To evaluate the influence of age, sex, body mass index (BMI), extent of meniscal resection, cartilage status, and knee load on the development of radiographically evident osteoarthritis (OA) of the knee and knee symptoms after meniscal resection.
We evaluated 317 patients with no cruciate ligament injury (mean ± SD age 54 ± 11 years) who had undergone meniscal resection 15–22 years earlier (followup rate 70%), with radiographic and clinical examination. The Knee injury and Osteoarthritis Outcome Score was used to quantify knee-related symptoms. Sixty-eight unoperated subjects identified from national population records were included as a reference group.
Symptomatic radiographic OA (corresponding to Kellgren/Lawrence grade ≥2) was present in 83 of 305 operated knees (27%) and 7 of 68 control knees (10%) (relative risk 2.6, 95% confidence interval [95% CI] 1.3–6.1). Patients who had undergone total meniscectomy and subjects with obesity (BMI ≥30) had a greater likelihood of tibiofemoral radiographic OA than those who had undergone partial meniscal resection and those with a BMI <25, respectively. Furthermore, degenerative meniscal tear, intraoperative cartilage changes, and lateral meniscectomy were associated with radiographic OA more frequently than were longitudinal tear, absence of cartilage changes, and medial meniscectomy, respectively. Symptomatic tibiofemoral or patellofemoral radiographic OA was associated with obesity, female sex, and degenerative meniscal tear.
Contributing risk factors for OA development after meniscal resection are similar to risk factors for common knee OA. Systemic factors and local biomechanical factors interact. Obesity, female sex, and preexisting early-stage OA are features associated with poor self-reported and radiographic outcome. Partial meniscal resection is associated with less radiographic OA over time than is total meniscectomy.
Meniscectomy is a risk factor for osteoarthritis (OA) of the knee (1–6). We have shown a 6-fold increased relative risk for development of radiographic OA after total meniscectomy, compared with unoperated controls (7, 8). Variations in disease prevalence between studies on this topic are likely due to heterogeneity of patient groups, high dropout rates, and varying classifications of radiographic OA (9). The contributions of and interaction between different risk factors for OA of the knee after meniscectomy are similarly difficult to evaluate and sometimes contradictory, probably owing to the same reasons, as well as to the absence of multivariate analysis (2, 3, 5, 6, 10). In addition, most outcome studies of meniscectomy are compromised because patient-relevant outcome instruments are not used, thus increasing the risk for introduction of bias by the interviewer (11, 12).
We have previously demonstrated an interaction between the endogenous risk factor heritable OA and exogenous OA risk in the form of meniscectomy (13), as well as the impact of the type of meniscal tear that may indicate preexisting early-stage knee OA (8). The present study was undertaken to evaluate the influence of additional risk factors for OA in the meniscectomized knee. We specifically evaluated the effect of age, sex, body mass index (BMI), arthroscopic or open surgical technique, extent of meniscal resection, intraoperative cartilage changes, and knee load during work and leisure on the development of radiographic OA and symptoms in the operated knee. We studied a well-defined and carefully characterized cohort, using a combination of validated self-administered questionnaires and standardized radiographic procedures.
PATIENTS AND METHODS
The Ethics Committee of the Lund University Faculty of Medicine approved the study, and informed consent was obtained from all participating subjects. Patients who underwent isolated meniscectomy at Lund University Hospital in 1973, 1978, or 1983–1985 were retrospectively identified through the surgical coding system or by manual search of surgical records. Surgical findings and other knee-relevant information were extracted from the surgical and medical records by 2 observers. If there was uncertainty about a finding listed in the record, a third observer was involved and consensus was reached. Meniscal tears and joint cartilage changes recorded at the time of surgery were classified as previously outlined (8). When there was no report of cartilage status, the cartilage was regarded as normal. 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 resection of more than one-third of the meniscal surface, but still leaving intact meniscal tissue (commonly, a peripheral rim) as subtotal. A meniscectomy carried out along the transitional zone between the meniscus and the joint capsule was recorded as total.
