-related quality of life (QOL). A score from 0 to 100 is calculated for each subscale, with 100 representing the best result. Patients were instructed to complete the KOOS form by considering their operated knee. Control subjects were asked to consider their knees in general. One patient did not complete the questionnaire.

Definition of a symptomatic knee.

Because no agreed-upon cutoff exists with regard to the definition of a symptomatic knee, we created a definition based on the patient's self-report from the KOOS questionnaire and consensus among the authors. This operational definition aimed at identifying individuals symptomatic enough to possibly seek medical care. The definition of a symptomatic knee required that the score for the KOOS subscale QOL and 2 of the 4 additional subscales should be equal to or less than the score obtained as follows: at least 50% of the questions within the subscale were answered with at least a 1-step decrease from the best response (indicating no pain/best possible function, etc.) on a 5-point Likert scale. After conversion to a 0–100 scale (0 = worst, 100 = best), the cutoffs were as follows: pain ≤ 86.1, symptoms ≤ 85.7, ADL ≤ 86.8, sport/rec ≤ 85.0, and QOL ≤ 87.5.

Radiographic imaging and OA scoring.

At Lund University Hospital, the same technician obtained standing anteroposterior (AP) radiographs of all patients and controls, with both knees in 15° of flexion and the same standardized knee positioning. For patients, radiographs were obtained using a CGR Phasix 60 generator at 70 kV, 16 mA, film-focus distance 1.5 meter (CGR, Liège, Belgium). For control subjects, radiographs were obtained using a Basic Radiographic System (Siemens, Erlangen, Germany) at 70 kV and 10 mA, film-focus distance 1.4 m. A fluoroscopically positioned x-ray beam was used in both situations. Compared with the control images, patient radiographs had a slightly increased magnification factor, which was accounted for when grading the films. All AP radiographs of the tibiofemoral joints obtained at followup were assessed for joint space narrowing (JSN) and osteophytes, according to the atlas from Osteoarthritis Research Society International (OARSI) (31). The presence of these features was graded on a 4-point scale (range 0–3, 0 = no evidence of bony changes or JSN).

One trained observer (ME) read all the radiographs within a period of 5 days, with films from patients and controls mixed together. The observer was blinded to surgical details but knew whether the image was obtained from a patient or a control (due to different size of the radiographic film used). To further ensure consistent assessment throughout the reading session, specific radiographs within the study material were selected as references. When a patient had undergone subsequent tibial osteotomy or arthroplasty for OA, JSN in the affected compartment was regarded as grade 3. When possible, JSN in the contralateral compartment and osteophytes were assessed on preoperative images, and otherwise were recorded as missing data. In a previous study (4), 2 orthopedic surgeons read the control images. The interrater reliability (kappa statistic) for their consensus reading and that of the present observer was 0.66 for JSN and 0.64 for osteophytes (maximum grade for each compartment), with complete agreement in 252 (93%) of 272 compartments for JSN and 250 (92%) of 272 for osteophytes. No discrepancy of more than 1 grade was observed.

A criterion for radiographic OA (ROA) in the knee was defined in the following manner: the 2 marginal osteophyte grades from the same compartment were added (sum osteophyte compartment score). We considered ROA to be present if any of the following criteria was achieved in any of the 2 tibiofemoral compartments: JSN of grade 2 or higher, sum osteophyte compartment score ≥2, or grade 1 JSN in combination with grade 1 osteophytes in the same compartment. This cutoff approximates grade 2 knee OA based on the Kellgren/Lawrence (K/L) scale (32).

We evaluated the sensitivity and specificity of the present definition of ROA, as well as several other definitions, to identify individuals who were symptomatic (as described above). Use of the definition for JSN alone yielded sensitivity of 24% and specificity of 92%; corresponding figures for use of our definition of osteophytes were 27% and 90%, respectively. Use of “any (one or both) of these 2 criteria” to define ROA yielded sensitivity of 34% and specificity of 86%. Our chosen definition, which also incorporated the third alternative (JSN grade 1 in combination with grade 1 osteophytes in the same compartment), yielded sensitivity of 47% and specificity of 77%, which resulted in the highest efficiency ([sensitivity + specificity]/2). Patients who fulfilled the criteria for both having ROA and being symptomatic were classified as having combined radiographic and symptomatic OA (RSOA).

