While knee osteoarthritis (OA) is now recognized as a complex disease affecting the whole joint, not just the cartilages, there remains a paucity of data regarding the interactions between knee components. One relationship of particular interest is between the spatial variations in cartilage thickness (CTh) and subchondral bone mineral density (BMD). Indeed, bone and cartilage are two mechanosensitive tissues that interact as a functional unit and there is evidence of a biomechanical coupling between both tissues. Particularly, a recent in vivo study has shown a positive relationship in non-OA knees with thicker cartilage where bone is denser, and an alteration of this relationship in OA knees. These observations support the concept of an osteochondral unit and warrant additional research to assess the influence of bone depth. Therefore, this study aimed to characterize the relationship between spatial variations in CTh and BMD measured at various depths below the bone surface.

CT-arthrography of 20 non-OA tibias and 20 severe medial-compartment OA tibias were segmented to build 3D mesh models of the bones and cartilages. Each individual tibia model was registered to a reference tibia, allowing to calculate BMD maps at 1, 3, 5 and 10mm below the bone-cartilage interface in the medial compartment. Pearson correlations between CTh maps and the four BMD maps were then calculated for each knee. Lastly, differences in correlation coefficients between successive bone layers were assessed using Wilcoxon signed-rank tests.

In both OA and non-OA tibias, the correlation coefficients were higher with the BMD measured in the 1mm layer, and followed a pattern of statistically significant decrease with bone layers of increasing depth (p < 0.021). In non-OA tibias, the median relationship was positive with a strong effect size in the 1, 3 and 5mm layers, while in OA tibias the median relationship was positive only in the 1mm layer and with a medium effect size. In the OA tibias, the median relationship was negative with a weak effect size in the 3 and 5mm layers, and it was negative with a medium effect size in the 10mm layer.

In conclusion, the results of the present study support the value of considering bone and cartilage as a unit, and more generally support OA pathophysiology models based on relationships among knee properties.

While osteophytes are a hallmark feature of knee osteoarthritis (OA), there is limited information regarding their location. In particular, it is unknown whether osteophytes develop in patient-specific locations or if there are consistent osteophyte locations among OA knees. This lack of data mainly stems from the fact that osteophytes have been mostly assessed with scores quantifying their size or severity but not their location. Given the important role that bone could play in OA development and the option it offers for OA treatment, there is a need to better understand the osteophyte locations. This study aimed to develop a method to compare osteophyte locations among knees and determine the overlapping ratio. CT arthrogram of 11 medial-compartment OA tibias (Kellgren-Lawrence grade ≥ 3) were segmented to locate the osteophytes and a bone matching technique was used to report the osteophyte locations of the 11 knees on a single reference tibia. This newly proposed method was highly reproducible (intra-operator ICC = 0.89). When used to compare the 11 tibias, it showed that more than 60% of the overall subosteophytal area, defined as the reference bone area covered by at least one osteophyte from one knee, was common to less than two tibias. Moreover, less than 20% of the overall subosteophytal area was common to five or more tibias. The results of this study suggest that osteophyte locations are specific to each knee. Future work should determine the relationships with mechanical loading, as this could explain the high inter-patient variability.

Background

Although knee osteoarthritis (OA) has first been thought to be a wear-and-tear disease of the articular cartilage, it is now established that it is a disease of the entire joint. Unfortunately, its complex physiopathology is still incompletely understood, limiting the design of efficient therapeutic options. Recently, it has been suggested that OA could be related to the breakdown of the relationships among knee tissues rather than to isolated alterations of one or more tissues. However, there is a paucity of data regarding relationships among knee tissues. Better understanding the bone and cartilage relationships is thus of particular interest as both tissues contribute to the transfer of the mechanical loads through the joint and altered loading is known to be a main factor of OA onset and progression. Specifically, there is an interest to determine if cartilage thickness (CTh) and subchondral bone mineral density (sBMD), two properties related to loading, are adapted in healthy femurs.

Knee osteoarthritis (OA) affects an estimated 250 million people worldwide, with a cure yet to be found. Consequently, there is an urgent need to improve our understanding of OA physiopathology. While knee OA has long been mostly described as a loss of cartilage thickness (CTh) and research has focused on this characteristic, the role of bone alterations is rapidly gaining in interest. Analyzing subchondral bone mineral density (sBMD) is particularly interesting because this could inform on the mechanical environment at the knee. However, there is a paucity of data on sBMD in literature mainly because of the lack of prior methods to measure this parameter. A method for 3D sBMD assessment based on computed tomography (CT) scans was recently proposed, thus allowing testing for sBMD differences in knee OA. This study aimed at comparing non-OA and medial OA knees in terms of tibial sBMD and CTh. Specifically, it was hypothesized that sBMD and CTh differ with OA.

Ten knees with severe medial OA and 10 matched non-OA knees were analyzed after ethical approval (50% male; 60 ± 3 years old). The arthro-CT scans of the 20 knees were segmented using custom software to build 3D mesh models of the tibial bone and cartilage. CTh maps were obtained by calculating the distance between cartilage and bone meshes, while sBMD maps were calculated based on the intensity of the CT in the first 3mm of bone. For each knee, the average CTh and sBMD values over the entire medial and lateral compartments were calculated and used to determine the medial-to-lateral (M/L) CTh and sBMD ratios. Unpaired t-tests and receiver operating characteristic (ROC) were used for statistical analysis.

The M/L sBMD ratio was significantly higher in OA compared to non-OA knees (1.14 ± 0.04 vs. 1.08 ± 0.03; p<0.01), whereas the CTh ratio was not significantly different between groups (0.70 ± 0.21 vs. 0.85 ± 0.10; p=0.06). No significant differences were found between OA and non-OA knees for the average medial CTh and sBMD (p>0.4). High classification performance was obtained for the sBMD ratio and low performance for the average sBMD in the medial compartment (areas under the ROC curve of 0.9 and 0.6, respectively). CTh ratio and medial compartment average provided medium classification performances (areas under the curve of 0.7).

This study showed that sBMD differed between non-OA and severe medial OA knees and that sBMD M/L ratio was more sensitive to OA severity than CTh variables. These results brought new insights into the pathogenesis of knee OA, by supporting the idea that sBMD is altered with OA and suggesting that sBMD could play a role in disease development. Indeed, the mechanical stresses on the cartilages are related to the mechanical characteristics of the bones. Indirectly, this study also demonstrated the value of arthro-CT scans to simultaneously assess sBMD and CTh. Additional studies with larger cohorts of patients at different stages of the disease are necessary to better understand when changes in sBMD occur.