General criteria for exclusion from the study were as follows: death or relocation outside the South Sweden Health Care Region since the time of surgery, age <10 years at the time of surgery, or a diagnosis of rheumatoid arthritis, crystal-associated arthritis, or psoriatic arthritis. Knee-specific criteria for exclusion (either knee) were previous knee surgery, meniscectomy in both compartments, osteochondritis dissecans, fracture in or adjacent to the knee, septic arthritis, osteonecrosis, major ligament injury before or at the time of assessment, or radiographic changes indicating knee OA at the time of index surgery. In the cohort of patients who had surgery between 1983 and 1985, 6 patients with hip arthroplasty and 4 with severe neurologic or muscular disease were excluded because their comorbid conditions affected the patient-relevant outcome (14). A total of 456 patients fulfilled the criteria and were invited to undergo radiographic and clinical assessment in 1994, 1995, or 2000 (Figure 1). Current addresses were obtained from national population records. Of the 329 responders, knee radiographs were not obtained in 12 cases, leaving 317 subjects included in the study, which represents 70% of the available cohort. Demographic data and findings at the time of the index surgery are displayed in Table 1.
Table 1. Characteristics of the patients and control subjects
Patients (n = 317)
Controls (n = 68)
Retrospective estimates of knee load not available for patients who underwent surgery between 1983 and 1985.
Joint cartilage changes noted at index surgery, no. (%)
During the followup period, 33 patients (10%) had undergone further meniscal surgery in the index knee, and 53 (17%) had meniscectomy performed in the contralateral knee. Ten patients (3%) had a high tibial osteotomy for OA in the index knee. Two of these patients and 3 additional patients underwent high tibial osteotomy of the contralateral knee. Four patients had knee arthroplasty, in the contralateral knee in 1. All of these patients remained in the study. Patients who had the index meniscectomy performed arthroscopically but who were later operated on with an open technique were all considered to have had open knee surgery. Information on index and subsequent surgeries was obtained from medical records at Lund University Hospital and by patient self-report.
To evaluate possible inclusion bias, the nonparticipants in the largest of the 3 subcohorts (patients operated on between 1983 and 1985; n = 155) were evaluated with respect to age, sex, open or arthroscopic technique of surgery, operated compartment, extent of meniscal resection, and type of meniscal tear. The only factor that differed significantly was lower mean age at surgery among the nonparticipants (33 years versus 38 years; P < 0.001). Furthermore, in the same cohort, the patients who participated in a mailed survey in 1998 (14) but did not attend the followup visit in 2000 (n = 54) did not differ significantly with respect to their self-reported outcome, compared with the self-reports from the participants.
The control group comprised 68 individuals who have not undergone meniscectomy and who had no clinical meniscal or cruciate ligament injury. Controls were identified using national population records (7). Age at examination, sex ratio, and general geographic living area were similar between the control group and the patient group.
Patient-relevant outcome measures.
To evaluate knee-specific patient-relevant outcome, we used the Swedish version of the Knee injury and Osteoarthritis Outcome Score (KOOS) (www.koos.nu) (15, 16). Patients were asked to complete the KOOS questionnaire with reference to their operated knee, while controls were asked to consider their knees in general. The KOOS was developed for short- and long-term followup studies of knee injury and knee OA. It comprises 5 subscales: pain, other symptoms, activities of daily living, function in sports and recreation, and knee-related quality of life. A score of 0–100 is calculated for each subscale, with 100 representing the best result. Patients who underwent meniscectomy in 1978 completed a visual analog scale version of the KOOS questionnaire. The other subjects used a version based on a 5-point Likert scale. Patients examined in 1994 completed the KOOS questionnaire by mail in 1996; when the prevalence of symptomatic radiographic OA in these patients was evaluated, the knee radiographic status was still considered to be relevant to their self-report 2 years later. The other subjects completed their questionnaires on their own in conjunction with the clinical and radiographic assessment. Self-reports of outcome were not obtained from 12 patients.
We created a definition of a symptomatic knee based on the patient's self-report from the KOOS questionnaire and consensus among the authors. This operational definition, described in detail in a previous report (8), aimed at identifying individuals who had symptoms severe enough or of long enough duration that they might seek medical care.
Information on occupational physical workload and leisure physical activity level was collected from patients and control subjects. The estimates were made retrospectively by the subject and divided into 5-year periods from the time of surgery until time of assessment. Occupational load was graded as clerical work/unemployed/retired, light labor, moderate labor, or heavy labor. Leisure physical activity level was graded as low, moderate, high, or very high. Examples from each category were given in the questionnaire.
Radiographic imaging and OA scoring.
