INTRODUCTION

Bone is a complex composite material made of a cellular and extracellular matrix. The extracellular matrix comprises of 40% organic components such as collagen type I, proteoglycan, proteins and cytokines, that contribute to bone structure and strength (Camozzi et al., 2010). The remaining 60% of the extracellular matrix is made from minerals including hydroxyapatite, calcium and phosphate, that help provide mechanical strength (Rusu et al., 2005). Both the cellular and extracellular matrices work together to maintain bone remodeling. Any disruption to these components will affect bone remodeling which will later lead to pathological bone conditions. The most common bone disease is osteoporosis. Osteoporosis is a progressive systemic skeletal disorder characterized by low bone mineral density (BMD), deterioration of the microarchitecture and susceptibility to fracture (WHO., 1994). It is a disease that causes bone loss and fractures which together lead to severe pain, deformity and in certain cases, secondary complications that result in death (Johnell and Kanis, 2006). As defined by the World Health Organization (WHO), osteoporosis occurs when BMD T-score is more than 2.5 standard deviation below the peak bone mass reference standard for young women (Nelson et al., 2002). The prevalence of osteoporosis is currently increasing globally due to the general increase in the population and in the proportion of aged individuals. This increase contributes to a higher economic burden due to the medical care expenditures.

Osteoporosis is classified clinically into primary and secondary osteoporosis. Primary osteoporosis refers to both bone loss due to sex hormone deficiency such as in post-menopausal women (type I) and bone loss due to the normal ageing process (type II) (McNamara, 2010). Secondary osteoporosis refers to bone loss that ensues as a secondary effect of other diseases or drug treatment. Post-menopausal osteoporosis is the most common form of the disease and is due to estrogen deficiency following menopause (Riggs et al., 2002). There are many factors that contribute to the reasons why women (75%) are affected by osteoporosis to a greater extent than men (25%). Firstly, women have a smaller skeletal size, lower bone mass (Nieves et al., 2005) and achieve a lower peak BMD compared with men (Avdagic et al., 2009). Secondly, women are prone to rapid bone loss due to estrogen reduction following menopause and thirdly, in almost every population, women have a longer life expectancy than men. As a consequence, there is a steadily increasing proportion of women at advanced ages (Barling, 2013).

Postmenopausal bone loss caused by both decreased ovarian production of sex steroids and an increase in Follicle Stimulation Hormone (FSH) secondary to estrogen deficiency (Sun et al., 2006). Estrogen is an important sex hormone that plays a fundamental role in modulating bone remodeling. Estrogen acts directly on bone via Estrogen Receptor (ER)-α and ER-β which are highly expressed on osteoblasts and osteoclasts (Komm and Bodine, 2001). Estrogen deficiency leads to accelerated bone resorption, primarily due to increased osteoclast differentiation and stimulation of osteoblast apoptosis. Consequently, high bone turnover stimulates osteoblastogenesis fuelled by an expansion of the pool of early mesenchymal progenitors and increased activities of pluripotent precursors toward the osteoblastic lineage (Jilka et al., 1998). Despite the stimulation of osteoblastogenesis, the net increase in bone formation is insufficient to compensate for the increase bone resorption due to the acceleration in osteoclast differentiation and osteoblast apoptosis (Kousteni et al., 2001). In addition to the direct effect of estrogen deficiency, an indirect effect due to an increase in FSH will lead to stimulation of tumor necrosis factor (TNF). The increased TNF production stimulates receptor activator of nuclear factor kappa-β ligand (RANKL) which further increases osteoclast formation. Simultaneously, a potent anti-osteoclastogenic factor, osteoprotegerin (OPG) will be inhibited (Cenci et al., 2000; Collin-Osdoby et al., 2001). The TNF also stimulates the production of inflammatory cytokines which further exacerbate bone loss by promoting osteoclastogenesis (Lorenzo, 2000; Wei et al., 2005). Estrogen deficiency also induces T-cell activation and osteoclastogenesis by downregulating the antioxidant defense pathway, leading to an upregulation of Reactive Oxygen Species (ROS). The ROS have been shown to be responsible in the development of bone loss in postmenopausal osteoporosis (Ozgocmen et al., 2007). Understanding the pathological mechanisms underlying postmenopausal osteoporosis will contribute to advances in the field of osteoporosis including its diagnostics and pharmacological interventions.

Bone loss affects both cortical and trabecular bone with trabecular bone loss more prominent in postmenopausal osteoporosis (Khosla et al., 2006). This is because women have thinner trabeculae and are more prone to trabecular thinning. Due to its large surface/volume ratio, trabecular bone shows a higher rate of turnover than cortical bone. Hence, trabecular bone is more responsive to risk factors causing deterioration and any interventions applied. Therefore, it has been studied more extensively to understand the mechanisms of osteoporosis, diagnostic methods and interventions with anti-osteoporotic agents. Experimentations on osteoporosis using both human and animal will lead to an improved understanding of this disease. It is important to not only understand its causes but also the mechanisms of bone deterioration, diagnostic methods and treatments, as well as preventative measures.

Experimentations using animal models not only discern bone loss mechanisms but also serve as a platform to investigate the efficacy of pharmacological interventions for osteoporosis. A wide variety of animals such as rodents, dogs and sheep have been used as animal models in osteoporosis studies. Laboratory rat is the most widely used in experimental protocols to induce bone loss using hormonal interventions (ovariectomy, orchidectomy, hyperphysectomy, parathyroidectomy) (An and Freidman, 1998; Iwamoto et al., 2004; Iwamoto et al., 2007) and dietary manipulations such as a low calcium diet (Koshihara et al., 2004). The ovariectomized (OVX) rat is commonly used as a postmenopausal osteoporosis model (Kalu, 1991; Turner, 2001). Following ovariectomy, the reduction in estrogen levels result in dramatic bone loss because bone resorption outweighs bone formation activity (Lelovas et al., 2008). The similarities in pathophysiologic responses between the human and rat skeleton, the fact that the rat is readily available and the financial advantages offered by the laboratory rat, have together made it a suitable model for osteoporosis research (Lelovas et al., 2008).

