Breast tissue stiffness plays a role in the development of breast cancer (13), with mammary tissue undergoing up to 20-fold stiffening during tumor progression (14). Non-tumorigenic MCF-10A human breast epithelial cells cultured in a matrix of a stiffness of ~1 kPa has been reported to cause tumor-like development in these cells compared with those cultured on native mammary tissue stiffness of ~150 Pa, which did not develop a cancerous phenotype (15). Matrix stiffness affects genome accessibility, mediating the induction of malignancy via increased matrix stiffness (16). Stiffening of the ECM causes non-tumorigenic mammary epithelial cells to develop an invasive phenotype and activate mesenchymal gene expression (17). Matrix stiffening from <150 Pa to >3,000 Pa, at specific time points, was reported to promote epithelial-mesenchymal transition (EMT) in mammary epithelial cells (14). Matrix stiffness-induced EMT was reported in breast epithelial cells when grown on 5.7 kPa gels that mimic tumor stiffness, but not when cultured on gels that mimic the natural stiffness of breast tissue, via a TWIST1–G3BP2 mechanotransduction pathway (18). Malignant BCCs have been shown to adapt and survive on collagen matrices of different densities, whereas non-metastatic BCCs cannot (19). Cancer progression is also associated with significant softening of tumor epithelial cells relative to normal mammary epithelium, as detected by indentation-type atomic force microscopy (AFM), as well as broadening stiffness distribution of the TME (20). This demonstrates the dynamic nature of extracellular matrix remodeling in cancer progression.

The stiffness gradient within a tumor also affects the migration ability of BCCs. Anisotropic stiffness gradients caused by remodeling of the ECM triggers directional cell migration and increases BCCs ability to migrate and metastasize (21). Different types of BCCs respond differently to various levels of matrix stiffness. Triple-negative BCCs favor honeycomb geometries with greater stiffness, adenocarcinoma MCF-7 cells prefer mesh scaffolds with low elastic modulus, and pre-malignant cells favor aligned scaffolds with high stiffness and contact guidance (22). Additionally, it has been reported that rigidity sensing by healthy mammary cells is impaired in BCCs but remains active in normal mammary cells (23). This is further supported by work by Tse et al., who showed that compressive stress that accumulates during tumor growth can promote the migration of BCCs by stimulating the formation of leader cells and increasing cell–substrate adhesion (24).

Bone is a highly heterogeneous tissue. Elastic and viscoelastic measurements on cortical and trabecular bone have reported varying Young's/shear moduli in the GPa range (25–27). Differences within bone and bone marrow, determined by cell type and ECM components, also plays a significant role in bone metastasis in vivo (28). The endosteal surface is reported to be over 35 kPa, marrow sinusoids are more compliant (2–10 kPa), while the densely-populated central marrow is much softer (0.3 kPa) (29). Matrix elasticity has been shown to drive the differentiation of mesenchymal stromal cells (MSCs) (30). Furthermore, matrix stiffness modulates cellular stiffness during the differentiation process: The rate of increase in cellular Young's modulus during osteogenic differentiation has been positively correlated with matrix stiffness (31). Bone matrix elasticity also plays a role in the regulation of immune response by MSCs: Soft extracellular matrices maximize the ability of MSCs to produce paracrine factors that are implicated in monocyte production and chemotaxis upon inflammatory stimulation by tumor necrosis factor–α (TNFα) (32).

