Knee osteoarthritis patients with more subchondral cysts have altered tibial subchondral bone mineral density

Knee osteoarthritis (OA) is a painful and debilitating disease characterized by cartilage deterioration and altered subchondral bone. Recently, there has been increasing interest in the role of subchondral bone cysts in OA progression; in particular how subchondral bone cysts may influence pain [1‐3], or how subchondral bone cysts influence subchondral bone mechanical behaviour [4].

Subchondral cysts are typically spherical or ellipsoidal cavities within the subchondral bone region and are related to both altered subchondral bone and cartilage degeneration in patients with OA [5‐7]. Recent studies have indicated associations between subchondral bone cysts and pain [8] as well as bone marrow lesions (BMLs) in patients with knee OA [6]; though, evidence of relationships between cysts and other patient characteristics (e.g., disease severity, joint space narrowing, alignment) is limited. A clear understanding of disease pathogenesis is crucial for rational therapeutic targeting [9], particularly understanding of which structures contribute to pain [9]. As subchondral cysts are related to OA progression [10, 11], it is meaningful to investigate associations between subchondral cyst parameters (e.g., number and size) and OA severity and related pain.

Subchondral cysts are associated with higher localized stress [4], which could stimulate bone remodelling or bone alterations. Ex vivo studies at both the hip using high-resolution computed tomography (HR-QCT) [5], and the tibia using micro-computed tomography (micro-CT) [7] report changes in bone mineral density (BMD), especially in regions adjacent to cysts. These ex vivo studies are able to evaluate cyst number and size, but are not able to correlate cyst properties with important clinical symptoms such as pain. Clinical techniques, such as magnetic resonance imaging (MRI) and quantitative computed tomography (QCT) have the potential to offer three-dimensional (3D) characterizations of cyst structure. Using MRI techniques, cysts can be distinguished from other bony features, such as BMLs [12, 13], but it is difficult to reliably quantify BMD. Clinical QCT has potential to characterize cysts in vivo, to explore relationships with clinical OA symptoms (such as pain), to determine tibial BMD, and could potentially be used to evaluate 3D cyst development throughout disease progression to determine the role of cysts in OA. However, it is unclear which specific cyst parameters (e.g., number, size) are associated with clinical symptoms, and which parameters are associated with BMD.

The objective of this study was to use QCT and image-processing techniques to determine relationships between subchondral cyst parameters and subchondral BMD as well as clinical characteristics of OA (OA severity, OA-related pain, alignment, joint space narrowing) in patients with knee OA.

Patient characteristics are found in Table 1. Cysts were present in 88% of participants (37/42). For the entire sample (n = 42), total cyst number varied from none (0) to up to 30 over the total proximal tibia and up to 29 in the medial compartment and 11 in the lateral compartment (Table 2). The ratio of cyst volume to tibial volume ranged from 0 to 14.8% over the total proximal tibia, and up to 24.5% in the medial compartment and 5.3% in the lateral compartments. For many cyst volume parameters, the standard deviation was similar to or greater than mean volume, indicating a large distribution in cyst volume within this sample (Table 2).

There were no significant associations between cyst parameters and age, total WOMAC pain, or nocturnal pain (Tables 3, 4 and 5). Over the total proximal tibia, we found significant positive relationships between (a) alignment and total cyst number and (b) alignment and cyst number per tibial volume (Table 3). At the medial region, higher medial BMD was associated with greater cyst number and volume (Table 4). At the lateral region (Table 5), greater cyst number and volume were also associated with higher BMD, as well as valgus alignment, lateral KL grade, lateral JSN, sex and BMI.

From the Self-Administered Comorbidity Questionnaire, four individuals had diabetes. Exclusion of these individuals did not affect or change associations. No additional comorbidities were reported.

In this exploratory study using 3D in vivo QCT for analysis of proximal tibial subchondral cysts, we report cyst characteristics as well as associations between total, medial, and lateral cyst parameters and patient characteristics. At both the medial and lateral region, cyst number and volume was related to BMD. At the lateral region, cyst number and volume were also related to alignment, KL grade, JSN, sex and BMI. We found no relationships between cyst parameters and total WOMAC pain or nocturnal pain. This is the first in vivo study to use clinical QCT imaging at the knee to evaluate associations between cyst parameters and WOMAC pain, OA severity, and volumetric subchondral BMD. Exclusion of individuals with comorbidities did not affect study findings.