In patients and controls, standing anteroposterior images of both knees in 15 degrees of flexion were obtained using a fluoroscopically positioned x-ray beam. Axial views of the patellofemoral joint were obtained with a vertical beam with the subject standing with the knee in 50 degrees of flexion. A Siemens Basic Radiological System (Siemens, Erlangen, Germany) with a film-focus distance of 1.4m at 70 kV and 10 mA was used for patients who were followed up in 1994 and 1995, and for the control subjects. For patients who were assessed in 2000, we used a Phasix 60 generator (CGR, Liege, Belgium) at 70 kV, 16 mA, film-focus distance 1.5m. Two patients did not undergo radiographic examination of the contralateral knee.
The tibiofemoral and patellofemoral joints were assessed for joint space narrowing (JSN) and osteophytes according to the atlas published by the Osteoarthritis Research Society International (17). The presence of these features was graded on a 4-point scale (range 0–3, with 0 representing no evidence of bony changes or JSN). We did not score sclerosis, attrition, malalignment, or patellar subluxation. When the patient had undergone subsequent tibial osteotomy or arthroplasty for OA, JSN in the affected compartment was scored as grade 3. In these cases, the contralateral and patellofemoral compartments (if total arthroplasty) and osteophytes were assessed on preoperative images when available; otherwise, “missing data” was recorded for these compartments. One trained observer (ME) read all knee radiographs within a period of 2 weeks, with radiographs from patients and controls mixed and with blinding to clinical details. The tibiofemoral images from the patients who underwent surgery in 1973 and 1978 had been read previously (7, 18); interrater reliability (kappa statistic) for these readings and the present grading was 0.64 for JSN and 0.61 for the maximum osteophyte grade in the compartment, with complete agreement of 85% for JSN and 81% for osteophytes.
We considered radiographic OA of the knee to be present if any of the following criteria was achieved in any of the 2 tibiofemoral compartments or in the patellofemoral compartment: JSN grade ≥2, the sum of the 2 marginal osteophyte grades from the same compartment ≥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 scale (19). Patients who fulfilled the criteria both for having radiographic OA and for being symptomatic were classified as having symptomatic radiographic OA.
P values for binary data were calculated by Fisher's exact test. The effects of the evaluated risk factors, using the presence or absence of radiographic OA, a symptomatic knee, or symptomatic radiographic OA as the dependent variable, were analyzed by logistic regression. Odds ratio (OR) estimates with 95% confidence intervals (95% CIs) were based on the models with all variables entered. Data on current BMI was not obtained in 7 subjects. In order to avoid exclusion of these subjects from the analysis, the average BMI of study patients of their sex was recorded for them. However, they were excluded from the model when the ORs for BMI itself were estimated. All tests were 2-tailed (SPSS for Windows, release 11.5.0), and P values less than or equal to 0.05 were considered significant.
Tibiofemoral radiographic OA was present in 6 control knees (9%) and in 152 index (operated) knees (48%) (relative risk [RR] 5.4, 95% CI 2.5–13) and 84 contralateral knees (27%) (RR 3.0, 95% CI 1.4–7.5) (Figure 2). Patellofemoral radiographic OA was present in 62 index knees (20%) and 31 contralateral knees (10%).
The univariate and multivariate effects of each investigated risk factor on the development of tibiofemoral radiographic OA in the operated knee are presented in Table 2. In summary, BMI was a significant factor, with obesity (BMI ≥30) associated with increased odds of development of radiographic OA. Patients who had undergone total meniscectomy had a higher likelihood of developing radiographic OA than did those who had had partial meniscal resection. Lateral meniscectomy, intraoperative cartilage changes, and degenerative meniscal tear were associated with worse prognosis compared with medial meniscectomy, absence of cartilage changes, and longitudinal tear, respectively. Intraoperative cartilage changes in the index compartment were noted more frequently in association with degenerative meniscal tears compared with longitudinal tears (23% versus 4%; P < 0.001).