Matrix elasticity has also been shown to prime breast epithelial cells and regulate their collective migration (33). Epithelial cells primed on stiff matrices were reported to migrate faster, display higher actomyosin expression, form larger focal adhesions, and retain nuclear Yes-associated protein (YAP), even when cells had arrived at a softer secondary matrix. This showed that epithelial cells have memory of past matrix stiffness (33). Bone matrices with high elastic moduli, e.g., 380 kPa, can impart a bone-destructive phenotype in triple-negative MDA-MB-231 breast cancer cells via regulating integrin-β3 (Iβ3) and transforming growth factor-β (TGF-β) (34). Furthermore, parathyroid hormone-related protein (associated with tumor destruction of bone) and Iβ3 expression increases with increasing substrate rigidity (34), promoting the progression and symptoms of bone metastasis. A reduction in matrix elasticity from 35 to 0.5 kPa was shown to increase MSCs homing to the breast TME (35). Migration of MSCs to the TME has been associated with an enhancement of tumor growth due to BCCs stimulating de novo secretion of the chemokine CCL5 from MSCs. CCL5 acts on BCCs to enhance their motility, invasion and metastatic capability (36). Migration rate of MDA-MB-231 cells toward a human trabecular bone explant was enhanced when cultured in stiffer gelatin hydrogels relative to softer ones, signifying an association between mechanical properties of the matrix and chemotactic signaling (37).

Human tissues exhibit viscoelasticity, which is a time-dependent response to loading or deformation. Viscoelastic materials display a combination of elastic and viscous characteristics (38). Breast cancer progression has been linked to variations in tissue viscoelasticity, with remarkable differences reported between benign and malignant tumors (39). Additionally, MSCs have been found to exhibit enhanced proliferation and osteogenic differentiation when cultured in 3D viscoelastic alginate hydrogels that exhibit faster stress relaxation. These effects were mediated by actomyosin contractility, adhesion ligand binding and ligand clustering (40).

Topography is defined as the microscopic surface features that cells interact with, and is determined by the hierarchical structure of the ECM and roughness (41). Topography of the breast TME is a mixture of ridges, pores and collagen fibers, which makes it challenging to mimic in vitro (42). Topography regulates the expression and translation of human epidermal growth factor receptor type 2 (HER-2) (42), which is a predictive biomarker of the clinical efficiency of chemotherapeutic agents (43). While normal breast epithelial cells has been reported to show increased proliferation on topographically-patterned surfaces, defined by the presence of polystyrene beads of 23, 300, or 400 nm in diameter, estrogen receptor-positive MCF-7 breast cancer cell line showed lower proliferation (44). Furthermore, different BCC populations show distinctive behaviors on varying patterned culture substrates, dependent on cancer subtype. Metastatic MDA-MB-231 cells display uneven vinculin distribution when cultured on patterned substrates but not flat ones. On the contrary, non-metastatic MCF-7 cells do not exhibit this non-uniform vinculin distribution regardless of the substrates used. Significant difference in motility was also reported between the two cell lines (45).

Trabecular bone consists of a highly porous lattice network encompassing marrow and determines the topography of the majority of bone metastatic sites (29). Topography can be more dominant than biochemical cues and matrix elasticity in determining behavior of MSCs (30), and has been used to guide their differentiation (46, 47). Adhesion of MSCs was also shown to be bi-phasically regulated by interfacial roughness (48). In the context of bone remodeling, previous studies have reported the impact of surface topography on osteoclast differentiation and resorption (49, 50).

Guidance from matrix topography is a key regulator of metastasis. Nuclei of BCCs undergo extensive deformation when migrating through tight interstitial spaces (51). Topotaxis (directed cell migration guided by a gradient of topographical features) has also been suggested to influence the invasiveness of cancers (52). Benign and metastatic BCCs favor movement parallel to nano-ridges, showing greater speeds relative to flat surfaces. In contrast, asymmetric sawtooth structures create a unidirectional, cell type-dependent bias in the movement of BCCs (53).

Shear stress, which is induced by fluid flow, contributes to proliferation and metastasis of tumor cells. Flow can either be through vasculature or through the ECM, the latter termed interstitial flow (54). Whilst different tissues support different flow rates, flow within solid tumors is generally much lower than in healthy tissue due to underdeveloped vasculature (55).