Our findings are consistent with prior research exploring ex vivo cyst characteristics at the hip [5] and at the tibia [7], where high cyst number per volume was also associated with high BV/TV (analogous to BMD) [5] and high trabecular thickness [7]. Cysts, which present as voids in bone, create stress concentrations [25]. High BMD is likely a response to higher stress, with the response being local bone remodelling and altered bone structure near the subchondral surface [7, 26]. High subchondral BMD, resulting from bone remodelling, may counterbalance structural instability due to cyst presence and higher stress. Surprisingly, although medial and lateral BMD were associated with medial and lateral cyst parameters, total BMD was not associated with cyst parameters over the total proximal tibia. This appears to be due to opposing findings in corresponding medial:lateral compartments (e.g., a high medial cyst number would coincide with a high medial BMD, which would coincide with a low lateral cyst number and low lateral BMD). By combining these measures, overall density was reduced but the total cyst property changed very little. This had the effect of reducing associations between total BMD and cyst properties measured over the total proximal tibia. These findings highlight the importance of assessing compartment-specific cyst properties.

Although the mechanism of cyst formation in patients with knee OA is still unknown, there are two primary hypotheses: the “bony contusion theory” [27, 28]—which proposes that excessive loading or trauma can lead to trabecular microfractures, necrotic bone, and focal bone resorption, eventually resulting in cyst development—and the “synovial fluid intrusion theory” [29, 30]—which proposes that the calcified barrier between cartilage and subchondral bone is damaged, allowing for fluid to seep into the subchondral bone, creating a fluid-filled cyst lesion.

In this work, there may be indications of each of these mechanisms in cyst formation, either independently or in combination. Over the total tibial region, the strongest associations were observed between cyst number and alignment, while at the lateral region similar associations were observed between cyst parameters and alignment as well as lateral JSN and lateral OA severity. These results suggest that cartilage degeneration may be associated with proportionally larger and more numerous cysts. Chen et al. [7] report similar findings, where subchondral cyst presence was associated with JSN and cartilage degeneration. Most likely, cyst development is a response to altered loading resulting in potential changes to bone congruence, contact forces, potentially changing load distribution through the proximal tibia [31, 32], possibly through JSN with disease progression or knee alignment. Although alignment was measured based on imaged reconstructions [3], it may be worthwhile hypothesizing why total cyst number and lateral cyst parameters were associated with valgus alignment. As the lateral compartment is predisposed to higher tibial loads in patients with valgus alignment [33, 34], patients with valgus alignment may be predisposed to higher cyst numbers and volume, even before clinical signs of OA, such as pain. However, based on this finding, it would be expected that varus alignment would be associated with medial cyst number and volume, but this relationship was not significant in our study. We recommend further studies using weight-bearing hip-to-toe radiographs to determine knee alignment to evaluate the effect of alignment on cyst characteristics and development in participants with OA.

In this study, sex and BMI were associated with cyst number and volume in the lateral compartment. To further investigate these findings related to sex, we performed comparisons to determine differences in cyst characteristics and participant characteristics between males and females (Additional file 1: Table S1 & Additional file 2: Table S2). Our results indicate that females had more and larger cysts in the lateral compartment than males (Additional file 1: Table S1). Surprisingly, of the 18 participants with lateral cysts, 17 were female and 1 was male. As lateral cyst parameters were dominated by females, an association with sex at the lateral compartment would be expected. This finding is possibly due to loading effects as females also exhibited more valgus loading (Additional file 2: Table S2), which is in line with previous research by Hvid et al. [35] and Sharma et al. [36]. Accordingly, sex and alignment may be factors to consider in future, larger scale studies evaluating proximal tibial subchondral cysts in patients with knee OA. With regards to associations between lower BMI and lateral cyst parameters, this also appears to be sex-related. Specifically, because the lateral cyst parameters were dominated by females, who had lower BMI than males (Additional file 2: Table S2), a negative association with BMI would be expected.

Interestingly, there were no associations between cyst parameters and OA severity at either the total proximal tibia or the medial compartment, but lateral OA severity was associated with all lateral cyst parameters (ρ ranged from 0.51 to 0.54, p < 0.01). As medial OA is more commonly reported [36‐38], this work may highlight a select sub-group of patients (mostly female, mostly with valgus alignment, most with lateral cyst presence) with a tendency for more severe OA at the commonly overlooked lateral compartment. As this is a small sample of participants with severe OA, we recommend further analysis in a larger sample of patients at various stages of OA to further evaluate if this observed phenomenon is a characteristic of this sample, or generalizable to a larger sample of individuals with OA.

