Sagittal T1-weighted fat-saturated 3-dimensional magnetic resonance images showing changes in knee cartilage defects (arrows). A, Increase of tibial cartilage defect from grade 1 at baseline to grade 3 at follow-up. B, Decrease of tibial cartilage defect from grade 3 adjacent to bone surface at baseline to grade 1 at follow-up.
Sagittal T1-weighted fat-saturated 3-dimensional magnetic resonance images showing measurements of knee cartilage volume, determined by summing all the pertinent voxels within the resultant binary volume. A, Lateral tibial cartilage. B, Patellar cartilage. ROI indicates region of interest.
Axial T1-weighted fat-saturated 3-dimensional magnetic resonance images showing measurements of tibial plateau bone area. The mean area of medial (region of interest [ROI]-1) and lateral (ROI-2) tibial plateau bone is measured manually on the 3 reformatted images closest to tibial cartilage. A, The first image. B, The second image.
Changes in knee cartilage defect scores (A) and percentage changes in knee cartilage defects (B) during 2.3 years.
Associations between age, body mass index (BMI, calculated as weight in kilograms divided by the square of height in meters), and change in total knee cartilage defects. The increase in total knee cartilage defects was higher in subjects 40 years and older (A) and in subjects with BMI of 25 or higher (B).
Ding C, Cicuttini F, Scott F, Cooley H, Boon C, Jones G. Natural History of Knee Cartilage Defects and Factors Affecting Change. Arch Intern Med. 2006;166(6):651-658. doi:10.1001/archinte.166.6.651
Knee cartilage defects may play an important role in early osteoarthritis, but little is known about their natural history.
Knee cartilage defect score (range, 0-4), cartilage volume, and bone surface area were determined using T1-weighted fat-saturated magnetic resonance imaging in 325 subjects (mean age, 45 years) at baseline and 2 years later.
Thirty-three percent of the subjects had a worsening (≥1-point increase) and 37% of the subjects had an improvement (≥1-point decrease) in cartilage defect score in any knee compartment during 2.3 years. A worsening in cartilage defect score was significantly associated with female sex (odds ratio [OR], 3.09 and 3.64 in the medial and lateral tibiofemoral compartments) and baseline factors, including age (OR, 1.05 per year in the medial tibiofemoral compartment), body mass index (OR, 1.08 in the lateral tibiofemoral compartment), tibiofemoral osteophytes (OR, 6.22 and 6.04 per grade), tibial bone area (OR, 1.24 and 2.07 per square centimeter), and cartilage volume (OR, 2.91 and 1.71 per milliliter in the medial tibiofemoral and patellar compartments). An improvement in cartilage defect score had similar but reversed associations with these factors (except for sex), including a decrease in body mass index (OR, 1.23 in the medial tibiofemoral compartment).
Knee cartilage defects are variable, and changes are associated with female sex, age, and body mass index. Increases are associated with baseline cartilage volume, bone size, and osteophytes, suggesting a role for these in the pathogenesis of cartilage defects. Interventions such as weight loss may improve knee cartilage defects.
Knee cartilage defects are commonly found by magnetic resonance imaging (MR imaging) in healthy subjects1 and by arthroscopy in symptomatic subjects,2 in whom they are thought to be largely traumatic,3 but little is known about their natural history. They lead to osteoarthritis (OA) in a surgical model of articular condylar defects in rabbits,4 but there are limited human data to support this.
Investigators in preliminary studies have reported that knee cartilage defects are associated with the Kellgren-Lawrence score in patients with advanced OA5 and with osteophytes in subjects with chronic knee pain.6 It is uncertain whether these are precursors of OA or a result of OA. It was reported in 2005 that the prevalence and severity of knee cartilage defects increase with increasing age7 and body mass index (BMI, calculated as weight in kilograms divided by the square of height in meters)8 in healthy younger subjects. Furthermore, their severity and prevalence were associated with tibiofemoral osteophytes, increased tibial bone area, decreased knee cartilage volume, and increased type II collagen breakdown.1 Knee cartilage defects were predictive of knee cartilage loss at the medial tibia9 during 2 years, suggesting an important role for knee cartilage defects in early knee OA.
