High‐resolution peripheral quantitative computed tomography for the evaluation of bone erosions of metatarsophalangeal joints in patients with rheumatoid arthritis

To compare if the 4th and 5th metatarsophalangeal (MTP) joints evaluated by high‐resolution peripheral quantitative computed tomography (HR‐pQCT) could classify more patients with erosive rheumatoid arthritis (RA) compared with conventional radiography (CR) of the hands, wrists, and feet. Furthermore, we characterize and quantify bone erosions in the two MTP joints by HR‐pQCT.


| INTRODUC TI ON
Rheumatoid arthritis (RA) is a chronic inflammatory joint disease.
Bone erosions are a hallmark of RA and occur early with a predilection for the wrist, metacarpophalangeal (MCP), and metatarsophalangeal (MTP) joints. [1][2][3] The size and number of erosions in a joint, and the number of joints involved, are associated with a poor functional outcome. 4 Hence, early diagnosis, frequent disease monitoring, and individualized treatment of RA are important to prevent disease progression. 5 In clinical practice, the presence of erosions in patients with RA is currently assessed by conventional radiography (CR) of the hands, wrists, and feet. 6,7 However, CR is known to have low sensitivity due to a low spatial resolution and superimposition of two-dimensional imaging for three-dimensional pathology. 8,9 In addition, recent disease-modifying anti-rheumatic drugs and the treat-to-target principle for RA have reduced the occurrence and progression of bone erosions and even documented a partial repair of erosions. 10,11 Therefore, more sensitive imaging methods to identify erosions, disease progression, and treatment response in patients with RA are warranted.
High-resolution peripheral quantitative computed tomography (HR-pQCT) is a three-dimensional imaging modality with a voxel size of 82 μm 3 or 60 μm 3 for the first and second generation, 12 respectively, and a superior spatial resolution (100-140 μm 3 ) compared with all other imaging modalities in vivo, including magnetic resonance imaging (MRI). 13 HR-pQCT is non-invasive, scan time is short, and the radiation exposure is approximately equal to CR. HR-pQCT has successfully been used to quantify bone erosions in the wrist and MCP joints of patients with RA. [14][15][16][17] However, despite erosions also being frequent in the MTP joints, no study has evaluated erosions in this location with HR-pQCT. Previous research indicates that the initial radiographic erosions in RA are likely to occur in the 4th and 5th MTP joints. 2,18 The objective of this study was to compare if the 4th and 5th MTP joints evaluated by HR-pQCT could classify more patients with erosive RA compared with CR of the hands, wrists, and feet. Furthermore, we aimed to characterize the localization, size, and frequency of erosions in the 4th and 5th MTP joints with HR-pQCT.

| Participants
In total, 44 patients with established RA were included and investigated for erosive disease by HR-pQCT and CR. Patients were recruited between June 2021 and November 2021 from the outpatient clinic at the Department of Rheumatology, Aarhus University Hospital, as part of the 1-year follow-up visit in an ongoing study. 17 All patients signed a written informed consent before inclusion.
Demographic and clinical data were collected at the follow-up visit including age, gender, disease duration, Health Assessment Questionnaire-Disability Index, Simplified Disease Activity Index, and C-reactive protein value with the Disease Activity Score of 28 joints. The body mass index, smoking status, disease-modifying antirheumatic drug treatments, anti-citrullinated protein antibodies, and rheumatoid factor status were collected at baseline.

| Eligibility criteria
Inclusion criteria for patients were established RA according to the ACR/EULAR (2010) classification criteria, age 18 years or older, disease duration at least 5 years, and the ability to give informed consent. 19 Exclusion criteria were conditions affecting the 4th and 5th MTP bone and joints, such as previous fracture, metal implants, new surgery, or prosthesis of the feet. Patients with ankle stiffness in both feet (n = 1), which made the scan procedure impossible, were also excluded. Other criteria for exclusion were factors with a possible negative influence on bone density, such as active malignant disease, untreated hypo-or hyperthyroidism, hypocalcemia, or impaired renal function (estimated glomerular filtration rate <35 mL/ min). Pregnant patients were also excluded.

| Conventional radiography
All patients had standard CR of the hands, wrists, and feet performed. A focus distance of 100-115 cm, and exposures of 50-55 kV and 2-12 mAs were used to generate images. The erosive score was evaluated by a single trained reader (RKJ) using the Sharp/van der Heijde (SHS) method, blinded to patient data. 7,17 Previously published baseline data in the ongoing study revealed a high intra-reader reproducibility of the SHS erosion scoring by RKJ (intraclass correlation coefficient [ICC] ≥ 0.990). 17

| HR-pQCT procedure
All patients were scanned using the first-generation Xtreme CT scanner (Scanco Medical AG, Brüttisellen, Switzerland). HR-pQCT imaging settings were adjusted to a 2.7 cm-long volume of interest, a voxel size of 82 μm 3 , an X-ray tube voltage of 59.4 kVp, a current of 900 μA, and an integration time of 100 ms. The scan was performed in the region of 80 slices (6.56 mm) distal and 250 slices (20.5 mm) proximal to the distal end of the 4th metatarsal head. The right foot was used as standard for imaging, but the left foot was used if the right foot fulfilled the exclusion criteria. Before the scan, the patients were asked to perform plantar flexion of the selected foot and place it in a custom positioning device ( Figure S1). The patients' knee was also flexed and supported during scanning. Scan duration was approximately 9 min.