To date, the gold standard used to assess the risk of osteoporosis is by the Bone Mineral Density (BMD) measurement using dual X-ray absorptiometry (DXA) which was initially proposed by WHO (1994), Siris et al. (2004), Winzenberg and Jones (2011). It is versatile due to its high precision, short scan time and low radiation dose and it may be used to assess bone mineral density/bone mineral content of the entire skeleton as well as specific sites, such as the hip and vertebrae which are the most vulnerable to fracture. Although BMD is the cornerstone for the diagnosis of osteoporosis, the use of BMD alone is less than optimal for use as an intervention threshold for several reasons. For some cases, such as those with osteomalacia, a complication of poor nutrition in the elderly, DXA may underestimate total bone matrix due to decreased mineralization of the bone. Osteoarthritis at the spine or hip is common in the elderly and contributes to density measurements, but not necessarily to skeletal strength (Kanis et al., 1997).

The operational definition of osteoporosis which involves BMD measurements using DXA, is discussed often because it focuses too much on bone density and bone mass rather than on structure. Although BMD is an important determinant of bone strength, it does not take into account the microarchitectural changes that occur in trabecular bone (Laib et al., 2001) hence, it is not an accurate predictor of the risk of osteoporotic fracture. It has been reported that the dominant features of the initial phase of rapid bone loss following the onset of estrogen deficiency is increased bone resorption, trabecular thinning and perforation and a loss in connections between remaining trabeculae. This phase is followed by a long-lasting period of slower bone loss where the dominant microarchitectural change is a loss of trabecular connections and trabecular thinning (Eriksen et al., 1990; Pacifici, 2008). This progression shows the importance of bone microarchitecture study in the field of osteoporosis. Regardless of the widespread use of DXA, its inability to analyze bone microarchitecture, has raised concerns over its reliability (Peter and Felix, 2008). Due to these limitations, ongoing studies endeavor to replace this conventional osteoporosis diagnostic tool with a more accurate bone assessment tool.

Assessment of the risk of osteoporosis should involve determining bone quality as a whole. To study bone quality which comprises bone mass, strength and microarchitecture, it is essential to develop an effective, sensitive and non-invasive tool that can analyze cortical and trabecular bone separately and detect early changes in bone. Previous studies have reported that bone microarchitectural structure is an important indicator of mass, strength and density which aids in diagnosing osteoporosis (Brandi, 2009; Neil, 2012). Traditionally, trabecular bone structure has been analyzed using histomorphometry which provides two-dimensional (2D) information on architectural parameters (Chappard et al., 2005). To overcome some of the limitations of 2D histological sections, several non-destructive three-dimensional (3D) techniques have been developed. In recent years, a highly-developed radiological tool, micro-computed tomography (μ-CT) has received attention in bone studies. This technique provides a new method of measuring bone microarchitecture and strength in 3D, replacing the tedious serial staining required for histomorphometric analyses. Measurement of 3D architecture provides improved insight into the underlying bone microarchitecture changes and on its biomechanical properties (Boyd et al., 2006). The development of μCT was first driven by the need for a highly precise and effective tool to reconstruct the complexity of bone architecture at high resolution. Later, it became a tool crucial for evaluating the pathophysiology of osteoporosis, to test the efficacy of pharmaceutical interventions and to estimate bone biomechanical properties.

High resolution μCT imaging is now becoming more applicable in the bone research field because it provides better and more accurate information on bone structural parameters. Compared to other radiological tools used in bone studies, μCT provides much greater accuracy when measuring bone mineralization. The μCT can measure changes in cortical thickness in the range of 10-20% which is undetectable when using other X-ray imaging tools. Previous studies have reported widely that μCT is capable of analyzing cortical and trabecular bone separately which cannot be done with other X-ray tools (Genant et al., 2008). This property is important for the investigation of osteoporosis. μCT may be used to analyze human bone and animal bone. Due to its high resolution, it is possible to analyze data from bone areas as small as the trabeculae of small rodents, such as mice and rats. It was originally developed for in vitro use which later led to an increase in the use of in vivo μCT.

The investigation of postmenopausal osteoporosis using μCT should be studied extensively in animals prior to embarking on clinical trials. The increasing number of bone studies using μCT which focus on the assessment of bone changes in the postmenopausal osteoporosis rat model, warrants a review. The aim of this evidence-based review is to explore original research articles to determine the efficacy and reliability of μCT in assessing bone changes in the postmenopausal osteoporosis rat model.

USE OF MICRO COMPUTED TOMOGRAPHY AND POSTMENOPAUSAL OSTEOPOROSIS

A systematic review of the literature was performed to identify relevant studies on the use of micro computed tomography and postmenopausal osteoporosis. To conduct a comprehensive search of health science journals, we used the Medline via Ovid Medline and Scopus databases (reports published between 1994 and 2014). The search strategy involved a combination of the following two sets of key words (1) Micro CT OR micro computed tomography and (2) Postmenopausal osteoporosis OR postmenopausal osteoporosis.

