Longitudinal assessment of cyst-like lesions of the knee and their relation to radiographic osteoarthritis and MRI-detected effusion and synovitis in patients with knee pain

Subjects included in the present study were participants in the Joints On Glucosamine (JOG) cohort. The JOG study is a 6-month, double-blind, randomized controlled trial to examine the efficacy of oral glucosamine supplementation. Two hundred and one participants, aged 35 to 65 with mild to moderate chronic, frequent knee pain (Western Ontario and McMaster Universities score ≥25), were recruited at the University of Pittsburgh, PA, USA. Institutional Review Board approval and written informed consent from all participants were obtained for the present study.

Fifteen types of lesions were studied, including popliteal cysts (Figure 2), proximal tibiofibular joint (PTFJ) cysts, medial and lateral meniscal cysts, and anterior and posterior cruciate ligament (ACL and PCL) and Hoffa's fat pad ganglion cysts. In addition, the following bursitides with cyst-like appearances were included: anserine, prepatellar, superficial and deep infrapatellar, iliotibial, SM-MCL, medial collateral ligament, and subgastrocnemius bursitides [7] (Figures 1 and 2).

Radiographic images of the knee were obtained using a standard radiographic technique according to published guidelines [14]. Radiographs were read according to the Kellgren-Lawrence (K-L) grading scheme [15] by one expert musculoskeletal radiologist (AG) with 11 years of experience in K-L grading of knee radiographs. A random set of 30 radiographs (60 knees) were re-read to calculate the intraobserver reliability for the K-L grading of the knee, and the weighted kappa value was 0.89.

Of 201 subjects in the JOG cohort, 38 were excluded because only a baseline MRI scan was available (24 subjects) or complete follow-up MRI assessment was not possible due to poor acquisition quality (14 subjects). Thus, 319 (161 left and 158 right) knees of 163 subjects were included in the present study.

MRI of each knee was performed using a 3 Tesla MRI (Siemens Trio, Erlangen, Germany) at baseline and at 6 months. The identical protocol used for the Osteoarthritis Initiative was applied in the JOG study, excluding the Fast Low Angle Shot sequence and the Multi-Echo Spin Echo T2 mapping sequence. Details of the full Osteoarthritis Initiative pulse sequence protocol and the sequence parameters have been published [16]. The protocol included the sagittal triplanar three-dimensional Dual Echo at Steady State sequence (slice thickness = 0.7 mm, interslice gap = 0 mm, repetition time = 16.3 ms, echo time = 4.7 ms, flip angle = 25°, field of view = 140 mm × 140 mm, matrix = 384 × 307 pixels, echo train length = 1, number of slices = 35, bandwidth = 185 Hz/pixel, number of excitations = 1, anterior/posterior phase encoding axis, acquisition time = 10 minutes 23 seconds) and the sagittal intermediate-weighted fat-suppressed sequence (slice thickness = 3 mm, interslice gap = 0 mm, repetition time = 30 ms, echo time = 3,200 ms, flip angle = 180°, field of view = 160 mm × 160 mm, matrix = 313 × 448 pixels, echo train length = 5, number of slices = 37, bandwidth = 248 Hz/pixel, number of excitations = 1, anterior/posterior phase encoding axis, acquisition time = 4 minutes 42 seconds). Axial and coronal images were reformatted from the sagittal three-dimensional Dual Echo at Steady State images.

One musculoskeletal radiologist (ZD) with 15 years of clinical experience, blinded to clinical data but not to the timepoint of the MRI examination, reviewed all images. The images were evaluated for the presence or absence of cystic lesions at baseline and at 6 months. Measurements of the largest diameter of each lesion were made to the nearest millimeter using a manual caliper on a standard DICOM viewer (eFilm Workstation, version 2.0.0; Merge Healthcare, Milwaukee, WI, USA). For the cystic lesions, unidimensional measurement was made based on the assumption that the size of a lesion increases or decreases in the same direction for all three axes - see Revised Evaluation Criteria in Solid Tumors (RECIST) (version 1.1) for size measurement of solid tumors [17] (in RECIST 1.1, cystic tumors are also considered a measurable lesion). The longest diameter of each lesion was measured on the sequence that showed the lesion in its largest dimension using the same plane (axial or sagittal) at baseline and at follow-up.

Of the lesions listed earlier, it was only possible to make meaningful size measurements on popliteal cysts, meniscal cysts, anterior cruciate ligament and PCL ganglion cysts, prepatellar bursitis, superficial infrapatellar bursitis and subgastrocnemius bursitis. These lesions had approximately circular or elliptical shape and their shape remained similar at two timepoints, enabling us to make meaningful size comparisons based on the assumption described earlier. However, it was not possible to measure the size of other lesions due to their completely irregular or undefinable shape. For these lesions, only the presence or the absence was noted. Development of a new lesion at follow-up was treated as an increase in size, and complete resolution of a pre-existing lesion at follow-up was treated as a decrease in size for the purpose of our analysis.

Severity of synovitis and effusion at baseline was semiquantitatively assessed according to the Whole Organ Magnetic Resonance Imaging Score (WORMS) system [18], taking into account all available sequences. Synovitis was graded 0 to 3 at infrapatellar and intercondylar sites (0, none; 1, mild; 2, moderate; 3, severe), and the maximum grade at either site was recorded. Effusion was also graded 0 to 3 (0, none; 1, ≤33% of maximum potential distention of the synovial cavity; 2, 33 to 66% of maximum potential distention; 3, ≥66% of maximum potential distention).

