Imaging technologies for preclinical models of bone and joint disorders

Overview of CT technology

The development of dedicated imaging equipment for small animals and, in particular, the implementation of computed tomography (CT) have revolutionised the use of animal models in musculoskeletal research, becoming the gold standard for evaluation of bone morphology and micro-architecture in animal models [3, 4]. While histomorphometric assessment has been extensively used as the main standard for investigating bone architecture, the development of 3D imaging techniques such as CT have provided an accurate non-invasive tool for directly measuring bone architecture. Indeed, since the development of clinical CT, the examination of small animals for research using purpose built CT has rapidly advanced providing high-quality resolution and fast reconstruction and assessment protocols for preclinical applications [5–7].

Micro-CT uses X-ray attenuation data acquired at multiple viewing angles to reconstruct a 3D representation of the imaged specimen, characterising the spatial distribution of the material density [8, 9]. Currently available micro-CT scanners can achieve high resolution with an isotopic voxel size of as low as a few micrometres (down to 5 μm; although new generation on nano-scanners can go down below 1 μm) [10]. There are, broadly speaking, two different micro-CT construction systems, one type in which the examined object is placed in the centre and the X-ray detector and radiation source is mounted in a gantry that rotates around it; in this system, the geometrical magnification is defined by the source-detector distance. This set-up is the one most commonly adapted for animal scanners. In the second type of CT scanner, the object is rotated within the course of the X-ray beam and the set-up allows the free positioning of the sample between the detector and the source, allowing the adjustment of the magnification level. This second construction is more often utilised in ex vivo custom-built systems. There are also differences on the beam geometry of the X-ray source used. Images can be acquired by using either a fan-shaped beam in which data are acquired through dynamic acquisition plane by plane or by a cone beam - this is also called 'volume-CT' [11] where the scanned subject is captured completely (based on the axial extent of the CT field of view) in one rotation, speeding up the imaging process. Furthermore, micro-CT systems can be fitted with a flat-panel-based detector system with slip ring technology that allows for very high-speed data collection [12]. The rapid acquisition times come at the expense of compromising spatial resolution, but this may be justifiable for in vivo applications that require rapid scan times such as perfusion imaging and high-throughput imaging. Similarly, tissue contrast enhancement can be induced by using a dual-energy X-ray computed tomography method in which the projection data are acquired by using two different X-ray spectra [13]. Figure 1 illustrates the different micro-CT systems technologies.

Current availability of multi-modality imaging platforms which can provide integrated positron emission tomography (PET)/single-photon emission computed tomography (SPECT)/computed tomography (CT) imaging and analysis are extremely useful for co-registering images within the same gantry, facilitating a multi-modal imaging approach within the same animal in vivo at the same time points during disease evolution [14]. Moreover, rapid image acquisition can also be facilitated by gating signal acquisition to cardiac and respiratory cycles by utilising a high-speed shutter system that allows image times as short as 10 ms [15].

Micro-CT imaging applications of preclinical models for bone and joint disorders

Micro-CT is extensively used to investigate the structure and density of bone in rodents. Micro-CT has high spatial resolution and contrast for imaging mineralised tissues and the ability to qualitatively and quantitatively assess 3D bone structures. This allows measurements of trabecular morphology such as thickness and separation. Micro-CT has been used for a wide range of bone studies, including bone anatomy and density to assess bone repair during fracture healing [16]; bone resorption, remodelling and regeneration [17]; bone neoplasm and metastases [18] and bone changes influenced by metabolic disorders, e.g. osteoporosis [19] and the characterisation of skeletal phenotypes from different mouse strains [20]. Figure 2 provides an example of different imaging acquisitions of various mouse bones used for skeleton phenotypes and bone repair studies.

The high isotropic resolution can also provide good information on trabecular bone spatial orientation patterns, density, and geometry and growth plate morphology. While much finer detailed imaging can be achieved in ex vivo samples, micro-CT imaging has proven very valuable for in vivo longitudinal studies. At least a 100 μm isotropic spatial resolution can be effectively achieved within a safe and reasonable acquisition time through longitudinal in vivo studies [21]. Some studies have reported in vivo isotropic resolutions down to 15 μm in serial imaging sessions using rats (seven sessions of 10 min acquisitions with an absorbed dose of 0.5 Gy-CTDI) [22], but there are concerns about the amount of ionizing radiation delivered during repetitive in vivo scanning. This radiation may introduce unwanted effects on the tissues or processes of interest or have an adverse effect on animal welfare [23].