Selection of research articles: The results were limited to full research articles that were published in English language. Studies that complied with these following inclusion criteria were included (1) Reported μCT analysis of the postmenopausal osteoporosis rat model and (2) The postmenopausal osteoporosis-related bone changes should relate to lifestyle variables, aging or experimentally-induced conditions. Studies were excluded if they were (1) Duplicated studies (2) Reviews, news, letter, editorials or case studies (3) Did not use a control group (4) μCT was not the primary outcome (v) did not use the OVX-induced postmenopausal rat model (5) related to other diseases (e.g., chronic obstructive pulmonary disease, osteoarthritis) (6) Related to bone fracture healing.

Data extraction and management: Papers included in this review were selected based on three phases. Firstly, we excluded the papers that did not match the inclusion criteria based solely on their titles. Secondly, the abstracts of the remaining papers were screened and papers that did not match the inclusion criteria were excluded. Lastly, we scrutinized the remaining papers to exclude a second group of papers that did not match our inclusion criteria. The remaining papers were again screened by two reviewers prior to data extraction phase. Any discrepancy that arose was settled in discussions between the reviewers. We recorded (1) the type of study and micro-CT used and provides; (2) a brief description of the sample/population; (3) a brief description of the methods used; (4) a brief description of the results and (5) outcomes and comments on the study.

SEARCH RESULTS

The literature searches identified 234 potentially relevant articles. Two reviewers independently assessed all articles for inclusion or exclusion based on the title and abstract. A total of 182 articles were retrieved for further assessment and data extraction. Following these assessments, 160 of these articles were excluded because they did not involve postmenopausal osteoporosis and μCT was not the primary outcome. Frequent discussions between the two reviewers took place to resolve differences in opinion on the inclusion or exclusion of full articles. The remaining 22 articles that fulfilled all inclusion and did not fulfill exclusion criteria were included in this review. The process of paper selection from the beginning of the identification of relevant articles to articles selection based on the inclusion and exclusion criteria is summarized in a flow chart shown in Fig. 1.

Study characteristics: The characteristics of all studies are summarized in Table 1. All animal studies were conducted between the year 2000 and 2013, with majority conducted within the past five years. All studies used female rats, because this review focused on postmenopause-induced osteoporosis. Fifteen studies used Sprague-Dawley rats and the remaining 7 studies used Wistar rats. Rats were of varying age, with the majority being mature and aged 3 months. There were also some studies that used aged rats of 8-10 months. The number of rats for each study ranged from 3 up to 152 and in the majority of studies, rat numbers were kept to a minimal number due to animal ethics requirements.

Bone microarchitecture, a highly reliable indicator of osteoporosis was the primary outcome measured in reports used in this review. Bone microarchitecture parameters were measured using μCT. Out of 22, a total number of 14 studies used in vitro μCT, 7 used in vivo μCT and only one study used both in vitro and in vivo μCT. Six out of seven in vivo μCT studies examined in this review performed a longitudinal study to monitor time-dependent bone changes at different periods. Rats were scanned under anesthesia at different time periods. In the study by Wu et al. (2012) using in vivo μCT, femoral bone were scanned only after the completion of treatment, at 8 weeks (Wu et al., 2012). Bone microarchitecture of trabecular bone is an important determinant of osteoporotic changes. However, cortical bone analysis also contributes to improved insight and understanding of the changes in bone structure following ovariectomized-induced osteoporosis. Of all the studies included in this review, a total of 15 studies focused only on trabecular bone as their primary parameter whereas the remaining seven studies measured both trabecular and cortical bone.

Eight studies used μCT alone to analyze bone loss in the postmenopausal osteoporotic rat model (Peyrin et al., 1998; Waarsing et al., 2004; Yoon et al., 2012; Jee et al., 2010; Waarsing et al., 2006; Park et al., 2008; Brouwers et al., 2009; Rhee et al., 2009). In the remaining studies, bone changes were studied using μCT, histomorphometry, DXA and biomechanical bone strength assessments. The outcomes of these methods were compared to determine the most efficient bone analysis method. Types of bone used for micro-CT analysis varied between the studies with the most widely used being the tibiae (Peyrin et al., 1998; Waarsing et al., 2004, 2006; Gasser et al., 2005; Jee et al., 2010; Brouwers et al., 2009, 2010; Zhao et al., 2011; Zhang et al., 2012). Seven studies used the femora as their sample (Park et al., 2008; Stunes et al., 2011; Montero et al., 2012; Wu et al., 2012; Sung et al., 2012; Li et al., 2013; Zhao et al., 2013), three used lumbar vertebrae (Yoon et al., 2012; Rhee et al., 2009; Bian et al., 2012) and the remaining three studies (Sampath et al., 2007; Kuber et al., 2008; Sharan et al., 2010) used both tibiae and femora. The Region of Interest (ROI) chosen in the majority of studies was below the growth plate extending proximally. However, in two of the studies, the ROI was not mentioned (Waarsing et al., 2006; Park et al., 2008). In studies by Kuber et al. (2008) and Li et al. (2013), the area scanned was mentioned, but not the specific ROI site.

The resolution used for μCT scanning in these studies ranged from 6-40 μm. However, in four studies, the resolution was not mentioned (Sampath et al., 2007; Park et al., 2008; Kuber et al., 2008; Bian et al., 2012). The majority studies in this review used protocols derived from Ruegsegger et al. (1996) with a resolution of 20 μm (Ruegsegger et al., 1996). Micro-CT analysis is capable of measuring bone structural parameters similar to histomorphometry analysis, such as bone volume fraction (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N) and trabecular separation (Tb.Sp). In contrast to histomorphometry, μCT provides a more accurate 3D microarchitecture measure by providing non-metric parameters such as Structural Model Index (SMI), connectivity density (Conn.D), degree of anisotropy (DA) and trabecular bone pattern factors (TbPfs). In seven studies used for this review, μCT analysis measured only the directly assessed metric indices which are BV/TV, Tb.Th, Tb.N and Tb.Sp (Waarsing et al., 2004; Gasser et al., 2005; Yoon et al., 2012; Park et al., 2008; Montero et al., 2012; Sung et al., 2012; Zhao et al., 2013). The remaining studies measured both metric and non-metric bone microarchitecture indices.

