HR-pQCT in vivo imaging of periarticular bone changes in chronic inflammatory diseases: Data from acquisition to impact on treatment indications

Abstract

Imaging is essential for the assessment of bone and inflammatory joint diseases. There are several imaging techniques available that differ regarding resolution, radiation exposure, time expending, precision, cost, availability or ability to predict disease progression. High-resolution peripheral quantitative computed tomography (HR-pQCT) that was introduced in 2004 allows the in vivo evaluation of peripheral bone microarchitecture and demonstrated high precision in assessing bone changes in inflammatory musculoskeletal diseases. This review summarizes the use of HR-pQCT for the evaluation of the hand skeleton in inflammatory joint diseases. We conducted a review of the literature regarding the protocols that involve hand joints assessment and evaluation of bone changes as erosions and osteophytes in chronic inflammatory diseases. Apart from measuring bone density and structure of the radius and the tibia, HR-pQCT has contributed to assessment of bone erosions and osteophytes, considered the hallmark of diseases as rheumatoid arthritis and psoriatic arthritis, respectively. In this way, there are some conventions recently established by rheumatic study groups that we just summarized here in order to standardize HR-pQCT measurements.

Keywords:

• Arthritis
• bone erosion
• bone spur
• computed tomography

Introduction

Imaging is a crucial technology to evaluate bone. Several different techniques have been developed to assess bone in vivo that differ in methodology (being based on X-ray, magnetic resonance imaging or ultrasound), assessable dimensions (producing either two- or three-dimensional [3D] images), spatial resolution, radiation exposure, acquisition time, parameters assessed, region of interest (ROI) analyzed, accuracy, precision, cost, availability and ability in predicting clinical outcomes [1,2]. One of the most exciting advances to assess bone microarchitecture over the past 10 years has been the introduction of high-resolution peripheral quantitative computed tomography (HR-pQCT), an innovative and advanced imaging technology available since 2004 [3] which has been used to study bone micro-architecture in humans in vivo [4–8].

In this review, the use of HR-pQCT in assessing the hand skeleton in patients with inflammatory joint diseases is described. The characteristics of HR-pQCT image acquisition, reconstruction and evaluation as well as the methods to evaluate local bone changes such as erosions and osteophytes are reported, showing the potential of HR-pQCT in rheumatology research and potential future use in routine clinical practice.

Methods

Search strategy and selection criteria

Articles included in this review were searched and limited full-text English language articles of PubMed with combination or isolate of following items: ‘HR-pQCT’, ‘bone erosions’, ‘osteophytes’, ‘hand bone destruction’, ‘bone proliferations’, ‘imaging’, ‘metacarpal joint space’, ‘rheumatoid arthritis’, ‘psoriatic arthritis’, ‘osteoarthritis’, ‘SPECTRA group’, ‘OMERACT’, ‘inflammatory disease’. The search strategy was done from 31 March 2019 until 30 September 2019.

Study selection

There were selected the most important papers regarding acquisition of images, standardized methods of evaluation from study groups (SPECTRA and OMERACT), as well as different methods that might be use to peripheral bone erosion and bone spur analysis. Moreover, some literature evidences of its use in treatment follow-ups were also included.

Main outcome variables

All analyzed topics reviewed here, they were chosen according to authors specialists in HR-pQCT technique, after intensive study and many previous publications, in order to summarize all information in a single article.

Risk of bias assessment

The assessment here focused only in hand’s peripheral bone changes assessment, and this review did not include the whole analysis that the machine is capable to do, as volumetric bone mineral density in tibia or even in radius.

Analysis

All literature selected was analyzed and summarized here without any statistical software.

Results

High-resolution peripheral quantitative computed tomography

HR-pQCT is part of different noninvasive techniques for three-dimensional (3D) assessment of human bone in vivo, which is important for improving fracture prediction and prevention [9]. In fact, dual X-ray absorptiometry (DXA), the widely used method to evaluate bone mass through areal bone mineral density (BMD), which represents a technique to assess osteoporosis and predict fracture risk accepted by the World Health Organization, is not able to assess bone microstructure or bone geometry. Moreover, BMD does not entirely explain fracture risk, because most of the fractured patients are osteopenic rather than osteoporotic in DXA measurements [10,11].

