A Machine Learning-Optimized System for Pulsatile, Photo- and Chemotherapeutic Treatment Using Near-Infrared Responsive MoS2-Based Microparticles in a Breast Cancer Model

Abstract

Multimodal cancer therapies are often required for progressive cancers due to the high persistence and mortality of the disease and the negative systemic side effects of traditional therapeutic methods. Thus, the development of less invasive modalities for recurring treatment cycles is of clinical significance. Herein, a light-activatable microparticle system was developed for localized, pulsatile delivery of anticancer drugs with simultaneous thermal ablation by applying controlled ON–OFF thermal cycles using near-infrared laser irradiation. The system is composed of poly(caprolactone) microparticles of 200 μm size containing molybdenum disulfide (MoS₂) nanosheets as the photothermal agent and hydrophilic doxorubicin or hydrophobic violacein, as model drugs. Upon irradiation, the nanosheets heat up to ≥50 °C leading to polymer softening and release of the drug. MoS₂ nanosheets exhibit high photothermal conversion efficiency and require low-power laser irradiation. A machine learning algorithm was applied to acquire the optimal laser operation conditions. In a mouse subcutaneous model of 4T1 triple-negative breast cancer, 25 microparticles were intratumorally administered, and after 3-cycle laser treatment, the system conferred synergistic phototherapeutic and chemotherapeutic effects. Our on-demand, pulsatile synergistic treatment resulted in increased median survival up to 39 days post start of treatment compared to untreated mice, with complete eradication of the tumors at the primary site. Such a system is therapeutically relevant for patients in need of recurring cycles of treatment on small tumors, since it provides precise localization and low invasiveness and is not cross-resistant with other treatments.