Micro-CT imaging has also proven to be very valuable in investigating morphometric changes in joints in osteoarthritis (OA) animal models [24–26]; it has been successfully applied to study changes in the subchondral bone architecture in excised osteoarthritic knees in collagenase-induced OA mice [27] and in surgical destabilisation models [28]. Figure 3 shows micro-CT athrograms of mouse knees from the surgical destabilisation of the medial meniscus (DMM) model. Recently, Stok and collaborators [29], by imaging excised knees ex vivo, showed how changes in subchondral bone seem to be inversely correlated to the ongoing degenerative changes in the articular cartilage in STR/ort mice (naturally occurring OA). Indeed, micro-CT imaging provides quantitative and qualitative 2D and 3D assessments of bony structures in the osteoarthritic joint such as subchondral bone morphology and bone mineral density, trabecular bone patterns, meniscus morphology, heterotopic ossification and subchondral cyst formation [30]. Figure 3 shows micro-CT athrograms of mouse knees in C57BL/6 mice, 8 weeks after surgical destabilisation of the medial meniscus to investigate changes in the subchondral bone (Tremoleda et al., unpublished data). Interestingly, such technology has also been successfully incorporated to monitor the progression of OA in longitudinal studies in rats [31].

One of the major challenges for imaging osteoarthitis models is that the micro-CT has low sensitivity for soft tissue; hence, compromising the visualisation of the degenerative changes in the articular cartilage. Contrast agents can be used to enhance the contrast resolution of CT imaging, and there have been a few attempts to indirectly visualise the cartilage morphology using these agents in ex vivo samples [32, 33]. To compensate for the poor radio-opacity of the cartilage, tissues to be imaged are equilibrated with an ionic contrast agent which is taken up by the cartilage matrix and distributed inversely to the density of the negatively charged glycosaminoglycans (GAGs). This contrast-enhanced CT technique which is based on the detection of the equilibrium partitioning of an ionic contrast agent (EPIC-micro-CT) provides good 3D characterisation of the articular surface morphology and ligament insertions in situ in ex vivo rabbit knees [31]. This technique allows for the monitoring of surface contours and healing processes in injured articular cartilage in defect repair studies as well as changes in cartilage thickness. Changes to other anatomical features such as ligaments and non-calcified menisci can also be detected. While the use of this methodology has great potential for assessing cartilage degeneration in in vivo longitudinal studies, the thin layer of articular cartilage and the small joint space in mice combined with diffusion variability between contrast agents remain challenges for accurate quantification of signal change.

Overview of the MRI technology

Magnetic resonance imaging (MRI) is a non-ionizing 3D imaging technique that has advantages over other methods that depend on ionizing radiation such as CT, SPECT and PET. It remains one of the main imaging applications of choice for assessment of musculoskeletal tissue structures such as tendons, cartilage, menisci and ligaments in the clinical setting and has also being successfully tested in animal models (mice [34], rats [35] and rabbits [36]).

MRI technology uses the magnetic properties of atoms and molecules of tissues to be imaged and their interactions with both a large external magnetic field and radiowaves. The proton ¹H is the nucleus mostly used for anatomical imaging because of its abundance of soft tissue structures where there is a high water content. MRI technology utilises a powerful magnetic field (non-ionizing radiation) to align the nuclear magnetization of hydrogen atoms, and then, radio frequency (RF) pulses (through RF coils) are used to systematically alter this alignment, causing the hydrogen nuclei to produce a rotating magnetic field that is detectable by the scanner. The application of radio frequency pulses rotates the magnetization by 90°. Then, the magnetization returns to its initial value, and the rate at which the magnetization decays away is characterised by two relaxation times named as T1 and T2. These signals are then digitalised and processed to build up enough information to construct an image of the targeted body area. It is this relationship between field strength and frequency implemented onto the hydrogen atoms within the tissue that allows the use of nuclear magnetic resonance for imaging.

MRI applications for imaging preclinical models for bone and joint disorders

MRI assessment of bone structure has remained a challenge due to its composite biomaterial characteristics. Bone is made up of an organic substrate (mostly collagen type I approximately 40% by volume) and mineral crystals of hydroxyapatite (approximately 45%). The remaining volume is occupied by water (approximately 15%). Due to this limited water composition, proton imaging of bone remains challenging. Advanced micro-MRI systems have been used successfully to image trabecular bone in rats in vivo [37]. These methods rely on indirectly imaging the trabecular bone structure that appears as a signal void surrounded by high-intensity signal from the fatty bone marrow. Because of the small trabecular thickness in small animals (50 to 100 μm in rats; 40 to 60 μm in mice), the resolution requirements are more stringent with a high risk of overestimation of the trabecular thickness depending on the volume of the trabeculae and therefore an adverse signal-to-noise ratio - reinforcing the need for longer acquisition times.

