New 3D Bioprinted Osteosarcoma Model for Drug Testing

Osteosarcoma research (like much of the cancer field in general) has had a persistent problem for years. The standard 2D dish is too simple, animal models are slow and messy in their own ways, and somewhere in between there is still a real need for a tumour model that behaves a little more like the disease researchers are actually trying to treat.

This study out of the University of Lisbon takes a serious run at that gap with a 3D bioprinted osteosarcoma model built from MG-63 spheroids embedded in a decellularized extracellular matrix (dECM)-enriched bioink. Early in the workflow, the team used a CellScale UniVert to measure compressive stiffness, which matters here because the mechanics of the scaffold sit in the background of everything else the model is supposed to do.

The dECM-Enriched Osteosarcoma Model

What makes this study stand out is how many “real-world” constraints it tries to keep in the same workflow. The authors combined tumour spheroids (to preserve cell-cell signalling), osteosarcoma-derived ECM cues (to bring in tumour-specific matrix context), and a bioink/scaffold format that can actually be printed the same way each time and then used for drug response testing with doxorubicin.

The result is a model that sits between the usual endpoints. It does not read out like a 2D monolayer, and it is not the same as a scaffold-free spheroid. In their drug experiments, the printed construct showed a stronger resistance pattern than the simpler formats, which is the kind of shift researchers are often aiming for in any model, and of course in a 3D bioprinted osteosarcoma model.

That also puts the work in the cancer mechanobiology lane, even though the endpoint is drug response. The mechanical side is baked into the model design: stiffness, structure, and the surrounding matrix are part of what the cells are “reading” during culture and treatment. If the goal is to say something credible about osteosarcoma in a 3D setting, the scaffold and bioink cannot be treated as background. They need enough basic measurements on the scaffold and bioink that the drug-response data can be discussed in the context of the physical setup, not treated as a black box.

Why build a 3D bioprinted osteosarcoma model at all?

A 3D Bioprinted osteosarcoma model is mainly about closing the gap between what we can test easily and what we are trying to predict. Osteosarcoma can behave very differently depending on local microenvironment and treatment history, and simple 2D readouts often flatten those differences. The authors frame that as one reason conventional preclinical systems keep coming up short. Their approach was to combine scaffold-free spheroids with a scaffold-based matrix, then organize that system using extrusion bioprinting. The idea is simple enough: preserve the 3D tumour structure and cell-cell interactions of spheroids, but also add some ECM context that scaffold-free systems do not really provide.

They generated MG-63-derived dECM, confirmed substantial DNA removal after decellularization, and retained key ECM proteins including collagen I, fibronectin, and laminin. From there, they compared two related inks, one standard and one dECM-enriched, then moved toward the final bioprinted construct.

There is a nice overlap here with our earlier post on scaffold stiffness for in vitro liver models. Different disease, different setup, but the same lesson applies. Once you leave flat culture, the scaffold stops being a neutral container and starts influencing the readout. Growth patterns, transport, and drug response can shift just because the physical context changed.

How the 3D bioprinted osteosarcoma model was built

The final 3D bioprinted osteosarcoma model was assembled in steps, not printed as a simple “cells in gel” construct. The authors first generated MG-63 spheroids, then chose a spheroid size that could pass through the nozzle reliably without blocking flow. Those spheroids were mixed into the candidate inks rather than being seeded onto a printed scaffold afterward. For the architecture, they printed a five-layer mesh with irregular polygonal pores, aiming for a trabecular-style layout instead of a uniform grid.

That choice gives the construct a more bone-like architectural feel, which is sensible for osteosarcoma even if it is still a simplified system.

They then printed the spheroid-laden constructs and checked whether the spheroids were still intact afterwards. Diameter, circularity, and viability were broadly maintained after bioprinting, which is one of the practical things that stands out in the paper. A lot of the interest in spheroid-laden bioinks tends to run into the same obvious concern: can you actually extrude them without damaging the model you are trying to preserve? Here, it appears the answer was mostly yes.

That connects well with our earlier highlight on mechanical analysis of tissue spheroids. The two platforms are different, but both point to the same idea: spheroids are being handled more like defined, repeatable test samples, not just a convenient way to cluster cells.