HR-pQCT is currently available from a single manufacturer (XtremeCT; Scanco Medical AG, Brüttisellen, Switzerland) and it was developed based on experimental micro-CT technology, considered the gold standard for bone microstructural analysis [12,13]. The device was designed to assess accurately bone microstructure at the micrometer level resembling a virtual bone biopsy, evaluating a voxel size of 82 µm with the first-generation device [14,15] and 61 µm with the second-generation device [16]. Aside the assessment of bone microstructure and geometric parameters, HR-pQCT allows quantification of volumetric bone mineral density (vBMD) and mechanical properties of bone in the upper and lower appendicular skeleton (distal radius/tibia) by micro-finite element analysis, assessing independent effects in cortical and trabecular bone compartments [7,17,18]. Trabecular number, trabecular thickness and trabecular separation [14] and cortical thickness and cortical porosity [19] are some of the bone structure parameters that can be assessed by HR-pQCT. The ability to study trabecular and cortical microarchitecture in humans makes this method interesting for evaluating therapeutic response to antiresorptive and anabolic agents, especially if considering that these agents may have site-specific skeletal effects [20]. Moreover, bone strength can be assessed at these peripheral sites using micro-finite element analysis.

Substantial efforts have been made to apply HR-pQCT not exclusively in studying bone diseases, but also to analyze bone structure in arthritis [21–24].

Imaging acquisition

Standardized imaging and image assessment is essential to accuracy and reproducibility for HR-pQCT, as for any other medical imaging device. To overcome motion artifacts, the respective limb is immobilized in a carbon fiber cast that is fixed within the gantry of the scanner [25,26]. Motion grade system is applied, where only images with high quality (scores 1–3) are evaluated, while low-quality images (scores 4–5) are usually excluded from the analysis [27], resulting in 82µm isotropic voxels images [28].

Aside imaging the distal radius in arthritis patients in order to measure the bone architecture at a site, which is not directly related to the joint, wrist, metacarpophalangeal (MCP) and proximal interphalangeal (PIP) joints (2nd and 3rd digits) are usually evaluated, because of the high frequency of bone changes in these joints. In contrast to the radius (or the tibia), in which the nondominant side is assessed [29], joint evaluation usually happens in the dominantly affected hand. Similar to the distal radius, a scout view is obtained, that is, a two-dimensional dorsal-palmar X-ray scan, so that the operator can define the tomographic region of 2nd and 3rd MCPs and PIPs joints, for the 3D measurement [30]. Standardized acquisition of images have been elaborated by the SPECTRA group [31].

A scout view is acquired, and both 2nd/3rd MCPs and PIPs are assessed [32]. The reference line is placed at the midpoint of the concave articular surface of the proximal phalanx. The scan extending 9.02 mm in distal direction, and 18.04 mm in proximal direction (27 mm length), which represents 330 slices in total, with a scan time being around eight minutes. In PIP evaluation, the reference line is positioned to the upper margin of the proximal phalangeal head (18 mm length), extending 9mm in both directions, which means that PIPs are imaged with two stacks (220 slices), which takes five minutes (Figure 1). The total equivalent radiation exposure is minimal (24µSv) if all sites (tibia, radius, MCP and PIP) are evaluated, which is five times less than a conventional chest X-ray [32,33] and 1000 times less than the recommended radiation yearly (radiation dose limit of 20mSv/year) [34].

Figure 1. Scout view (radiographs) generated by HR-pQCT for region of interest acquisition (region between the dot lines). The operator places the reference line (continuous line) at the midpoint of the concave articular surface at the base of the 2nd metacarpal head, extending 9.02 mm in distal direction and 18.04 mm in the proximal direction (A). For proximal interphalangeal joints (PIP) evaluation, the reference line is placed at the 2nd proximal phalanx, extending 9.02 mm in both directions (B).

Once the images have been acquired, a 3D reconstruction is made using software provided by manufacturer. After that, the images are exported as DICOM files, volume rendered and usually analyzed by the open source viewer OsiriX software, a program that use a threshold-based algorithm, installed on external workstations [30]. Then, the bone changes are identified by OsiriX as following: erosions are defined as juxta-articular breaks within the cortical shell, and osteophytes as bony protrusions emerging from the cortical bone shell, both must be seen at least in two consecutive slices and in two perpendicular plans. The MCP and PIP joints are divided into four quadrants, palmar (I), ulnar (II), dorsal (III) and radial (IV) allowing to define the spatial distribution of the lesions (Figure 2) [31].

Figure 2. Axial plane of 2nd metacarpal head HR-pQCT scan with the anatomic division in four quadrants: I (palmar), II (ulnar), III (dorsal), IV (radial).

