Association between knee osteoarthritis and volumetric bone

mineral density

Duy K. Hoang1,2, Linh N. Luu2, An T. Truong2, Tan D. Nguyen1,2, Nhan M. Le2,

Huy G. Nguyen1,2, Lan T Ho-Pham2, Tuan V. Nguyen1,3,4
1
University of Technology Sydney, Sydney, Australia;
2
Saigon Precision Medicine Research Center, Vietnam;
3
School of Population Health, UNSW Sydney, Australia;
4
Tam Anh Research Institute, Vietnam
Abstract

Backgrounds: Although patients with radiographic knee osteoarthritis (OA) have a higher
areal bone mineral density (BMD) compared to non-OA individuals, their fracture risk was
not significantly different. This study sought to define the association between radiographic
knee OA and volumetric BMD.

Methods: The study was part of the Vietnam Osteoporosis Study, in which 944 men and
1506 women aged ≥ 40 years were randomly recruited from Ho Chi Minh City (Vietnam).
Radiographs of the knee were evaluated using the Kellgren and Lawrence scale, with grades
ranging from 0 to 4. Knee OA was defined as the presence of radiographic grades of 2 or
higher in a knee joint. Trabecular and cortical volumetric bone density (vBMD) was
measured in the tibia bone by a pQCT XCT2000 (Stratec, Germany). Linear regression
model was used to analyze the association between pQCT parameters and knee OA.

Results: The prevalence of radiographic knee OA was approximately 31% (n = 755), and it
increased with advancing age. In comparison to non-OA individuals, those with knee OA
exhibited higher femoral neck aBMD (effect size [ES] = 0.05, 95% CI: 0.01 to 0.09; P = 0.02
in men vs. ES = 0.02, 95% CI: 0.0004 to 0.04; P = 0.02 in women); however, they had lower
vBMD at the cortical tibia bone (ES = -16, 95% CI: -27 to -4.3; P < 0.001 in men vs. ES = -
11, 95% CI: -2.2 to -2.6; P = 0.01 in women).

Conclusion: These data indicate that approximately a third of Vietnamese people have
radiographic knee OA and that the cortical vBMD was lower in knee OA patients.
Introduction

Osteoarthritis (OA) and its consequence of fragility are increasingly recognized as a major
public health burden in contemporary populations. The prevalence of OA increases with age
and usually affects women rather than men (1). Asymptomatic (radiographic) OA is more
popular than symptomatic OA (2). OA affects multiple joints in the body, but it is commonly
found in the knee. The prevalence of knee OA in an Asian aged 60 years was 22% in men
and 43% in women, which was 45% greater than the US white population (3). Furthermore,
OA is regarded as the most common cause of disability in the elderly (4), accounting for over
three-quarters of all knee and hip replacements due to OA (5).

Previous cross-sectional studies have found that knee OA patients have higher bone mineral
density (BMD) than non-knee OA individuals (6–9); it is expected that knee OA patients are
likely to have a decreased risk of fracture. Nevertheless, some cross-sectional studies have
indicated that the greater BMD changes seen in knee OA do not appear to confer a lower
fracture risk (10–12). Data from Rotterdam Study on 2745 people showed that radiographic
OA was associated with high BMD and increased bone loss (10). Consequently, the
association between knee OA and bone strength should be explained further.

Previous studies tried identifying the association between bone composition and knee OA
using dual X-ray absorptiometry (DXA) to measure areal BMD (13–15). However, this
technology found distinguishing between cortical and trabecular BMD difficult. A key factor
in determining bone strength is cortical bone density. With the ability to estimate cortical and
trabecular bone separately, peripheral quantitative computed tomography (pQCT) has shown
to be an excellent method for measuring the bone area and volumetric BMD (vBMD)
(16–18)
. From these parameters, bone strength can be calculated. As a result, finding biomaterial
of bone composition will help researchers better understand the relationship between knee
OA and bone health.

Even though knee OA and bone fractures are recognized as a public health burden in
Vietnam and developing countries, the prevalence of knee OA and its association with bone
health remained undocumented. We hypothesize that individuals with diagnosed radiographic
knee OA have lower cortical bone density than non-radiographic knee OA individuals. In this
study, we sought to test the hypothesis by pursuing the following specific aims: (i) to estimate
the prevalence of knee OA in Vietnamese older adults and (ii) to investigate the association
between radiological knee OA and vBMD measured by pQCT.
Methods

Study design
This study was part of the Vietnam Osteoporosis Study (VOS) project that was initiated in
mid-2015 and involved more than 4000 men and women aged 18 years and older in Ho Chi
Minh City (formerly Saigon). The study's rationale, protocol, and procedure have been
described elsewhere (19). The study's procedure and protocol were approved by the research
and ethics committee of the People's Hospital 115 on 6 August 2015 (Approval Number
297/BV-NCKH). The study was conducted according to the ethical principles of the
Declaration of Helsinki, and all participants gave written informed consent.