Fig. 1:	Flow chart of the selection process of the articles used in this review

The directly assessed metric structural parameters provided by μCT are similar to histomorphometry. However, six studies used for this review performed the conventional bone histomorphometry analysis regardless of results obtained from μCT (Sampath et al., 2007; Kuber et al., 2008; Wu et al., 2012; Bian et al., 2012; Li et al., 2013; Zhao et al., 2013). In addition to structural parameters, bone histomorphometry analysis provides information on dynamic and static parameters. Dynamic parameters yield information on bone turnover such as Bone Formation Rate (BFR), mineral apposition rate (MAR), mineral formation rate, single labeled surfaces (sLs) and double labeled surfaces (dLs). In contrast, static parameters reflect bone cellular conditions, such as osteoblast and osteoclast volume. Out of these six studies, two measured only the structural bone parameters which paralleled the μCT analysis (Wu et al., 2012; Zhao et al., 2013). Two studies performed histomorphometry to measure structural and static parameters (Bian et al., 2012; Li et al., 2013). Sampath et al. (2007) measured both structural and dynamic parameters, whereas the remaining study by Kuber et al. (2008) measured only the dynamic bone parameters.

Apart from μCT scanning and histomorphometry, some studies performed dual X-ray absorptiometry (DXA) scanning to measure BMD. This analysis provides better insight into bone changes in the postmenopausal osteoporosis rat model. Ten studies used DXA as one of their parameters. In three other studies however, BMD was measured using the μCT itself (Yoon et al., 2012; Brouwers et al., 2009; Zhao et al., 2013). Yoon et al. (2012) measured Bone Mineral Content (BMC) and BMD of both trabecular and cortical bone and OVX rats showed a significant reduction in BMD whereas, BMC was not affected (Yoon et al., 2012). In another study by Brouwers et al. (2009), μCT was used to measure BMD of both cortical and trabecular bone (Brouwers et al., 2009). Furthermore, Zhao et al. (2013) reported that BMD analysis using μCT revealed a significant increase in the treated group compared to OVX control group (Zhao et al., 2011). Based on the ability of μCT to measure BMD, these data may lead to the rise of μCT as the gold standard in accessing bone status.

Out of 22 studies used for this review, three studies demonstrated the use of pQCT as an outcome measurement. Gasser et al. (2005) employed pQCT and DXA to measure BMD of both cortical and cancellous bone of OVX rats (Gasser et al., 2005). pQCT showed a decrease in BMD. In an additional two studies by Sampath et al. (2007) and Kuber et al. (2008), pQCT was used to measure the volumetric content and density of both cortical and cancellous bone. However, these results were not reported. In recent years, the rise of a new computational engineering method has led to the development of Finite Element Analysis (FEA) by μCT. The study by Rhee et al. (2009), Sharan et al. (2010) reported the use of FEA to convert 3D reconstructed μCT images into micro-finite elements. Indirectly, this technique enabled the measurement of bone mechanical properties, an indicator of bone strength. In addition to FEA, some studies performed conventional bone mechanical tests. Apart from μCT as their primary outcome, a total number of 11 studies used for this review performed the conventional 3-point bending biomechanical bone test.

ASSESSMENT OF BONE MICROARCHITECTURAL CHANGES USING MICRO-CT VARIABLES IN IN VITRO μCT

A total of 14 studies used this review used in vitro μCT. In general, bone structural parameters, such as BV/TV, Tb.Th, Tb.Sp and Tb.N were analyzed. In the majority of studies, BV/TV,Tb.Th,Tb.N were decreased whereas Tb.Sp increased in the untreated ovariectomized group, indicating deterioration in bone microarchitecture. For example, in the studies by Sung et al. (2012) and Li et al. (2013), the untreated OVX group showed significantly decreased BV/TV, Tb.Th, Tb.N and increased Tb.Sp (Sung et al., 2012; Li et al., 2013). In addition to measuring these typical variables, some studies used for this review measured additional variables, such as BMD and BMC (Yoon et al., 2012; Brouwers et al., 2009; Zhao et al., 2013), TbPfs (Sampath et al., 2007; Kuber et al., 2008; Stunes et al., 2011; Montero et al., 2012), Conn.D (Peyrin et al., 1998; Zhao et al., 2011; Zhang et al., 2012; Wu et al., 2012), DA (Peyrin et al., 1998), SMI (Peyrin et al., 1998; Sampath et al., 2007; Kuber et al., 2008; Sharan et al., 2010; Brouwers et al., 2010; Wu et al., 2012; Li et al., 2013) and FEA (Rhee et al., 2009). In the majority of studies, BMC, BMD, TbPfs, Conn. D. and FEA were reduced whereas SMI was increased (rod-like structure) in the untreated OVX group.

According to Laib et al. (2001) and Peyrin et al. (1998), in vitro μCT was capable of detecting bone microarchitectural changes in OVX rats as early as 12 days after surgery. Bone microarchitectural changes were measured over time because scans were performed at 35, 60 and 110 days post-ovariectomy. Within the first 12 days post-ovariectomy, all trabecular structural parameters in the OVX group showed a dramatic decrease which was then followed by a slower decline. In the study by Bian et al. (2012), bone microarchitectural changes were reported at different times; at 2 weeks and 3-months, after treatment of OVX-induced postmenopausal osteoporosis rats with Oleanolic Acid (OA). In this study, μCT analysis was capable of showing a reduction in metric structural indices (i.e., BV/TV,Tb.N and Tb.Th) and an increase in non-metric index (Conn. D.) as early as 2 weeks after surgery.

