Chondrocalcinosis is Associated with Increased Knee Joint Degeneration Over 4 Years: Data from the Osteoarthritis Initiative

Abstract

Objective

To determine if presence of calcium-containing crystals (CaC) is associated with increased knee joint degeneration over four years and assess if total number of CaCs deposited is a useful measure of disease burden.

Design

Seventy subjects with CaCs in right knees at baseline were selected from the Osteoarthritis Initiative and matched to 70 subjects without evidence of CaCs. T1-weighted gradient-echo sequences were used to confirm presence of CaCs and count the numbers of distinct circumscribed CaCs. Morphological abnormalities were assessed at baseline and four-year follow-up using the modified semi-quantitative Whole-Organ Magnetic Resonance Imaging Score (WORMS). Linear regression models were used to analyze the associations between presence of CaCs at baseline and changes in WORMS and to analyze the associations between numbers of circumscribed CaCs at baseline and changes in WORMS.

Results

Presence of CaCs was associated with increased cartilage degeneration in the patella (coefficient: 0.33; 95% confidence interval (CI): 0.04–0.63), the medial femur (coefficient: 0.51; 95% CI: 0.18–0.83), the lateral tibia (coefficient: 0.36; 95% CI: 0.01–0.71) as well as the medial and lateral meniscus (coefficient: 0.38; 95% CI: 0.00–0.75 and coefficient: 0.72; 95% CI: 0.121.32). Knees with higher numbers of CaCs had increased cartilage degeneration in the patella and medial femur (coefficient: 0.09; 95% CI: 0.05–0.14; p<0.001 and coefficient: 0.08; 95% CI: 0.020.14; p=0.005).

Conclusions

CaCs were associated with increased cartilage and meniscus degeneration over a period of four years. Assessing the number of CaC depositions may be useful to evaluate risk of onset and worsening of degenerative disease.

Introduction

Chondrocalcinosis is an arthropathy defined as the presence of calcium-containing crystal (CaC) depositions on radiographs in cartilage or other soft-tissue structures of the joint [1–3]. The main two types of CaCs are basic calcium phosphate (BCP) and calcium pyrophosphate deposition (CPPD) crystals [1, 4, 5]. Since neither CT [6] nor MRI techniques are able to differentiate between these crystal types, we used the term CaCs to encompass both types in this study. CaCs have previously been described to be associated with symptomatic and asymptomatic radiographic knee osteoarthritis (OA), though CaCs and knee OA were both found to increase with age [7]. Currently, it remains unclear if CaCs are the cause or result from OA [8, 9].

Few MRI studies assessed the association of CaCs and structural knee joint degeneration. Recently a cross-sectional MRI study demonstrated that CaCs are associated with a higher prevalence of cartilage and meniscal damage [10]. A longitudinal knee MRI study evaluated synovitis at baseline and four-year follow-up in knees with and without CaCs but detected no statistically significant difference in change of synovitis between both groups [11]. Therefore, it would be of interest to determine how the presence of CaCs is associated with changes of other morphological knee joint structures using state of the art 3T MR imaging.

A previous CT study has shown that CaCs rarely occur as a solitary deposition but are typically ubiquitously present throughout the knee joint [6]. Yet to date it is unknown if these varying numbers of CaC depositions are related to disease onset or severity. Detecting circumscribed CaC depositions on MR imaging is challenging, however, a recent study has shown 3D gradient-echo MRI sequences to be useful for the visualization and quantification of circumscribed CaC depositions [10].

Therefore, the purpose of this study was firstly, to determine if presence of CaCs, confirmed by radiographs and MR sequences, is associated with increased structural knee joint degeneration over four years compared to controls without CaCs, and secondly, to assess if the total number of CaC depositions, assessed with 3D gradient echo MRI sequences, correlates with OA structural damage and could be used by clinicians to evaluate the disease burden.