Statistical analysis.

P values for 2 × 2 frequency tables were calculated with Fisher's exact test. The effects of the specified explanatory factors were analyzed by means of logistic regression. In the univariate analyses, the odds ratios with 95% confidence intervals (95% CIs) were estimated. Additionally, those factors that showed a tendency of affecting the outcome (by indicating P < 0.10 on the likelihood ratio test) were forwarded in a multivariate analysis. A final multivariate model included only significant factors. The outcome of the matched patient–control pairs was compared using McNemar's exact test. At followup, we accepted an age difference between the matched patient–control pair of up to 5 years and a body mass index (BMI) difference of up to 5 units (in 84% of the pairs, the difference was within 3 units). In the matched comparisons, we used the knee of the control subject that corresponded to the knee of the patient. When estimating the risk ratio of ROA for all patients compared with the control group (unmatched analyses), we used the mean ROA prevalence of right and left control knees. The RRs with 95% CIs, comparing patients with matched controls, was estimated by the Mantel-Haenszel test. The 4 matched patient–control analyses of the subgroups (degenerative and traumatic type of meniscal tear, and partial and subtotal resection) were performed separately. P values less than or equal to 0.05 were considered significant, and all tests were 2-tailed (SPSS for Windows, version 11.0.0. Chicago: SPSS; 2001).


OA prevalence in general.

Radiographic tibiofemoral OA as defined here was observed in 66 index (operated) knees (43%) and 45 contralateral knees (29%) of patients, and in 6 knees (the mean number of right/left knees) of control subjects (9%). Significantly more ROA was observed in the contralateral knees than in control knees, even when patients who had undergone subsequent meniscectomy in the contralateral knees were excluded from the analysis (P = 0.008) (Figure 3).

Figure 3.

Prevalence of radiographic changes in index (operated) knees (•; n = 155) and contralateral knees (○; n = 154) 16 years after meniscectomy, and in controls (▵; n = 68). P values were determined by Fisher's exact test and indicate the significance of the difference compared with control subjects. See Patients and Methods for the definition of radiographic osteoarthritis (OA). JSN = joint space narrowing.

At followup, 77 patients reported having enough knee disability for their index knee to be classified as symptomatic (according to our operational definition); 39 of these patients had ROA of the knee. Among the 13 control subjects who were symptomatic, 5 fulfilled the criteria for ROA (in either knee) (Figure 4).

Figure 4.

Prevalence of and relationship between radiographic osteoarthritis (OA) (operated knee) and symptomatic knees 16 years after isolated meniscectomy (n =154; the self-reported outcome for one subject is missing). Values within parentheses are the corresponding percentages of control subjects (n = 68) (any control knee). RSOA = radiographic and symptomatic OA.

Based on an unmatched analysis including all patients and control subjects, the RR for patients (compared with controls) to develop ROA in the index knee was 4.8 (95% CI 2.2–12.0). The corresponding risk regarding the contralateral knee was also increased, even when patients who had undergone meniscectomy in that knee were excluded (RR 2.8, 95% CI 1.2–7.2). The RR for a meniscectomized patient to be symptomatic compared with control subjects was 2.6 (95% CI 1.6–4.7), and the RR for the index knee to fulfill criteria for classic OA (i.e., by both symptomatology and radiography [RSOA]) was 3.4 (95% CI 1.4–9.7).

Explanatory factors for JSN ≥2, sum osteophyte compartment score ≥ 2, and for ROA and RSOA of the index knee were tested in a multivariate model. Compared with traumatic tears, degenerative meniscal tears were associated with a higher incidence of JSN ≥ grade 2 (P = 0.032), ROA (P = 0.030), and RSOA (P = 0.015). There was a trend toward greater osteophyte prevalence (P = 0.078) in women, who also had increased JSN (P = 0.033) and a greater prevalence of RSOA (P = 0.030) than did men. In addition, a trend for an increased prevalence of ROA and RSOA with higher age and increased BMI was suggested (Table 2). No significant differences in outcome between arthroscopic or open surgical techniques were detected (P ≥ 0.11).

