Background/Purpose:

Osteophyte formation and evolution is a hallmark of knee OA, and their radiographic identification and progression is fundamental in clinical practice, observational research and randomized clinical trials. The most commonly used grading scales are ordinal, requiring subjective judgment and expert reading time potentially hampering precision and responsiveness. We have developed a semi-automated method for osteophyte volumetric analysis on MRI. Our objective was to evaluate the sensitivity and precision of this method to osteophyte volume change over four years in subjects with established knee OA according to Kellgren and Lawrence (KL) grade.

Methods: Ninety subjects (51 KL 2 and 39 KL 3) were selected from the Osteoarthritis Initiative (OAI) Progression Cohort. Double echo steady state 3D sagittal images were obtained on a 3T Siemens Trio MR system. Measurements were performed on coronal reformatted series. A reader (MH) used software to segment marginal femoral and tibial osteophytes of all 90 knees at baseline and 48 months, blinded to order of visit. The first and last slice of the central weight bearing region were identified and an edge detection algorithm demarcated the bone margins. The reader then ‘closed off’ each osteophyte by marking the expected normal bone contour. The software calculated the total volume (V) for each compartment, bone and knee.  Reliability was assessed using an experienced MSK radiologist (MJ) on a random sub-sample of 20 subjects.  

The primary outcome was change in osteophyte volume (ΔV).  Statistics used were the average change (ΔV), the standard deviation of ΔV (SD), the standardized response mean of (SRM, defined as ΔV/SD), and the percentage of subjects with net increase in V.  Intraclass correlation coefficient (ICC) and root mean square of the standard deviation (RMSSD) were used to assess reliability.

Results: The average change in osteophyte volume (ΔV) was 196 mm³(272), and the SRM was 0.72. A net increase in osteophyte volume from baseline to 48 months was observed for 84% (76/90, 40 KL 2 and 36 KL 3) of the subjects. The average reading time was approximately 10 minutes per knee.

Baseline KL grade (n)	Net Increase	Mean ΔV	SD ΔV	SRM
KL 2 + KL 3 (90)	76 (84%)	196 mm3	272 mm3	0.72
KL 2  (51)	40 (78%)	155 mm3	233 mm3	0.67
KL 3 (39)	36 (92%)	250 mm3	309 mm3	0.81

Table 1 – Responsiveness to change over 48 months

 

	Mean ΔV	SD ΔV	SRM
Medial Compartment	53 mm3	88 mm3	0.60
Lateral Compartment	45 mm3	95 mm3	0.48
Medial Femur	72 mm3	99 mm3	0.72
Lateral Femur	60 mm3	115 mm3	0.52
Medial Tibia	34 mm3	72 mm3	0.47
Lateral Tibia	30 mm3	66 mm3	0.45

Table 2 – Responsiveness to change by compartment

The intra- reader ICC was 0.98, and RMSSD 82 mm³, while inter- reader ICC was 0.97 and RMSSD 91 mm³.

Conclusion: A new computer-assisted method of osteophyte volume measurement is reliable and sensitive. It may provide a more responsive and rapid method for osteophyte change than traditional ordinal methods, making it feasible to assess a large number of knees in a short period of time.

This study was funded by NIH/NIAMS R01AR056664

---

« Back to 2014 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/sensitivity-and-precision-of-automated-osteophyte-volumetric-measurement-in-knee-osteoarthritis-over-four-years/