Flow within breast tissue has a significant impact on BCCs within the primary tumor. Higher flow rates promote EMT and increase cell metastatic capability (56), which promotes the conversion of BCCs into cancer stem-like cells and tumor-initiating cells by the suppression of extracellular signal-related kinase/glycogen synthase kinase 3β pathway (56). Interstitial fluid flow also generates mechanical forces that modulate tumor growth (57). Interstitial shear stress within breast tumor tissues is reported to create pressures of around 0.001 Pa (58) while circulating cancer cells can experience up to 3 Pa (59). When extravasation of cancer cell clusters (up to 20 cells) were investigated in a microfluidic device containing micro-channels with narrow constrictions of 5–10 μm, most were observed to migrate through the narrowest constrictions of 5 μm, unfolding into chain-like arrangements as they approached them and forming clusters as they exit. Clusters of over five cells displayed a migration velocity similar to the sum of resistances of the cluster's cells (60). Only 0.02% of circulating tumor cells survive to successfully undergo metastasis owing to anoikis, natural killer cells and mechanical damage by shear stress (61). Fluid flow also creates shear stress on endothelial cells, which leads to vasculature remodeling, cytoskeletal rearrangement and changes in the expression of transcriptional genes (62).

Interstitial fluid flow in tissues exposed to repeated mechanical loading, such as bone, induces variable levels of fluid shear stress of up to 3 Pa (63). Flow within bone causes osteocytes to release signaling factors, including nitric oxide and prostaglandin E₂. These influence osteoblast behavior and trigger osteocytes to release receptor activator of nuclear factor kappa-B ligand (RANKL) and osteoprotegerin, the ratio of which is indicative of the amount of bone resorption (64). Laminar flow within bone is altered by compressive loading, muscle contractions, blood pressure changes and mechanical loading, and this generates varying shear stress on cells (65). Additionally, different cell types experience flow differently within the bone. Osteoclasts experience high flow and shear stress as they are directly exposed to flow through the lacunar-canalicular system (66). Osteoblasts, however, experience less flow and shear stress (67). Flow within the bone is also altered by the degree of mineralization, which changes the shear stress forces that cells experience (64).

Flow within bone has a significant impact on metastatic BCCs present, as increased bone remodeling can promote proliferation of BCCs (68). The “vicious cycle theory” states that osteoclastic bone resorption releases bone-derived growth factors, including TGF-β and insulin-like growth factor-1, which stimulates metastatic cancer cells to release osteoclast-stimulating factors, such as parathyroid hormone-related protein. These in turn causes further bone destruction through upregulated RANKL expression in MSCs and osteoblasts. This promotes the differentiation of RANK-expressing osteoclast precursor cells into mature osteoclasts, which resorb bone and so the cycle continues (69).

Bone ECM comprises collagen, laminin, fibronectin, adhesive proteins, adipocytes, proteoglycans (70), and natural bone apatite (71). Around 60% of the dry weight of adult human cortical bone is mineral (71, 72). Mineral in bone is documented to be chemically similar to hydroxyapatite [HAp; Ca₁₀(PO₄)₆(OH)₂] although it tends to be less crystalline, more soluble, and highly substituted in the human body (73). HAp is known to interact with a variety of proteins, such as osteopontin, to support cell adhesion (74). Crystallinity, particle size, and substitution of HAp all affect cellular behavior. Highly crystalline HAp exhibits increased osteoconductivity (75). Smaller, less crystalline HAp enhances cell adhesion and proliferation whereas larger, more crystalline HAp particles enhance expression of the osteolytic factor interleukin-8 (IL-8) in BCCs (76), but not in the more benign MCF-7 cells (77). It has also been revealed that HAp has a role in the adhesion of BCCs (78) and causes BCCs to secrete higher levels of cytokines, including TGF-β and pro-osteoclastic IL-8, relative to non-mineralized scaffolds (79). IL-8 drives bone degradation (76) and results in the clinical symptoms of bone metastases, namely bone pain and pathological fractures (80).