Key structural lesions in arthritis and their evaluation by HR-pQCT

Erosions

Bone erosions are a central feature of rheumatoid arthritis (RA). Not surprisingly, RA has been chosen as the main disease to HR-pQCT studies. Accuracy is crucial to measure erosion volume [35], especially because bone erosions characterize disease severity and are associated with impairment of physical function and increased mortality in RA [36,37]. Techniques for manual and semi-automated segmentation of erosions have been described [38,39]. The manual approach for erosion volume calculation using a semi-ellipsoid formula had been used in many studies [40]. Nevertheless, manual segmentation underestimates erosion size especially in case of large erosions, which can be better quantified by semi-automated segmentation algorithms [41].

Two semi-automated techniques for erosions segmentation have been developed that precisely assess the volume of the erosion: a previously described semi-automated 3D Medical Image Analysis Framework software (MIAF) [39,42] and a ‘modified Evaluation Script for Erosions (mESE)’ tool, that uses the native ISQ image data format provided as analysis software in XtremeCT scanners. mESE uses standard segmentation algorithms implemented for segmenting radius or tibia, applying a slice by slice segmentation and using fixed thresholds, which explains that the segmented erosion often seems to be slightly removed from the cortical break. Differently from mESE, MIAF applies a periosteal surface segmentation through a sophisticated 3D segmentation algorithm that virtually ‘closes’ the cortical breaks, restoring the original cortical surface, and then depicts the erosion volume. After that, the operator places a seed point in each cortical break, and the erosions are segmented using the level-set method stopping at surrounding trabecular bone while the periosteal contouring works as edge at the cortical break [41]. Both software algorithms define the volume of erosions in mm³, but often the operator still has to decide whether regions with low absorption are still part of the erosion or are related to a cyst or other bone marrow changes [41]. MIAF also provides BMD values in four different layers around erosion and in total MCP head (Figure 3). However, none of the techniques automatically identifies the location of an erosion; so far, this remains the task of the expert rheumatologist or radiologist [41].

Figure 3. HR-pQCT scan of a 2nd metacarpal head of a patient with rheumatoid arthritis in an axial plane. (A) Erosion at quadrant IV; (B) MIAF bone segmentation as a continuous line around metacarpal head and erosion segmentation as a line inside metacarpal head; (C) Four different layers of bone mineral density (BMD) values around erosion (lines around bone erosion) provided by Medical Image Analysis Framework (MIAF); (D) BMD of total metacarpal head (MCP) analyzed by MIAF as a continuos line at the edge of metacarpal head and without the area around the bone erosion.

Bone proliferation

Proliferative bone changes (enthesiophytes), typically found in PsA and SpA, and that might occur independent of bone erosions, may also be found as a feature of secondary osteoarthritis in RA (osteophytes) [43], but here it is usually linked to bone erosions (Figure 4).

Figure 4. HR-pQCT scan of 2nd metacarpal heads. (A) and (B) - rheumatoid arthritis patient: in (A) axial slice showing close association between osteophyte (white arrow) and erosion (white head arrow); (B) the same bone lesions in a 3D reconstruction, erosion (black arrow) and osteophyte (white arrow) represented. (C) and (D) - psoriatic arthritis patient: (C) axial slice demonstrating enthesiophytes (white arrows) without erosions, and (D) the same bone proliferations (white and black arrows) in a 3D image.

Similarly, to erosions, HR-pQCT allows precise assessment of bony proliferation. However, the accurate measurement of bone protrusion depends on the used method to evaluate the precise volume of osteophytes.

In some HR-pQCT studies, osteophytes are analyzed semi-quantitatively through four grades (0–3), based on the maximum distance between the ‘original’ and new cortical surface [30,38]. However, a quantitative method was recently developed to assess osteophytes [43].

Joint space narrowing

The first study that demonstrated HR-pQCT capability of quantifying joint space width (JSW) at the MCP and PIP joints by HR-pQCT was from Barnabe et al. [32]. The joints were recognized in the scan and periosteal surfaces were automatically identified through a modified contouring algorithm [44]. After, a 3D image analysis established the joint space width using Image Processing Language. Here, the joint space was mapped in a 3D setting, characterizing joint morphology, as well as the distribution of widths was described using histograms [25].