In the preclinical setting, the acquisition of high-resolution MRI within an acceptable time frame remains an important drawback due to the relatively small size of anatomical structures in rodents; long acquisition is needed to generate sufficient isotropic (3D) resolution with a small-bore MRI and with high field strength. This needs to be performed with the animal under general anaesthesia.

To overcome the limited detection sensitivity and to increase the signal-to-noise ratio, enhanced pulse sequences have been developed. These allow adequate sample volumes to be scanned with scan times maintained within the limits of in vivo tolerance. 3D gradient-echo sequences allow for a shorter TE than spin-echo or fast spin-echo sequences, allowing for faster data acquisition compared to fast spin-echo sequences, and therefore allow more data averaging within a fixed duration of scan time [38].

One of the most critical aspects of preclinical MRI is the radio frequency coil which drives both the signal excitation and reception. Because the coil's sensitivity increases as its volume decreases, animal coils are smaller than humans, improving the signal gain significantly while still scanning the anatomical area of interest. Takahashi and collaborators [39] successfully applied MRI technology using a 3D spin-echo pulse sequence in conjunction with spectroscopy to investigate the longitudinal changes in trabecular bone architecture in response to pharmacological interventions in rabbits. Another promising approach is based on the use of ultra-short echo-time (UTE) radial acquisition sequences with hard pulse excitation, which can detect subtle changes in mineral phosphorus and water contents of cortical bone (³¹P and ¹H), as applied for investigating changes in bone mineralization in ex vivo bones from oestrogen deficiency-induced ovariectomised rats [40]. The quantification of water content through UTE-MRI provides a good technique for measuring bone's hydration state in situ with great potential for investigating bone metabolic disorders in vivo. Figure 4 displays the magnetic resonance (MR) images of rat and mice knees obtained ex vivo in a 9.4-T MRI scanner using a 3D gradient echo and fast spin-echo multi-slice acquisitions, respectively.

MRI is extensively used in clinics for assessment of articular cartilage in joint disorders. As in clinical MRI, the MR sequence that best delineated the cartilage from the surrounding tissues, i.e. 3D fat-suppressed spoiled gradient echo, has also been successfully applied for depicting changes in the rat knee of OA models [41]. Moreover, some studies in ex vivo rat knees have measured the spatial distribution of T2 relaxation times as a function of the water content and the collagen ultrastructure correlating with the structural integrity of the articular cartilage [42]. These measurements may permit early detection of changes of cartilage matrix integrity, which could lead to osteoarthritis, and thus would be a relevant preclinical model for the development of clinical treatment before any cartilage morphological alterations occur.

Another commonly used approach is the use of contrast agents such as gadolinium (Gd-DTPA, gadopentetate dimeglumine or Magnevist™ Schering, Berlin, Germany). The Gd-DTPA penetrates into cartilage and distributes preferentially to areas of the cartilage in which the GAGs are low. Subsequently, this will induce a decrease in T1 relaxation that reflects the Gd-DTPA concentration and thus the distribution of GAGs. This clinically validated technique is referred to as delayed gadolinium-enhanced MRI of cartilage and has also been investigated in preclinical models [43]. Another application of contrast-enhanced MRI is the assessment of blood flow perfusion and permeability which has important physiopathological relevance as reduced bone blood perfusion is also an indicator of disease progression and severity. Contrast agents like gadolinium (Gd) have been applied for imaging bone blood flow using dynamic contrast-enhanced MRI techniques in animal models of OA [41]. These techniques may be very valuable to assess the healing process and estimate the risk of avascular necrosis.

Finally, ongoing developments in preclinical MRI technology, with the implementation of phased-array coils which by contrast with the traditional single-channel surface coil can increase the signal-to-noise ratio, thereby providing superior image sensitivity covering a specific field of view, have been successfully used for imaging the knee joint in rats [44].

