What kind of CRO work does an oncology program need?
Oncology is the most crowded therapeutic area in drug development, and the outsourcing map is wide because the science is. Early on you need tumor biology and target validation: confirming the target actually drives the cancer you care about, profiling expression across tumor types, and running CRISPR or knockdown work to show dependency. From there a program moves into the part of oncology preclinical work that almost everyone outsources, the in vivo efficacy studies, because few biotechs keep a colony of immunodeficient mice and a working bank of tumor models in-house.
The model you pick is not a detail, it is the whole readout. Cell-line-derived xenografts in immunodeficient mice are the fast, cheap workhorse for showing a small molecule or antibody shrinks a tumor. Syngeneic models in immunocompetent mice are what you need for immuno-oncology, because an anti-PD-1 or a bispecific does nothing without a working immune system to engage. Patient-derived xenografts (PDX) keep the original tumor's heterogeneity and are closer to the clinic but slower and pricier. Humanized-mouse and orthotopic models, plus metastasis and patient-derived organoid work, fill in the harder questions. A CRO that runs your specific model with historical control data is worth far more than one spinning it up for the first time on your budget.
Then come the standard preclinical workstreams, read through an oncology lens: DMPK and ADME, dose-range-finding, and tox, where the safety bar differs from other areas because cytotoxic and targeted oncology agents are often allowed a narrower therapeutic window and a higher tolerated toxicity. Biomarker and companion-diagnostic work tends to start earlier here than anywhere else, since modern oncology approvals lean on patient selection. On the clinical side the work is GCP clinical operations built around oncology endpoints, overall response rate (ORR) by RECIST, progression-free survival (PFS), overall survival (OS), and duration of response, run through specialist sites, central imaging and independent review, and safety reporting. The modality range is enormous: small molecules and kinase inhibitors, monoclonal and bispecific antibodies, antibody-drug conjugates (ADCs), CAR-T and other cell therapies, radioligands, and cancer vaccines, each of which changes which suppliers can actually do the work.
How do you choose a CRO for oncology?
The logo on the contract matters less than whether the team has run your exact kind of program before. Oncology is broad enough that genuine expertise is narrow: a CRO that is excellent at solid-tumor xenografts may have never touched a hematologic disseminated model or a CAR-T persistence study. Score two or three suppliers against the same written scope, and weight relevant, recent, in-modality experience above the headline quote. Below are the filters that actually separate a clean program from a frustrating one.
- Therapeutic-area and tumor-type fit: ask specifically about the cancer type and modality you are working in (solid versus heme, the indication, immuno-oncology versus cytotoxic versus targeted), not generic oncology experience.
- Relevant models or patient access: for preclinical, confirm the specific xenograft, syngeneic, PDX, or orthotopic model is already validated in-house with historical control data. For clinical, ask about access to oncology investigator sites and realistic enrollment in your indication, which is the usual rate-limiter.
- Regulatory and quality track record: GLP for the pivotal tox, GCP and a clean inspection history for clinical work, and experience with the regulatory pathway your program is aiming at, including any breakthrough, fast track, or accelerated-approval context.
- Endpoint and data quality: for clinical, how they handle RECIST assessment, central imaging, blinded independent review, and CDISC-conformant data. For preclinical, clean tumor-growth data, honest reporting of the cohorts that failed, and clear statistics.
- Capacity and timeline: a strong lab booked solid for months can be slower than a good lab with an open slot. Confirm current queue, report turnaround (not just bench time), and how change orders are priced when a study or a cohort slips.
- Biomarker and CDx capability: if your program depends on patient selection, confirm the supplier can run the biomarker assay or companion-diagnostic development you will need, ideally the same team that carries it toward the clinic.