It has been long understood that HAp, in the form of microcalcifications, is also found within breast tissue of breast cancer patients (81). The processes by which these microcalcifications form are largely unknown despite being positively correlated with BCCs mitogenesis (81). Microcalcifications with larger HAp particles have been more strongly associated with malignant breast cancer than microcalcifications made up of smaller particles (82). HAp has been implicated in upregulating the expression of several matrix metalloproteinases (MMPs), including MMP-2, MMP-9, and MMP-13, which promote BCCs migration due to decreased ECM elasticity (81). The existence of a direct relationship between mineral deposition and the ability of BCCs to metastasize to distant organs suggests a link between mineral deposition in the breast and metastasis. This has been suggested to be through the regulation of osteopontin expression (83).

HAp has also been implicated in promoting the colonization of metastatic BCCs in bone (84, 85). Although processes by which BCCs preferentially target bone and induce pathological remodeling remain unclear, there is increasing evidence that HAp is involved in this. HAp increases the expression of stromal-derived factor-1 (SDF-1) which promotes the activation of the migratory axis in BCCs thus directing their circulation to bone via increased signaling through the chemoattractant CXCR4 (85). Mineral-mediated changes in the collagen network were reported to result in increased cell motility (86). However, HAp inhibited directed migration of BCCs (86). These results suggest that mineralization of collagen fibrils reduces tumor cell adhesion, which may affect skeletal homing of disseminated tumor cells in the early stages of breast cancer metastasis (86).

Studies into the influence of matrix elasticity on breast and bone tissues gives rise to fundamental knowledge that can be applied to OOAC platform design. Methacrylated hyaluronic acid hydrogels have been used to mimic the stiffening of breast tissue in microwell plates to model epithelial-mesenchymal plasticity and the metastatic capability of BCCs (14). Polyethylene glycol (PEG) have also been introduced into collagen gels to alter their stiffnesses without altering their microstructures (109). These can be applied within OOAC models, where PEG is mixed with cells and pre-gelled collagen solution before inserting into the platform. Gel formation would then be induced by changing pH. PEG changes the non-covalent forces that are responsible for collagen self-assembly, so varying the concentration of PEG can vary the stiffness of collagen gel (109). With UV-crosslinked hydrogels, such as PEG and its derivatives, stiffness can be tailored by modifying macromer content and molecular weight (30). Another example are PEG-based hydrogels containing enzymatically degradable peptide sequences with varying concentrations of the integrin ligating peptide RGDS, and the non-degradable, co-monomer N-vinyl pyrrolidinone (NVP) (110). This allows tuning of cross-linking density and degradation, and has been shown to generate different phenotypic states in BCCs (110). However, changes in the content of the monomer and crosslinking will alter the porosity of the matrix; this impacts protein tethering, which in turn influences stem cell differentiation (111). Whilst this is an effective method for the inclusion of customizable matrix stiffness within OOAC platforms, it does not permit dynamic changes in stiffness of the TME as seen in cancer progression.

Conventional hydrogels offer inadequate dynamic signals, such as spatial or temporal control of formation/degradation, and the mechanical strength required to mimic bone stiffness. One way to mimic the dynamic TME is to use tunable hydrogels, which allow flexible control of the biophysical features of the gels in question to study how changes in matrix stiffness affect a cell population over time. One example is photodegradable hydrogels. Light-responsive biomaterials are of interest as an approach for spatiotemporal regulation of cell dynamics. These can be modified in a controlled manner by illuminating with light of specific intensity and wavelength (112). Progress has been made in the synthesis of dynamically tunable photo-responsive biomaterials (113, 114). For 4D cell culture (where 3D culture substrates are used to present changing biophysical or biochemical properties over time), chemically-decorated hydrogels have been fabricated with a combination of photochemistry and other orthogonal click reactions (115, 116). Temporal and spatial control of in vivo presentation of cell-adhesive RGD peptides has also been reported using a protecting group that can be removed via transdermal light exposure to activate the peptide (117). Temporal stiffening of substrates is key to mimicking breast cancer progression in vitro, and was reported to induce invasion from MCF-10A mammary acini (15). Innovative engineering of the mechanics and multi-scale architecture of tunable hydrogels is needed to achieve physiologically relevant features for in vitro modeling.