Methods

Study subjects

We conducted a nested double cohort study selecting subjects based on presence or absence of CaCs at baseline from the Osteoarthritis Initiative (OAI; https://www.niams.nih.gov/grants-funding/funded-research/osteoarthritis-initiative); a longitudinal, multicenter study enrolling a total of 4796 participants with or at risk of developing knee OA. Kellgren-Lawrence (KL)-grades and presence of chondrocalcinosis, on radiographs were assessed by the central OAI radiograph readings. We used radiographs as the gold standard imaging modality for confirming presence of chondrocalcinosis. Radiographic chondrocalcinosis was defined as a definite linear cartilage calcification on the PA view in a compartment-specific manner as reported previously [9]. Moreover, we used 3D gradient-echo sequences to assess distinct circumscribed CaCs. Since these were only available in the right knee, we focused on the evaluation of the right knee of the participants in this study.

We selected subjects with no or mild radiographic OA in the right knee at baseline (KL-grade ≤2, n=3692), and excluded subjects with a history of inflammatory disorders (n=17) and no right knee MRI available at baseline or four-year follow-up (n=817). We excluded subjects with advanced or end-stage OA of the right knee (KL-grad 3 or 4), to avoid not being able to detect differences in changes between groups caused by score limitations for morphological changes at the top of the scale (ceiling effect). From the 2858 subjects who were eligible, we identified 70 subjects that showed radiographic chondrocalcinosis of the right knee at baseline based on central readings and confirmed by our review of radiographs. Moreover, we evaluated the appearance of CaCs on MRI as described previously, defined as presence of punctate foci of hypointensity on gradient-echo sequences along the cartilage or menisci [10]. We confirmed the presence of multiple hypointense foci on gradient-echo MRI sequences in all 70 subjects with radiographic chondrocalcinosis. For comparison, we selected 70 subjects without evidence of CaCs in the right or left knee, based on the central readings and confirmed by our review of radiographs. We then reviewed gradient-echo MRI sequences in all subjects without radiographic chondrocalcinosis and confirmed the absence of punctate foci of hypointensity in all subjects. To protect against confounding, we assessed the distribution of age, race, KL-grades and BMI in the CaC cohort and compared it to the distribution in the 2779 subjects who were eligible but did not have CaCs. Statistically significant differences were found for age, BMI and KL-grades, but not for sex and race. Therefore, the subjects in the comparison group were frequency matched to those with CaCs by randomly selecting 70 subjects from subgroups simultaneously defined by age (5-year intervals), BMI (2.5 kg/m² intervals), and KL grade (0, 1, and 2), Fig 1. We also analyzed knee trauma in all subjects assessed by the question: “have you injured your right knee badly enough to limit your ability to walk for at least two days since the last visit?”.

Figure 1:

Subject selection flowchart

MR imaging

MR images were acquired with 3T scanners (Siemens Magnetom Trio; Siemens, Erlangen, Germany) using standard transmit-receive knee coils (USA Instruments, Aurora, Ohio, USA). The following sequences from the MRI imaging protocol [12] were used: a 3D Tl-weighted fast low-angle shot (FLASH) gradient-echo sequence in a coronal plane (TlwGE) (20/7.6 /12°, TR/TE/flip angle) was used to count the numbers of distinct circumscribed CaCs [10]. Two 2D intermediate-weighted/proton density sequences in a sagittal plane (3200/30, TR/TE, with fat suppression) and coronal plane (3700/29, TR/TE) and a 3D dual-echo steady state (DESS) with water excitation (WE) sequence in a sagittal plane (16.3/4.7/25°, TR/TE/flip angle) were used to analyze morphological/structural abnormalities of the knee.

MR image analysis

MRIs were analyzed by two radiologists (S.C.F., A.S.G., 3 and 7 years of experience, respectively) in consensus and under supervision of a board-certified musculoskeletal radiologist (T.M.L., 25 years of experience). Images were analyzed in a randomized order and radiologists were blinded to clinical and radiographic findings.