Table 2. Results of logistic regression analysis of risk factors for tibiofemoral radiographic osteoarthritis (OA) in the meniscectomized knee (317 patients, of whom 152 developed radiographic OA)
Prevalence of tibiofemoral radiographic OA, no. (%)
Logistic regression analysis demonstrated several risk factors that were significantly associated with a symptomatic knee (as determined by the KOOS). These factors were female sex (OR 2.4, 95% CI 1.2–4.6), overweight (BMI 25–29) (OR 2.6, 95% CI 1.5–4.5) or obesity (OR 3.0, 95% CI 1.3–7.0), and a degenerative type of meniscal tear (OR 1.9, 95% CI 1.1–3.5) as opposed to a traumatic tear.
Symptomatic radiographic OA.
We also evaluated the presence of tibiofemoral and/or patellofemoral radiographic OA in combination with knee symptoms (symptomatic radiographic OA) and found a substantial discrepancy. Nearly half of the patients with radiographic OA did not have symptomatic disease as assessed by the KOOS (Figure 3). Symptomatic radiographic OA in the index knee was present in 83 of 305 patients (27%) and 7 of 68 control subjects (10%) (RR 2.6, 95% CI 1.3–6.1). Characteristics of the subjects with healthy knees, those whose knees were symptomatic according to the KOOS, those with asymptomatic radiographic OA, and those with symptomatic radiographic OA are shown in Table 3.
Table 3. Main characteristics of the patients according to the outcome in their meniscectomized knee*
“Healthy” (n = 88)
Symptomatic, no radiographic OA (n = 64)
Asymptomatic radiographic OA (n = 70)
Symptomatic radiographic OA (n = 83)
See Patients and Methods and ref. 8 for definition of symptomatic knee; see Patients and Methods for definition of radiographic osteoarthritis (OA) (includes either the tibiofemoral or the patellofemoral compartment).
Women, no. (%)
Age at followup, mean ± SD years
51 ± 10
53 ± 11
54 ± 12
57 ± 12
Body mass index, mean ± SD kg/m2
25 ± 3
26 ± 3
26 ± 4
27 ± 4
Open surgery, no. (%)
Total meniscectomy, no. (%)
Lateral meniscectomy, no. (%)
Tibiofemoral cartilage changes, no. (%)
Degenerative tear, no. (%)
The relationship of each risk factor to the development of symptomatic radiographic OA in the operated knee was also investigated (Table 4). When the major findings in terms of risk factors for symptomatic radiographic OA were compared with risk factors that had been identified for tibiofemoral radiographic OA only (Table 2), we found that high BMI and degenerative meniscal tear were factors that remained statistically significant, while a trend toward significance remained for the operated compartment and the presence of intraoperative cartilage changes. In contrast to findings in the analysis of radiographic OA only, female sex was found to be associated with symptomatic radiographic OA.
Table 4. Results of logistic regression analysis of risk factors for symptomatic radiographic osteoarthritis (OA) in the meniscectomized knee (305 patients, of whom 83 developed symptomatic radiographic OA)*
Prevalence of symptomatic radiographic OA, no. (%)
Women had slightly lower BMI at followup compared with men (BMI 24.6 versus 26.5; P = 0.001). Moreover, lateral meniscectomy was more common in women (38% versus 16%; P < 0.001), while total meniscectomy was less frequent (45% versus 61%; P = 0.04). Other demographic and surgical characteristics did not differ by sex.
Information regarding retrospectively estimated occupational workload and spare-time physical activity level was not obtained in patients who underwent surgery between 1983 and 1985. However, in a separate analysis of the patients whose surgery took place in 1973 or 1978, the knee load was not found to have a significant effect on the outcome (data not shown), and was therefore omitted from further multivariate analyses using all patient subjects.
In the multivariate analyses (for radiographic OA, symptomatic knee, and symptomatic radiographic OA), there was association between type of resection and technique of surgery, and between cartilage changes and type of meniscal tear (i.e., total meniscectomy was more frequent in conjunction with open surgery than arthroscopy, and cartilage changes were more frequently noted if a degenerative tear was present). Therefore, we evaluated alternative models in which only one of these variables was entered at a time. In these models, the results remained essentially the same.
Knee injury is an important risk factor for knee OA (20–22), but little is known of the mechanisms leading to osteoarthritic disease in the injured joint, or how injury interacts with other risk factors. In this study we investigated a well-defined and carefully characterized cohort of 317 subjects who had undergone meniscal resection 15–22 years previously. Among all patients who had undergone this procedure, the loss to followup was low.