Cells often exhibit directed migration in response to the rigidity of the surrounding environment, migrating toward stiffer regions. This mechanosensitive behavior is termed durotaxis. Matrix elasticity gradients can be created to mimic in vivo stiffness gradients by exploiting chemical and mechanical gradients within OOAC platforms. This has been previously used to investigate haptotactic behavior in neurite growth, where a matrix elasticity gradient was created across an H-shaped collagen gel to investigate neurite growth (118). Additionally, matrix stiffness of the secondary site needs to be considered as bone ECM is much stiffer than breast tumor ECM.

A number of attempts have been made to incorporate topography in microfluidic devices. Replication of bioinspired surfaces with tunable 3D topography has been achieved using microinjection compression molding with novel dual-layer molds to create open microfluidic devices (119). Yang et al. fabricated nano-patterned surfaces using photolithography and electron beam lithography and stitched them together to assemble PDMS-based microfluidic platforms (120). Integration of patterned electrospun fibers into microfluidic systems to create aligned and random patterns in PDMS microfluidic chips was also used to create a complex microenvironment to mimic that in vivo, as validated by neural stem cell alignment (121). Another method to integrate topography is by surface patterning, for example PDMS grates. Grates with widths from 2 to 4 μm have been shown to cause MDA-MB-231 cells to align along the grating length and to increase the extension and spreading of these cells compared to planar controls (122). Topography can also be exploited in OOAC platforms to mimic BCCs migration seen in vivo. For example, nanoscale ridges have been shown to enhance the movement of benign and metastatic BCCs relative to flat surfaces (53).

Polyacrylamide hydrogels with different patterns, including spirals, stars, and squares, have been employed to study the concept of ‘topographical memory', showing that topography can affect histone modifications and prime cancer cells to a tumorigenic state (123). Different shaped fibers have been tested for their topographical effect on BCCs, with curved polystyrene fibers causing MDA-MB-231 cells to exhibit sensitivity to curvature, in terms of eccentricity (measurement of protrusion width), compared to flat ribbons (124). Varying the diameter of type I collagen fibers has also been investigated, where BCCs exhibited ~20% increase in cell spreading on type I collagen fibers with a diameter of 850 nm compared to 550 nm. Cell invasiveness also increased with fiber diameter, although proliferation was unchanged (125). This shows that cell response and tumor invasiveness are dependent on substrate surface patterns and collagen fiber diameter, and highlights the importance of considering topography of the microenvironment created within OOAC platforms.

Constructing appropriate topographical cues within metastasis models is essential, in terms of both the surfaces which cells have to travel over and the environment which they must travel through. The dimensionality, length and scale of the physical microenvironment has been reported to influence MDA-MB-231 triple-negative carcinoma cell signaling and decision-making at intersections of micro-contact printed lines (126). Cells confined in narrow micro-channels, and therefore had fully explored their microenvironment, favored entry into wider branches at bifurcation points, whereas cells in wider channels made pore size–independent decisions (126). Cancer cells traveling through ECM pores, modeled using a microfluidic device containing micro-channels of varying widths (3–50 μm), was shown to prevent the normal increase in cellular size during cell cycle progression, resulting in a reduction in the frequency of cell division and an increase in the frequency of abnormal division events (127). Pore size reduction within collagen-PEG mesh networks causes BCCs to exhibit reduced cell spreading, increased cell-cell adhesion protein expression, larger cell aggregates and triggers morphogenesis (128).