CaC MRI gradings were performed at baseline. We evaluated MRI artifacts (e.g. motion and pulsation artefacts) on a scale from 5–1 (5, none; 4, mild, not affecting diagnostic value; 3, moderate, minor impact expected on diagnostic value; 2, pronounced, major impact on diagnostic value; 1, severe, no diagnostic value) [13] in all knees to assure sufficient image quality for the assessment of CaCs. All images were graded as ≥4, therefore no knees were excluded due to insufficient image quality. Presence or absence of CaCs was graded on T1wGE sequences and the numbers of distinct circumscribed CaCs were counted within the following cartilage regions as described previously [10]: patella (PAT), trochlea (TRO), medial femur (MF), lateral femur (LF), medial tibia (MT), lateral tibia (LT). One CaC lesion was defined as one distinct circumscribed hypointense focal area on 3D T1wGE sequences. Fig. 2 illustrates these lesions within hyaline cartilage.

Figure 2:

Illustration of lesion count (white arrows) in the hyaline cartilage using T1-weighted gradient-echo sequences; (A) one lesion, (B) two lesions, (C) three lesions, (D) four lesions.

Morphological abnormalities were assessed at baseline and four-year follow-up using the modified semi-quantitative Whole-Organ Magnetic Resonance Imaging Score (WORMS) of the knee [14]. Cartilage lesions were scored on a scale from 0–6 in six subregions (PAT, TRO, MF, LF, MT, LT). Meniscal lesions were graded on IW- and PD-images together with T1wGE sequences to verify meniscal lesions were actual lesions and not mimicked by CaC depositions [10, 15]. Meniscal lesions were scored from 0–4 in each of the three subregions of the medial and lateral meniscus (anterior/body/posterior). BMEP and subchondral cysts were each scored from 0–3 in six subregions (PAT, TRO, MF, LF, MT, LT). Ligament abnormalities and effusion were graded as described previously [16, 17]. A sum score was calculated for each WORMS feature by adding the scores across all subregions of each knee [18].

Statistical analysis

The statistical analysis was performed with Stata software, version 14 (StataCorp, College Station, TX) using a two-sided 0.05 level of significance. We used linear regression models adjusted for baseline age, BMI, KL-scores, sex, and race to analyze the associations between presence of CaCs at baseline and changes in WORMS scores over four years, and to analyze the associations between total number of distinct circumscribed CaCs at baseline and changes in WORMS scores over four years (unit of increase for the predictor was +10 CaCs).

The variables baseline age, BMI, KL-scores, sex, and race were included for adjustment in the regression models, since these factors could be causal antecedents for OA and CaCs and therefore suspected confounders. Advanced radiographic OA is one of the strongest predictors of fast OA progression, with knees with higher KL-grades displaying greater rates of change in cartilage thickness than those with lower KL-grades [19]. Moreover, knees with higher KL-grades are more likely to have CaCs [10]. Obesity is considered the primary modifiable risk factor for knee OA [20] and patients with CaCs have been shown to have a higher average BMI compared to patients without chondrocalcinosis [21]. Age is another important risk factor for CaCs and OA, with the prevalence of OA increasing from 13.5% in adults of 25 years and older, to 33.5% in adults above 65 years [5, 22]. Moreover, race and sex have also been associated with radiographic worsening of knee OA and CaCs [10, 23].

To check if the underlying assumptions for linear regression models were fulfilled, we conducted the following tests: we tested for linearity by including a quadratic term in the regression model and found linear relationships for all features. We tested for normality by using a qnorm plot and found a normal distribution of the residuals for all features. We checked for homoscedasticity and outliers by analyzing the residual plot versus the predicted values. We also checked for absence of multicollinearity using variance inflation factor values and found all values ≤ 1.2.

Inter-/intrareader reproducibility

For CaC MRI gradings the inter- and intraobserver agreement were assessed using Bland-Altman analysis [24]. The MRIs of all subjects were evaluated by both radiologists for presence of CaCs and number of CaCs. For the intraobserver agreement, the evaluation of 70 subjects was repeated with 14 days in between readings. The Bland-Altman analysis showed no statistically significant bias for the interobserver agreement between both radiologists, nor between the intraobserver readings of each reader [10].