So far, there are few longitudinal data to describe the changes in knee cartilage defects. Results of a retrospective cohort MR imaging–based study10 in 43 patients during 1.8 years suggest that 38% of grade 1 knee cartilage lesions progressed to higher-grade lesions, while 23% of grade 1 knee cartilage lesions reverted to grade 0. The presence of meniscal and anterior cruciate ligament tears was associated with more rapid progression of cartilage lesions.
However, there are no reports about the effects of sex, age, BMI, cartilage volume, and subchondral bone on knee cartilage defect changes, to our knowledge. The objective of this longitudinal study was to describe the natural history of knee cartilage defects and the factors associated with change in a convenience sample of adults.
The study was carried out in Southern Tasmania from June 26, 2000, until December 1, 2001, and the follow-up study was conducted approximately 2 years later. Subjects were selected from 2 sources. Half the subjects were the adult children of subjects (offspring) who had a knee replacement performed for primary knee OA in 1996 through 2000. This diagnosis was confirmed by reference to the medical records of the orthopedic surgeon and to the original radiograph, where possible. The other half were randomly selected control subjects without this history, chosen by computer-generated random numbers from the electoral roll. Subjects from either group were excluded on the basis of contraindication to MR imaging (including metal sutures, the presence of shrapnel, iron filing in the eye, and claustrophobia). The study was approved by the Southern Tasmanian Health and Medical Human Research Ethics Committee, Hobart, and all subjects provided informed written consent.
Height and weight were measured as previously described.1,7,8 Body mass index was calculated. Knee pain at baseline was assessed by questionnaire and was defined as pain lasting longer than 24 hours in the past 12 months or daily pain lasting longer than 30 days in the past year.1
Magnetic resonance imaging of the right knee was performed on 2 occasions, at baseline and at follow-up. Knees were imaged in the sagittal plane using a 1.5-T whole-body MR unit (Picker International, Cleveland, Ohio) with use of a commercial transmit/receive extremity coil. The image data were transferred to a workstation using the software program Osiris (University of Geneva, Geneva, Switzerland). The cartilage defects (score range, 0-4) were graded by 1 of us (C.D.) at baseline and at follow-up (Figure 1) at medial tibiofemoral, lateral tibiofemoral, and patellar sites as previously described1,7- 9 as follows: grade 0, normal cartilage; grade 1, focal blistering and intracartilaginous low-signal intensity area with an intact surface and bottom; grade 2, irregularities on the surface or bottom and less than 50% loss of thickness; grade 3, deep ulceration with 50% loss of thickness or greater; and grade 4, full-thickness chondral wear with exposure of subchondral bone. The reader was unaware of the initial result at the time of the second reading. A cartilage defect also had to be present in at least 2 consecutive sections. If multiple defects existed at 1 site (uncommon in this sample), we selected the highest grade to represent the cartilage defect score at that site. This type of defect evaluation is correlated with histological11 and arthroscopic12 findings. The cartilage defects were regraded 1 month later, and the mean cartilage defect scores at the medial tibiofemoral (range, 0-8), lateral tibiofemoral (range, 0-8), patellar (range, 0-4), and whole (range, 0-20) compartments were used in the study. Intraobserver reliability (expressed as intraclass correlation coefficient) was 0.89 to 0.94, and interobserver reliability (assessed in 50 images by 2 of us [C.D. and H.C.]) was 0.85 to 0.93.1,7- 9
Changes in cartilage defects were calculated by subtracting the cartilage defect scores at baseline from the cartilage defect scores at follow-up. An increase in cartilage defect score of 1 or more was defined as an increase in cartilage defects, and a decrease in cartilage defect score of 1 or more was defined as a decrease in cartilage defects. These changes were in excess of the less significant change in an individual as calculated by a standard formula.13 All increases and decreases in knee cartilage defect scores were confirmed by paired reading in 150 subjects by 1 of us (C.D.) who was unaware of the subject's score at the time of reading.