In studies by Yoon et al. (2012) and Zhao et al. (2013), BMD and BMC were determined using in vitro μCT. According to Yoon et al. (2012), BMD and BMC of the OVX group were significantly reduced when compared with the Sham group (Yoon et al., 2012). According to Zhao et al. (2013), the OVX rats treated with bone-seeking estrogen compound (SE₂) showed similar BMD to the estrogen-treated and Sham-operated groups and this outcome was significantly increased compared with the OVX group (Zhao et al., 2013).

Only two studies used for this review measured TbPfs and SMI simultaneously (Sampath et al., 2007; Kuber et al., 2008). In the study by Sampath et al. (2007), TbPf and SMI were measured in rats given various doses of Thyroid Stimulating Hormone (TSH) for the prevention and restoration modes of bone loss. The TbPfs and SMI values at all the TSH doses were decreased in both the prevention and restoration modes (Sampath et al., 2007). In the study by Kuber et al. (2008), TbPfs and SMI were measured in rats given Sevalamer and TbPfs was significantly decreased compared with the OVX group, whereas the SMI value tended to decrease but was not significantly different from the OVX group (Kuber et al., 2008). The TbPfs value also decreased in OVX rats treated with Kalsis, an antioxidant dietary supplement. The decrease in TbPfs attenuated the bone disconnection effect (Montero et al., 2012). SMI was also measured in the study by Sharan et al. (2010), where SMI was shown to increase (more rod-like) in the OVX group and was lower in the 5.0 mg kg^-1 GTDF-treated rats (Sharan et al., 2010). According to Stunes et al. (2011), the ovariectomized group exhibited a significantly higher SMI value compared with the Sham group. The OVX group that was treated with Wyeth at 90 mg kg^-1 showed a significantly higher Conn.D. value compared with the OVX group (Stunes et al., 2011).

Conn. D. Which indicates connections between trabecular structures was measured in a study by Zhao et al (2011) where OVX rats treated with Cibotium Barometz Extract (CBE) showed a significant increase in Conn.D compared with the OVX control group. Other bone structural parameters, such as BV/TV, Tb.N and Tb.Th were also significantly increased when compared with the OVX which denotes an improvement in bone microarchitectural (Zhao et al., 2011). In addition, a study by Zhang et al. (2012) reported a significant decreased in Conn.D. in the OVX group compared with the Sham group. Other bone structural parameters, such as BV/TV, Tb.N and Tb.Th were also significantly decreased in OVX rats compared with the Sham rats. These data indicate trabecular deterioration in OVX rats (Zhang et al., 2012).

Laib et al. (2001) and Rhee et al. (2009) were the only two studies used for this review that used in vitro μCT to measure the Degree of Anisotropy (DA). According to Laib et al. (2001), DA was calculated by projecting the triangles of bone surfaces onto an ellipse, based on the methods described by Laib et al. (2000) and Harrigan and Mann (1984). DA in the OVX group at 12, 35, 60 and 110 days was significantly different from the Sham group (Peyrin et al., 1998). In the study by Rhee et al. (2009), DA was not significantly different between the control and treatment groups. These authors also performed a Finite Element Analysis (FEA) to measure mechanical parameters, such as stiffness and elastic modulus. The FEA of the OVX group was not significantly different from the Sham group for both mechanical parameters (p>0.05). PTH-treated rats showed higher mechanical properties than Sham rats (Rhee et al., 2009).

ASSESSMENT OF BONE MICROARCHITECTURAL CHANGES USING IN VIVO μCT

A total of 7 studies used for this review in vivo μCT. Notably, μCT variables measured using in vivo μCT were similar to those measured using in vitro μCT. By using in vivo μCT, longitudinal studies may be performed to determine bone microarchitectural changes overtime. In the study by Waarsing et al. (2004), a dramatic and progressive loss in epiphyseal and metaphyseal of the trabeculae bone in OVX group was reported during week 4 to week 14 after surgery. In the Sham group, no changes in bone volume were noted, however, after 14 weeks, thinning of the trabeculae in metaphyseal region may be observed (Waarsing et al., 2004). Gasser et al., (2005) performed a longitudinal study using in vivo μCT and reported significant microarchitectural changes at as early as 2 weeks after surgery and gradual changes of BV/TV and Tb.Th were observed for up to 12 weeks in the OVX group (Gasser et al., 2005). A longitudinal study by Boyd et al. (2006) used in vivo μCT over a longer period, by scanning only sham-operated and OVX groups for 6 months at 1-month intervals. The OVX group showed significant changes in all structural parameters, including BV/TV, BS/BV, Tb.Th, Tb.Sp and Tb.N. However, the Sham group also showed reduction in these structural parameters that were less prominent than in the OVX group (Jee et al., 2010).

Waarsing et al. (2006) performed a longitudinal study for over 54 weeks, where scanning was performed at week 0 (prior to OVX) and at weeks 4, 34 and 54 post-OVX. Changes in bone microarchitectural were obvious during the first 14 weeks with the trabecular bone of the OVX group showing a dramatic reduction in Conn.D. In addition, this study also reported on the microarchitectural changes in cortical bone, with cortical thickness of the OVX group being significantly reduced compared with the previous time point. Because in vivo μCT exposes live rats to radiation, the radiation may affect the results of the study and these authors also performed a comparison between irradiated and non-irradiated rats. In this study, small differences were observed in SMI and trabecular thickness. The non-irradiated controls showed slightly more rod-like trabeculae than age-matched irradiated animals and trabecular thickness in noon-irradiated controls was not significantly different from baseline at week 0 (Waarsing et al., 2006).