Intra-and interreader reproducibility of WORMS grading by our group have been validated in multiple previous studies [10, 25–28]. In these studies, intraclass correlation coefficients were calculated in order to compare WORMS subscores for the meniscus and cartilage. Intraclass correlation coefficients for intrareader reproducibility ranged between 0.80 (0.69–0.95) [10] and 0.96 (0.94–0.97) [26] for the meniscus and between 0.81 (0.68–0.91) [10] and 0.99 (0.98–0.99) [26] for the cartilage. Interreader intraclass correlation coefficients ranged between 0.81 (0.76–0.88) [10] and 0.97 (0.95–0.98) [26] for the meniscus and between 0.79 (0.72–0.868) [27] and 0.97 (0.95–0.98) [26] for the cartilage.

Results

Subject characteristics

The baseline age range for our cohort of subjects with CaCs spanned 48–78 years (mean ± standard deviation (SD) = 66.86±7.2 years). The baseline BMI range for subjects with CaCs spanned 18.2–36.7 kg/m² (mean ± SD = 27.60±3.9 kg/m²). The baseline age range for our control cohort spanned 48–79 years (mean ± SD = 66.97±7.8 years) and the baseline BMI spanned 19.9–36.4 kg/m² (mean ± SD = 27.76±3.5 kg/m²). Moreover, both the CaC and control cohort had similar distributions for sex, KL-grades, and race. There were more females (39/70. 56%) than males (31/70, 44%) in the CaC cohort and most subjects had a baseline right knee KL-grade of 2 (50/70, 71%), followed by a KL-grade of 0 (12/70, 17%) and a KL-grade of 1 (8/70, 11%). Knee trauma during the observation period occurred 11 subjects, of these 6 were in the CaC cohort, 5 in the cohort without CaCs. Subject characteristics are also summarized in supplementary material Table 1.

MR imaging findings

We analyzed the associations between presence of CaCs at baseline and changes in WORMS over four years. Overall, changes in WORMS scores reflected increased knee joint degeneration in knees with CaCs compared to controls without CaCs (Table 1). Progression of cartilage lesions was found in more knees with CaCs (53/70; 76%) than in controls without CaCs (36/70; 51%). In knees with CaCs, cartilage lesions most frequently progressed in the PAT (n=33) and MF (n=24), in knees without CaCs cartilage lesions most frequently progressed in the PAT (n=17) and TRO (n=15). Development of a new full-thickness cartilage lesion was found in 34/70 (49%) CaC knees and 22/70 (31%) without CaCs. Compared to knees without CaCs, cartilage lesions increased statistically significantly more in the PAT (coefficient: 0.33; 95% confidence interval (CI): 0.04–0.63; p=0.024), the MF (coefficient: 0.51; 95% CI: 0.18–0.83; p=0.003) and the LT (coefficient: 0.36; 95% CI: 0.01–0.71; p=0.044) compartment in the knees with CaCs as shown in Fig. 3. Moreover, knees with CaCs showed statistically significantly more progression of medial and lateral meniscus lesions (coefficient: 0.38; 95% CI: 0.00–0.75; p=0.049 and coefficient: 0.72; 95% CI: 0.12–1.32; p=0.020, respectively). Subchondral cysts increased more in knees with CaCs compared to knees without CaCs (coefficient: 0.64; 95% CI: 0.19–1.10; p=0.006). BMEP, ligamentous changes and effusion did not progress more in those with CaCs compared to those without CaCs.

Table 1:

Association of presence or absence of calcium crystals at baseline and changes in WORMS scores over 4 years

