Objective: Recent studies using various standardized radiographic acquisition techniques have demonstrated the necessity of reproducible radioanatomic alignment of the knee to assure precise measurements of medial tibiofemoral joint space width (JSW). The objective of the present study was to characterize the longitudinal performance of several acquisition techniques with respect to long-term reproducibility of positioning of the knee, and the impact of changes in positioning on the rate and variability of joint space narrowing (JSN).
Methods: Eighty subjects were randomly selected from each of three cohorts followed in recent studies of the radiographic progression of knee osteoarthritis (OA): the Health ABC study (paired fixed-flexion [FF] radiographs taken at a 36-month interval); the Glucosamine Arthritis Intervention Trial (GAIT) (paired metatarsophalangeal [MTP] radiographs obtained at a 12-month interval), and a randomized clinical trial of doxycycline (fluoroscopically assisted semiflexed anteroposterior (AP) radiographs taken at a 16-month interval). Manual measurements were obtained from each radiograph to represent markers of radioanatomic positioning of the knee (alignment of the medial tibial plateau and X-ray beam, knee rotation, femorotibial angle) and to evaluate minimum JSW (mJSW) in the medial tibiofemoral compartment. The effects on the mean annualized rate of JSN and on the variability of that rate of highly reproduced vs variable positioning of the knee in serial radiographs were evaluated.
Results: Parallel or near-parallel alignment was achieved significantly more frequently with the fluoroscopically guided positioning used in the semiflexed AP protocol than with either the non-fluoroscopic FF or MTP protocol (68% vs 14% for both FF and MTP protocols when measured at the midpoint of the medial compartment; 75% vs 26% and 34% for the FF and MTP protocols, respectively, when measured at the site of mJSW; P<0.001 for each). Knee rotation was reproduced more frequently in semiflexed AP radiographs than in FF radiographs (66% vs 45%, P<0.01). In contrast, the FF technique yielded a greater proportion of paired radiographs in which the femorotibial angle was accurately reproduced than the semiflexed AP or MTP protocol (78% vs 59% and 56%, respectively, P<0.01 for each). Notably, only paired radiographs with parallel or near-parallel alignment exhibited a mean rate of JSN (+/-SD) in the OA knee that was more rapid and less variable than that measured in all knees (0.186+/-0.274 mm/year, standardized response to mean [SRM]=0.68 vs 0.128+/-0.291 mm/year, SRM=0.44).
Conclusion: This study confirms the importance of parallel radioanatomic alignment of the anterior and posterior margins of the medial tibial plateau in detecting JSN in subjects with knee OA. The use of radiographic methods that assure parallel alignment during serial X-ray examinations will permit the design of more efficient studies of biomarkers of OA progression and of structure modification in knee OA.