Beyond conceptual advances in cancer therapies

For papers describing cancer therapies, we often look beyond conceptual advances to prioritize studies that pair strong ideas with rigorous in vivo evidence, clinically coherent models and a credible path to translation. Here, we summarize the major criteria that make cancer therapies truly stand out.

At Nature Biomedical Engineering, years of handling manuscripts proposing cancer therapies have helped to shape our editorial criteria to identify therapies that solve clinically relevant problems, move the field closer to effectively treating cancer and are likely to pass regulatory scrutiny. In this Editorial, we hope to offer a constructive direction for researchers preparing manuscripts in this space.

Editorial expectations now emphasize demonstrable in vivo potency, reflecting higher standards for what constitutes meaningful antitumour activity. In mouse models, complete eradication of solid tumours is now achievable, particularly through combinatory therapeutic approaches. For example, macrophage-mediated therapies have led to full tumour regression in stringent settings¹, and optimized STING agonists have produced similarly compelling outcomes when combined with immune checkpoint blockade in a single protein-based construct². By contrast, a recurring issue in our submissions is that the therapeutic effects demonstrated in vivo often do not reach this level of potency. To support a credible translational trajectory, new therapies must provide clear evidence of clinically meaningful benefits. When in vivo activity is limited to slowed tumour growth, without achieving substantive tumour control or regression, it becomes challenging to argue for translational relevance.

The adequacy of the in vivo experimental design is another critical factor; preclinical models should reflect how therapies would be used clinically. Many solid tumours, including melanoma, colorectal cancer and brain cancers, are first managed by surgical resection. Accordingly, models of postsurgical recurrence or minimal residual disease are often more informative and better reflect realistic opportunities for intervention. In highly translational fields, such as neuro-oncology, postsurgical models that benchmark and combine therapies against standard of care can be especially valuable³.

Benchmarking therapeutics against relevant standards can only increase the translational impact of studies⁴. In many cases, demonstrating superiority or at least additive benefit is necessary for a therapy to make a meaningful clinical argument. For example, comparisons of mRNA-based cancer vaccines with immune checkpoint blockade have been critical to demonstrate clinical utility⁵. When the innovation lies primarily in the delivery vehicle rather than in the therapeutic payload, benchmarking may instead focus on compatibility and performance relative to established targeted nanoparticle formulations⁶. Still, there are notable exceptions. Studies that identify new cancer biomarkers, particularly for malignancies where existing biomarkers perform poorly, often warrant more flexible benchmarking expectations because the contribution lies in revealing a new biological handle that can be leveraged therapeutically⁷.

Proper validation and good experimental design are important to showcase the applicability or superiority of new technologies. Recurrent shortcomings related to poor experimental design of in vivo experiments include small cohort size, reliance on endpoint tumour weight rather than tumour growth kinetics (for most murine solid tumour models, we believe that tumours need to be monitored for at least 14 days), absence of survival analyses, or the use of only one or a non-representative tumour model. We believe that it is good practice to match the standards of reporting in line with published studies that use the same models and with the same interventional mechanism.

Likewise, mechanistic studies often benefit from validation across multiple tumour models to avoid conclusions that rely too heavily on a single genetic or immunological background, as illustrated in recent methodological work⁸. The choice of tumour source can also determine the strength of a manuscript’s conclusions. For example, therapies designed to enhance cytotoxicity, such as oncolytic pathogens, toxins, nanomedicines, gene therapies or immune engagers, often gain translational robustness from validation in patient-derived xenograft models that better capture tumour heterogeneity and mice with humanized immune systems that better recapitulate human immunology. Often, testing cancer therapies in these animals increases the chances of “ethics committees and regulators approving early-phase trials”, as mentioned by Yang Shi (RWTH Aachen University) during an informal exchange.

In parallel, delivery strategies should be aligned with clinical feasibility. Intratumoural injection, while practical for early-stage proof-of-concept studies, is not broadly applicable to most solid or metastatic tumours. Innovations in nanotechnology and bioorthogonal chemistry now allow for highly specific tumour targeting — including approaches that leverage tumour-specific amino acids⁹ or enable in situ assembly of immune engagers and targeted degraders¹⁰. Clearly articulating how a delivery route fits within or benefits from these modern capabilities helps to contextualize a therapy’s translational potential.

Mechanistic evidence can certainly strengthen therapeutic claims when authors aim to establish a biological mode of action. For papers describing new therapies, mechanistic studies confirming that the approach is having the desired impact at the molecular level can add clear value to a study. However, lack of mechanistic depth is rarely a primary reason for editorial rejection. At Nature Biomedical Engineering, we recognize that many engineering-led or materials-focused studies serve practical and logistical purposes rather than benefits to basic research. For example, improvements in the manufacturing, handling or storage of nanotechnology-based nucleic acid therapeutics, as reported by Jeffrey A. Hubbell and colleagues in this issue, do not hinge on mechanistic dissection; hence, we do not expect such studies to provide extensive mechanistic insight. In these contexts, demonstrating clear practical benefits, reliability and usability is more relevant than elucidating molecular pathways.

In summary, often our strongest submissions in cancer therapy offer a conceptual and generalizable advance, show meaningfully beneficial therapeutic performance over the state of the art in relevant animal and cancer models over meaningful time periods, reflect current clinical workflows and make a compelling case for translatability.

Researchers are encouraged to engage closely with the recent publications of their target journal, which provide a clear reflection of the level of mechanistic support, translational relevance and therapeutic robustness expected for manuscripts to proceed to peer review. Doing so not only helps to align study design with editorial expectations but also increases the likelihood that the work will resonate with reviewers and readers alike. Ultimately, our editorial criteria are established with the aim of increasing the real-world impact of research.