Bone characteristics, vitamin D and osteoarthritis

Osteoarthritis (OA) is a disabling joint disease that results from degeneration of joint cartilage and underlying bone. It is the most common form of arthritis, and the leading cause of disability in the elderly. Despite numerous studies, the etiology of OA is still unclear, leaving much questions how to treat and prevent OA unanswered. In this thesis, the possible role of several bone quality related factors on the development and progression of different types of OA is investigated. The studies were mainly performed within the Rotterdam Study, a large prospective population-based cohort study in the Netherlands.
First, the association between OA and fracture risk was examined. In Chapter 2.1 the results of a longitudinal study among 2,773 men and women are shown. Despite a higher bone mineral density (BMD), subjects with prevalent radiographic OA (ROA) had a 2 times increased risk for an incident vertebral fracture compared to subjects without ROA, and a 50% increased risk for a non-vertebral fracture. Stratification in subgroups showed that this risk was even higher in subjects with high BMD compared to those with low BMD. Adjustment for potential confounding factors, including parameters of postural stability, did not change the results.
Chapter 2.2 describes the results of a study among 1,403 subjects on the association between BMD, fractures and ROA-risk. High systemic BMD at baseline was associated with increased incidence of knee ROA in men and women: subjects in the highest femoral neck BMD (FN-BMD) quartile had a nearly 3 times increased risk of incident knee ROA, and the risk of those in the highest lumbar spine BMD (LS-BMD) almost 5 times, compared to subjects in the lowest quartiles. We also examined whether having a fracture at baseline changed the risk for and OA. We observed that the presence of a vertebral fracture had a protective effect on the incidence and progression of knee ROA.
Over time, more data became available: the study population expanded from 1,403 to 4,154 participants, while also a longer follow-up time was available. In addition, data on the development of hip and hand ROA, besides knee ROA, were collected. Chapter 2.3 shows the results of this expanded study. It confirmed that high baseline BMD is a risk factor for the development of knee OA, and also showed it to be a risk factor for the development of hip OA; subjects in the highest quartile of FN-BMD had a 60% increased risk of incident knee and of incident hip ROA as well compared to the lowest quartile. No relation between high BMD and progression of knee or hip ROA, or incident hand ROA, was found. And, in contrast to the findings of Chapter 2.2, a protective effect of a prevalent fracture on development knee ROA was not observed anymore. On the contrary, a vertebral fracture was found to be a risk factor for incident hand ROA: a subject with a vertebral fracture at baseline had a more than 70% increased risk of developing hand OA. Thus, the association between baseline BMD as a marker for bone quality and the development of hip and knee OA seem robust, but the association between fractures and OA is less clear.
Another factor influencing bone quality is vitamin D status. In Chapter 3.1 the results of our study on the association between vitamin D status, BMD and the development of knee ROA are presented. The mean vitamin D intake was 64 IU/day and the mean 25(OH)vitamin D level 66 nmol/l, the mean follow-up time was 6.5 years. Progressive knee ROA occurred in 5.1% of the subjects in the highest tertile of vitamin D intake against 12.6% in the lowest tertile, resulting in a crude odds ratio (OR) of 2.7 (95% confidence interval (CI) 0.8 to 8.8), and an adjusted OR of 7.7 (95% CI 1.3 to 43.5). No association between vitamin D intake and incident knee ROA was seen. However, we found a significant interaction between vitamin D intake and BMD in the association with incident knee ROA. In subjects with low LS-BMD at baseline we observed an increasing incidence of knee ROA with decreasing vitamin D intake and serum levels. Vitamin D levels were not associated with incident or progressive knee OA. Thus, in this study we could confirm the results of earlier studies that found an association of progressive knee OA with vitamin D intake, but not with 25(OH)vitamin D levels. In addition, the observed interaction between vitamin D intake and BMD in the association with incident knee ROA could indicate that subgroup analyses (in this case subjects with low baseline BMD) might lead to altered insights.
The association between vitamin D intake and OA is more prone to confounding by lifestyle influences than between vitamin D levels and OA. In addition, sample size and replication of association across different cohorts can result in more firm and robust conclusions. Therefore, we conducted a meta-analysis on the relation between vitamin D levels and OA of the knee, hip and hand, with up-dated and expanded results of our previous study (Chapter 3.2). In this meta-analysis, 6 cross-sectional and 6 longitudinal studies were included. No clear association between vitamin serum levels and prevalent, incident or progressive knee, hip or hand OA was observed. However, the quality of most studies was low, and the results conflicting. Meta-analysis of three cross-sectional studies on vitamin D levels and knee joint space narrowing (JSN) showed an increased risk of prevalent JSN with decreasing vitamin D levels (OR 1.52, 95% CI 1.15 to 2.01). The association observed in the meta-analysis of three studies on low vitamin D levels and incident and progressive knee OA was not significant (OR 1.37, 95% CI 0.97 to 1.92), however when considering solely progressive knee OA, the risk was significantly increased (OR 2.40, 95% CI 1.22 to 4.72). In conclusion, this meta-analysis did not provide evidence of an independent association between 25(OH) vitamin D serum levels with hip or hand OA. When analyzing subgroups of knee OA, significant associations of low vitamin D levels with prevalent knee JSN and with progressive knee OA were observed.
Since the role of vitamin D in the development of OA remained unclear, a two sample Mendelian randomization (MR) study was conducted to investigate the causal relationship between genetically determined serum vitamin D levels and hip and knee OA (Chapter 4.1). MR is a study design in which genetic variants are used as proxy for modifiable exposure to test the unconfounded effect of the exposure on a specific outcome. Six single nucleotide polymorphisms (SNPs) associated with vitamin D levels were selected as instrumental variables. Summary statistics of the SNPs effects on OA were derived from the Iceland and UK Biobank, comprising 23,877 knee OA cases, 17,151 hip OA cases and over 562,000 controls. The control samples match the osteoarthritis cases in age, sex, and county of origin. The MR analyses showed no causal association between genetically determined vitamin D levels and knee OA (OR 1.03, 95% CI 0.84 to 1.26) or hip OA (OR 1.06, 95% CI 0.83 to 1.35). So, genetic variations associated with low vitamin D serum levels are not associated with increased risk of OA of hip or knee. This suggests that lifetime decreased vitamin D levels do not increase the risk of knee or hip OA.
In Chapter 5, the results of our investigations are discussed. We observed that OA is a risk factor for future fracture, despite high BMD, suggesting altered bone quality in the presence of OA. Another finding was that high systemic BMD increases the risk of developing knee and hip OA, also implying that higher BMD does not always reflect better bone quality. When studying the possible association of vitamin D status and the development of OA in various ways (prospective epidemiological study, meta-analysis and MR), no consistent association was found. This implicates that supplementing vitamin D does not prevent the onset or worsening of OA. While studying the possible association between different bone-related factors and OA, the importance of considering analysis of subgroups became clear: stratification of OA cases into more homogeneous phenotypes is needed to identify specific factors possibly related to the development of OA. Furthermore, genetic studies like MR are useful to to test the unconfounded effect of a possible factor on the development op OA, or to exclude its effect on OA. When underlying factors associated with OA can be identified, subgroups-specific pharmacotherapeutic treatments could be developed to prevent development or worsening of OA at an early stage.