Park et al. (2008) used in vivo μCT to measure only three variables; BV/TV,Tb.Th and BMC to determine the effects of apigenin (API) and estrogen (E2) on OVX-induced osteoporotic rats. The rats were scanned at 7, 12 and 23 weeks. At week 7, the BV/TV and Tb.Th values of the OVX group were significantly decreased compared with the Sham, indicating deterioration in bone microarchitecture and proving that the induction of osteoporosis was successful. The BMC of the API group was increased compared to OVX group (Park et al., 2008). In vivo μCT can also be used to measure BMD, as studied by Brouwers et al. (2009). In this paper, the authors measured BMD of both cortical and trabecular bone. They reported that the PTH-treated group showed an increase in trabecular BMD whereas the cortical BMD remained unaffected (Brouwers et al., 2009).

Brouwers et al. (2010) used both types of μCT in their study. For the in vivo μCT, scanning was performed at weeks 8, 10, 12 and 14 post-OVX on the proximal tibia of rats. At week 8, both tibial meta-and epiphysis of the OVX group showed a deterioration in bone microarchitecture which were demonstrated by a decrease in BV/TV,Conn. D, Tb.N, Tb.Th and an increase SMI and Tb.Sp. Beyond 8 weeks, all variables worsen further, except in the case of Tb.Th which improved. For the in vitro μCT, epiphysis region of the femora was scanned to determine microarchitectural changes. All structural parameters including BV/TV, Tb.N, Tb.Th and Tb.Sp were significantly different in the OVX group when compared with the Sham group. However, the Conn. D. Value of the epiphysis region in OVX group was not significantly different from the Sham group (Brouwers et al., 2010). These data indicate that the epiphysis of the femora shows much slower changes compared with the epiphysis region of the tibiae.

In a study by Wu et al. (2012), in vivo μCT was not used for longitudinal study. Instead, the OVX rats treated with zoledronate-impregnated calcium phosphate (ZLN/CPC) were euthanized and femora were scanned after the completion of the treatment to measure bone structural parameters including BV/TV, Tb.N, Tb. Sp, SMI and Conn. D. The ZLN/CPC-treated group showed a significant increase in BV/TV, Tb.N and Conn. D and a decrease in Tb.Sp and SMI compared with the OVX group (Wu et al., 2012) (Fig. 1, Table 1).

Table 1:	Review of literature for the type of study and type of μCt

USE OF μCT IN BONE ASSESSMENTS

Our primary aim is to review the use of μCT in bone assessments of ovariectomized-induced osteoporosis rat models. Based on the chosen studies, μCT has shown beneficial advantages as a bone-assessment tool. The initial type of μCT made commercially available for bone studies was in vitro μCT. Osteoporotic studies using in vitro μCT are typically performed using the proximal tibia and distal femur. In this review, the majority of studies performed μCT analysis using tibiae although seven studies used femora, three studies used lumbar and the remaining studies analyzed a combination of these three types of bones. The most common complication of postmenopausal osteoporosis is hip fracture (Cummings and Melton, 2002). The annual incidence of hip fractures has been increased worldwide during the last five decades and that the current total number of women who sustain a hip fracture is estimated to be one million annually (Gullberg et al., 1997). In comparison with other types of fractures, such as vertebral and wrist fractures, hip fracture is associated with serious disability and excess mortality. Women who have sustained a hip fracture have a 10-20% higher mortality for their age (Cummings and Melton, 2002). This fact underlies the wide used of tibiae and femora in osteoporotic studies. Tibiae or femora area, in particular, the area approximately 1.5 mm below the epiphyseal growth plate that extends towards the proximal direction is the region of interest selected most often for osteoporotic studies (Martin et al., 2003). This ROI is the trabecular region rich in blood supply and with high bone turnover activity, thus, it is sensitive to changes caused by any stimulation (Judith, 2009).

This review demonstrated that in vitro μCT bone assessment requires rats to be euthanized at the end of the study, followed by excision of the bone for scanning. This protocol demands indirectly that a large number of animals must be used to overcome variability to obtain statistically significant results. The results of in vitro μCT are often consistent with structural parameters assessed using traditional histomorphometric method. In vitro μCT is commonly used to assess the effects of therapeutic interventions such as in the studies by Sampath et al. (2007) and Rhee et al. (2009) which investigate the effects of Thyroid Stimulating Hormone (TSH) and zoledronic acid respectively. In contrast, in vivo μCT is always used to perform longitudinal studies such asin the study by Waarsing et al. (2004). The OVX rats were scanned at baseline and at, 4 and 14 weeks later to observe changes in trabecular structure. This study reported a progressive dramatic loss of both epiphyseal and metaphyseal trabeculae over weeks 4-14. In another study by Gasser et al. (2005), OVX rats were scanned at baseline and at, 1, 2, 4, 8 and 12 weeks. Significant gradual changes were detected at 2 weeks and up to 12 weeks (Gasser et al., 2005).