PET/SPECT systems technology

Bone scintigraphy is extensively used as one of the most common diagnostic techniques to investigate bone lesions and metastases in various musculoskeletal conditions and in diagnostic orthopaedic medicine. Further innovations such as single-photon emission computed tomography (SPECT) and positron emission tomography (PET) have allowed the acquisitions of whole-body images of the entire skeleton. They provide increased sensitivity for lesion detection and, importantly, a 3D localization of the radiation emitted by radionuclide imaging agents or biomarkers with very high detection sensitivity down to nano- or picomolar concentration.

The growth in clinical nuclear imaging applications has led to the development of SPECT and PET scanners dedicated for small animal imaging [45, 46]. These systems have to cope with the small size of rodents; hence, they have to achieve enhanced spatial resolution and high sensitivity for the targeted biomarker. The application of this technology in preclinical models has a significant scope for non-invasively studying dynamic biological processes at the molecular and cellular level. It provides good functional information through the detection of onset and progression of a given biological process, and can assist in the development of biomarkers and in measuring the effectiveness of new treatments.

SPECT systems record gamma rays directly after radionuclide emission. The system uses a gamma camera to acquire projections data that are detected using a parallel hole collimator. However, most of the preclinical SPECT scanners are equipped with multipinhole collimators to acquire high spatial resolution of foci of gamma-emitting tracers within the subject volume. Then, a tomographic reconstruction of the data is acquired yielding a 3D dataset that can then be manipulated to show any particular axis of the body. PET systems also detect gamma rays that are emitted by a biomarker tracer labelled to a positron emitter (F-18, C-11, N-13, O-15). These radio-nuclides emit positrons which cause two gamma photons to be emitted in opposite directions by annihilation with an electron. The two photons are accepted to be in coincidence if their incidence on a detector pair is within a predefined timing resolution window. The scanner detects these dual emissions 'coincident in time', providing a higher radiation on a given location and thus higher resolution images. This is in contrast to SPECT systems that rely on hardware collimation. The resolution of preclinical PET scanner lies in the range of 1 to 2 mm, but some state-of-the-art dedicated preclinical SPECT systems can provide better resolution capabilities (down to the sub-millimetre range) [43]. SPECT radiopharmaceuticals have a long half-life, and they are routinely produced for clinical nuclear medicine, making SPECT isotopes easily available and more cost-effective than PET tracers. Nevertheless, the high sensitivity of PET tracers and their integration as biomarkers with multi-capability applications makes them very well suited for small animal imaging. Tracers such as ¹⁸F-FDG, a glucose analogue, and ¹⁸F-FLT, a pyrimidine analogue are used as biomarkers of tissue metabolic activity and inflammation, and cell proliferation, respectively.

PET/SPECT imaging applications

Several studies have successfully reported the use of Tc-99 m-labelled diphosphonate compounds, e.g. methylene diphosphonate (MDP), hydroxymethane diphosphonate and hydroxyethylidene diphosphonate, to detect changes in bone turnover and cartilage composition in osteoarthritis models in rodents [47]. The co-registration of micro-SPECT/micro-CT images allows the detection of high Tc-99 m MDP uptake, depicting areas of high bone turnover, e.g. joints (knees, shoulders), spine and skull (Figure 5). SPECT imaging of articular cartilage in rodents has been reported using N-[triethylamonium]-3-propyl-[15]ane-N (NTP) bound to Tc-99 m. NTP has a high affinity for binding to negatively charged sulphate groups of the glycosaminoglycans, a major component of the cartilage matrix, and it has been successfully used for in vivo imaging of a cartilage tumoural model in rats [48].

For PET imaging, ¹⁸F-fluoride (NaF) is being extensively used for assessing bone metabolism, and it has a similar uptake mechanism to ^{99 m}Tc-MDP, being absorbed onto bone surfaces. This tracer diffuses well through capillaries reaching bone extracellular fluid where it is rapidly incorporated in the bone hydroxyapatite. Its negligible binding to plasma proteins, rapid blood and renal clearance with the high uptake after injection (40 to 60 min) provides a higher accuracy than SPECT imaging. Similar to MDP, it is taken up preferentially in malignant cancerous bone lesions such as sclerotic metastases and areas of altered osteogenic activity, reflecting an increase in regional blood flow and bone turnover [49] and for detecting bone micro-damage [50] in preclinical models.

