Sex Specific Determinants in Osteoarthritis: A Systematic Review of Preclinical Studies

Osteoarthritis (OA) is a highly prevalent joint disease that primarily affects about 10% of the world’s population over 60 years old. The purpose of this study is to systematically review the preclinical studies regarding sex differences in OA, with particular attention to the molecular aspect and gene expression, but also to the histopathological aspects. Three databases (PubMed, Scopus, and Web of Knowledge) were screened for eligible studies. In vitro and in vivo papers written in English, published in the last 11 years (2009–2020) were eligible. Participants were preclinical studies, including cell cultures and animal models of OA, evaluating sex differences. Independent extraction of articles and quality assessments were performed by two authors using predefined data fields and specific tools (Animals in Research Reporting In Vivo Experiments (ARRIVE) guideline and Systematic Review Centre for Laboratory animal Experimentation (SYRCLE) tool). Twenty-three studies were included in the review: 4 in vitro studies, 18 in vivo studies, and 1 both in vitro and in vivo study. From in vitro works, sex differences were found in the gene expression of inflammatory molecules, hormonal receptors, and in responsiveness to hormonal stimulation. In vivo research showed a great heterogeneity of animal models mainly focused on the histopathological aspects rather than on the analysis of sex-related molecular mechanisms. This review highlights that many gaps in knowledge still exist; improvementsin the selection and reporting of animal models, the use of advanced in vitro models, and multiomics analyses might contribute to developing a personalized gender-based medicine.


Introduction
Sex has an important role in the pathophysiology of several diseases, including osteoarthritis (OA). OA is a highly prevalent joint disease that primarily affects the elderly population causing frailty in patients [1,2]. The World Health Organization (WHO) Scientific Group on Rheumatic Diseases estimated that 10% of the world's population who are 60 years or older present significant clinical problems attributed to OA [3]. Although OA may affect any joint in the body, it most commonly affects the hips, knees, hands, and spine, causing musculoskeletal disability and resulting in limited activities of daily living in the adult population, reduced mobility, and increased levels of healthcare utilization [4][5][6][7]. OA is defined as a multifactorial condition of joint failure mainly characterized by articular cartilage loss, subchondral bone sclerosis, and inflammation, leading to progressive joint degradation, structural alterations, loss of mobility and function, pain, and decreased quality of life [8][9][10]. The risk factors associated with OA are multiple and can be genetic, metabolic, biochemical and biomechanical in nature, and patient related factors, including age, gender, obesity, and hormone levels, thus contributing to the unbalance of the delicate equilibrium between anabolic and catabolic

Preclinical In Vitro Studies
The in vitro studies were analyzed based on the cell source (Table 1). Most in vitro studies were performed by isolating cells from human tissues (3 papers out of 5; 60%), then, in the same way, by canine tissues (1 paper out of 5; 20%) and rats (1 out of 5; 20%).

Preclinical in Vitro Studies
The in vitro studies were analyzed based on the cell source (Table 1).