Using UniVert compression testing to characterize the bioink

Before the drug testing results, there is a quieter but important part of the study: the mechanical characterization. The authors used a UniVert with a 10 N load cell to perform unconfined compression testing on the hydrogel formulations and calculate compressive modulus from the initial linear region of the stress-strain response. These UniVert measurements are a key “baseline check” for the model. If you do not know the stiffness range of the printed matrix, it becomes hard to separate biological effects from material effects.

The two inks, GMA and GMA-d, were essentially matched in compressive modulus (about 7.27 kPa and 7.24 kPa). So adding dECM did not change bulk stiffness in compression, even though it improved printability and supported good viability. That is a useful result in its own right. It means the later differences in model performance were not just a simple story of one ink being mechanically stiffer than the other.

The paper also measured gelation, swelling, degradation, filament width, and printability factor, but the UniVert result anchors the mechanics. It gives the model a defined stiffness range and lets the authors discuss drug-response behaviour with at least some mechanical context rather than hand-waving around it. In cancer mechanobiology, that sort of context matters.

What the 3D bioprinted osteosarcoma model did under doxorubicin

This is where the paper starts to justify the effort. To validate the 3D bioprinted osteosarcoma model, the authors compared doxorubicin response across three systems: 2D monolayer cells, scaffold-free spheroids, and the bioprinted model. The broad pattern was clear. Across the three conditions, the bioprinted group stayed “alive” at doses where the other formats dropped off. The spheroids inside the printed scaffold also kept more of their projected area under treatment than the free spheroids.

The viability values in the paper lay that out. At 0.1, 0.5, and 1 µg mL⁻¹ doxorubicin, 2D cells were reported at ~55.3%, 30.4%, and 15.24%. Scaffold-free spheroids were ~72.3%, 40.4%, and 28.6%. The 3D bioprinted osteosarcoma model was ~78.4%, 69.0%, and 63.1%.

The gene expression data moved in the same direction. The bioprinted model showed elevated expression of VEGF and RUNX2, and along with scaffold-free spheroids also showed elevated MMP9 relative to 2D cells. The authors interpret that as a sign that the model better reproduces osteosarcoma-relevant behaviour and a more tumour-like microenvironment. There is always some caution needed with that kind of claim, but it is hard to miss the pattern.

That makes this a natural companion to our previous post on tumour model stiffness in prostate cancer research. The tumour types are different, but the comparison point is similar. Once microenvironment and physical context are treated as part of the experiment, the drug-response curves tend to move away from the 2D baseline.

Why this matters for cancer mechanobiology

One thing that stands out is that the paper does not argue that stiffness alone explains the result. In fact, the compressive modulus between inks barely changed. The stronger case is that architecture, matrix composition, and transport barriers all work together. The scaffold likely limits diffusion to some extent, the spheroids preserve 3D organisation, and the dECM adds at least some osteosarcoma-relevant biochemical cues. That combination seems to matter more than any single design choice on its own.

That is really where the cancer mechanobiology angle comes in. A 3D bioprinted osteosarcoma model is not just useful because it is 3D. It is useful when the structure, mechanics, and matrix environment begin to shape cell behaviour in ways that look less like a dish and a bit more like a tumour.

About the UniVert

In this study, the UniVert was used to perform unconfined compression testing on the hydrogel formulations that became the basis of the osteosarcoma model. That testing established that both candidate inks sat in a similar compressive stiffness range, around 7.2 kPa, while the dECM-enriched formulation improved printability and supported strong viability without introducing a major stiffness shift.

For studies like this, the UniVert fits into the workflow at an early but important point. Before a printed tumour model is used to interpret cell behaviour or drug response, the material itself needs to be measured. Compression testing helps place that model in a defined mechanical range and makes it easier to compare formulations without guessing what the scaffold is contributing.

Learn more about the CellScale UniVert.

Closing thoughts

This 3D bioprinted osteosarcoma model is interesting partly because it does not overpromise. The authors are fairly open about the fact that the dECM fraction was low and that the scaffold stiffness still leaves room for refinement. But even with those limitations, the model behaved differently enough under doxorubicin to suggest that the extra structural and matrix context is doing real work.

Citation