Table 2. Explanatory factors predicting the incidence of ROA and RSOA for the index knee 16 years after meniscectomy*
FactorRadiographic OARadiographic and symptomatic OA
ncases/ntotalUnivariate analysesncases/ntotalUnivariate analysesMultivariate analysis
  • *

    Results based on logistic regression analysis. ROA = radiographic osteoarthritis; RSOA = radiographic and symptomatic OA; OR = odds ratio; 95% CI = 95% confidence interval; Ref. = reference category; BMI = body mass index.

  • One subject did not report patient-relevant outcome (n = 154).

  • No significant interaction between sex and type of meniscal tear.

  • §

    A multivariate model with sex, age at surgery, BMI, and type of meniscal tear yielded P = 0.41 for age at surgery and P = 0.071 for BMI.

  • Values for some patients were missing.

  • #

    A multivariate model including the factors BMI and type of meniscal tear yielded P = 0.24 for BMI and P = 0.037 for type of meniscal tear.

Sex   0.14   0.047  0.030
 Male51/1281.0Ref. 28/1271.0Ref. 1.0Ref. 
 Female15/271.90.8–4.4 11/272.41.0–5.8 2.91.1–7.8 
Age at surgery, years   0.20   0.090Not included in model§
 <3519/571.0Ref. 9/571.0Ref.    
 35–4425/541.70.8–3.7 15/542.10.8–5.2    
 ≥4522/442.00.9–4.5 15/432.91.1–7.4    
BMI, kg/m2   0.063#   0.039Not included in model§
 <2522/601.0Ref. 9/591.0Ref.    
 25–2930/691.30.7–2.7 21/692.41.0–5.8    
 ≥3013/193.71.2–11.2 8/194.01.3–12.8    
Type of meniscectomy   0.61   0.51   
 Medial52/1251.0Ref. 30/1241.0Ref.    
 Lateral14/301.20.6–2.7 9/301.30.6–3.2    
Type of resection   0.84   0.62   
 Partial27/621.0Ref. 17/621.0Ref.    
 Subtotal39/930.90.5–1.8 22/920.80.4–1.7    
Type of meniscal tear   0.030#   0.017  0.015
 Traumatic21/631.0Ref. 10/631.0Ref. 1.0Ref. 
 Degenerative38/712.21.1–4.4 24/702.81.2–6.4 2.91.2–6.8 

Type of meniscal tear and OA.

Thirty-seven patients with degenerative tears (52%), 21 with traumatic tears (33%), 5 with radial tears, and 3 with unclassified tears had ROA. In unmatched analyses comparing all patients with degenerative or traumatic types of meniscal tear with the control group, patients had an increased RR for ROA in the index knee irrespective of the type of meniscal tear. The risk of being symptomatic or having RSOA was increased after a degenerative tear (P < 0.001), but not after a traumatic tear (P ≥ 0.07). Patients with ROA following a degenerative meniscal tear had the worst patient-relevant outcome (Figure 5). To eliminate a confounding effect of age, sex, and BMI, we repeated the analyses with control subjects who were individually matched with patients for age, sex, and BMI. The results were robust in that degenerative tears remained more strongly associated with increased knee disability and RSOA than did traumatic meniscal tears (Table 3). At the time of index surgery, cartilage changes in the index compartment were more frequently noted if a degenerative tear rather than a traumatic tear was present (P = 0.010). All 8 patients (5 of whom were women) who underwent subsequent OA-related surgery in any knee had degenerative meniscal tears.

Figure 5.