Human Tissues
The majority of studies (n = 3) used human cells isolated from different tissues of male and female patients affected by OA and subjected to total knee arthroplasty (TKA) [25][26][27].
Overall, these three studies analyzed samples from 400 patients, of which 146 were males and 254 females (64% female prevalence). Retrieved tissues were used both for cell isolation and for a wide plethora of assays (histological, immunohistochemistry (IHC), immunoblotting, microarray, fluorescent in situ hybridization (FISH), gene expression, and ELISA analyses). The works of Koelling and Pan were well structured with a clearly reported diagnosis of OA (Kellgren-Lawrence grading scale or the American College of Rheumatology (ACR) classification) [25,26]; while the work of Stumm, besides having the lowest number of patients, did not give evidence of OA diagnosis [27]. All studies harvested control tissues or from minimally eroded surfaces or from healthy donors, thus including a control, not OA; moreover, they specified that chondrocytes used for experiments were at passage one or cultured in three-dimensional (3D) culture conditions in order to retain the chondrocyte phenotype.
Sex differences were observed in synovial fluid composition: males had increased levels of testosterone, MMPs, sulphated glycosaminoglycans (sGAGs), transforming growth factor (TGF)-β1, TGF-β2, and hepatocyte growth factor (HGF), whereas females had higher levels of macrophage stimulators (leukemia inhibitory factor (LIF), macrophage colony-stimulating factor (M-CSF), macrophage migration inhibitory factor (MIF)), pro-inflammatory mediators (growth-regulated oncogene α (GRO-α), monocyte chemotactic protein-3 (MCP-3), and monokine induced by gamma interferon (MIG)), inflammatory interleukins (IL2α, IL3, IL12p40, IL16, IL18, and TNF-β), TNF-α, and TNF-related apoptosis-inducing ligand (TRAIL) [25,26]. In chondroblast and chondrocyte cultures, female OA cells had lower mRNA expression of vitamin D 3 receptors (VDRs) and of protein disulfide isomerase A3 (PDIA3), and higher expression of estrogen receptors (ERs) compared to male cells. Female cells from minimally eroded surfaces had less VDR mRNA than male cells but comparable levels of mRNAs for the other three receptors, and higher mRNA levels of pro-inflammatory interleukins (IL1A, IL6, and IL8) in comparison to male cells. Both OA and non OA female cells had reduced expression of signaling molecules (WNT5A and DKK2) than male cells [26]. Moreover, Koelling analyzed by microarray the expression activities of many genes and found that 4.9% of genes (ESR1, ESR2, AR, Sox9, Runx2, Col I, and Col II) were differently expressed between the sexes [25]. Cells differed also in their responsiveness to sexual hormones: in general, cultures from females exerted responsiveness to estrogens and males to androgens and vitamin D 3 . Sexual hormones differently regulate the chondrogenic gene expression in a sex-specific and dose-dependent manner; their actions are also modulated by an upregulation and downregulation of receptors. In a study by Stummet al., the authors found an age dependent effect on loss of the Y-chromosome in chondrocyte cultures from male donors, but no sex differences in gene expression between cells from male and female donors; although their samples size (eight patients) was low and limited [27].
In osteoblast cultures, ER expression was increased in female cells in comparison to male cells, while VDR expression was increased in males. Treatment with 17β-estradiol (E2) induced an increase in alkaline phosphatase (ALP) and osteocalcin (OC) and a decrease in osteoprotegerin (OP) in females compared to males, while treatment with 1α,25(OH)2D3 induced an increase in ALP in males compared to females.

Canine Tissues
One study used osteoblasts isolated from subchondral, trabecular, and cortical bone from the femoral head of male and female dogs undergoing total hip replacement (THR) (36% female prevalence) [28]. Cells cultured in mineralizing medium were evaluated for cell viability and number, mineralization, and tissue non-specific alkaline phosphatase (TNAP) activity. Results showed no differences in any of the analysis parameters irrespective of the bone tissue types between males and females. Asonly one study, the results lack a suitable evidence-based level.

Rats
One study used fibroblast-like synoviocyte (FLS) isolated from the synovial membrane of female and male Sprague-Dawley rats (n =16 animals, 50% female prevalence) with temporomandibular joint (TMJ) OA chemically induced by Freund's adjuvant combined with monosodium iodoacetate (MIA) [29]. Cells cultured in a co-culture system with a macrophage cell line (NR8383) were treated with or without TNF-α, and evaluated through ELISA, Western blot, and RT-PCR. Results showed increasedinducible nitric oxide synthase (iNOS), IL-1β, and monocyte chemoattractant protein-1 (MCP-1) expression, and an increase in macrophages number and migration in female cells compared to males in the presence of TNF-α stimulation, while no difference without treatment was observed.

