One year into the MOSAIC project: key learnings from a multidisciplinary cancer research project

In June 2023, at the Annual meeting of the American Society for Clinical Oncology (ASCO) we publicly launched MOSAIC, a multi-center initiative to map up to 7,000 cancer patients and build the first large-scale spatial omics dataset. As we head into the second year post-launch, we reflect on the ground that we have covered, both operationally and technically. Unsurprisingly, throughout this process we have encountered various challenges and learned a number of lessons associated with managing a project of this scale, fostering collaborations with a distributed network of partner hospitals, working with technologies that are not fully established yet (i.e. spatial omics), and navigating the complexities of working with patients data.

Choosing the right partners and technologies

While conducting a pilot may seem like it slows a project down initially, it is, in fact, a critical opportunity to identify and address potential issues early on. For example, it allowed us to explore and evaluate the most suitable spatial technologies for a project of the scale of MOSAIC, involving samples from up to 7,000 patients across five different hospitals. The technology we selected had to meet several key criteria:

• Relevance for the MOSAIC’s discovery approach, favoring full transcriptome technology
• Suitability for a wide range of tumor types (MOSAIC indications)
• Reproducibility between centers and scalability of throughput
• Best compromise between feasibility (material requirements, throughput and cost) and resolution (as close to single-cell as possible)
• Ability to integrate into multi-omics workflow and Owkin’s AI capabilities
• Security of reagent supply in the selected geographies

These are complex considerations, many of which go beyond individual experiments and require thinking about the project holistically.

Such considerations were not limited to spatial omics but also applied to all technology providers and CROs (contract research organizations). For a project that blends innovation with large-scale operations, the flexibility and collaborative mindset of a partner are as important as their technical expertise and the quality of their products and services.

Choosing the right partnering hospitals has been instrumental to the success of the MOSAIC project. Each of our partner hospitals—Uniklinikum Erlangen, Gustave Roussy, University of Pittsburgh, Charité Comprehensive Cancer Center, and Centre hospitalier universitaire vaudois (CHUV)—is a center of excellence for cancer research, both nationally and internationally. Collaborating with these institutions not only provides the project with key opinion leaders in the field but also ensures that we adopt best practices and insights from top-tier cancer research environments.

This strategic selection of partners fosters an environment of innovation and excellence, which is essential to the success of MOSAIC.

Keeping a large-scale, multi-site project running smoothly

Having everything in place from a technical perspective is not enough to ensure the success of such a large-scale initiative. Equally critical is establishing a clear governance structure where accountability is well-defined, and teams and individuals are empowered to make decisions. This framework enables smooth operations and ensures timely responses, allowing the organization to act swiftly when challenges arise.

MOSAIC operates under a joint steering committee comprising representatives from all founding members, who regularly meet to make higher-level decisions, such as the scientific direction of the project and the inclusion of new cohorts and therapeutic areas. In parallel, more focused groups are in place to address day-to-day operations, as well as subject-specific matters. For example, the initiative is organized into specialized work streams—such as data generation and operations—each managed by dedicated subject matter experts and technical specialists to ensure effective oversight.

Given the multicentric nature of MOSAIC, a balance of site-specific and cross-site meetings has been essential. This approach has allowed for effective troubleshooting and local issue resolution while fostering knowledge-sharing across institutes and avoiding the creation of silos. For instance, one potential silo could have arisen between the data generation and data analysis phases. However, maintaining a constant feedback loop between these steps has been key to developing a workflow that guarantees the data generated is of the highest quality and ready for analysis.

Moreover, the multidisciplinary nature of MOSAIC requires close coordination across various teams, including clinical oncology, pathology, data science, legal, data privacy, and IT. Involving these groups from the start is crucial, especially for those whose work, while downstream, is essential to early planning. One of the early challenges we faced was the delayed involvement of pathology teams, whose contributions are critical to both the scientific and operational success of the project.

Another set of challenges associated with managing a large-scale, multi-site project is ensuring smooth logistics and operations. One of the key lessons we’ve learned in this regard is the importance of implementing the right tracking tools. By utilizing uniform tools across all sites—ranging from supply chain management to cohort tracking—we can simplify data centralization and facilitate seamless information sharing between teams. This capability is crucial for navigating the complexities of a project like MOSAIC and enhancing collaboration among teams with diverse remits.

Working with hospitals and patients

Collaboration with our partner hospitals has been instrumental to the success of MOSAIC thus far. A key lesson we’ve learned is the importance of deeply understanding the operational processes of each hospital— screening patients based on clinical inclusion criteria, verifying the presence of patient's consents and informing patients when needed, de-archiving FFPE blocks from different biobanks, up to checking each block from a pathology perspective—to allow MOSAIC to seamlessly integrate into each clinical setting. This is enabling us to achieve the project’s goals without disrupting existing procedures the partnering hospitals.

We also recognized the delicate balance between standardization and customization. While lab operations across participating hospitals must be standardized to maintain consistently high-quality data, patient selection processes often require customization to align with each center’s unique tools, procedures, and governance structures.

Equally critical to the project’s success is the involvement of patients, their samples, and their consent. The MOSAIC Study was approved by the Ethic Committees (ECs) and Institutional Review Boards (IRBs) of the hospitals involved in the study. Early engagement with local IRBs and ECs has proven vital. Navigating the regulatory approval process at each center can be time-consuming, but initiating these discussions early has minimized delays and ensured the project remains on track.

With regards to patient consent, one of our most successful strategies has been developing a more patient-centric approach. Perhaps unsurprisingly, we found that when patients are actively engaged in the consent process and understand the importance of the research, consent rates rise significantly—reaching over 50% in some cases—which has been invaluable to the project’s progress.

Conclusion

The first year of MOSAIC has been a remarkable journey. From selecting the right partners and technologies to ensuring seamless coordination across multiple hospitals and disciplines, we’ve gained invaluable insights into managing large-scale, multidisciplinary research. To date, we’ve successfully included over 1,300 patient samples from across our five partnering hospitals, marking significant progress and making MOSAIC the world’s largest spatial omics dataset. As we move into the second year, we remain committed to fostering collaboration, maintaining data integrity, and continuing our pursuit of groundbreaking discoveries in cancer research. With these pillars in place, we are well-positioned to achieve MOSAIC’s ultimate goal: advancing our understanding of the tumor microenvironment through cutting-edge spatial omics.