(575a) Multiscale Healthcare Supply Chains: Patient in the Loop

Personalised therapies offer ground-breaking opportunities in the treatment of life-threatening diseases, including cancer. Currently, clinically successful personalised therapies involve infusing previously engineered cells to the target patient; those are then used to produce therapeutic proteins and/or other factors which serve as the treatment itself. In the instance of autologous therapies, the patient’s own tissue/cells are collected, modified and re-administered to the same patient ^[1][2]. The patient-specific nature of autologous therapies makes them a unique class of therapeutics to manufacture and distribute, coupled with a consumer-centric pharmaceutical supply chain ^[3].

Manufacturers in the space of personalised medicine must carefully orchestrate resources to ensure product availability in the right place and at the right time. This translates into: (a) patient-specific batches and distribution nodes and (b) challenges in demand forecasting which impact raw material planning activities. This is further highlighted for next-generation vaccines and therapeutics, which use viral vectors as means to genetically modify the therapy in-manufacturing. Vectors are equally coupled with a lengthy and complex production process, leading to a interconnected network of manufacturing and distribution of complex biologics.

Furthermore, individual product orders vary on a day-to-day basis and short product shelf-life restricts waiting and storage times. This highlights the need for responsive and scalable supply chains that readily adapt to demand dynamics. Constraints on the flexibility of the manufacturing process are also imposed by strategic decisions, which range from mid-term acquisition of raw materials to long-term investments in additional capacity and infrastructure. In this space, computer-aided tools can support supply chain design and planning under uncertainty, whilst capturing interconnectivity of strategic and operational decision-making processes ^[4].

We propose a multiscale mixed-integer linear programming (MILP) framework for the design and optimisation of end-to-end CAR T cell therapy supply chains (Figure 1). Given a set of location-specific product demands, candidate location for network nodes (e.g. hospitals, manufacturing facilities and Quality Control sites), capital and operating expenditures, the optimisation determines suitable candidate supply chain structures, investment plans, production schedules and distribution plans. Production and inventory levels of required raw materials are also suggested by the optimisation algorithm. The candidate networks are assessed for cost, responsiveness and resilience with respect to the uncertain demand, as well as unexpected delays in the end-to-end product lifecycle.

^[1] Novartis (2018), KYMRIAH Treatment Process, Dosing & Administration | HCP.

^[2] Kite Pharma (2018), First CAR T Therapy for Certain Types of Relapsed or Refractory B-Cell Lymphoma.

^[3] Papathanasiou, M.M.; Stamatis, C.; Lakelin, M.; Farid, S.; Titchener-Hooker, N.; Shah, N. (2020). Autologous CAR T-cell therapies supply chain: challenges and opportunities? Cancer Gene Therapy, 27, 799-809.

^[4] Sarkis, M.; Bernardi, A.; Shah, N.; Papathanasiou, M. (2021). Decision support tools for next-generation vaccines and advanced therapy medicinal products: present and future, Current Opinion in Chemical Engineering, 32, 100689.