Preclinical in Vivo Studies
The results of the Systematic Review Centre for Laboratory animal Experimentation (SYRCLE) showed that higher risk of bias was observed in most papers at the items 1 "allocation bias", 3 "allocation concealment", and 6 "random selection for outcome assessment", with frequencies of 95%, 79%, and 89%, respectively. Items 2 "baseline characteristic" and 7 "blinding of outcome assessor" presented low risk of bias, with frequencies of 63% and 53%, respectively. The remaining items (4 "random housing of animals", 5 "blinding of caregivers and investigators", 8 "incomplete outcome data", 9 "selective outcome reporting", and 10 "other bias") presented a low-to-unclear risk of bias.
In two studies STR/Ort mice were used as spontaneous OA models naturally occurring: both authors observed that differences in bone structure between female and male mice may explain the OA development in this strain [33,39] with an age-related decline of bone marrow space in both males and females, adecrease in cancellous bone mineral density (BMD) in males compared to females, and an higher trabecular bone mass and greater cross-sectional of cortical bone in females than males [33,39].
By resorting to genetic modifications to induce OA, one study used mice lacking frizzled-related protein (Frzb), a gene associated to OA, to study its effect as well as the addition of voluntary running wheel exercise performance and OA development [35]. At 6 months, both female and male Frzb−/− mice showed reduced exercise performance compared tothe control. However, running exercise did not significantly affect severity of OA either in males or females.
Another study used mice with a mutation of matricellular protein NOV (nephroblastoma overexpressed) (NOV del3 ), commonly implicated in the maintenance of joint homeostasis, to determine if its disruption causes joint degeneration [38]. Results showed that compared to females, males exhibited severe OA-like pathology at 12 months (with high Osteoarthritis Research Society International (OARSI) score), affecting all tissues of the joint. The expression of collagen I (Col I), collagen X (Col X), and proliferating cell nuclear antigen (PCNA) in male mice was greater compared to females. The density of articular cartilage cells of control mice is maintained at a constant level from 2 to 12 months of age, whereas this is not the case in Nov del3 −/− mice, in which a significant increase in cell density was observed in males at 2 months, and a decrease at 6 and 12 months in both males and females. Therefore, NOV is necessary for maintenance of articular cartilage and for joint homeostasis, its disruption causes OA-like disease in Nov del3 −/− mice, with greater severity in males compared to females.  µCT (subchondral bone plate thickness and BV/TV of osteophytes) ↑OA severity and subchondral bone plate thickening in males (12-19+ months) and in females (21 months).
↑OA severity and subchondral bone plate thickening in males vs. females (4.5 and 12 months). =BV/TV of osteophytes in males and females (12 months) Huang et al. 2017 [32] Male and female CD-1, wild-type, and ADAMTS-5KO mice (lacking Adamts5 gene) on congenic C57BL/6J background (8-12 weeks-old)      One study used Atg5cKO mice lacking the autophagy-indispensable Atg5 gene in their chondrocytes, to analyze the development of spontaneous OA, both upon aging and post-traumatic surgically induced (partial MMT) [30]. With aging, results showed a development of OA particularly severe, with cell death of articular chondrocytes, in males compared to females. Upon surgery, males developed a more severe OA compared to females, however, in both sexesno differences were seen between the control and Atg5cKO animals. Therefore, autophagy protects articular cartilage and the abrogation of this process leads to the development of OA upon aging, especially in male mice, while it has no effect in the development of post-traumatic OA.
One study used C57BL/6J mice to investigate sex effects on pain-related behaviors and knee damage following MMT [40]. Results showed more severe cartilage damage in males than in females, butcomparable results between sexes in osteophyte formation. In addition, both sexes developed comparable heat hypersensitivity in the acute phase, although this was delayed in females compared to males. Locomotor activity was higher in females compared to males in the acute phase, indicating that females were more active at the beginning of the test, regardless of the knee condition.