Mean Knee Injury and Osteoarthritis Outcome Score according to type of meniscal tear for patients with radiographic osteoarthritis (ROA) in the operated knee and control subjects with no ROA (either knee). Seven patients had undergone subsequent OA-related surgery in the index knee, and 1 patient had undergone surgery in the contralateral knee; all of them had degenerative meniscal tears. No significant change was detected when these patients were excluded from analysis. Bars show the 95% confidence interval. ADL = activities of daily living; Sport/Rec = sport and recreation function; QOL = knee-related quality of life.

Table 3. Relative risk of radiographic and patient-relevant outcome 16 years after meniscectomy compared with nonoperated controls*
OutcomeType of meniscal tearType of meniscal resection
Degenerative (n = 59 pairs)Traumatic (n = 57 pairs)Partial (n = 55 pairs)Subtotal (n = 63 pairs)
RR95% CIRR95% CIRR95% CIRR95% CI
  • *

    Relative risks (RRs) determined with McNemar's exact test and 95% confidence intervals (95% CIs) were obtained by comparing patients with controls matched for age, sex, and body mass index. If not indicated otherwise, outcome refers to the operated knee. For tibiofemoral joint space narrowing (JSN) and sum osteophyte compartment score, the most affected compartment was used for analysis. RSOA = radiographic and symptomatic osteoarthritis.

  • Value for 1 patient was missing.

Radiographic OA5.32.4––––10.5
JSN ≥ grade 29.52.2––––34.0
Sum osteophyte compartment score ≥210.02.3––––33.2
Radiographic OA of contralateral knee4.21.6––––8.4

Extent of meniscal resection and OA.

In an unmatched analysis and in matched patient–control pairs, patients had a significantly increased risk of developing ROA as well as RSOA in the index knee, irrespective of the extent of meniscal resection (Table 3). No significant differences in outcome between partial and subtotal meniscectomy were suggested by logistic regression analysis (Table 2). However, the proportion of degenerative tears was substantially greater in patients who underwent partial resection (71%) than in those who had larger resections (29%), which may have acted as a confounding factor. When analyzing only patients with a degenerative tear, JSN of grade 2 or higher was more common after subtotal resection than after partial resection (P = 0.004). Furthermore, patients with a more extensive surgical resection of these tears scored significantly worse on the KOOS subscales for pain, sport/rec, and QOL (P ≤ 0.037). A corresponding analysis for patients with traumatic tears was not performed due to the small number treated with partial meniscectomy (n = 6). Among the 11 patients who underwent total meniscectomy (which in this study was considered to be a subtotal resection), the outcome was essentially the same as that after limited resections; 6 subjects had ROA and 5 had RSOA (5 of the 11 had degenerative meniscal tears). Of the 8 patients who had undergone subsequent OA-related surgery, 4 had partial meniscectomy, and the other 4 had subtotal meniscectomy (none were total meniscectomies).

Compartment and location of tear and OA.

No significant differences in radiographic outcome or the KOOS were detected, regardless of whether medial or lateral meniscectomy had been performed. The proportion of women was higher among patients who had undergone lateral meniscectomy than among those who had undergone medial meniscectomy (37% versus 13%; P = 0.005). The proportion of degenerative tears was greater in the medial meniscus compared with the lateral meniscus (50% versus 27%; P = 0.024). Both degenerative and traumatic tears commonly lead to meniscal resections involving some of the posterior third of the meniscus (85% and 84%, respectively).


In this 16-year followup study, we observed that a meniscal tear and ensuing meniscectomy were associated with an increased risk of radiographic tibiofemoral OA, irrespective of the type of tear and extent of surgical resection. However, compared with other types of tears, a degenerative type of meniscal tear was associated with a higher frequency of both ROA and knee symptoms, which is consistent with our previous report showing a worse patient-relevant outcome for this type of tear (26). Patients exhibited an increased frequency of ROA in the contralateral knee, compared with controls.

The classification of meniscal tears used here was based on morphologic appearance and is supported by previous studies (19–21). This definition does not necessarily correlate with the patient's onset of symptoms, because a degenerative tear may have an insidious onset or present as an acute tear due to minor trauma. Partial meniscectomy has traditionally been regarded as resection of the injured region of the meniscus only, preserving a stable remnant. A precise and commonly accepted definition of a subtotal resection has not been presented. Total meniscectomy, which rarely is used today, involves incision along the transitional zone and the joint capsule, preserving only a small rim of meniscal tissue (18).