| HR-pQCT analysis
Digital Imaging and Communications in Medicine were exported from the HR-pQCT scanner and evaluated by OsiriX software (Version 9.0.1; Pixmeo, Bernex, Switzerland) on a 27-inch screen iMac. The quality of each scan was evaluated with a 1-5 scale based on visual grading and motion artifacts. 20 Scans with a visual grading and motion-induced artifacts corresponding to a score greater than 3 were excluded from the study. Erosions were defined according to the Study grouP for xtrEme-Computed Tomography in Rheumatoid Arthritis (SPECTRA). 21 In brief, SPECTRA describes bone erosions as a definite cortical break in two consecutive slices in at least two perpendicular planes, nonlinear shape and underlying loss of trabecular structure. 22 All erosions in the 4th and 5th MTP joints were assessed manually by one trained reader (JMM), blinded to patient data. The dorsal, lateral, plantar, and medial quadrants of the metatarsal heads and proximal phalangeal base were evaluated. The number and size of each erosion were evaluated by measuring the maximal axial width with perpendicular depth and length. Volume was measured by manual segmentation. As MTP joints have not been evaluated previously by HR-pQCT, problematic lesions, ie lesions with indistinct cortical breaks, were classified after consensus between JMM, KKK, and RKJ.

| Statistical methods
Data were analyzed using R (version 4.2.1) and R studio (version 7.0.548) software. Histograms and QQ plots were used to investigate the distribution of the data. All data were non-normally distributed and presented as median (interquartile range [IQR]). Statistical significance was tested using the Mann-Whitney U test, while the McNemar's χ 2 test was used to test if there was a significant difference between the proportion of patients classified with erosive RA by HR-pQCT and CR. 23 Furthermore, we used STATA 13 (StataCorp LP, College Station, TX, USA) to assess optimal cut-off values for the number of erosions and total erosive volume per patient evaluated by HR-pQCT for predicting erosive disease in the hands, wrists, and feet by CR using empirical estimations for optimal outcome prediction. 24 Sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were reported. The intra-reader reliability was investigated using the ICC on the number of erosions, average width, average depth, average length, average erosive volume, and total erosive volume per patient for 50% of the HR-pQCT scans.
In addition, Cohen's κ coefficient was used to investigate the intrareader agreement for classifying the scans as erosive or non-erosive by HR-pQCT.

| RE SULTS
A flow-chart of patient inclusion is illustrated in Figure 1. The left foot was imaged in 4 out of 44 (9.1%) patients, while 2 out of 44 (4.5%) patients were excluded because of motion-induced image degradation of the HR-pQCT scans. In total, 42 patients with established RA were included in the analysis. The patient demographics, clinical data, SHS, and HR-pQCT erosive scores are reported in Table 1.
Conversely, the sensitivity and specificity of CR of 44 joints in the The sensitivity and specificity for HR-pQCT of two MTP joints for classifying patients with erosive RA were 100% (95% CI 75.3%-100%) and 13.8% (95% CI 3.9%-31.7%), respectively when CR of the 4th and 5th MTP joint was used as a reference (Table S1).
Meanwhile, the sensitivity and specificity for identifying erosive MTP joints by CR were 34.2% (95% CI 19.6%-51.4%) and 100% (95% CI 39.8%-100%), respectively, when HR-pQCT was used as a reference. McNemar's χ 2 test showed a statistically significant larger proportion of patients classified with erosive RA by HR-pQCT compared with CR of 4th and 5th MTP joints (p < .01). An example of an erosion identified by HR-pQCT and not by CR is illustrated in Figure 2.
The empirical estimation of the optimal cut-off value for the number of erosions per patient in the 4th and 5th MTP joints by HR-pQCT was 7.5 (95% CI 2.9-12.1) erosions for detecting erosive disease by CR in 44 joints of the hands, wrists, and feet. The sensitivity and specificity at the cut-point were 0.60 and 0.83, respectively, with an AUC of 0.72. In addition, the optimal cut-off value for the total erosive volume per patient by HR-pQCT was 11.7 mm 3 (95% CI −6.0 to 29. The number of erosions and total erosive volume, but not the average erosive volume, identified by HR-pQCT were significantly larger in patients classified as having erosive RA than in patients determined as having non-erosive RA by CR (Table 3). Furthermore, the average width and length, but not the depth, of erosions were significantly larger in patients with erosive RA compared with patients with non-erosive RA by CR (Table 3). No significant differences were found regarding the demographic or clinical characteristics in Table 1 between patients classified with erosive RA and non-erosive RA by CR.
A complete agreement was measured by Cohen's κ (κ = 1) for classifying the images as erosive or non-erosive by HR-pQCT. The     The funding sources did not have any role in the collection, analysis, or interpretation of data. The Financial contributors had no role in the study design, collection, analysis, and interpretation of data, the writing of the manuscript or in the decision to submit the manuscript for publication.