When the surrogate endpoint moves months before imaging does, adaptive designs get the evidence to drop a losing arm sooner. The cost saving is not a discount on a per-sample price; it is arm-level, from running a tighter, faster trial.

For early-phase work with low tumour burden, the sensitivity floor is also what determines whether ctDNA is usable as an endpoint at all.

Pharma partners often run exploratory and discovery samples in-house for speed and control, then route confirmatory or regulatory-grade MRD work to the accredited lab so the result sits inside an ISO 15189 envelope. The choice is operational, not scientific: the chemistry, sensitivity and data quality are identical either way.

It also means a multi-site trial can standardise on one method without forcing every site through one central lab.

For a sponsor, that is biomarker and resistance evidence accruing alongside the efficacy readout, the kind that supports the next protocol, the label conversation and the publication. Each variant carries a clinical rationale your medical and biomarker teams can defend.
If a regulator or a reviewer asks why a patient was scored positive, there is an answer beyond a score.

For a regulated submission, reproducibility is not a nice-to-have. You need to re-run the call, validate it against a second pipeline, and stand behind the method in front of an authority without a vendor sitting between you and the data.

We work that way by default, and we get told when our pipeline is wrong. We prefer that to the alternative.

An early conversation on assay and sampling design can save a sample run, and an underpowered endpoint, later. Co-authorship where the science supports it. Interpretation help when a result is unusual and you want a second set of eyes before it reaches a steering committee.

We are not the right partner for every trial. We are an exceptionally good partner for the ones we are right for..