Two studies used mice with a C57BL/6J background: wild-type, CD-1, and ADAMTS-5 (a disintegrin and metalloproteinase with thrombospondin motifs), the latter genetically modified lacking the Adamts5 gene responsible for damage to joint tissues, to evaluate OA development and mechanical allodynia, following DMM [32,36]. Results showed severe OA and mechanical allodynia in male wild-type mice compared to females, with thicker subchondral bone plates, cartilage degradation and bone volume fraction higher in old than young. In CD-1 mice, both sexes developed OA changes with concurrent mechanical allodynia; while both male and female ADAMTS-5 mice did not develop OA changes or mechanical allodynia. Therefore, these genetically modified mice are resistant to DMM-induced OA-like lesions and to the associated mechanical allodynia. Finally, one study used wild-type mice and integrin α1-deficient BALB/c mice to determine in serum metabolic changes and OA development following DMM surgery and treatment with the epidermal growth factor receptor (EGFR) inhibitor erlotinib [37]. Erlotinib treatment resulted in a decrease of multiple OA signs in the female mice including bone volume and density, subchondral bone thickness, and cartilage degradation, but not in males. Therefore, the effects of erlotinib treatment were 'on target' and sex dependent.
Moreover, two studies, always using genetically modified mice, evaluated metabolic effects of high-caloric diets on inflammatory processes and OA development [31,34]. Results from the first study showed that reduction into the ratio of polyunsaturated fatty acids (PUFAs) caused a modest decrease in the inflammatory cytokine IL6 both in males and females (52% vs. 46%), while TNF-α decreased to a lesser extent in male mice compared to female mice (19% vs. 41%). However, these systemic changes did not reduce osteophyte development, synovial hyperplasia, or cartilage degeneration [31]. In a second study, transgenic hCRP, LDLr-/-, and ApoE*3Leiden.CETP mice based on C57BL/6J background were used [34]. Results showed that long-term feeding of high-caloric diets showed significant OA aggravation upon feeding, whereas females gained less weight and did not develop diet-induced OA. ApoE*3Leiden.CETP mice rapidly developed obesity and were highly prone to cartilage degradation. Specifically, in males, OA scores were consistently higher than in females. In addition, ApoE*3Leiden.CETP males showed high scores for both osteophyte formation and synovitis, while females developed almost no osteophytes but demonstrated high synovitis scores and significantly higher plasma cholesterol levels. Therefore, diet-induced metabolic dysfunction per se does not necessarily lead to aggravated articular cartilage degradation and to OA development in mice on C57BL/6J background.
In the majority of studies (n = 3), OA was chemically induced by an injection of MIA or MIA combined with complete Freund's adjuvant (CFA) in the knee or TMJ [29,41,42].
In the remaining study, OA was surgically induced in the TMJ through the experimental creation of malocclusion [43].
Overall, in a range from 2 to 4 weeks, results showed larger OA-like degenerative regions in female rats compared to male rats, with cartilage loss and thinning, reduced bone volume, subchondral bone deterioration, osteophyte formation, chondrocytes loss, and synovial inflammation. In addition, from a molecular point of view, the gene expression and immunohistochemical signals of insulin-like growth factor-1 (IGF1), its type 1 receptor (IGFR1), IGF binding protein-3 (IGFBP3), extracellular protein, and matrix proteins, mostly Col II, were reduced in the female group compared to the male group, consistent with the observation of greater OA severity, cartilage damage, and disrupted reparative response in females [43]. Moreover, increased IHC staining and mRNA expression of iNOS, IL-1β, MCP-1, and greater amounts of CD68-positive macrophages infiltration were detected in the synovial membrane of females compared to that of males. These results indicate that the female synovial membrane is responsible for secreting greater amounts of pro-inflammatory factors, showing more severe synovitis, with hyperplastic synovial lining, as compared to males. Another important aspect considered by two of the mentioned studies was the sex and age differences in response to pain and tactility [41,42]. Results clearly demonstrated that older rats exhibited significantly more OA-like pain and a more pronounced and longer-lasting hyperalgesia than young rats; and female rats showed greater pain responses, higher susceptibility, and central sensitization as compared to males. Therefore, these observations indicate that females represent the most vulnerable group to OA and to related pain and inflammation conditions.