Based on this review, it can be postulated that in vivo μCT is designed to enable monitoring of bone changes over time without euthanizing the animal. The advantage of longitudinal studies is that each rat acts as its own control, hence, control groups are not required (Woo et al., 2005). This design indirectly reduces the number of animals required for study. Brouwers et al. (2010) employed the use of both in vitro and in vivo μCT to monitor the bone changes following OVX and the effect of Whole Body Vibration (WBV) treatment (Brouwers et al., 2010). In vivo μCT analysis revealed a significant difference between the OVX and the treated group at only week 8. Over the period of week 8-14, there was no significant difference found between the two groups. This finding is in contrast to the in vitro μCT analysis which revealed a significant difference between the groups at 14 weeks after OVX. This finding may be due to several reasons. Compared with in vitro scanning, it is difficult to obtain a consistent orientation during in vivo μCT because the rat leg cannot be positioned exactly the same during each scan (Waarsing et al., 2005). Another limitation of in vivo μCT is that the presence of soft tissues around the bone during scanning may reduce image quality and the actual volume of the bone may be difficult to ascertain (Cendre et al., 2002; Nadia et al., 2013). Both in vitro and in vivo μCT have their own distinct advantages and disadvantages that make them effective tools in monitoring bone changes in postmenopausal osteoporosis.

In osteoporosis studies, it is important to determine bone quality as a whole which comprises all characteristics, such as bone mass, microarchitecture and strength. Although BMD has long been known as an important determinant of bone status and is a gold standard for assessing the risk of osteoporosis, bone microarchitecture also plays an important factor in bone studies. DXA limitations that focused too much on bone mass have led to the rise of other radiological imaging tools. Prior to the development μCT, the radiological tools used for bone assessment included peripheral quantitative CT (pQCT), high resolution CT (hrCT) and Magnetic Resonance Imaging (MRI). Many previous studies have reported the use of pQCT in longitudinal osteoporotic studies due to its advantages over DXA (Brunader and Shelton, 2002).

In this review, three studies used pQCT as one of their imaging protocols. Gasser et al. (2005) reported on the use of DXA, pQCT and μCT in assessing osteoporotic rats. DXA measurements showed a decrease in BMD of cortical tibia whereas pQCT measurements revealed a decrease in cancellous BMD and cortical thickness in OVX rats. In contrast, μCT analysis revealed a significant osteoporotic changes in microarchitectural parameters which were reversed by parathyroid hormone (PTH) and Zoledronic Acid (ZA). The strong anabolic effect of PTH also caused a reduction in SMI values, indicating that the bone is no longer plate-like but more like a solid block of bone with pores (Gasser et al., 2005). Other studies by Sampath et al. (2007) and Kuber et al. (2008) also reported on the use of DXA and pQCT together with μCT and showed decreased BMD in the proximal tibia, distal femur and the spine of OVX rats. These changes were reversed following treatment. The high resolution provided by pQCT is sufficient to distinguish cortical and trabecular bone but remain too low to resolve trabecular microarchitecture which consequently may result in underestimations of trabecular thickness (Hangartner, 2007). Due to the high resolution provided by μCT (1-100 μm), it is feasible to analyze areas of bone as small as the trabeculae of small rodents. μCT analysis in these studies showed restoration of structural parameters which also included SMI and trabecular bone pattern factors (TbPf), following treatment. In a latter study, μCT analysis also revealed an increase in cortical thickness (Ct.Th).

In the majority of studies, μCT analysis also revealed non-metric indices. SMI indicates the relative prevalence of rods and plates in a 3D trabecular structure and involves measurement of the surface convexity (Hildebrand and Ruegsegger, 1997). SMI values of a pure plate-shaped bone and pure rod-shaped bone are 0 and 3, respectively. A negative SMI value indicates a solid and very dense trabecular structure. TbPf is a quantitative ratio of inter-trabecular connectivity (Hahn et al., 1992), where positive values denote concave structures and negative values denote convex structures (Odgaard, 1997). In addition to SMI and TbPf, another non-metric index that may be derived from μCT analysis is the Degree of Anisotropy (DA). DA is a measure of how substructures are oriented within a volume and it is calculated using Mean Intercept Length (MIL) method (Van der Linden and Weihans, 2007). Bone is known to be anisotropic, meaning that it is stronger when loaded in one direction. Higher DA suggests greater anisotropy resulting from loss of directional trabeculae. This is in accordance with previous studies which reported that increase of bone resorption in osteoporosis leads to thinner trabeculae, resulting in an increase of anisotropy (Newitt et al., 2002; Odgaard et al., 1990). In this review, the two studies by Laib et al. (2001) and Rhee et al. (2009) demonstrated DA in OVX rats using μCT (Peyrin et al., 1998; Rhee et al., 2009).

In this review, some studies also demonstrated the use of histomorphometry and biomechanical bone analyses. Bone architecture has long been performed using conventional histomorphometric methods. In contrast to μCT, histomorphometry provides only two-dimensional (2D) information on trabecular bone parameters (Iwaniec et al., 2008). Histomorphometry is also time-consuming because it requires tedious processing of thin bone sections and serial staining (Vidal et al., 2012). In comparison between 2D histomorphometry and 3D μCT, many previous studies have demonstrated the advantages of μCT in detecting earlier changes in bone architecture (Jiang et al., 2005). These data are consistent with data presented by Laib et al. (2001) which reported a dramatic decrease in trabecular structural parameters of OVX rats as early as 12 days post-OVX. The trabeculae also began to change into rod-like structure after 12 days post-OVX (Peyrin et al., 1998). Another study by Gasser et al. (2005) reported significant trabecular changes at 2 weeks post-OVX that progressed gradually up to 12 weeks (Gasser et al., 2005). Bian et al. (2012) who used both histomorphometric and μCT analyses, reported significant changes in trabecular parameters using μCT as early as 2 weeks post-OVX. In contrast, histomorphometric analysis only showed a reduction in Tb.N at 3 months after OVX (Bian et al., 2012). Another report by Sampath et al. (2007) also revealed that histomorphometric analysis showed that the Tb.N and Tb.Th were maintained where, μCT showed an increase in trabecular structural parameters (Sampath et al., 2007). These results support the fact that μCT is capable of detecting early changes in bone structure compared with conventional histomorphometric analyses.