Radiography (a standing anteroposterior semiflexed view of the right knee) was performed in all subjects. This method has been previously described.14
Knee cartilage volume was determined (Figure 2) by means of image processing on an independent workstation by 1 of us (C.D.) as previously described.1,7- 9,13- 16 The volumes of individual cartilage plates (medial tibial, lateral tibial, and patella) were isolated from the total volume by manually drawing disarticulation contours around the cartilage boundaries on a section-by-section basis. These data were then resampled by means of bilinear and cubic interpolation (continuous sections with 312 × 312-μm area and 1.5-mm thickness) for the final 3-dimensional rendering. The coefficients of variation for cartilage volume measures were 2.1% to 2.6%.16
Knee tibial plateau bone area (Figure 3) and patellar bone volume were determined as previously described.1,7- 9,14- 16 The area of medial and lateral tibial plateau bone is measured manually on the 3 reformatted images closest to tibial cartilage. The mean of these 3 areas is used as an estimate of the tibial plateau bone area. The coefficients of variation for these measures were 2.2% to 2.6%.16
Paired t tests were used to compare means. Logistic regression analysis was used to examine the associations between individual knee cartilage defect increase (increase vs stable and decrease) or decrease (decrease vs stable and increase) and other variables. Higher baseline cartilage defect scores were considered less likely to progress than lower baseline cartilage defect scores; therefore, we adjusted all associations for baseline cartilage defect scores in logistic regression analyses. Although associations did not differ in offspring and controls, we adjusted all associations for offspring and control status because of the convenience nature of the sample. P<.05 (2-tailed) was regarded as statistically significant. All statistical analyses were performed using SPSS version 10.0 for Windows (SPSS Inc, Chicago, Ill).
Three hundred twenty-five subjects (190 women and 135 men, 87% of those originally studied) completed the study. The reasons for loss to follow-up were as follows: 2 died, 5 moved out of state, 3 were claustrophobic, 4 dropped out because of illness, and others gave no reason for their discontinuation. This was a young sample, with a mean age of 45 years (age range, 26-61 years) at baseline. Although radiographic OA was uncommon (17%) and was predominantly grade 1, knee cartilage defects were common, varying from grade 1 to grade 4, but grade 1 defects were the most common at each knee compartment site. After a mean of 2.3 years (range, 1.8-2.6 years), the patellar cartilage defect score increased significantly, whereas there were no significant changes in cartilage defect scores in other compartments (Figure 4A). However, 21%, 21%, 22%, and 33% of subjects had increases and 27%, 26%, 13%, and 37% of subjects had decreases in cartilage defect scores in the medial tibiofemoral compartment, lateral tibiofemoral compartment, patellar compartment, and any knee compartment, respectively (Figure 4B). Subjects with an increase in knee cartilage defects in any compartment had increased height, weight, change in weight, osteophytes, cartilage volume, and bone size compared with subjects who had a decrease in knee cartilage defects in any compartment (Table 1). There was no significant difference in chronic knee pain or past knee injury between those who had increases and those who had decreases.
Women had smaller increases in tibiofemoral cartilage defect scores than men before (Table 2) and after adjustment for age, BMI, and radiographic OA (data not shown); however, increases in knee cartilage defect scores became larger in women after further adjustment for baseline cartilage volume and tibial bone area. Decreases in knee cartilage defects did not differ between women and men in multivariate analyses (Table 3). The associations between sex and an increase (odds ratio [OR], 1.18; P = .69) or a decrease (OR, 1.24; P = .70) in patellar cartilage defects were also not significant.
The rate of increase in knee cartilage defects at any site was higher in subjects 40 years and older (37%) than in subjects younger than 40 years (19%) (P = .003), while the rate of decrease in knee cartilage defects at any site was not different between the groups (Figure 5A). Age was positively associated with increases in tibiofemoral defects (Table 2) and was negatively associated with decreases in tibiofemoral defects (Table 3). No significant associations were found between age and an increase (OR, 1.00 per year; P = .75) or a decrease (OR, 1.00 per year; P = .87) in patellar cartilage defects.