One way of incorporating topographical cues within OOAC platforms to mimic topographical variations sensed by metastatic BCCs would be to use temperature- or pH-sensitive hydrogels. Tunable topographies can be created using shape-changing, stimuli-responsive polymers or self-folding films to provide cells with an appropriate spatial environment for cells (129). Thermoresponsive hydrogels have been created with poly(N-isopropylacryalmide)-based copolymers (130–132) and exhibit low critical solution temperature behavior, swelling in water at lower temperatures and contract with increasing temperature (129). Temperature-sensitive hydrogels can be used for the fabrication of bilayers, which allow reversible folding and unfolding at low and elevated temperature, respectively, and have been used for cell encapsulation (130, 131). The topography of these hydrogels can be altered: Swelling will stretch the culture surface by expanding it. By utilizing shape transformation of these tunable hydrogels there is greater potential for the creation of dynamic structures with high resolution, which are difficult to achieve by other biofabrication techniques, such as bioprinting (129). Topography can also be created and controlled by utilizing culture substrates such as polymeric microparticles within OOAC platforms. Topographically-designed microparticles have been used to drive MSCs down an osteogenic lineage without the use of exogenous osteo-inductive factors (47). In addition, surface roughness is important in bone OOAC platforms as osteogenic differentiation can be enhanced in MSCs by culturing them on growth surfaces with intermediate roughness (48). This is controlled through roughness-regulated expression of YAP (48). Substrate roughness is another design parameter to consider. Sandblasting of the surface of poly(methyl methacrylate) films with alumina grains has been used to alter roughness, resulting in increased adhesion and migration of vascular cells with higher surface roughness. Increased cell adhesion was attributed to higher adsorption of proteins, such as fibronectin and collagen I, on the surface of the films (133).

Topography can also be controlled on the surface of the membranes used within OOAC platforms. These membranes are often suboptimal, as they are typically flat and do not replicate the shape or surface morphology of a tissue (134). Microfabrication techniques can be applied to create porous membranes with suitable porosity, shape and surface morphology to match the requirements of the tissue being modeled. Porous membranes with micron-sized features can be made by electrospinning (135). Soft lithography and thermoforming are used to prepare porous, micro-structured membranes for other applications, where thermoforming is combined with ion track etching to create microstructures and pores, respectively (134). Phase separation micro-molding is able to create patterns and pores in membranes in one step by inducing phase separation (136).

Flow of cell culture media within OOAC platforms generates pressures and oxygen gradients which affect cell behavior and survival (65). As there are different flow rates within the body, different organs modeled within OOAC platforms require different flow rates. There are several ways to recreate flow in these devices, including the use of peristaltic pumps, syringe pumps, or rocker plates. These different methods can create significantly different flow rates, with peristaltic pumps creating the strongest rates and rocker plates the slowest ones (137). The technique for perfusing an OOAC platform therefore needs to be carefully selected for each organ component.

Microfluidic channels filled with gels can be used to study the effects of interstitial flow. Haessler et al. used this technique to study the effects of 10 μm s⁻¹ interstitial flow on MDA-MB-231 cells (138). Gel-free systems or channels coated in a thin layer of gel only can be used to study laminar flow. For example, a channel can be lined with gelatin before allowing fluid to flow through the channel with endothelial cells to form a single channel blood vessel (139). Lanz et al. seeded metastatic triple negative BCCs in an artificial ECM under static and dynamic conditions, with cells displaying a proliferation rate of 80% in the perfused system compared to 60% in the static one (104). Pradhan et al. fabricated a high and low perfusion chip to mimic cancer-ECM-endothelial interactions. This showed that BCCs elongated forming colonies similar to those formed in vivo in high perfusion regions, whereas BCCs appeared rounded, relatively dark and unhealthy in low perfusion region, possibly indicative of a quiescent state due to lack of nutrient availability (140). Extreme flow rates are undesirable within breast OOAC platforms, as only peripheral and circulating BCC should be exposed to high flow rates (56). BCCs cultured within 3D ECM are more representative of in vivo BCCs, displaying greater proliferation, invasiveness, chemoresistance and higher plasminogen activator urokinase signaling when subjected to interstitial flow than cells in static conditions (141).

