Introduction

Psoriatic arthritis (PsA) is a chronic inflammatory arthritis characterized by synovitis and enthesitis [1,2,3]. Inflammatory arthritis, including rheumatoid arthritis (RA) and PsA, leads to structural damage at the metacarpophalangeal joint [4, 5]. High-resolution peripheral quantitative computed tomography (HR-pQCT) enables the assessment of bone erosions, new bone formation, and cortical and trabecular microarchitecture [6]. HR-pQCT-based studies have shown that entheseal new bone formation is an early sign of bone change in PsA [7, 8]. This entheseal new bone formation represents a dominant structural feature of disease in established PsA which is largely absent in RA [4]. Much of the structural change seen in PsA seems to be driven by entheseal inflammation [4]. Erosions, on the other hand, reflect mechanical and inflammatory joint damage and are more prevalent in RA rather than PsA [4]. An HR-pQCT-based study has demonstrated that tumor necrosis factor (TNF) inhibition arrests the progression of bone erosion but not enthesiophye formation [9], while interleukin (IL)-17 inhibition arrests the progression of both bone erosion and enthesiophyte formation [10], suggesting fundamental differences in the pathophysiology of bone erosions and enthesiophytes.

The entire metacarpal head (MCH) is located inside the joint capsule and is therefore exposed to the intra-articular inflammatory milieu in inflammatory arthritides [11]. Within the MCH, bone loss has been found in both the trabecular and cortical compartments in RA patients [11]. Whether the predominantly entheseal-driven inflammatory process of PsA is associated with intra-articular bone loss similar to that seen in RA patients is uncertain. Our aims were (1) to compare the intra-articular bone structure and microstructure between PsA patients and healthy control subjects; and (2) to ascertain the demographic and clinical factors associated with compromised bone microstructure in PsA patients.

Methods

Patients

Seventy-seven PsA patients and 62 healthy controls were studied in this cross-sectional study. Distal radial densitometric and microstructural features in 53 of the 77 PsA patients and 53 of the 62 controls were published previously [12]. All PsA patients fulfilled the ClASsification for Psoriatic ARthritis (CASPAR) criteria [13] and were recruited from the Rheumatology Clinic at the Prince of Wales Hospital. Exclusion criteria for patients were as follows: (1) HR-pQCT image showing joint destruction (bone erosions and enthesiophytes leading to destruction of the normal joint structure, sFigure 1) [14]; (2) presence of an underlying disorder that could affect bone metabolism such as chronic renal impairment (chronic kidney disease stage IV or V), type 1 diabetes mellitus (DM), unstable cardiovascular disease, thyroid or parathyroid disease, malignancy, or chronic liver disease; (3) current or past usage of anticonvulsant therapy, thyroid or parathyroid hormone, or antiosteoporosis therapy; and (4) pregnancy or breastfeeding. Treatment with glucocorticoids, calcium, and/or vitamin supplement was allowed. After PsA patients with joint destruction were excluded, sixty-two age- and sex-matched healthy control subjects were selected from our previous study (sFigure 2) [12]. Exclusion criteria for control subjects include all of the above exclusion criteria, as well as a history of any autoimmune disease or any other major illness, with the exception of hypertension, type II DM, and dyslipidemia. Ethics approval (CRE-2012.082) was obtained from the Ethics Committee of The Chinese University of Hong Kong-New Territories East Cluster Hospitals with written informed consent obtained from all participants.

Clinical assessment

Age, body weight, body height, and smoking habit were recorded and body mass index (BMI) was calculated for all participants. Clinical assessment performed by rheumatologists included the number of tender (0–68) and swollen (0–66) joints, permanently deformed joints and the presence of dactylitis. Psoriatic nail involvements, including pitting, hyperkeratosis, and onycholysis, were assessed. The levels of erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) were measured. Disease activity was assessed using the Disease Activity index for PsA (DAPSA) [15] and Psoriasis Area and Severity Index (PASI); while physical function was assessed using the Health Assessment Questionnaire (HAQ) disability index. Also recorded were the presence of rheumatoid factor (RF), current use of conventional synthetic disease-modifying anti-rheumatic drugs (csDMARDs), non-steroidal anti-inflammatory drugs (NSAIDs), and biologic DMARDs (bDMARDs).

High-resolution peripheral quantitative CT

All patients underwent an HR-pQCT (Xtreme-CT scanner, Scanco Medical AG, Brüttisellen, Switzerland) examination of the second and third MCH (MCH 2&3) of the non-dominant forearm [12]. The patient’s forearm was immobilized in a carbon fiber cast fixed within the scanner gantry. A dorso-palmar projection image was obtained to define the tomographic scan plane. The scan region started at the distal end of the MCH 2&3 and spanned proximally 9.02 mm (110 slices).