The inclusion criteria of VOS were men and women aged 18 years and older, who agreed to
participate in the study. However, for the current study, only individuals aged 40 years and
older were included because OA of the knee mainly affects people in that age range (20). The
following exclusion criteria were applied: inability to stand or walk without aids, previous
knee injuries and individuals who were wholly paralyzed and could not give informed
consent.

Data collection and measurements

Height and body weight were measured by an electronic portable, wall-mounted stadiometer
(Seca Model 769; Seca Corp, CA, USA) without shoes, hats, ornaments, or heavy layers of
clothing. Body mass index was calculated as weight (kg) divided by the square of height
(kg/m2).

We measured BMD at the femoral neck and lumbar spine (L2-L4) with a Hologic Horizon
densitometer (Hologic Corp., Bedford, MA, USA). BMD at the femoral neck (FNBMD) and
lumbar spine (LSBMD) was measured in gram per cm 2. The densitometer was standardized
before each measurement with a phantom. A qualified radiology technologist did the process.
The coefficient of variation of BMD measurements was 1.5% for the lumbar spine and 1.7%
for the femoral neck. Fat mass and lean mass were also derived from the whole body scan.

Ascertainment of radiographic knee osteoarthritis

The ascertainment of OA was based on a radiographic assessment using the Kellgren -
Lawrence scoring system recommended by the WHO as a standard method for studying OA
in epidemiologic studies (21). Anterior-posterior radiographs of both knees were taken from
all participants. The radiographs will be read by two radiologists who were blinded to each
other’s readings and will be completely unaware of the clinical conditions of the participants.
In the case of discrepancy, readings were adjudicated by consensus with a third reader with
more than 20 years of experience in rheumatology practice. In each knee, the presence or
absence of osteophytes, joint space narrowing, sclerosis and cysts was examined for each
hand joint using the Kellgren-Lawrence system of scoring: 0 = none, 1 = possible osteophytes
only, 2 = definite osteophytes and possible joint space narrowing, and 4 = large osteophytes,
severe joint space narrowing, and/or bony sclerosis. The presence of radiographic OA was
defined if the grade was two or more in at least one joint.

Measurements of bone architecture

We used a pQCT XCT2000 (Stratec, Germany) to measure bone volume and bone geometric
parameters, including the cortical and trabecular compartments of the lower leg. Three slices
were taken in the lower leg at the 4, 38, and 66% sites. Cortical vBMD and cortical bone area
were evaluated at the 38% portion of the tibia. The linear correlation between vBMD at the
38% site and 66% site was ~ 0.96. Trabecular vBMD and trabecular bone area were
evaluated at the 4% site for the tibia. Based on 20 volunteers, the coefficient of
reproducibility for the two sites ranged between 1 and 3%.

Data analysis
The analysis plan was initiated prior to the data collection and ascertainment of OA of the
knee. In the descriptive analysis, we determined the prevalence of radiographic OA in the
knee by age group. We grouped participants into three 10-year age groups: under 50, 50 to
59, 60 to 69 and 70 years or above.
The primary analysis method was the multiple linear regression model, in which bone
parameters were considered dependent variables, and the knee OA group was the
independent variable. Radiographic knee OA was classified into three groups: KL grade 0&1,
KL grade 2 and KL grade 3&4.
We considered four bone measures: areal BMD measured at the lumbar spine and femoral
neck, trabecular vBMD, and cortical vBMD at the proximal tibia. Age was the covariate in
the linear regression model. Before the formal analysis, all outcome variables (e.g., bone
parameters) were standardized to have zero mean and unit variance. The test of the null
hypothesis with P values being less than 0.05 was considered “statistically significant.”
All analyses were done within the R statistical environment.

Results

Our study included 2450 (944 men and 1506 women) aged 40 years and older, whose
demographic and lifestyle characteristics are shown in Table 1. The average of participants
was 55.8 years (SD 9.41), and patients with knee OA were older than non-knee OA
individuals. Men and women with knee OA also had significantly higher percent body fat,
hence, making the proportion of Overweight + Obesity higher than non-knee OA
people. Furthermore, the prevalence of alcohol drinking and smoking was significantly
higher in men than in women (41.6% vs. 2.61% and 35.8% vs. 0.94%, p < 0.001).

