Reproducibility and Diagnostic Accuracy of Kellgren-Lawrence Grading for Osteoarthritis Using Radiographs and Dual-Energy X-ray Absorptiometry Images

Abstract

Advances in image quality from modern dual-energy X-ray absorptiometry (DXA) scanners now allow near radiograph-like quality images at a low radiation dose. This opens potential new applications for the use of DXA scanners to study other musculoskeletal conditions, such as osteoarthritis, which is often investigated by visual assessment of radiographs. Together, osteoporosis and osteoarthritis are the 2 most common musculoskeletal conditions, both of which primarily affect older people. The aim of this study was to determine whether Kellgren-Lawrence grading of DXA images can be used to grade hip osteoarthritis as effectively as radiographs. People who had attended for recent pelvic radiographs underwent DXA of hips (50 hips from 25 people) using a GE Healthcare iDXA scanner. Three observers assigned Kellgren-Lawrence grades to each image, and grading was repeated at least 1 week apart. Intraobserver and interobserver reliability for radiographs and DXA images were calculated using quadratic-weighted kappa (QWK). People were recalled 12 months later, and the tests were repeated with both the radiograph and DXA scans taken within 2 weeks of each other. Hip DXA intraobserver reproducibility achieved a QWK range of 0.88-0.95 and interobserver reproducibility of 0.85-0.88, similar to QWK from hip radiographs. Intraobserver reliability between subject-matched radiograph and iDXA images revealed QWK ranging between 0.80 and 0.88. Reproducibility of hip osteoarthritis grading using DXA was comparable with that of radiographs in this study and similar to repeatability scores previously published in literature. Given the lower radiation dose and the opportunity to simultaneously investigate osteoporosis, DXA presents an attractive imaging option for osteoarthritis.

MRI and the distribution of bone marrow fat in hip osteoarthritis

“…Precision of KL grading was determined in a previous study using quadratic-weighted kappa (QWK) with an intraobserver repeatability for this reader of 0.88 and interobserver repeatability of 0.81 against a radiologist with more than 10 years' experience. 37 For analysis, KLG0 and 1 were combined, as were KLG3 and 4. Table 1 shows the groups and results of tests for differences in the distribution of radiographic severity with sex, age, and body mass index (BMI).…”

“…While it provides initial data on possible associations between marrow fractional-fat content and OA severity, a longitudinal study would be needed to show whether more subtle changes could be detected in individuals with the incidence and progression of OA. Radiographs were graded by only one reader, not directly involved in this study, which may have led to some ROA cases being misclassified, although calibration in previous studies showed intra-and interobserver repeatability for this reader 37 to fall into the "almost perfect" (QWK !0.81) as defined by Landis and Koch. 45 Study numbers are also relatively small, especially in the KLG2 and 4 groups, and when further subdivided by sex led to some results being close to, but greater than, traditional measures denoting significance.…”

Age at Onset of Walking in Infancy Is Associated With Hip Shape in Early Old Age

“…To examine whether associations between walking age and hip shape were secondary to OA, DXA images were graded for hip OA and assigned Kellgren‐Lawrence grades as previously described . Associations between walking age and OA were assessed in the whole cohort, then associations between walking age and hip shape modes were repeated in only the 707 individuals with no evidence of OA.…”

Trabecular bone score (TBS) as a new complementary approach for osteoporosis evaluation in clinical practice

“…Both points were demonstrated in a recent study by Kolta et al [80] in which 1254 postmenopausal women in the European OPUS cohort, mean age 67 years, underwent TBS measurement at baseline and again at a mean of six years later. The severity of spinal osteoarthritis was graded according to Kellgren and Lawrence [81]. The investigators found that TBS was no different in women with, versus without, osteoarthritic changes in the spine at baseline, while lumbar spine aBMD measurements averaged 5.7% higher in the former group (p< 0.003).…”