Detection of erosion and enthesiophyte

MCH 2&3 was divided into palmar, ulnar, dorsal, and radial quadrants [14, 16] to evaluate the number and size of erosions and enthesiophytes within each quadrant (sFigure 3). Erosions were defined as a clear break in the outer cortical margin evident on at least two consecutive slices and in two orthogonal planes [4, 7, 14]. A cortical break was considered a vascular channel if it was tubular in outline, regularly delineated, without loss of surrounding trabecular structure [17,18,19]. Pseudo-erosions formed by two osteophytes configured like an open forceps were excluded [20]. To increase the specificity of identifying pathological erosions, only cortical breaks wider than 1.9 mm [14] were included for further analysis.

Enthesiophytes were defined as new bone formation arising from the periosteal bone cortex at the insertion sites of the capsule, ligament, or tendons or at the location of functional enthesis [7, 21]. The maximal distance between the superficial surface of the enthesiophyte and the missing true cortical boundary was defined as the maximal height (mm) (sFigure 4) [4, 5, 7, 21]. Erosions and enthesiophytes were identified and evaluated by two independent, blinded readers (DZW, SHL). When more than one erosion or enthesiophyte was present in a single quadrant, the one with the largest volume or height was selected [7].

Image analysis for erosion and enthesiophyte at the second and third metacarpal head

Images were analyzed using the open-source application ITK-SNAP interactive image visualization and segmentation tool [22,23,24], validated for volume measurement of hard and soft tissue [24,25,26,27,28]. The technique comprised four major steps: (1) metacarpal bone cropping, (2) periosteal surface segmentation, (3) restoration of the missing true periosteal cortical boundary, and (4) volume calculation (Fig. 1A, B). Periosteal surface restoration was performed using a threshold-based semi-automatic segmentation method with supervised learning [23]. Any missing periosteal cortex margin was manually restored based on the anatomic curve. Restoration of the periosteal cortical margin was an important step in the analytical process given that enthesiophytes and erosions were bone deposited on and resorbed from the periosteal cortical margin respectively [29]. As determined by intraclass correlation coefficient (ICC), inter-observer reliability for bone erosion volume was 0.996 (95%CI: 0.992, 0.998) and for enthesiophyte maximal height and volume were 0.741 (95%CI: 0.419, 0.885) and 0.986 (95%CI: 0.968, 0.994).

Fig. 1
figure 1

Image analysis on structural and microstructural bone changes. Quantification of bone erosion or enthesiophyte volume (A, B): crop of metacarpal bone (a, b), segmentation of periosteal surface (c, e), restoration of the missing true cortical boundary based on anatomic curve, (d, f), and three-dimensional calculation of volume (g). Calculation of bone density and microstructure after the exclusion of bone erosion and enthesiophyte (C, D): contour before exclusion of bone erosion and enthesiophyte (a, c) and contour after exclusion of bone erosion and enthesiophyte (b, d)

Full size image

Image analysis for vBMD and microstructure at the second and third metacarpal head

MCH 2&3 were chosen as they are the most commonly affected finger joints among MCH 2–4, [4]. Images were analyzed using a standard protocol provided by the manufacturer [30] after exclusion of bone erosion and enthesiophyte (Fig. 1C, D) [31]. The following parameters were measured: average vBMD, trabecular (Tb) vBMD, meta-trabecular (mTb, inner 60% of the medullary region) vBMD, peri-trabecular (pTb, outer 40% of the medullary region) vBMD. Trabecular bone structure was evaluated by determining the trabecular bone volume fraction (BV/TV), trabecular number, thickness, separation, and trabecular inhomogeneity (standard deviation of 1/trabecular number). Cortical bone vBMD, thickness, and perimeter were also measured. Our short-term HR-pQCT reproducibility, expressed as coefficient of variance, ranges from 0.38 to 1.03% for density measures and from 0.80 to 3.73% for microstructural measures [32].