The prevalence of radiographic knee OA in our study was 30.81% (n = 755), with women
having a higher prevalence than men at any age (36.6% in women and 21.6% in men). The
prevalence of radiographic OA of the knee was associated with advancing age. Among those
50 years of age or younger, the prevalence of knee OA was around 10% in both genders. This
rate increased to 30% in men and 50% in women between 50 and 59 before increasing to
40.58% in men and up to 70.54% among women aged 70 years and older. (Figure 1)

Almost bone parameters were significantly associated with age except for the trabecular bone
area in both genders. After controlling for age, in both gender, the KL grade 3&4 in men
(effect size [ES] = 0.05, 95% CI: 0.01 to 0.09; P = 0.02) and women (ES = 0.02, 95% CI:
0.0004 to 0.04; P = 0.02) were associated with higher aBMD at the femoral neck. Despite
higher aBMD, individuals with knee OA KL grade 3&4 had significantly lower cortical
vBMD at the tibia cortical bone than non-knee OA individuals (ES = -16, 95% CI: -27 to -
4.3; P < 0.001 in men vs. ES = -11, 95% CI: -2.2 to -2.6; P = 0.01 in women) . However, the
trabecular vBMD tends to increase in knee OA individuals (p = 0.6). (Table 2)

Discussion
Osteoarthritis of the knee and osteoporosis represent significant public health issues
worldwide because of the increasing substantial disability and healthcare costs. However,
there is a lack of information available about the prevalence of knee OA in Asian
populations. We identified that the prevalence of radiographic knee OA in a Vietnamese
community is as high as in other Caucasian populations. On the other hand, although most
previous studies have found that knee OA patients have high aBMD, the risk of fragility
fracture in these people is unchanged. According to our hypothesis that the risk of fracture
among knee OA individuals is attributable to their low vBMD, our study found that despite
the high aBMD in the patients with knee OA, they had lower cortical vBMD on the tibia
bone. These findings merit further elaboration.

In this study, we found that the prevalence of radiographic knee osteoarthritis was nearly
31% which was as high as in other Asian countries. The previous study found that the
prevalence of knee OA at the continent level in Asia was higher than in North America and
Europe (2). In Asia, Thailand, Japan and Korea had a higher prevalence than Vietnam
(22–24)
. Using the same Kellgren-Lawrence criteria for finding knee OA individuals, the
difference between our study with others could be attributable to the relatively young age of
participants, who are generally healthy. Another study showed that the prevalence of knee
OA was more popular in Asian than Caucasians (25). The bending knee and squatting habit
of the Asian could explain this issue (26). Finally, taken together, it appears that the
prevalence of knee OA in Asian populations is higher than that in Caucasian populations.

There is a lack of information about bone microarchitecture in knee OA patients. Although it

is well-known that knee OA was associated with high aBMD, the prevalence and incidence
of fracture are unchanged. Arden et al. found that despite increased aBMD in knee OA
patients, the risk of osteoporotic fractures was not significantly decreased (27) in 5552 older
women from the Study of Osteoporotic Fractures. In addition, patients with knee OA and
higher aBMD in Dubbo Osteoporosis Epidemiology Study were not helped against the
nonvertebral osteoporotic fractures (28). This issue could be explained by reducing postural
stability, then increasing their propensity to fall. Because of the paradox of knee OA,
measuring the bone microarchitecture may provide some information into the bone structure
of knee OA patients. In the Hertfordshire Cohort Study data, Abdin-Mohamed et al. observed
that 87 knee OA patients had greater cross-sectional bone area; hence, the increase in aBMD
(29). However, they could not find an association between knee OA with an increase in
trabecular or cortical volumetric bone density at the tibia after adjusting for age and BMI
(29).

Our major finding is that there is an association between Tibia cortical vBMD and knee OA.
We found that in knee OA grade 3&4 patients, despite increasing aBMD, cortical vBMD
declined in both genders. Compared with a study by Bennell et al. also showed that in those
with moderate OA, lower vBMD was seen compared with controls and those with mild knee
OA (30). The cortical bone is the main component that contributes to bone strength. Taken
together, our findings and previous research contribute to why, even though aBMD is
increased in people with osteoarthritis, the fracture rate remains unchanged.

Our results align with several DXA studies that have found increased and reduced aBMD in
people with knee OA (31,32). Whereas comparing DXA and pQCT is difficult. Above all, the
difference may be attributed to DXA's inability to differentiate between trabecular and
cortical bone as well as its assessment of aBMD. We suggested that the superiority of the
pQCT is utilizing 3-dimensional imaging to evaluate bone architecture and true vBMD for
cortical and trabecular bone separately (33). Therefore, the difference in individuals' bone
size and bone size variations due to knee OA disease could be significant and may
consequently affect the results of BMD from a DXA scan. Altogether, we propose that pQCT
provides a more accurate reflection of bone mineral density than DXA, especially in knee OA
patients.

Different from our study just adjusted for age, Abdin-Mohamed et al. adjusted for age and
BMI in the linear regression analysis in their study and could not find any association (29).
This difference suggested that the association in vBMD among knee OA patients did not
observe was probably due to the effect of body mass index. This problem is explained by the
association of BMI with greater BMD (34) and a higher risk of knee OA (35). The key
conclusion of our findings is that confounders should think carefully before putting into
adjustment methods such as linear regression.