To standardize joint space width measurement, a study comparing these fully automated methods previously published to develop a consensus were performed. The study was running within different RA spectrum (early to late), with scan/rescan procedures (positioning/repositioning), besides using two machines available: XtremeCT and XtremeCT II. It was shown that the evaluation of JSW by all 3D automated methods is excellent, comparable, with high reproducibility and is minimally sensitive to positioning [45]. HR-pQCT joint space assessment was also compared to conventional radiographs assessed using the joint space domain of the van der Heijde-modified total Sharp score system using the OMERACT 2.0 filter [46] and outcomes were validated following the OMERACT Working Group [47]. Although HR-pQCT provides 3D visualization of joint space, which is not possible with conventional radiography, longitudinal studies are still necessary to understand the sensitivity of HR-pQCT in detecting changes in the joint space width.

HR-pQCT studies detecting bone changes in arthritis

Chronic inflammatory arthritis is associated with substantial bone damage. For instance, RA is characterized by chronic inflammation of the synovial membrane that leads to destruction of periarticular bone and impaired joint function [48–50], as evidenced by bone erosions, periarticular osteopenia, and generalized osteoporosis [51]. RA is also appreciated as an independent risk factor of secondary osteoporosis [52,53]. In RA, there is substantial interest in using sensitive imaging tools to diagnose structural bone damage and monitor the progression of structural bone damage during anti-rheumatic treatment [54,55]. Conventional radiographs are considered the gold standard for detection of bone erosions in daily clinic practice, but they have limitations in detecting small erosions, as well as to characterize the nature and extent of bone damage in RA [56,57]. Other imaging methods, such as ultrasound and magnetic resonance imaging are used in RA, however mainly emphasizing on the detection of inflammatory lesions [58] while having limitations to visualize bone [59,60]. Stach et al. first applied HR-pQCT in RA patients [30] and visualized bone erosions in metacarpophalangeal (MCP) joints with great accuracy and in three dimensions. Shortly thereafter, Fouque-Aubert et al. assessed hand bone loss by HR-pQCT in RA patients, evaluating vBMD and microarchitecture [33].

Subsequent studies focused on differences of bone disease in various forms of arthritis such as RA and psoriatic arthritis (PsA). While, osteophytes in RA typically resemble secondary osteoarthritis, lesions in PsA typically affect entheseal insertion sites and sometimes are very extensive affecting the entire surface of bone (periostitis). Moreover, erosive lesions in RA are larger in size than those found in PsA, and RA shows U-shaped erosions, PsA typically has Ω-shaped lesions, with a narrower neck and wider base [38]. Also, with respect to erosions distribution, in RA the radial side of the MCP head is typically affected [61,62], while in PsA, involvement of phalangeal base and palmar surface is also common [38]. Concerning PsA and osteoarthritis (OA) anabolic bone changes (bone spurs), the major finding in both diseases, the literature has shown that number and size of bone spurs were comparable. However, the lesions in OA primarily emerged from the cartilage affecting palmar and dorsal sites, while in PsA they were predominantly related to entheseal sites including phalangeal bases, which are spared in OA [63].

Comparative analyses between changes in the distal radius and joint have also been made in RA patients. Hence, there is a significant correlation between vBMD, as well as microarchitecture parameters between metacarpal head and distal radius [64].

Moreover, HR-pQCT studies in RA have shown that bone changes are related to autoantibodies and occur very early in the disease course. Kleyer et al. showed that individuals with anti-citrullinated protein antibodies (ACPA), who have not yet developed RA show cortical bone changes characterized by the appearance of so-called cortical micro-channels (CoMiCs) suggesting that initial bone changes precede the onset of clinical RA [65,66]. Pre-disease bone changes have also been described in PsA. Thus, a cross-sectional study on psoriasis patients without any current or past symptoms of PsA showed that bone changes such as small enthesiophytes (Deep Koebner Phenomenon) can be detected in psoriasis patients while being virtually absent in healthy individuals [67].

Comparison of HR-pQCT imaging with other imaging modalities

Plain radiographies are widely used for detection of structural bone damage in inflammatory arthritis. However, because of the high resolution of the HR-pQCT, the sensitivity in detecting bone changes in small joints is clearly higher with this technique. Barnabe et al. showed that the probability of agreement between both methods is 67.5%, with a large number of erosions being detected only by HR-pQCT scans [32].

Another study compared HR-pQCT images with ultrasound (US) results. The data showed that most of bone erosions detected by US are true erosions. US sensitivity for erosions was high, and a good correlation was also found between severity of bone erosions assessed by US and HR-pQCT. However, the specificity of US in detecting bone catabolic changes is markedly lower, showing that minor erosions identified on US do not always represent a genuine cortical break [68].