Statistical analysis

Data are presented as number (percentage) for categorical data, mean ± SD for normally distributed data, or median and interquartile range (IQR) for non-normally distributed data. Normality testing was performed using a combination of Kolmogorov-Smirnov test and histogram. Differences between PsA patients and controls in terms of demographic and HR-pQCT parameters (erosion and enthesiophyte [total number and total volume at MCH 2&3], mean vBMD, and bone microstructure at MCH 2&3) were assessed using the chi-square test for categorical variables, Mann–Whitney U test, or Student’s t test for continuous variables depending on the data distribution. In PsA patients, Pearson’s and Spearman’s correlation analyses were used to determine the association between demographic and clinical characteristics (age, psoriasis and PsA disease duration, BMI, CRP and DAPSA) and HR-pQCT parameters (erosion and enthesiophyte volume, vBMD, and bone microstructure) for normally or non-normally distributed data respectively. Potential variables associated with microstructural measures were examined first by univariate correlation analysis and subsequently by multiple linear regression analysis. Independent explanatory variables associated with compromised microstructure in PsA were assessed using multivariable linear regression models with adjustment for covariates (including all the potential explanatory variables associated with compromised microstructure identified from the univariate analyses with a p value < 0.05). A histogram was used to show the anatomic distribution of bone erosion and enthesiophyte. A p value of < 0.05 was considered statistically significant throughout the analysis. All statistical analyses were performed using the Statistics Package for Social Sciences (IBM SPSS V.21.0, IBM Corporation, Armonk, NY, USA).

Results

Demographic and clinical features

Fifteen out of 77 (19.5%) PsA patients were excluded due to joint destruction. The remaining 62 patients and 62 healthy control subjects were comparable for gender, age, body mass index (BMI), and smoking habit (Table 1). PsA disease duration was 17.3 ± 14.5 years, with a low to moderate disease activity (DAPSA: 10.8 ± 8.1; ESR: 20.5 ± 20.6 mm/h; CRP: 5.2 ± 8.4 mg/L). Three (4.8%), 33 (53.2%), and 11 (17.7%) patients were receiving corticosteroids, conventional synthetic disease-modifying anti-rheumatic drugs (csDMARDs), and biologic DMARDs respectively. None of the patients had a positive RF.

Table 1 Demographic and clinical characteristics of patients with psoriatic arthritis and controls
Full size table

Comparison of bone structure at the second and third metacarpal heads in PsA and controls

All subjects except one (a PsA patient) had at least one bone erosion. In the PsA group, 85 erosions were identified, while only 50 erosions were observed in the control group. At MCH 2&3, the mean number of erosion (PsA: 1.4 ± 1.2 vs control: 1.1 ± 1.0, p = 0.110) was similar between the two groups, yet the total erosion volume per person was significantly greater in PsA compared with controls (PsA: 5.2 ± 6.0 mm3 vs control: 2.8 ± 4.5 mm3, p = 0.003) (Table 2). Bone erosion in PsA patients had a typical predilection for the radial and dorsal site (Fig. 2A).

Table 2 Number and size of erosions and enthesiophytes in psoriatic arthritis (PsA) patients and controls
Full size table
Fig. 2
figure 2

Anatomic distribution of bone erosion and enthesiophyte. Number and quadrant of bone erosion in PsA patients and healthy controls (A). Number and quadrant of enthesiophyte in PsA patients and healthy controls (B). The metacarpal head was divided into palmar, ulnar, dorsal, and radial quadrants. MCH metacarpal head, PsA psoriatic arthritis, Ctrl healthy control

Full size image

Four PsA patients and 15 healthy controls had no enthesiophyte. In the PsA group, 150 enthesiophytes were identified, while only 80 enthesiophytes were observed in the PsA group. At MCH 2&3, the number (mean number per patient 2.4 ± 1.4 vs 1.3 ± 1.1, p = 0.000), height (PsA: 2.5 ± 1.6 mm vs control: 1.4 ± 1.3 mm, p = 0.000), and volume (PsA: 8.8 ± 7.0 mm3 vs control: 4.4 ± 4.9 mm3, p = 0.000) of enthesiophytes in PsA patients were greater than controls (Table 2). Enthesiophytes had a predilection for the palmar sites (Fig. 2B). In PsA patients, older age correlated with larger erosion and enthesiophyte volume (sTable 1).

Comparison of vBMD and microstructure at the second and third metacarpal head in PsA and controls

Table 3 summarizes the mean vBMD and microstructure at MCH 2&3. Significantly lower mean vBMD (average vBMD: − 6.9%, p = 0.007, Tb vBMD: − 8.8%, p < 0.001, pTb vBMD: − 7.7%, p = 0.001, mTb vBMD: − 9.8%, p < 0.001), average BV/TV (− 8.8%, p < 0.001), and average Tb thickness (− 8.1%, p = 0.003) were observed at the MCH 2&3 in PsA patients compared with control subjects (Table 3). Multivariable linear regression analysis revealed that older age and a higher CRP level were independently associated with lower mean vBMD (average, Tb, and mTb) and average BV/TV at MCH 2&3 (Table 4).