Guinea Pigs
Only one study used the guinea pig as an animal model (male and female Hartley, and female strain 13) to examine sexdifferences in the development of spontaneous knee OA [44]. Cartilage degeneration and subchondral BMD and meniscal mineral density (MD) in guinea pigs 12 months of age were examined. Results showed that male Hartley guinea pigs and female strain 13 guinea pigs had more severe cartilage degeneration and higher subchondral BMD than female Hartley guinea pigs. Meniscal MD in female Hartley guinea pigs and in female strain 13 guinea pigs was lower than the meniscal MD in the male Hartley guinea pigs. Finally, no significant differences were found for meniscal calcium content between males and females. In conclusion, the study demonstrated that higher subchondral BMD, but not meniscal MD, is associated with more severe cartilage degeneration in guinea pigs, suggesting that subchondral bone metabolisms may play a role in the development and/or progression of OA. However, being the only study performed in guinea pigs, no relevant evidences on sex differences can be extrapolated for this kind of model and OA.

Pigs/Mini-Pigs
In two 2015 studies, Kiapour et al. employed the pig (Yorkshire) or mini-pig (Yucatan) animal model to evaluate the role of sex on biomechanical outcomes in a model of surgically induced OA through the anterior cruciate ligament (ACL) transaction and repair [45,46]. Female and male animals underwent ACL reconstruction using conventional (bone-patellar tendon-bone graft) or bio-enhanced allografts with an extracellular matrix (ECM)-based scaffold and autologous platelet-rich plasma (PRP) (with absorbable or nonabsorbable sutures). In both studies, ACL structural, biomechanical, and histological properties, knee laxity, and cartilage damage were assessed after 15 weeks. Female pigs treated with bio-enhanced allografts and with absorbable sutures had poorer results from a biomechanical point of view with a lower linear stiffness, yield, and maximum load, and a greater knee laxity as compared to males. In addition, ACL grafts in males showed significantly greater vascularization when compared to females, while other histological parameters (cellular and collagen contents) showed no sex-related differences. Lastly, female pigs had a significantly larger area of cartilage damage after conventional ACL reconstruction than their male counterparts. This difference was not seen for the groups treated with bio-enhanced allografts, in particular with ECM-based scaffold and PRP, suggesting a greater protective effect of the bio-enhancement on the cartilage of the female knees. In conclusion, these studies showed that female pigs had significantly poorer biomechanical outcomes compared to males after 15 weeks of healing, when repaired with an absorbable suture. Finally, no significant differences were found for weight and range of motion (from pre-injury to post-injury) between males and females.

Baboons
One study used the baboon (Papiohamadryas) as an animal model to investigate sex-specific variation and effect of age on occurrence and severity of spontaneous knee OA in animals aged from 4.5 to 33 years [47]. By analyzing the joint of female and male baboons, an OA severity macroscopic score was assigned, recording articular cartilage degradation and presence/absence of osteophytes. Data showed a spontaneous OA more frequent and severe in older animals; males developed OA earlier with cartilage degeneration, while females progressed more quickly to advanced disease. However, as the only study performed in baboons, no relevant evidences on sex differences can be extrapolated for this kind of model and OA.