Intraobserver reliability for the detection of cystic lesions and their size measurement was assessed on a random sample of 20 knee MRI scans with all acquired sequences, and was excellent (κ = 0.88 for detection of cystic lesions; intraclass correlation coefficient for their size measurement = 0.94). Assessment of interobserver reliability was performed on a random sample of 50 knees by a second expert musculoskeletal radiologist (AG) with 11 years of experience in semiquantitative analysis of knee OA. Agreement for detection of cystic lesions was high (κ = 1.00), and the intraclass correlation coefficient for size measurements on the same sample was 0.99.

During our intraobserver and interobserver reliability exercises, it was noted that the maximum difference between two readings for the measurement of the same lesion (either at baseline or at follow-up) was 4 mm for the popliteal cyst and 1 mm for other measurable lesions. We thus considered that any changes within these ranges might be attributable to measurement error. We defined the significant change of size to be ≥5 mm for popliteal cyst and ≥2 mm for other measurable lesions.

The K-L grading was used to categorize knees according to the severity of radiographic OA. When there were eight or more of a particular type of lesion, the association between K-L grade and the prevalence of those lesions was tested using logistic regression controlling for clustering by person [19, 20]. For cyst-like lesions that are lined by synovium and known to have a communication with the joint capsule [4] (that is, popliteal cysts, subgastrocnemius bursitis and PTFJ cysts), we also analyzed whether there was a linear trend between their prevalence and increasing severity of MRI-detected effusion and synovitis using logistic regression controlling for clustering by person.

Finally, the Wilcoxon signed-rank test was used to determine whether there was a significant change in the size of lesions between baseline and follow-up.

All statistical calculations were performed using SAS^® software (version 9.1 for Windows; SAS Institute, Cary, NC, USA), except for the logistic regression that was performed using STATA (version 11.0; Statacorp LP, College Station, TX, USA).

The present study included 163 participants with a mean age of 52 years (range 35 to 65 years, standard deviation ± 6) and a mean body mass index of 29 ± 4 kg/m². Subjects were predominantly white (92%, 150/163) and approximately one-half were women (46%, 75/163). There were no statistically significant trends in the prevalence of any type of lesions with respect to age, gender or body mass index (data not shown), except for PCL ganglion cysts. Nine subjects (82%) who had PCL ganglion cysts were women, whereas 66 (43%) of those without the lesion were women (P = 0.024).

Table 1 summarizes the prevalence of all lesions. At least one fluid-filled lesion of any type was present in 208 knees (65%). The most prevalent type was popliteal cysts (Figure 2), found in 128 knees (40%). Subgastrocnemius bursitis, PTFJ cysts and Hoffa's fat pad ganglion cysts were observed in 49 knees (15%), 26 knees (8%) and 25 knees (8%), respectively. Nineteen (39%) subgastrocnemius bursitides co-existed with a communicating popliteal cyst (Figure 2). Other types such as SM-MCL bursitis, ACL and PCL ganglion cysts, and medial meniscal cysts were observed less frequently. Superficial and deep infrapatellar, prepatellar, anserine, and iliotibial bursitides were rarely observed but seemed to be more common in knees with OA than in those without OA. No lateral meniscal cysts were detected in any knee at baseline but one lesion had appeared at 6-month follow-up.

Of the 319 knees, 103 (32%) were K-L grade 0, 27 (9%) were K-L grade 1, 37 (12%) were K-L grade 2, 136 (43%) were K-L grade 3, and 16 (5%) were K-L grade 4 (Table 1). A linear trend was found between increasing prevalence of fluid-filled lesions of any type and increasing K-L grade (K-L 0 = 54%, K-L 1 = 70%, K-L 2 = 78%, K-L 3 = 66% and K-L 4 = 88%; P = 0.014). This trend, however, did not extend to any individual type of lesions.

Increasing prevalence of subgastrocnemius bursitis was associated with increasing severity of effusion (P = 0.0072) and synovitis (P = 0.0060) after controlling for clustering by person (Table 2). For popliteal cysts and PTFJ cysts, such a trend was not observed.

Changes in the size of the lesion were significant only for popliteal cysts and subgastrocnemius bursitides (P < 0.001 for both) (Table 3). Of the 128 popliteal cysts identified at baseline, 95 (74%) exhibited a change in size over 6 months. Twenty-three increased in size (including the development of six new lesions) and 39 decreased in size. Of the 49 subgastrocnemius bursitides, 13 (27%) exhibited a significant change in size over 6 months (seven increased and six decreased). The remaining lesions did not show notable fluctuation in size.

Seventy-six (49%) of the 156 subjects with readings for both knees exhibited at least one fluid-filled lesion (of any type) in both knees (Table 4). Bilateral popliteal cysts were observed in 34 subjects (22%) and bilateral subgastrocnemius bursitis in 11 subjects (7%). Bilaterality was observed to a lesser extent for PTFJ cysts, Hoffa's fat pad ganglion cysts, medial meniscal cysts and SM-MCL bursitis.