Changes in WORMS scores over 4 years	Subjects with calcium crystals a (n=70)	Subjects without calcium crystals a (n=70)	Coefficient (95% CI)	p-value*b
Cartilage
Cartilage sum score	0.54 (0.43–0.65)	0.26 (0.15–0.37)	0.28 (0.12–0.44)	0.001
PAT	0.72 (0.52–0.93)	0.39 (0.18–0.59)	0.33 (0.04–0.63)	0.024
TRO	0.40 (0.17–0.63)	0.35 (0.12–0.56)	0.06 (−0.27–0.38)	0.736
MF	0.81 (0.58–1.04)	0.30 (0.07–0.53)	0.51 (0.18–0.83)	0.003
LF	0.45 (0.26–0.63)	0.24 (0.05–0.43)	0.21 (−0.06–0.47)	0.124
MT	0.31 (0.15–0.47)	0.11 (0.00–0.27)	0.20 (−0.02–0.42)	0.081
LT	0.54 (0.29–0.79)	0.18 (0.00–0.43)	0.36 (0.01–0.71)	0.044
Meniscus
Sum of bilateral meniscus lesions	0.30 (0.21–0.38)	0.11 (0.03–0.20)	0.18 (0.06–0.30)	0.003
Sum of medial meniscus lesions	0.68 (0.42–0.95)	0.30 (0.39–0.57)	0.38 (0.00–0.75)	0.049
Sum of lateral meniscus lesions	1.10 (0.67–1.52)	0.38 (0.00–0.80)	0.72 (0.12–1.32)	0.020
BMEP	0.13 (0.07–0.19)	0.10 (0.04–0.16)	0.03 (−0.06–0.11)	0.512
Subchondral cysts	0.63 (0.31–0.96)	−0.01 (−0.33–0.31)	0.64 (0.19–1.10)	0.006
Ligaments	0.10 (−0.06–0.26)	0.02 (−0.13–0.19)	0.07 (−0.16–0.29)	0.564
Effusion	0.28 (0.13–0.43)	0.15 (0.00–0.29)	0.14 (−0.73–0.35)	0.201

^*Significant values are in bold

^aNumbers are given as adjusted means (95% CI)

^bMultivariable linear regression adjusting for age, sex, BMI, KL and race

Abbreviations: patella (PAT), trochlea (TRO), medial femur (MF), lateral femur (LF), medial tibia (MT), lateral tibia (LT)

Figure 3:

Mean and standard error of change in WORMS cartilage score of the patella (PAT), medial femur (MF), and lateral tibia (LT) over 4 years in knees with and without calcium crystals. Knees with calcium crystals had statistically significantly more cartilage degeneration in the PAT, MF and LT compared to knees without calcium crystals.

We assessed the association of total number of distinct circumscribed CaCs at baseline and changes in WORMS over four years (Table 2). The cartilage compartment with the highest mean±SD number of CaCs at baseline was the PAT (9.4±11), followed by the LT (8.1±8.6), LF (6.4±9.0), MF (6.2±8.9), TRO (4.9±6.2), and MT compartment (0.8±1.9). Higher numbers of circumscribed CaCs at baseline was associated with increased morphological knee joint damage over four years. In the analysis of the subregions, knees with higher numbers of CaCs had increased cartilage degeneration in the PAT and the MF compartment (coefficient: 0.09; 95% CI: 0.05–0.14; p<0.001 and coefficient: 0.08; 95% CI: 0.02–0.14; p=0.005, unit of increase for predictor +10 CaCs) and statistically significantly more changes of the cartilage sum score (coefficient: 0.03; 95% CI: 0.01–0.06; p=0.016, unit of increase for predictor +10 CaCs) compared to knees with lower numbers of CaCs at baseline. Changes in meniscal lesions, BMEP, subchondral cysts, ligamentous changes, and effusion WORMS subscores over four years were not associated with number of baseline CaCs.

Table 2:

Association of total number of distinet circumscribed calcium crystals at baseline and changes in WORMS scores over 4 years.

Changes in WORMS scores over 4 years	Coefficient (95% CI)	p-value*a
Cartilage
Cartilage sum score	0.03 (0.01–0.06)	0.016
PAT	0.09 (0.05–0.14)	<0.001
TRO	0.00 (−0.06–0.05)	0.943
MF	0.08 (0.02–0.14)	0.005
LF	0.00 (−0.04–0.05)	0.868
MT	0.01 (−0.03–0.05)	0.716
LT	0.02 (−0.04–0.08)	0.492
Meniscus
Sum of bilateral meniscus lesions	0.02 (0.00–0.04)	0.108
Sum of medial meniscus lesions	0.02 (−0.04–0.09)	0.483
Sum of lateral meniscus lesions	0.08 (−0.03–0.18)	0.136
Bone marrow edema pattern	−0.01 (−0.02–0.01)	0.491
Subchondral cysts	0.05 (−0.03–0.13)	0.190
Ligaments	0.02 (−0.02–0.06)	0.329
Effusion	0.03 (−0.01–0.06)	0.167