Discussion
Osteoarthritis is a serious pathology that causes persistent morbidity such as pain, fatigue, and disability, and has an enormous burden on a person's day-to-day quality of life [2]. Currently available treatments consist of analgesics, opioids, glucocorticoids, non-steroidal anti-inflammatory drugs (NSAIDs), viscosupplementation, physical exercise, diet, and surgery in the final stages of the disease [48,49]. However, these strategies are able to delay but not stop the progression of OA, with pain and inflammation relief and symptoms improvement. The OA pathophysiological mechanisms are difficult to understand because of the complexity of the disease in terms of quantity and the diversity of risk factors and tissues involved [10].
For years, OA was considered as an age-related pathology, but during the past decade, it has been observed that sex represents a well-known variable that influences its development. The lack of knowledge related to the sex differences in OA pathophysiology, especially at the molecular level, is partly due to the deficiency in studies that evaluate and examine this aspect, but also to the lack of relevant preclinical models. In this context, the present a review aimed to describe the results of in vitro and in vivo studies taking into consideration the gender balance and the related differences between sexes in the field of OA to highlight their possible implications in therapeutic approaches. Taking into account the numbers, in vitro studies analyzed 426 samples in total, of which 161 were retrieved from male and 266 from female subjects (62% female prevalence, independently from the type of tissue and human or animal donors); in vivo studies performed in mice used 458 animals of which 232 were male and 226 female (51% male prevalence), in rats 294 animals of which 153 were male and 141 female (52% male prevalence), in pigs/mini-pigs 58 animals of which 31 were male and 27 female (53% male prevalence), and in guinea pigs 15 animals of which 5 were male and 15 were female (75% female prevalence). Studies using baboons were gender balanced using 153 subjects. In total, the retrieved studies analyzed 1563 samples of which 735 were male and 828 were female (overall 53% female prevalence). These numbers, apart those in vitro in which samples from female subjects are prevalent, reflect the quite homogeneous distribution of sex (550 males vs. 562 females). However, preclinical studies testing on both sexes remain the minority of available studies in the literature.
In the in vitro studies, human or animal derived samples were used for histology, immunohistochemical analyses, growth factor, hormone measurements, and cell isolation, with the chondrocyte phenotype prevalently investigated. Apart from the scarce number of studies retrieved on this topic, sex differences regarded the molecular composition of synovial fluid, as highlighted in particular by the work of Pan [26]; males expressed more catabolic enzymes, such as MMPs, but also expressed many anabolic growth factors and sGAGs, while females had higher values of inflammatory cytokines that could explain the higher pain they experienced. This finding was confirmed in the work of Xue on rats [29]: they found greater pro-inflammatory cytokine production and gene expression (particularly, iNOS and IL-1β) in female compared to male synoviocytes, together with an increase of recruited macrophages and related MCP-1 production. These papers suppose that estrogens have an impact on an increased sensitivity to inflammatory stimuli and reactions in female, and that males have better compensatory and anabolic pathways. However, in these studies, estrogen evaluations were not performed. As far as cell cultures (chondrocytes and osteoblasts) were concerned, they differed both in terms of expression of hormone receptors and consequently in their responsiveness to hormonal stimulation. Generally, physiologic concentrations of estrogen help the regenerative potential of cartilage in females but not in males, while testosterone does it in males and not in females [25]. This suggests a sexual dimorphism in the development and regeneration mechanisms in OA. Although, a possible conversion of testosterone into estradiol, via the aromatase pathway, is presumed, thus, the regenerative effects of aromatase inhibitors (particularly in men) should be deeply investigated. Moreover, chondrocytes isolated from 20 OA patients exhibited sex differences in the expression of many genes: inflammatory cytokines, enzymes involved in matrix degradation, and WNT signaling molecules [26]. It should be noted that, when samples were retrieved from a smaller number of subjects, as in the work of Stumm or Meeson [27,28], no statistical evidence could be highlighted between sexes. So far, to highlight statistically significant differences related to sex, a large number of patients, samples, and cell culture replicates should be considered. From in vitro studies using samples from animal tissues, no clear conclusion can be drawn because of the lack of evidence due to the scarce number of utilized samples.
Benefits of the in vitro studies are the ease of performing cultures that in turn allow to study many aspects, from basal expression of phenotypic markers to gene expression and secretion of cartilage matrix molecules or pro-inflammatory mediators or DNA and sex chromosome differences, both as their intrinsic properties and in responses to hormonal fluctuation or stimulation. Moreover, the possibility of isolating cells from pathological OA tissues is of relevance to in vitro set up features typically expressed in OA pathology in order to improve the OA basic knowledge and to test regenerative/reparative therapeutic innovative strategies.
The drawbacks of in vitro studies are the possibility of cell transformation or dedifferentiation, the lack of microenvironment and surrounding matrix and, in primis, when dealing with pathological cultures, the difficulties in obtaining samples from healthy donors to have comparable controls. Many authors retrieved samples from non-loading areas or areas not affected by OA; however, it should be emphasized that they were harvested from a chronic inflamed microenvironment [25,26].
However, when testing new therapeutic approaches (i.e., biomaterials, cells, growth factors, drugs, biophysical therapies, and medical devices or their associations) for the treatment of OA, in vivo models cannot be ignored.
In the in vivo studies analyzed in this review, to evaluate the sex differences in OA development, the most commonly used animal model was the mouse. This model is a primary choice because of low maintenance costs, the progress in mice genetics, and easy availability, which allow the evaluation of changes in OA joints and allow the possibility to induce several genetic modifications [50]. However, other animals, such as rats, pigs/mini-pigs, guinea pigs, and baboons were also employed; the great heterogeneity of studies makes the comparison difficult. In addition to the animal species, another aspect to be considered is the choice of the model for OA induction.