This study has several strengths and potential limitations. The study was based on a large
sample size, and the participants were randomly selected using a rigorous random sampling
technique to ensure the representativeness of the general population. The study sample is
highly homogeneous, which reduces the impact of potential confounders that could
compromise the estimates. However, the participants in this study were sampled from an
urban population; therefore, the study's findings may not be generalizable to the rural
populations. Because participants in this sample were usually healthier volunteers than their
counterparts in the general population, the present results could have been biased.

Our present finding has essential implications in managing knee OA and osteoporosis
patients. Both of these diseases affect a huge number of older adults, not only in developing
countries but also in developed countries. The association between knee OA and bone
fracture indicates that the burden of bone fracture could be increased due to the rapid aging in
the population. Furthermore, our results suggest that areal bone mineral density (by DXA)
could not be able to find knee OA individuals who have a high risk of fracture. As a result,
we should include cortical volumetric BMD in assessing the fracture risk among knee OA
patients.

Conclusion
In summary, based on a population-based study, we found that 31% of Vietnam older adults
have knee OA, and this estimate is comparable with Caucasian populations. We also found
that although the knee OA patients had high aBMD, they had lower cortical vBMD on the
tibia bone in both genders. Therefore, it can be postulated that the increase in areal BMD
demonstrated by DXA is likely to be artefactual.
Figure 1. Prevalence of osteoarthritis of the knee classified by age group.
Table 1. Baseline characteristics of the enrolled 2450 participants by knee OA.

Men Women
Non-knee OA Knee OA Non-knee OA Knee OA
N=740 N=204 N=955 N=551
Age 54.3 (9.13) 59.8 (8.58) 53.1 (8.38) 61.1 (9.15)

Weight 62.4 (10.0) 65.1 (9.25) 53.8 (7.49) 57.3 (9.39)

Height 163 (6.15) 162 (6.81) 153 (5.39) 152 (5.54)

Body mass index 23.5 (3.22) 24.7 (3.11) 23.0 (2.87) 24.5 (3.28)

Lumbar spine BMD 0.92 (0.14) 0.99 (0.16) 0.87 (0.14) 0.82 (0.14)

Femoral neck BMD 0.73 (0.12) 0.76 (0.13) 0.66 (0.11) 0.63 (0.12)

Lean mass 42.80 (5.80) 43.10 (5.90) 31.30 (4.30) 31.65 (4.50)

Fat mass 19.56 (6.0) 21.57 (5.73) 22.44 (4.80) 25.09 (5.20)

Percent body fat 30.9 (5.47) 33.0 (5.72) 41.5 (4.54) 44.0 (4.19)

Overweight + Obesity 779 (54.4%) 505 (59.2%) 426 (46.7%) 353 (68.0%)

Alcohol consumption 322 (44.0%) 66 (32.7%) 32 (3.39%) 7 (1.28%)

Cigarette smoking 276 (37.8%) 58 (28.7%) 11 (1.17%) 3 (0.55%)

Tibia trab. vBMD 202 (31.1) 202 (33.0) 185 (32.1) 175 (33.5)

Tibia cort. vBMD 1186 (32.1) 1183 (29.8) 1130 (47.5) 1100 (50.1)
Parameter
Regression coefficient (and standard error) associated with

KL grade 2 KL grade 3&4 Age

Men

Lumbar spine BMD 0.08 (0.02)** 0.07 (0.03)* -0.002 (0.0006)**

Femoral neck BMD 0.06 (0.01)** 0.05 (0.02)* -0.005 (0.0005)**

Tibia trab. bone area 16.28 (7.87)* -27.06 (13.01)* -0.12 (0.32)

Tibia trab. vBMD 3.83 (3.57) 3.09 (5.90) -0.76 (0.14)**

Tibia cort. bone area 13.73 (4.24)** -1.37 (7.01) -0.58 (0.17)**

Tibia cort. vBMD 5.62 (3.57) -15.86 (5.91)** -0.53 (0.14)**

Women

Lumbar spine BMD -0.005 (0.01) 0.02 (0.014) -0.008 (0.0005)**

 Femoral neck BMD 0.02 (0.01) 0.02 (0.01)* -0.007 (0.0004)**

Tibia trab. bone area 2.42 (4.76) 2.19 (6.03) 0.38 (0.22)

Tibia trab. vBMD 1.19 (2.72) 1.34 (3.45) -1.49 (0.13)**

Tibia cort. bone area 4.85 (2.72) 4.92 (3.44) -1.33 (0.13)**

Tibia cort. vBMD -4.12 (3.54) -11.41 (4.51)* -2.87 (0.16)**

Table 2 Association between knee OA and bone parameters after adjusting for age:
results of multiple linear regression analysis