Knee OA after meniscectomy is traditionally considered a result of joint injury and increased cartilage contact stress due to the loss of meniscal tissue (23). “Wear and tear” is a straightforward explanation for OA, and surgery in which an intact peripheral rim of the meniscus is preserved would thus be expected to produce better long-term results. If there is a substantial intact portion of the circumferentially oriented matrix fibers, hoop tension, which counteracts meniscal extrusion when the knee is loaded, may still develop. Substantial function of the residual meniscus in shock absorption and load transmission would thus remain.
However, the evidence for improved long-term outcome after partial meniscal resection is limited (2, 8, 24–26). To our knowledge, the present study is the first, using a large number of subjects, to show that partial meniscal resection induces less radiographic OA over time than does total meniscectomy. The frequency of subjects experiencing symptoms was not substantially reduced, however. Although, the statistical power in our sample for detecting smaller proportional differences in numbers of symptomatic subjects was low, the raw KOOS data (not shown) supported our finding of a very limited effect of resection size on the patient-relevant outcome. Therefore, in view of the present and previous results, evidence for a greatly improved long-term patient-relevant outcome with the use of partial meniscectomy still appears weak.
Obesity is an established risk factor for knee OA (27, 28). We cannot rule out the possibility that some subjects in our study could first have developed OA and then become sedentary and overweight. However, using retrospective estimates of weight from the patients who underwent surgery in 1973 and 1978, we did not find any evidence for such a cause and effect. Therefore, our results strongly suggest that obesity is an important risk factor for the development of knee OA also in meniscectomized subjects. Consequently, it is important to encourage weight loss in obese patients who have undergone this surgical procedure and are at very high risk of developing OA (29).
Our finding of worse radiographic outcome in association with lateral meniscectomy is consistent with other reports (3, 5, 6, 25, 30). The lateral meniscus has been reported to carry a higher load in the knee compared with the medial meniscus. Consequently, its loss may result in increased cartilage contact stress (31, 32).
In community-based studies the prevalence of knee OA steadily increases with age (33, 34). Although there was an increase in prevalence with patient age in our study, this pattern seems less evident in knees that have undergone meniscectomy. The reason may be the introduction of an additional biomechanical risk factor (the loss of meniscal function). The limited time range since surgery in the present study limited the possibility to evaluate the influence of time.
Women appear more likely to develop symptomatic OA (27), suggesting that the complex relationship between biological, physiologic, and psychosocial factors to explain knee symptoms in OA needs further attention. In our study, just over half of the patients with radiographic OA had symptomatic disease, corroborating earlier observations (35).
Even if an arthroscopic surgical technique and a limited meniscal resection were used, patients still had a high risk of developing symptomatic or asymptomatic radiographic OA. Other risk factors thus need to be addressed. The contributing influence of heredity has been suggested by the association between radiographic OA of the hand and radiographic OA of the knee found after meniscectomy (13, 36). We have previously demonstrated the impact of preexisting early-stage OA (8); in further support of this is the higher frequency of intraoperative cartilage changes seen in conjunction with the degenerative type of meniscal tear in the present study, consistent with previous findings (37, 38). Patients with meniscal symptoms due to a degenerative tear may thus constitute a subpopulation enriched in individuals with incipient OA. However, the relationship between meniscal tears and pain is questionable (39), and in joints without specific mechanical symptoms, meniscal surgery may therefore often not be indicated. The intervention merely removes evidence of the disorder, while the OA joint degradation proceeds (40).
This was a retrospective cohort study with the limitations associated with such studies. Although the loss to followup of 30% was low compared with other studies of meniscectomy, selection bias may have occurred in both the patient group and the control group; the presence of knee symptoms among those invited could have generated a greater interest to participate. In contrast to findings in previous studies (41), our results did not reveal significant differences with respect to knee load. One explanation could be that it is difficult for patients to accurately estimate their load retrospectively. A further possible limitation of our study was that information on knee alignment, smoking history, and chondrocalcinosis was not recorded.
In conclusion, the contributing risk factors associated with the development of knee OA following meniscectomy are similar to those for knee OA in general. Systemic and local biomechanical risk factors interact. In addition to preexisting early-stage knee OA at the time of the index surgery, the demographic factors associated with the highest likelihood of symptomatic disease were obesity and female sex. When surgical repair of a symptomatic meniscal tear is not possible, a limited resection should be used, because total meniscectomy was associated with more radiographic changes of OA.