In half of the studies included in this review, spontaneous models of OA were used, including naturally occurring osteoarthritis (22%), genetic modifications (16%), and metabolically induced models by high-caloric diets (11%); the other half of the studies comprised the surgical induction of OA by MMT, DMM or ACL (42%) and chemical induction (16%) by an injection of MIA in the joint. Knee OA was prevalently studied (84%) followed by TMJ (16%). Macroscopic and histological scores, X-ray, micro-computed tomography (µCT), biomechanical testing, and immunohistochemical and histological analyses were performed. Main sex differences regarded the OA development and severity in terms of bone architecture, osteophytes formation, synovial inflammation, cartilage degeneration, BMD, subchondral bone deterioration, and pain development. In particular, female rats showed a more severe synovial inflammation, cartilage destruction, and subchondral bone deterioration than males. Females were more sensitive to inflammatory stimulus and showed more extensive CD68-positive macrophage infiltration around the synovial membrane [29,43]. This may explain the higher pain in females compared to males [41,42]. In addition, molecular investigations show that the expression of IGF1, IGFR1, and Col II is reduced in females, consistent with the observation of greater OA severity compared to males [29,43]. Sex differences were also reported in a large animal model as pig/mini-pig. In this case, females showed worse biomechanical outcomes (i.e., stiffness, yield, and maximum load), weaker and less vascular grafts, more lax knees, and greater cartilage damage than males following ACL repair; these factors, however, improved with the addition of an ECM-based scaffold loaded with autologous PRP [45,46]. Naturally-occurring knee OA is common in mice, guinea pigs, and baboons. Despite the scarce number of papers retrieved on these models, sex differences in OA development highlighted that male baboons begin to show earlier and mild articular cartilage degeneration, while females progress more rapidly to advanced stages of disease [4]. Contrarily, for guinea pig and mouse models an inverted situation was observed, with bone architectural changes (for example, greater tibia curvature and ellipticity), more severe cartilage degeneration, and higher subchondral BMD in males compared to females [33,39,44]. This association between subchondral and cartilage changes suggests that subchondral bone metabolism could play a role in the development and/or progression of OA, affecting the biomechanical properties of the bone, and the release of cytokines and growth factors, which may promote cartilage destruction. Finally, also following surgical induction of OA, greater damages with higher allodynia and subchondral bone plate thickening in male than female mice were observed [32,36,40]. Using genetic modifications in mice, such as deletion of the Frzb and Atg5 gene, or Nov targeted mutation, a depletion of articular cartilage in cell density was observed, as well as an increase in collagen fiber thickness and more Col I and Col X positive cells in the damaged areas, which indicate abnormal differentiation of chondrocytes with a more evident effect in males than females [30,35,38]. Thus, specific genes are essential for protecting and maintaining the morphology and cell density of articular cartilage; their abrogation leads to the development of OA, especially in male mice. However, when an animal model is adopted, it should be taken into account that sex affects OA development; it is well known that in male mice spontaneous and induced OA occurs more frequently than in female mice, probably due to sexual hormonal variations, although their mechanisms remain unclear [24].
From this review it could be concluded that from the in vitro viewpoint there is a lack of representative and advanced 3D in vitro models of OA investigated to give evidences of sex and gender related differences. In the field of OA, many advanced in vitro models exist to study this pathology: they comprise direct and indirect bi-and tri-cultures, by cultivating two or three different cell phenotypes, 3D cell cultures, as chondrocyte micromasses cultured in the presence of an external pathological stimulus as IL1-β, TNF-α or synovial fluid from OA patients, or tissue and organ culture, as cartilage or osteochondral explant culture. These models have numerous advantages in comparison with two-dimensional (2D) or monolayer cultures, such as the retention of ECM and intrinsic growth factors, the possibility to extend culture experimental times up to many weeks, and the possibility to perform histology and biomechanics on samples [51].
Moreover, when we talk of sexual hormones and their actions in contributing to sex-related differences in OA, it should be empathized that many factors should be considered as relevant: (i) circulating levels of sex hormones, (ii) local production, and (iii) presence and functionality of receptors and related signaling pathways [19]. In some studies, the gene expression or protein levels of receptors have been evaluated but in the absence of their functional assays; so far, the detection and quantification of a molecule (at both protein or gene expression levels) should be coupled with functional tests or inhibition tests.
Of main importance, by setting robust and representative in vitro models, the reduction and/or replacement of animal use could be pursued, thus being compliant with ethical principles of the three R's of Russell and Burch (reduction, refinement, and replacement) and legislative decree (Directive 2010/63/EU). However, in this field, animal models remain irreplaceable as alternative models since in vitro, in silico, and robotic models cannot recreate the complete biological setting of an organism particularly in a trauma or pathology.
From the in vivo viewpoint, there is no "gold standard" animal model used in OA and sex related researches, but several models aim to reproduce processes and OA lesions [24,52]. Most of the studies analyzed in this review used small animals such as mice, rats, and guinea pigs, due to cost-effectiveness, ease of handling and housing, and opportunity for genetic manipulations. No studies have used medium and large animals such as rabbits, sheep, goat, orhorses [50,53]. These models allowed to identify sex differences, mainly in the histopathological aspects, showing a higher prevalence of OA and greater bone and cartilage architecture changes in males compared to females, and particularly in mice, regardless of the OA induction method. However, in contrary to in vitro works, the analysis of sex-related molecular mechanisms isstill lacking. The evident heterogeneity of the retrieved papers demonstrated the lack of consistency and evidence-based results that could help in understanding sex differences in the etiology, molecular events, and signaling pathways which ultimate in OA development. On the other hand, this is not surprising because of the limits in the identification of an appropriate animal model, especially in complex diseases, such as OA [54]. Since differences in type of model, animal species, and OA induction methods can lead to different experimental outcomes, there is the need to reduce experimental variability and to increase the reliability of data by boosting sample sizes, adopting in vitro advanced models, and revealing basic and molecular aspects involved, in particular in the era of omics, that can lead to a better understanding of how cellular processes are differently regulated in males and females.