Table 3 Comparison of mean vBMD and microstructure at the second and third metacarpal heads in PsA patients and controls
Full size table
Table 4 Univariate and multivariable analysis for mean vBMD and microstructure at the second and third metacarpal head in PsA patients
Full size table

Discussion

The study quantitatively and simultaneously compares the bone structure and microstructure in the metacarpal head between PsA patients and age-, gender-, and BMI-matched healthy control. To achieve this aim, we quantified both enthesiophyte and bone erosion volume, and assessed vBMD and microstructure after exclusion of bone erosion and enthesiophyte. The methodology for measuring erosion volume was modified for the quantification of enthesiophyte volume in the current study, and we have shown that this was feasible and reliable.

In line with the only study addressing structural bone damage in PsA patients compared with control subjects [8], we found larger bone erosion and enthesiophyte volume and more enthesiophytes in PsA patients than control subjects. Similar to previous study [7, 8], enthesiophytes are more common in PsA compared with control, supporting the hypothesis that PsA is predominantly an enthesitis-driven disease. Unfortunately, only a small number of HR-pQCT studies showed that IL-17 inhibition might halt the progression of both erosion and enthesiophytes [10]. The anatomic distribution of bone erosion and enthesiophyte was consistent with previous HR-pQCT studies for patients with PsA and psoriasis patients without PsA [4, 7].

Contrary to the only other HR-pQCT-based study comparing PsA patients and control subjects, which did not show any differences in bone microarchitecture between PsA patients and control subjects [33], our study found that mean average vBMD, peripheral and medullary trabecular vBMD, and trabecular bone volume fraction at MCH 2&3 were significantly decreased in patients with PsA compared with healthy control subjects. This discrepancy may be related to a shorter PsA disease duration (4.6 ± 6.4 years) of the previous study as well as the methodology used [33]. While erosions were excluded during analysis in the previous study, enthesiophytes were not excluded. The inclusion of enthesiophytes would clearly increase perceived cortical bone density. In addition, the previous study deployed a technique based on contouring the metacarpal head to adapt the number of slices rather than using a pre-determined, fixed number of slices as in the present study [11, 33].

In patients with PsA, this study suggests that inflammation might aggravate trabecular bone loss in the intra-articular metacarpal bone. Data from this study demonstrated that in patients with PsA, deterioration in average vBMD and predominantly trabecular vBMD (and trabecular bone volume fraction which was derived from trabecular vBMD) were significantly associated with an increased CRP level, which provided preliminary evidence to suggest that inflammation may contribute to bone loss in PsA patients. However, multivariable analysis did not find any association between the CRP level and trabecular number and separation in our PsA cohort, which is consistent with our previous study in a group of patients with RA [34]. Whether and how inflammation accelerates loss of trabeculae in PsA would need to be addressed in future studies. Similarly, we did not observe any association between CRP level and the volume of enthesiophyte and erosion. Duration of PsA was reported to be significantly correlated with the burden of enthesiophytes and erosions in a previous study which included patients with a shorter disease duration (mean PsA duration of 6.4 years compared with 13.1 years in the current study) [8]. These bony damages reflect accumulating mechanical and inflammatory damage in the joints. The inflammatory burden may not be captured in a snapshot provided by the single CRP level. The effects of inflammation accelerating bone damage may be more prominent in patients with shorter disease duration and therefore even disease duration was not identified as an explanatory variable associated with bone damages in our cohort.

Our findings revealed decreased average and trabecular vBMD but not cortical vBMD in metacarpal head in PsA patients, which might be explained by the absence of anti-citrullinated protein antibody (ACPA) in PsA. In ACPA-positive healthy individuals compared with ACPA-negative healthy individuals without clinical signs of arthritis, bone loss tends to affect the cortical but not trabecular compartment of the metacarpal head [35]. Similarly, compared with healthy control subjects, ACPA-positive RA patients had significantly reduced cortical and trabecular vBMD [11, 33], though only decreased trabecular vBMD is seen in RA patients with fewer ACPA-positive patients [36].

There are a few limitations to our study. First, almost one-fifth (19.5%) of PsA patients with joint destruction were excluded from structural and microstructural analysis as hand positioning and image analysis can be difficult in patients with severe joint deformity. Second, the cross-sectional design could only establish an association between clinical characteristics and bone damage, but not a causal effect relationship. Future longitudinal study is needed to confirm the effect of inflammation on structural and microstructural bone change.

In conclusion, the burden of bone erosions, enthesiophytes, and trabecular bone loss is increased in PsA patients compared with matched healthy controls. PsA patients had a predominant deterioration in trabecular microstructure in the metacarpal head which was related to ongoing disease inflammation.