The criteria for ROA remain unsettled, and several grading systems are in use; the most widely adopted system is that described by Kellgren and Lawrence (32). We used the OARSI atlas for reading the films, scoring JSN and marginal osteophytes separately (31). An OA definition incorporating these features has been proven to offer the most precise estimation of the association of risk factors with disease worsening and clinical OA (33, 34). This finding was verified in the present study. The presence of JSN on conventional radiographs with a semiflexed knee might reflect the loss and (or) subluxation of meniscal tissue, not only loss of joint cartilage (35). However, meniscal extrusion is associated with OA of the knee (22, 36). In the present study we regarded the presence of either JSN grade 1 or a single grade 1 osteophyte as too nonspecific to be classified as radiographic OA. However, recent findings suggest that K/L grade 1 may reflect emergent knee OA (37). We considered JSN grade 1 and a single grade 1 osteophyte within the same compartment as a sufficient criterion, consistent with K/L grade ∼2. In addition, the presence of a sum osteophyte compartment score ≥2 in the absence of JSN was defined as ROA, which is consistent with previous suggestions (34, 38). Thus, we have slightly adjusted the criteria for ROA compared with our own previous studies (4).

This was a retrospective cohort study, with the weaknesses associated with such studies. The nonstandardized surgical reports regarding cartilage status and type and location of meniscal tear were, on occasion, difficult to classify. However, cases in which consensus could not be reached were recorded as unclassified. The radiologist's written statement for the preoperative knee radiographs was used to exclude patients with preoperative signs of OA. Thus, the investigators did not read these images separately. In 13 patients (8%), preoperative knee radiographs were not available. Knee alignment was not recorded, and no adjustment was made for workload or leisure physical activity level. The proportion of patients participating in the radiographic examination at 16-year followup was 61% of those available. Some selection bias may have occurred among both patients and control subjects, in that the presence of knee symptoms among those invited to participate could generate a greater interest in participating. However, the prevalence of ROA in the control group was comparable with that observed in a general population in the same area (39).

The single most important cause of exclusion was cruciate ligament injury. By focusing on subjects with isolated meniscal injuries, it is likely that we increased the proportion of degenerative meniscal lesions in our study population, as compared with patients undergoing meniscectomy in general. For the KOOS questionnaire, control subjects were asked to consider their knees in general, while patients were instructed to consider only the knee on which index surgery was performed. This may have caused underestimation of the patients' risk of outcomes involving KOOS (of being symptomatic and of having RSOA) compared with controls. One single observer read all radiographs, including those of controls (which were mixed in), and interrater reliability compared with earlier observers was good. Due to the limited number of control subjects and patients, the same patient may be involved in 2 of the 4 matched patient–control analyses (once for type of tear and once for type of resection), while a single control subject may be represented in all 4 analyses.

Epidemiologic studies have demonstrated that knee trauma is a significant risk factor for subsequent knee OA (15–17). However, the reported risks vary (for male patients, the range is 3.5–8.7), probably depending on different definitions of knee trauma and the wide range of followup times and criteria for OA. In the present study, we used the exposure definition of meniscectomy as documented by surgical records, not a self-report of knee trauma. A previous 21-year followup study of patients who underwent isolated total meniscectomy showed a relative risk of radiographic OA of 4.0–6.4, which is consistent with the risk increase demonstrated in the present study (4).*

There are, to our knowledge, no published followup reports describing individuals with a combination of well-defined knee disability and structural change consistent with a definition of classic OA. Most likely, these are the patients who will seek health care and require pharmacologic, surgical, or other treatment.