Search Strategy
The present systematic literature review involved a systematic search carried out according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement in three electronic databases (PubMed, Scopus, and Web of Knowledge www.pubmed.com, www.scopus.com and www.webofknowledge.com). The search was applied with the following keywords: "osteoarthritis AND (gender OR sex) AND (in vitro model OR in vitro culture OR in vivo model OR animal model OR preclinical model)". The screening process and analysis were conducted separately by two independent observers (D.C. and M.T.). Firstly, the articles were screened by title and abstract, using the following inclusion criteria: in vitro and in vivo reports about gender differences in osteoarthritis models, full text, written in English language, published in the last 11 years (2009-2020). Exclusion criteria were articles written in other languages, reviews, abstracts, book chapters, clinical and cadaver studies, reports in which the model of osteoarthritis was absent and articles in which gender differences were not taken into consideration. Secondly, the reference lists of the included papers were screened to obtain further studies. Included papers were grouped according to the test model used, whether in vitro, in vivo or both. Thirdly, duplicates were removed. Disagreements were resolved by discussion and where resolution was not possible, a third reviewer was consulted (M.F.).

Data Extraction
General characteristics were extracted by D.C. and M.T., including details of the study (the first author's name and year of publication) as well as main findings. The data were extracted into a structured data collection form by one reviewer (D.C.) and checked for accuracy and completeness by a second one (M.T.). From each article dealing with an animal model, the following information was extracted: animal number and strain (age), osteoarthritis model, aim, and follow-up evaluations.
From each article dealing with an in vitro model the following information wasextracted: cell source and phenotype, osteoarthritis diagnosis, culture conditions, analyses, and experimental times. Disagreements were resolved by discussion and where resolution was not possible, a third reviewer was consulted (M.F.). We did not contact any study authors.

Methodologic Quality Assessment and Risk of Bias
In this systematic review, quality assessment of the in vitrostudies was not performed since there is no validated grading scale. In all in vivo studies, quality assessment was performed according to the Animals in Research Reporting In Vivo Experiments (ARRIVE) guidelines [55]. They consist of a 20-item checklist including all details for a minimum quality in reporting researches performed on animals.
Riskof bias in all in vivo studies was assessed using the SYRCLE tool for animal studies that consists of a 10-item checklist [56].Overall, studies were considered at low risk of bias if all criteria were satisfied, at unclear risk of bias if one or more criteria were partly satisfied, or at high risk of bias if one or more criteria were not met. Quality criteria and risk of bias assessments were performed by two independent authors (D.C. and M.T.). Any disagreement was resolved by consensus with a third reviewer (M.F.).

Conclusions
In conclusion, there is a growing body of evidence suggesting that sex differences exist in clinical and preclinical manifestations, symptoms, and OA progression. However, at present the conflicting results obtained in animal models show several limits because of the variety of OA animal models, and the difficulty in identifying the appropriate model [54]. At the molecular level, the available data reviewed in this paper, only seem to indicate that inflammation is higher in female than males, with poor information concerning specific molecules and targets and signaling pathways involved in OA pathogenesis and progression. These data need to be highly implemented. Further challenges need to be addressed, such as the major inclusion of sex in preclinical studies, the use of advanced in vitro models and, in the era of big data, innovative technological and methodological approaches able to profile patients' genome, transcriptome, proteome, and metabolome through multiomics analyses [57,58]. All of these approaches might help to deeply understand molecular mechanisms involved in OA development and progression, to identify key molecular players in the disease and to clarify any unrevealed male and female difference, with the final aim to characterize patient profiles and achieve a fully personalized gender-based medicine [59]. Funding: This research was carried out thanks to the funds from the Network IRCCS AGING "Rete Nazionale di Ricerca sull'invecchiamento e la longevità attiva-Implementazione della RoadMap nella ricerca sull'Aging (IRMA)".

Conflicts of Interest:
The authors declare no conflicts of interest.