Flow within bone and vasculature representations is also of great importance. Fluid shear stress governs the binding of tumor cells to endothelial cells. Breast cancer cells were reported to not adhere to endothelial cells directly under low shear stress of 0.5–2 dyn cm⁻², instead forming a tumor-monocyte complex before binding to endothelial cells (142). MSCs grown on a bone-on-a-chip system under flow (30 μL/h; 0.346 mPa) showed improved survival and proliferation relative to cells grown under static conditions (100). Fluid shear stress was reported to enhance osteogenic differentiation in an osteogenesis-on-a-chip device, but was not sufficient to induce it on its own (143). Conditioned medium from oscillatory flow-stimulated osteocytes significantly increased migration and reduced apoptosis of BCCs (144). In another study, MDA-MB-231 BCCs were cultured inside a microfluidic channel lined with human endothelial cells and adjacent to another channel containing osteocyte-like cells. Physiologically relevant oscillatory fluid flow (1 Pa, 1 Hz) was applied and hydrogel-filled side channels allowed real-time extravasation to be observed. The applied fluid flow induced intracellular calcium responses in osteocytes (3.71-fold increase), with extravasation distance significantly reduced by mechanically-stimulated osteocytes relative to static osteocytes (145). Additionally, a physiologically relevant perfusable vasculature to investigate intravasation, migration, and extravasation, cannot be achieved without flow (62) and shear stress is required to maintain the system (146). Once vasculature has developed, flow continues to have a vital role in decreasing the permeability and increasing the stability of vessel walls within the vasculature (146).

Materials used to create OOAC platforms, such as elastomers, are commonly formed from acrylate monomers, which can adsorb different amounts of fibronectin. This affects cell attachment (147) and integrin signaling, which can promote osteoblast differentiation (148). Furthermore, elastomers can affect the composition of culture medium. PDMS, for example, has been shown both to adsorb small molecules and release other chemicals into the device (149). This can be overcome by priming devices with ECM components such as collagen, or by using a lipophilic coating (150).

A key part of the design of the bone component of a metastasis-on-a-chip platform is the inclusion of a mineral component. HAp-coated ceramic scaffolds with MSCs and hematopoietic stem/progenitor cells have already been successfully integrated into a microfluidic device (102). HAp has been shown to affect viability, proliferation and cytoplasmic volume of metastatic colorectal and gastric cancer cells when incorporated in a bone-mimicking OOAC model (151). HAp can be precipitated in bulk solution as anisotropic needles, and 50 nm pores support the formation of HAp rods that are comparably oriented to native mineralized collagen fibrils (152). If precipitated within the confines of cylindrical pores that offer physiologically relevant gap regions (>100 nm diameter), polycrystalline HAp rods are formed (152, 153).

The inclusion of HAp must be carefully considered, as it may have a negative impact on vascularization. While Jusoh et al. reported that HAp has a positive effect on sprouting angiogenesis (154), increasing HAp concentration was reported to result in a decrease of the number of blood vessel sprouts, with angiogenesis sprouting length displaying a biphasic response to increasing HAp concentration (151).

It has been demonstrated that epidermal growth factor and CXCL12 gradients cooperatively increase tumor cell motility and are important in regulating cell migration in 3D environments (155). Chemical gradients can be included within OOAC platforms by introducing these molecules through microfluidic gradient generators (156). Shared media and interconnecting chambers allow media exchange within the device and permit natural gradients to form. In this way, cytokines and growth factors released from cells in secondary locations, such as MSCs in bone microenvironment, can spread through the OOAC to create physiologically relevant gradients.

Components of the ECM can be studied within OOAC platforms. One example of this is a compartmentalized platform developed by Sung et al., which enabled the analysis of the intrinsic second harmonic generation signal of collagen. This offered a label-free assessment of collagen remodeling in the model (157).