In the present study, 59% of the meniscectomized patients defined as having ROA were also classified as being symptomatic according to our definition. Even though advanced radiographic features of OA were noted on the radiographs of several of the remaining 41%, the disease was silent and did not influence patient-relevant knee function as assessed by the KOOS. Furthermore, 49% of the patients who were symptomatic did not have ROA. This result is consistent with previous findings of a limited correlation between radiographic signs of OA and patient-relevant outcome (40–42). Several causes other than tibiofemoral OA could explain knee symptoms, but some of the patients may have symptoms of early-stage OA in the absence of definite features on conventional radiographs (37).

Most followup studies of meniscectomy have failed to show a difference in long-term radiographic or patient-relevant outcome according to the type of tear. However, small numbers of subjects, total meniscectomy, associated ligament injuries, or varying classifications of OA and of meniscal tears may have confounded interpretation of results (3, 11, 13, 23–25). We observed a higher incidence of ROA in patients with degenerative tears, and it is noteworthy that all 8 subjects who had undergone subsequent OA-related surgery (either high tibial osteotomy or arthroplasty) had degenerative meniscal tears. The worse outcome of degenerative tears is consistent with previous observations by our own and other groups (9, 14, 26, 43).

Some earlier studies have suggested a better long-term result after partial compared with more extensive meniscectomy, but the evidence for an improved radiographic outcome remains limited (8, 11–13, 24). No difference in long-term radiographic or symptomatic outcome was observed between patients who underwent open surgery and those who had arthroscopy. Thus, factors other than surgical technique or the extent of resection may be more important for the long-term outcome. The extent of resection is not an independent variable but is influenced by the extent and type of the meniscal tear.

Based on our results, we suggest 2 major pathways to knee OA in individuals with meniscus injuries. In one group, a tear in a healthy knee is initiated by a significant trauma, and OA development is catalyzed by the initial joint cartilage trauma and altered loading patterns due to the loss of the meniscus. A second group appears to experience more or less spontaneous meniscus tears in the presence of preexisting degenerative changes in both the meniscus and other knee joint structures. Such changes in the meniscus could be regarded as a signal of incipient OA. Compared with the latter group, the first-mentioned group reports fewer long-term patient-relevant knee problems. Even within the group of subjects with ROA, patients with degenerative meniscal tears and meniscectomy had the worst patient-relevant outcome (Figure 5), and their long-term results are comparable with preoperative self-reported outcomes in patients with isolated meniscal tears (44). Greater extent of soft tissue involvement or bone marrow lesions are factors that could explain the worse patient-relevant outcome for this patient category (45, 46). If the degenerative meniscus lesion is regarded as representing incipient OA, it is not surprising that surgical intervention directed to the meniscus has only limited influence on long-term symptomatic and radiographic outcome (47–49).

We observed a higher prevalence of ROA in the contralateral knees of patients, compared with control knees, even when patients undergoing meniscectomy in contralateral knees were excluded. There may be several reasons for this increase. First, the contralateral knee may have been subjected to trauma to a greater extent than were control knees. Second, the presence of injury in one knee may change joint loading and gait patterns, leading to an overload of the contralateral knee (50). Third, injury and synovitis in one knee may increase the risk of arthritic changes in the contralateral knee through neurogenic mechanisms (51). Fourth, an increased prevalence of ROA in the contralateral knee may reflect an endogenous (hereditary) increased risk of OA in some individuals (52). These different causes are not mutually exclusive, and their relative importance remains to be further explored.

In conclusion, an isolated meniscal tear followed by surgical meniscectomy imposes a high risk of both radiographic and classic (combined radiographic and symptomatic) knee OA, even if a limited meniscal resection is performed. We suggest that a degenerative meniscal tear may be the first signal announcing a more widespread osteoarthritic disease in the knee joint.

  • 1

    The relative risks of ROA for the matched patient-control pairs reported by Roos et al in 1998 (4) are incorrect, because the figures published actually represent odds ratios, not relative risks. However, this does not change the overall conclusion of that study. For the index knee, the relative risks, recalculated from original data in (4), should read 4.0 (95% CI 2.3–6.9) for “grade A” OA changes and 6.4 (95% CI 2.7–15.2) for the more advanced “grade B” changes.