^*Significant values are in bold

^aMultivariable linear regression adjusting for age, sex, BMI, KL and race. Unit of increase for the predictor is +10 calcium crystals

Abbreviations: patella (PAT), trochlea (TRO), medial femur (MF), lateral femur (LF), medial tibia (MT), lateral tibia (LT)

Discussion

In this longitudinal study we analyzed the association of CaCs and structural knee joint degeneration over four years, using 3T MRI to evaluate progression of cartilage lesions and damage to adjacent joint structures. We found that presence of CaCs was associated with increased knee joint degeneration compared to knees without CaCs. Cartilage compartments with statistically significantly increased degeneration included the PAT, the MF, and the LT compartment. Moreover, higher numbers of distinct circumscribed CaCs at baseline was associated with increased cartilage loss.

The pathophysiology of CaC depositions and the relationship to knee OA is unclear. CaCs have been shown to lower the synthesis of proteinase inhibitors such as tissue inhibitors of metalloproteinases causing increased tissue damage [29]. In addition, CaCs cause mechanical damage to joint structures by increasing the wear of the articular surface [29]. While synovitis is frequently associated with CaCs and is assumed to cause more damage to knee joint structures [30], a recent study found no difference in synovitis between subjects with radiographic chondrocalcinosis compared to controls without radiographic chondrocalcinosis [11]. These findings are in line with our results that showed no statistically significant differences between changes of effusion scores in knees with or without CaCs; though effusion is likely to fluctuate to show a steady increase.

To date, to the best of our knowledge, only two published studies have reported an association of CaCs and cartilage degeneration assessed using MRI [9, 10], and the results are different. Gersing et al reported that CaCs, confirmed on radiographs and MR images, are associated with a higher prevalence of cartilage and meniscal damage assessed using 3T MRI [10]. More previous to the mentioned study, Neogi et al found no association between presence of radiographic chondrocalcinosis and increased cartilage loss using 1.5T MRI [9]. We found that prevalent CaCs at baseline increased structural knee joint degeneration over four years, using 3T MR imaging to assess morphological changes. Moreover, our results indicate that presence of CaCs is also associated with increased damage of adjacent joint structures such as meniscal damage and subchondral cysts.

Interestingly, a higher total number of distinct circumscribed CaCs at baseline was associated with increased cartilage loss over four years, however no associations were found for BMEP, meniscal lesions, subchondral cysts, ligamentous changes and effusion WORMS subscores. These findings may have clinical implications since the total number of CaC depositions could be a useful measure to assess disease burden and risk of OA onset and worsening. Moreover, these results suggest a role for MR evaluation in subjects with chondrocalcinosis to assess both burden of CaCs and structural knee changes.

Our study has some limitations. Since no synovial fluid was available for analysis, the type of CaCs remained unknown. Therefore, we cannot differentiate between specific types of CaCs and differences regarding their association with OA onset and worsening. Moreover, while 3D TlwGE sequences have previously been correlated with CaC depositions radiographs, to the best of our knowledge to date no histopathological correlation has been performed of meniscal or cartilage tissue appearing as focal hypointense lesions on 3D TlwGE sequences. Therefore, it would be of interest to determine the differential diagnostic possibilities for focal hypointense lesions on gradient-echo sequences in the meniscus or cartilage. Our study was also limited by our sample size and the number of structural features we explored. Further research with a larger sample size of adults with chondrocalcinosis may help further refine our understanding of chondrocalcinosis and help develop a clinically useful predictive model. Despite these limitations, our study has a number of strengths including the use of state of the art 3T MR images to assess morphological changes and determining quantity of CaC depositions using gradient echo MRI sequences in addition to radiographs.

In conclusion, knees with CaCs have increased knee joint degeneration over four years compared to knees without CaCs. Cartilage compartments mainly affected by presence of CaCs were the PAT, the MF, and the LT compartment. Moreover, in knees with CaCs, a higher number of distinct circumscribed CaCs was associated with increased morphological knee joint degeneration suggesting that the number of CaC depositions could be a useful measure to assess disease burden and risk of OA onset and worsening.