The rate of increase in knee cartilage defects at any site was higher in subjects with a BMI of 25 or higher (38%) than in subjects with a BMI less than 25 (25%) (P = .02) (Figure 5B). Body mass index was positively associated with an increase in lateral tibiofemoral cartilage defects (Table 2) and was negatively associated with a decrease in medial cartilage defects (Table 3). Furthermore, BMI loss was associated with a decrease in medial cartilage defects. No significant associations were detected between BMI and an increase (OR, 1.03; P = .38) or a decrease (OR, 0.99; P = .72) in patellar cartilage defects.
Baseline knee cartilage defect scores were negatively associated with increases in tibiofemoral (Table 2) and patellar (OR, 0.53 per grade; P<.001) cartilage defects and were positively associated with decreases in tibiofemoral (Table 3) and patellar (OR, 1.66 per grade; P = .02) cartilage defects. Baseline knee cartilage volume was positively associated with increases in medial tibiofemoral (Table 2) and patellar (OR = 1.71 per milliliter, P = .02) cartilage defects and was negatively associated with decreases in medial tibiofemoral cartilage defects (Table 3). Joint space narrowing was not associated with changes in knee cartilage defects (Table 2 and Table 3).
Baseline tibial bone area was positively associated with increases in tibiofemoral cartilage defects (Table 2). Baseline lateral tibial bone area was negatively associated with decreases in lateral tibiofemoral cartilage defects (Table 3). Baseline osteophytes were strongly positively associated with increases in knee cartilage defects in tibiofemoral compartments (Table 2) and were negatively associated with decreases in knee cartilage defects in medial tibiofemoral compartments (Table 3). The association between baseline osteophytes and decreases in knee cartilage defects in lateral tibiofemoral compartments was not quantifiable, as there were no subjects with baseline osteophytes and a decrease in knee cartilage defects in a lateral tibiofemoral compartment.
Women vs men and offspring vs controls were analyzed separately. Similar results were obtained (data not shown).
To our knowledge, this is the largest study to describe the natural history of knee cartilage defects and factors affecting change. In this young sample, a substantial proportion had an increase or a decrease in knee cartilage defects that was greater than that expected because of measurement error. The risk factors for OA (age, BMI, and female sex) were associated with increases in knee cartilage defect scores; conversely, weight loss was associated with decreases in knee cartilage defect scores during 2.3 years. Increased baseline cartilage volume, tibial bone area, and osteophytes predicted increases in knee cartilage defects and vice versa, suggesting that increasing cartilage volume and subchondral bone expansion play roles in the pathogenesis of knee cartilage defects.
Our cross-sectional study showed that knee cartilage defects were common, with 44% of subjects having cartilage defects of grade 2 or higher at any site in the knee, while grade 1 defects are predominant at each site.1 Prevalent knee cartilage defects predict knee cartilage loss during 2 years in healthy adults,9 indicating the clinical importance of knee cartilage defects. Consistent with a previous report,10 the results of the present longitudinal study suggest that cartilage defects are not static. This may be due to measurement issues; however, a 1-U increase or decrease is greater than that expected because of measurement error, suggesting that these are real changes. The decrease in cartilage defects may represent cartilage repair and healing. This suggests that knee cartilage defects are reversible and may represent an intermediate factor to study early in the natural history of knee OA.
It is well recognized that age, BMI, and female sex are risk factors for knee OA.17 While women are more often affected with knee OA after about age 50 years, the incidence of knee OA is the same or even higher in men before age 50 years.17 Incident knee joint space narrowing was inconsistently associated with age and BMI.18,19 Although increases in tibiofemoral cartilage defects were 3.1- to 3.6-fold higher in women, age and BMI were associated with increases in knee cartilage defects in the young sample. These are consistent with other cross-sectional results,1,7,8,20 suggesting that age, BMI, and female sex are risk factors for knee cartilage degeneration. The rate of increase in knee cartilage defects in subjects 40 years and older was higher than that in subjects younger than 40 years, suggesting that knee cartilage defects are more likely to progress after this age. Moreover, age was negatively associated with decreases in knee cartilage defects in tibiofemoral compartments, and BMI loss was positively associated with decreases in knee cartilage defects in medial tibiofemoral compartment. Although an increase in knee cartilage defects with age may be inevitable, weight loss can be an important strategy to decrease knee cartilage defect progression. We failed to find significant associations between patellar cartilage defect changes and age or BMI, possibly because the higher baseline patellar cartilage defect scores may result in less change at this site.