¹⁸F-FDG is another PET tracer that can be used to assess inflammatory activity indirectly in bone. ¹⁸F-FDG is not specifically targeted to bone, but as a glucose analogue, it provides a sensitive complementary functional biomarker in defining areas of inflammation in the axial and appendicular skeleton. PET scanning with ¹⁸F-FDG has great potential for assessing fracture healing as it can provide a direct quantitative non-invasive assessment of the metabolic activity in the region of interest and therefore measure bone repair in fracture models [51], providing information on the treatment and prognosis of delayed fracture healing. ¹⁸F-FDG is also widely used in clinical oncology in musculoskeletal sarcomas, as the tracer is directly taken up into tumour cells and is often used to detect metastases. It has been used successfully to localise and quantify skeletal metabolic activity in the study of preclinical cancer metastasis in bone [52].

Fluorescence technology and applications

Fluorescence imaging relies on the detection of light emission of specific fluorophore when excited by appropriate wavelength energy. Several approaches have been described in vivo, including the use of non-specific dye tracking (e.g. for detection of non-specific inflammatory sites), the use of fluorophores tagged-antibody-based targets for specific cell molecules and/or metabolic pathways and the development of fluorescence reporter proteins (e.g. GFP) for the detection of gene expression. Targeted fluorescence imaging of arthritic joints has shown an increase in fluorescence signal in the arthritic joints in disease models. This is in part associated with increased blood perfusion and vascular leakiness, which are recognised to occur in areas of inflamed synovium [53]. In vivo targeting of specific key components identified during the inflammatory process such as the labelling of the F4/80 antigen on macrophages has been achieved [54]. Similarly, other components of the inflammatory cascade can be traced in vivo. For example, E-selectin-targeted in vivo imaging is a quantifiable method of detecting endothelial activation. E-selectin or endothelial adhesion molecule-1 is a 115-kDa glycoprotein induced on endothelial cells in response to pro-inflammatory cytokines involved in rheumatoid arthritis such as interleukin-1β and tumour necrosis factor alpha (TNF-α) [55]. E-selectin has been well validated as a potential biomarker of disease activity in rheumatoid arthritis [56].

Anti-E-selectin antibody labelled with NIR fluorophore has demonstrated specific signal increases compared to control antibody in a mouse model of paw swelling induced by the intra-plantar injection of TNF-α. This has also been demonstrated in acute collagen-induced arthritis, a widely used model of RA. Mapping of fluorescent E-selectin-specific signal in difference to the signal returned from fluorescently labelled control antibody is demonstrated in Figure 6 [57]. Utilising fluorescent compounds to label antibodies allows for a more targeted and specific approach to identify different components of disease, quantify levels of inflammation and assess the effect of novel therapies. Similarly, activity-based fluorophores where signal becomes amplified at the site of inflammation have also been used successfully in mouse models of arthritis demonstrating increases in signal intensity in the injured joint [58]. This type of approach may be particularly useful for optical imaging in OA, since local perturbations in proteolytic activity may be small and require amplification for adequate signal detection.

Fluorescence is also extensively used to image the skeleton, with a multitude of commercially available bone specific fluorophores (e.g. OsteoSense™ from PerkinElmer, Waltham, MA, USA), fluorescently labelled alendronate from Caliper^®LifeSciences (Hopkinton, MA, USA) and BoneTag™ from LI-COR^®Biosciences (Lincoln, NE, USA) [59], which are incorporated in the calcified bone matrix at spots with high bone turnover and are therefore good indicators of bone remodelling in sites of bone damage such as fractures and cancer-induced osteolytic/blastic lesions [60].

Bioluminescence technology and applications

Bioluminescence imaging is based on the detection of photons produced after an enzymatic reaction in living organisms. It is commonly used in reporter gene assays, where the promoters of genes under study have been linked to the luciferase gene, which in the presence of its substrate (luciferin) generates light. The most commonly used luciferase enzyme used for bioluminescence imaging purposes comes from the firefly (emitting light around 560 nm). This approach requires the luciferase enzyme to be transfected into the living cells and must be tightly controlled with the appropriate promoter. Also, a substrate needs to be delivered where the enzyme may be expressed. This may be challenging for certain organs and may affect the dynamics of the emission response due to different pharmacokinetic profiles related to altered tissue properties. Despite these conditions, its high specificity, low background signal, high signal-to-noise ratio and the ability to study gene expression over the lifetime of the animal are reasons why this technology is used extensively for in vivo imaging studies. Bioluminescence imaging has been successfully utilised to study the development of arthritis in transgenic mouse models [61], and because of its high sensitivity, it is also used widely in bone metastasis studies, to monitor the development and progression of luciferase positive tumour cell lines within the whole body [62].