Previous studies have reported that BMD and structural parameters measured by DXA, histomorphometry and μCT correlate significantly (Barou et al., 2002; Yeom et al., 2008). Although μCT is well known for its primary use in generating information on bone structure, in recent years, it has been proposed for use for non-destructive, 3D measurements of bone mineralization using a linear calibration of mineral density and x-ray attenuation. This is to measure BMD which is typically measured using DXA. In contrast to DXA which produces two different energy peaks to distinguish between soft tissue and bone absorption, μCT uses mixtures of polymers with hydroxyapatite (HA) crystals for calibration of x-ray attenuation to density (Kazakia et al., 2008; Burghardt et al., 2008). Phantom models comprising HA are desirable because HA exhibits a similar composition and x-ray attenuation to bone mineral. The measured x-ray attenuation versus HA density then undergo in linear regression which will then be calibrated to estimate bone mineral density (Nazarian et al., 2008). In this review, three studies measured BMD using μCT rather than DXA. The study by Yoon et al. (2012) showed a significant BMD reduction in both the cortical and trabecular bone of OVX animals and studies by Brouwers et al. (2009) and Zhao et al. (2013) demonstrated a significant increase in BMD only in trabecular bone of the treated group. Based on these results, it may be postulated that μCT is capable of replacing DXA as the gold standard for assessing bone mineral density.

BMD and bone microarchitecture are not the sole determinant of osteoporosis and fracture risks. A combination of bone mass, microarchitecture and intrinsic material determine a bone’s ability to withstand loading (Dempster, 2003). In some studies, bone strength was reported to play an important role in predicting early fracture risk (Sambrook et al., 2007). Hence, it is important to assess bone strength to have a better insight into bone quality. DXA, the gold standard for assessing osteoporotic risk does not measure bone strength accurately. Bone strength has long been measured using biomechanical tests. The gold standard for bone-strength assessment is performed by exerting a load on the bone until it fractures (Voide et al., 2006). In this review, a total of 11 studies performed biomechanical bone tests using the 3-point bending method. As expected, in the majority of papers, the OVX group showed weaker mechanical bone properties. In animal models, biomechanical bone testing, where the bone can be dissected out and tested biomechanically until it fractures, presents no problem. Another limitation of this conventional test is that it produces large errors and significant uncertainty due to the sensitivity of the mechanical measurements to friction between the sample and the load. Therefore, this test may not detect small or even large changes in mechanical properties of bone. In human studies, this method is not a viable option. Bone strength may only be assessed indirectly using computer software such as Finite Element Analysis (FEA) via μCT (9 Joshua and Steven, 2008; Grant and Tony, 2009). These limitations have led to an increase in effort to developing this feasible and advanced technique.

In this review, we noted that one study, by Rhee et al. (2009), performed a finite element analysis using in vitro μCT Skyscan 1072. All treated rats showed higher mechanical properties (stiffness and elastic modulus) compared with the sham and OVX rats (Rhee et al., 2009). For this technique, micro-images of the bone obtained from μCT scanning are converted into Finite Element (FE) models, presenting the bone tissue as equally shaped 8-node brick elements using the hexahedron meshing technique to stimulate real mechanical tests (Keyak and Rossi, 2000). These FE elements are described as elastic properties. Simulated compression tests of FE models are performed following a compressive displacement of 0.5% strain application (Torcasio et al., 2012). This advanced technique has enabled the calculations of bone mechanical properties, such as load, stress, strain and Young modulus which may replace conventional biomechanical bone testing. FEA has not been feasible for bone in vivo due to insufficient resolution, however, recently, in vivo imaging has been introduced with FEA (Ulrich et al., 19993). A combination of bone microarchitecture and bone biomechanical properties may reflect bone turnover and bone microdamage more effectively.

STRENGTHS AND LIMITATIONS OF THIS REVIEW

The μCT assessments of the postmenopausal osteoporosis rat model have revealed the capability of μCT to measure bone microarchitecture changes in several variables. To seek alternative treatments for postmenopausal osteoporosis, numerous studies have been conducted using rat osteoporosis model that used μCT for evaluation. Thus, a critical review is highly relevant to identify relevant articles published so far. Our search identified 22 research articles that were included in this systematic review. To the best of our knowledge, this is the first evidence-based review that focuses on the bone microarchitectural changes in the OVX-induced postmenopausal osteoporosis rat model. We have included both types of μCT in vivo and in vitro μCTs to provide a better overview of the most recent and reliable evidence presented on this subject. In one of the studies examined, investigations using the in vivo μCT were followed with in vitro μCT to compare the differences in μCT analysis methods (Brouwers et al., 2010).

A number of limitations were identified in this review. Many studies used different resolutions for their μCT protocols. Due to the differences in resolution used, the outcomes of μCT images are not uniform and may influence the interpretation of the results. In addition, the number of the animals per group was quite limited. For example, only one rat per group was used in Waarsing et al. (2004) and this limited number of rats may not be sufficient to examine bone microarchitectural changes. We have restricted our study solely to animal studies due to the limited use of μCT in humans to date. Thus, we cannot study μCT assessment of bone changes in humans systematically.

RECOMMENDATIONS

Based on the μCT protocols examined in this review, in the future, it will be important to use a standardized protocol for each specific μCT model to determine bone microarchitectural changes in rodents more uniformly. Furthermore, the use of the efficient, high resolution and sensitive μCT technique should not be restricted to animal experiments. It should be used widely on humans to provide better understanding and a true picture of microarchitectural changes in human bone.