Magnetic resonance imaging (MRI) is an excellent technique to analyze inflammation in RA patients. Albrecht et al. studied 50 RA individuals and compared MRI erosions images, analyzed by RAMRIS scoring system, with HR-pQCT scans. Although MRI is usually not the imaging modality of choice to evaluate bone structure, it provides excellent opportunities to identify inflamed tissue in arthritis, including juxta-articular bone marrow, visualizing tiny signal changes that are indicative of erosions lesions. The authors showed that both methods are comparable in identifying bone erosions, however very small lesions may not be detected by MRI, despite a high correlation between the techniques in quantifying 3D size of erosions [40].

Scharmga et al. evaluated patients with RA and healthy subjects by radiographs, MRI and HR-pQCT [69]. Interestingly, structural damage on MRI or radiographs was significantly associated with the presence of erosions in HR-pQCT. Similarly, when bone marrow edema was present, more erosions were found in HR-pQCT. Last, Peters and colleagues showed that HR-pQCT might add information to MRI images regarding bone damage, since HR-pQCT could detect very small cortical interruptions in RA patients [70]. In this way, HR-pQCT arrives in clinical practice as a sensitivity image tool in detecting early peripheral bone damage in inflammatory bone diseases as rheumatoid arthritis, psoriatic arthritis and osteoarthritis, which might help to improve treatment in these conditions, in order to suppress radiologic progression as well as deformities and impairment of physical function.

Effects of anti-rheumatic therapy on HR-pQCT changes

The effects of anti-osteoporotic drugs analyzed through HR-pQCT parameters in radius and tibia have been studied in many trials [71–73]. Some studies have also been carried out in arthritis, especially with respect to erosion repair. An HR-pQCT study compared RA patients in methotrexate (MTX) monotherapy with individuals treated with tumor necrosis factor inhibitors (TNFi), and after a 1-year follow-up the authors found evidence of erosion repair within the TNFi population, demonstrated by decrease in depth of lesions and sclerosis at the base of erosions, but not in patients treated with MTX monotherapy [74]. Similar results were also obtained when analyzing RA patients treated with the interleukin-6 receptor (IL-6R) inhibitor tocilizumab. Treatment with tocilizumab was associated with an improvement in the width and depth of cortical breaks, while complete healing was rarely observed [61].

Another HR-pQCT study performed by Shimzu et al. [75] showed that anti-TNFα treatment completely prevented erosion progression and deterioration of bone microarchitecture within the first 3 months of treatment, while patients with low disease activity scores treated with MTX experienced continuous progression of erosive disease.

Recently, another HR-pQCT study compared the effects of tocilizumab monotherapy with the anti-TNFα adalimumab in combination to MTX on bone erosion repair in RA patients, at baseline and after 52 weeks. The authors found that the tocilizumab monotherapy group achieved more repair of bone erosions than adalimumab plus MTX [76].

In PsA patients, longitudinal analyses of bone proliferations (enthesiophytes) progression were done by HR-pQCT. PsA patients treated with either methotrexate or TNFi showed that progression of new bone formation still occurred while the progression of erosions was stopped [77]. On the other hand, treatment with the IL-17 inhibiter secukinumab showed no progression of bone erosions and enthesiophyte formation in PsA [78].

Limitations

Although the HR-pQCT provides excellent opportunities to assess bone structure in joints of arthritis patients, there are still some limitations to be mentioned. Trabeculae parameters might therefore be underestimated at the resolution of 82 μm [79]. Image registration can bear some challenges, since HR-pQCT software compares slices by matching their total cross-sectional area, which is a challenge if bone size is affected by periosteal apposition. Another limitation of HR-pQCT is the fact that measurements of feet are still challenging due to the lack of a suitable cast. Besides, HR-pQCT is only available in some research centers, that is a great limitation.

Future possibilities

HR-pQCT is currently a research tool to study bone microarchitecture. Nonetheless this technique has the potential for wider clinical use as radiation dose is very low, short-time scan and the potential to evaluate bone with high resolution and accuracy, a further increase of resolution will improve the accuracy in assessing bone trabecular parameters. In addition, a reduction of scanning time is always attempted, seeking for lower radiation and motion grade artefacts. Furthermore, feet analysis with a suitable cast and proper scripts will be an important goal for future HR-pQCT studies.

Conflict of interest

None.