Emergence of drug-resistant cancer phenotypes is a challenge for anti-cancer therapy. Cancer stem cells are identified as one of the ways by which chemoresistance develops. We investigated the anti-inflammatory combinatorial treatment (DA) of doxorubicin and aspirin using a preclinical microfluidic model on cancer cell lines and patient-derived circulating tumour cell clusters. The model had been previously demonstrated to predict patient overall prognosis. We demonstrated that low-dose aspirin with a sub-optimal dose of doxorubicin for 72 h could generate higher killing efficacy and enhanced apoptosis. Seven days of DA treatment significantly reduced the proportion of cancer stem cells and colony-forming ability. DA treatment delayed the inhibition of interleukin-6 secretion, which is mediated by both COX-dependent and independent pathways. The response of patients varied due to clinical heterogeneity, with 62.5% and 64.7% of samples demonstrating higher killing efficacy or reduction in cancer stem cell (CSC) proportions after DA treatment, respectively. These results highlight the importance of using patient-derived models for drug discovery. This preclinical proof of concept seeks to reduce the onset of CSCs generated post treatment by stressful stimuli. Our study will promote a better understanding of anti-inflammatory treatments for cancer and reduce the risk of relapse in patients.