Although our previous work evaluating relationships between BMD and pain in this sample reported an association with high lateral focal BMD and pain [2], we found no relationships between cyst parameters and WOMAC pain. This was surprising as necrotic bone, which is found around cysts [10], is thought to contribute to pain [39]. Intra-osseous stress concentrations associated with cysts (which could approach or exceed bone’s yield strength) are also thought to lead to pain [4]. In patients with late-stage OA, cyst presence may not be related to pain in later stages of cyst development. In these cases, bone remodelling levels may be reduced or equilibrated such that the cyst structure is well formed and resistant to the effects of high stress (higher BMD found around the cysts supports this premise). We recommend further work with patients with varying stages of OA and degrees of pain to evaluate associations between cysts, BMD and pain during disease progression.

The findings of this study present various potential clinical impacts. Using clinical QCT we were able to measure relationships between cyst parameters (number, size, volume) and participant characteristics in vivo as had previously only been done ex vivo using non-clinical imaging tools such as micro-CT [7] and HR-QCT [5]. This presents a promising clinical QCT-based technique for monitoring similar cyst parameters and regional BMD in a clinical sample and the ability to incorporate clinical symptoms such as pain. Clinical studies using MRI-based techniques present that cyst presence may [8, 40] or may not [12, 41, 42] be related to pain in patients with OA. Although these studies show that it is possible to distinguish cysts and cyst-like lesions using MR, resolution limitations inhibit the ability to distinguish and measure smaller subchondral cysts. QCT provides higher resolution, as well as the ability to easily quantify cyst-adjacent BMD. This work further demonstrates the complexities of cyst presence in relation to pain in patients with OA, especially in late-stage OA, but may provide a comprehensive technique able to distinguish cysts, as well as regional or localized BMD in vivo.

This study has certain limitations to consider. First, our measurements use a larger voxel size than ex vivo approaches such as micro-CT and may ignore smaller cysts. It was difficult to reliably quantify any cysts smaller than 8 voxels (2x2x2 voxels or 1.95mm³). It was also challenging to differentiate between small cysts and surrounding bone. Second, it was challenging to determine individual cysts with adjacent large cysts that would sometimes connect with one another in some participants. In these cases, these were counted as a single large cyst, but could also be regarded as multiple smaller cysts, which could have merged into a large cyst-like void. This could account for multiple large cysts within our study. Third, our alignment measurement was a custom in-house approach using CT reconstructions and relative joint space widths to approximate mechanical alignment. Although this technique has not been validated against full-limb radiographs, in these patients with late OA malalignment was evident with values ranging from − 18° to 8° from neutral position [3]. Fourth, as this was an in vivo study, we did not include biochemical or histological analysis, and are thus uncertain of cyst genesis or development. Fifth, our sample has various limitations including participants at late stage of OA severity and pain measurement, as well as wide and non-normal variable distribution. Our sample represented patients at late stages of OA and findings may not be applicable to patients with early OA. Also, as this sample did not include healthy control subjects, it is difficult to determine if observed associations are specific to individuals with OA, or pertain to the healthy population as well. WOMAC pain severity and assessment was based on the entire knee joint, including all joint surfaces (tibiofemoral and patellofemoral) and tissues (e.g., bone, menisci, synovium), and it is uncertain if pain originated within the proximal tibial bony structure, other tissues, or a combination of tissues. It is uncertain if wide and non-normal distribution of cyst characteristics may be an element specific to this small sample of participants with severe OA and high amounts of pain, or if this is common over a larger sample with more varied pain or disease progression. Given that cyst parameters of the proximal femur reported by Chiba et al. also exhibited non-normal distributions [5], we believe non-normal distributions are common. However, further prospective evidence from participants at varying initial stages of OA and OA-related pain is needed to complement these preliminary findings. Such research will help to clarify the relationship between cyst parameters and patient characteristics, as well as the role of subchondral cysts in knee OA and OA-related pain.

In this exploratory study, we used clinical QCT to analyze subchondral cysts at the proximal tibia of OA patients to determine relationships between subchondral cyst parameters and clinical characteristics of OA. There was a large range in cyst number and volume in our sample, suggesting that cyst development and progression may vary from patient to patient. Greater cyst number and volume were associated with higher BMD at both the medial and lateral compartments of the proximal tibia. At the lateral region, cyst number and volume were also strongly associated with several patient characteristics including joint alignment, sex, lateral OA severity and lateral JSN, suggesting that there may be disease-associated changes in tibial loading distribution leading to cyst development and further joint deterioration. We found no associations between cyst number or cyst volume with OA-related knee pain in patients with late-stage OA. As such, it may be further worthwhile to explore other bone-related outcomes (e.g., BMLs, BMD), when investigating which structures are associated with pain.