Triple-Cell Pancreatic Spheroids Show That Stromal Context Can Shift Drug Response in Different Directions

Four Defined Cell-Line Microenvironments

University of Antwerp researchers assembled pancreatic ductal adenocarcinoma spheroids from three cell types to study how cellular context changes drug evaluation. The models combined either BxPC-3 or MiaPaCa-2 cancer cells, one of two pancreatic stellate lines and HMEC-1 endothelial cells. They were defined cell-line cocultures, rather than patient-derived tumor organoids.

The stellate sources differed: RLT-PSC originated from chronic pancreatitis, while hPSC21 originated from pancreatic cancer. HMEC-1 is an immortalized dermal microvascular endothelial line, rather than pancreas-derived endothelium. The team selected MCDB131 medium because it supported all components better than the tested alternative conditions.

Four combinations were optimized by measuring cell populations after three days. The starting ratios did not translate directly into the same final cell fractions, illustrating how growth in coculture changes composition. Matured spheroids were approximately 300–500 μm in diameter and differed in compactness, peripheral stellate-cell organization and invasive edges.

The four starting cell counts ranged from 3,500 to 7,000 per spheroid, depending on the combination, rather than using one common total. The authors based the compositions on literature-informed tumor proportions and refined them by flow cytometry. This design creates contrasting model contexts without making their geometry or initial cell number identical.

Architecture and Drug Response Do Not Follow One Simple Rule

MiaPaCa-2 alone formed loose aggregates around 1,000 μm across, whereas coculture produced more compact structures around 400 μm. BxPC-3 already formed compact monoculture spheroids. Adding other cell types therefore changed the two cancer backgrounds from different starting structures.

The drug experiment exposed three-day-old spheroids to gemcitabine, paclitaxel or their combination for five days, then measured ATP-based viability. Some curves did not cross 50% viability, so the authors compared response-curve area rather than forcing an IC50 estimate. A larger area represented greater residual viability over the tested concentration range.

BxPC-3 monocultures were more resistant than MiaPaCa-2 monocultures. BxPC-3 triple cultures were less resistant than their monoculture counterpart, while the addition of stellate and endothelial cells did not significantly change overall chemotherapy sensitivity in the MiaPaCa-2 comparisons. The data therefore do not support a universal increase in resistance whenever stroma is added.

Within the coculture panel, BxPC-3 with RLT-PSC and HMEC-1 was relatively resistant, whereas MiaPaCa-2 with hPSC21 and HMEC-1 provided a more sensitive comparison. Cancer background, cell density and spatial organization all contribute to interpreting those aggregate responses. The ATP assay did not separately identify which component cell type survived treatment.

Secreted Factors Produce Distinct Endothelial Responses

The investigators also collected medium from untreated spheroids and measured VEGF. MiaPaCa-2 combinations produced more VEGF than corresponding BxPC-3 combinations, but VEGF concentration alone did not explain the full ordering of endothelial responses.

Conditioned medium was applied to a separate six-hour HMEC-1 tube-formation assay. The IKOSA image-analysis software quantified branch points, total tube length, covered area and loops. BxPC-3/RLT-PSC and MiaPaCa-2/hPSC21 media produced comparatively larger branching, area and length measures, while the other two combinations produced smaller responses.

None of the untreated spheroid-conditioned media generated many closed loops, unlike the positive medium control. This is a short-term endothelial organization assay, not a demonstration of a mature or perfused vascular network within the spheroid. The study establishes differing baseline secreted effects; it did not test an antiangiogenic treatment in that comparison.

A Controlled Panel for Comparing Context

Drug-response results used at least seven measurements from three independent experiments. Other measurements had distinct replication: VEGF used three measurements from one experiment, and tube analysis at least eight from three experiments. These units should remain separate from the four cell-line combinations.

The authors propose the panel for secondary drug evaluation and for further extension to patient-derived or immune-containing systems. The current work shows how changing a defined tumor microenvironment can alter structure, drug response and secreted-factor activity in different ways. It does not establish a patient-specific prediction rule, and the dermal endothelial source and aggregate viability endpoint limit direct translation to pancreatic tumor physiology.