The underlying structural mechanisms associated with progression of knee cartilage defects are obscure. Although joint space narrowing was unassociated with knee cartilage defect changes, baseline cartilage volume was positively associated with increases in medial tibiofemoral and patellar cartilage defects and was negatively associated with decreases in medial tibiofemoral cartilage defects. Combined with previous findings that there was a trend to higher cartilage volume in the offspring of subjects who had severe knee OA than in controls15 and that initial cartilage volume was associated with tibial cartilage volume loss,21 these results suggest that higher knee cartilage volume, possibly due to swelling at an early stage of disease, increases the risk of cartilage defects or imply that they are part of the same pathogenetic mechanism. This is consistent with an experimental observation that tibial cartilage swelling (expressed as greater cartilage volume) occurs in early OA, followed by cartilage fragmentation and degeneration.22 Furthermore, subjects with lower baseline knee cartilage defect scores were more likely to have increases in cartilage defects, and subjects with higher baseline knee cartilage defect scores were more likely to have decreases in cartilage defects. This probably reflects that subjects with lower baseline scores have more scope for progression than subjects with higher baseline scores and vice versa for regression. Moreover, previous studies showed that knee cartilage defects detected on MR imaging were significantly associated with osteophytes6 and that the presence of subchondral bone marrow edema on MR imaging at baseline predicted worsening of cartilage defects after 1 year.23 Results of an earlier cross-sectional study1 suggest that osteophytes and increased knee bone size may be causally related to knee cartilage defects. The present longitudinal study confirms these results with strong associations for both factors, supporting a role for subchondral bone expansion in the origin of tibiofemoral cartilage defects.
The results of this study suggest that decreases in knee cartilage defects or healing is more likely in subjects with younger age, lower body weight, higher weight loss, and no radiographic OA. Decreased joint surface area and lower cartilage volume also appear to be protective. These associations may be causal, because they are prospective predictors, but the mechanism is unclear, and this work should stimulate research on cartilage healing.
This study has several potential limitations. First, the study was primarily designed to evaluate genetic mechanisms of knee OA and used a matched design. The matching was broken for the present study, but adjustment for family history did not alter the results. Although the sample is a convenience sample, Miettinen24 states that for associations to be generalizable to other populations 3 key criteria need to be met regarding selection, sample size, and adequate distribution of study factors, all of which are met in the present study. Nevertheless, these data need to be confirmed in other populations. Second, the semiquantitative method used to assess knee cartilage defects may result in a ceiling effect because of severe cartilage defects or a floor effect because of no cartilage defects. However, no subjects had a minimal score of 0 or a maximal score of 8 in tibiofemoral compartment defects at baseline in this study, so ceiling or floor effects had little effect on the results. Third, although medication use was uncommon in this sample, we cannot comment on the effect of agents such as glucosamine hydrochloride, which may affect cartilage repair.
The results of this longitudinal study suggest that knee cartilage defects are variable and that changes are associated with female sex, age, and BMI. In addition, increases in knee cartilage defects are associated with knee structural alteration such as increased cartilage volume, subchondral bone size, and osteophytes, suggesting a role for these in the pathogenesis of cartilage defects. Finally, interventions such as weight loss may improve knee cartilage defects.
Correspondence: Changhai Ding, MD, Menzies Research Institute, University of Tasmania, Private Bag 23, Hobart, Tasmania 7000, Australia (firstname.lastname@example.org).
Accepted for Publication: October 13, 2005.
Author Contributions: Dr Ding had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Financial Disclosure: None.
Funding/Support: This study was supported by the National Health and Medical Research Council of Australia, Canberra, and by the Tasmanian Masonic Centenary Medical Research Foundation, Hobart.
Acknowledgment: We thank the subjects and orthopedic surgeons who made this study possible, Martin Rush, a member of the Australian Institute of Radiography, who performed the MR imaging, and Kevin Morris, PhD, for technical support.