Congress settled the FY26 question in February, with NCI receiving $7.352 billion, an increase of $128 million over the prior year.1 After a year in which cancer center directors spent an extraordinary share of their leadership attention defending the research enterprise, that outcome deserves to be called what it was: a win.
It also did not change the math on anyone’s desk. A 1.8% increase does not keep pace with what it now costs to run a trial. Protocol complexity and staffing costs have kept climbing. The practical result is that cancer centers are being asked to grow their trial portfolios without growing the teams that execute them.
NCI designation makes this sharper still. It carries an obligation to run trials that will never be profitable. That obligation has to be funded by running the rest of the portfolio well enough to offset it. The question facing directors is not whether to be more efficient. It is where efficiency is actually available, and at what scale.
More money, more people. Repeat.
The approach of scaling trial operations by scaling people has been the operating model of clinical research for two decades, and it worked while the money grew. As trials exploded in number and complexity, a cottage industry became a behemoth. This was clearly seen in contract research organizations, which grew by consolidating smaller companies and adding thousands of people. The category is usually described as a tech-enabled service. In practice, the technology was most often open source, homegrown, or decades old. They made the money and did the work using legions of people.
Cancer centers have historically used the same approach. When they got financial breathing room, they quickly assigned it to a new headcount. Adding people feels good at first, until we realize it’s shuffling tasks and dividing workloads rather than delivering real gains in scalability and efficiency for the program.
As Albert Einstein famously said, we cannot solve our problems with the same thinking we used to create them.
A trials program is a lot like a startup
A cancer center clinical trials program does genuinely hard things at impossibly small scale. That is close to a working definition of a startup, and it is worth borrowing how startups think.
Peter Thiel asks a useful question in Zero to One. What important truth do very few people agree with you on?2
Here is mine: You don’t need more people to run more trials, and incremental cost savings won’t save the day. Trimming a few percent from a process that was never designed to scale only makes a cumbersome workflow slightly cheaper. The opportunity is to solve the biggest problem differently, not to marginally fix the small ones.
The hardest thing was left for last
With respect to imaging in trials, as with many things in life, the hardest thing was left for last. Nearly every other data stream in a trial has been standardized, integrated, or automated to some degree. Imaging has not.
Tumor progression drives eligibility and continuation decisions, and the workflow producing those decisions still runs largely on manual methods that are neither FDA Good Clinical Practice (GCP)-compliant nor traceable, relying on emails, phone calls, paper forms, or spreadsheets. That is not a small gap in trial operations. It sits directly on the critical path. The consequences show up in two places.
With respect to imaging in trials, as with many things in life, the hardest thing was left for last. Nearly every other data stream in a trial has been standardized, integrated, or automated to some degree. Imaging has not.
First, accuracy: Data presented at the AACI Clinical Research Innovation meeting examined compliance with trial imaging protocols at three major NCI-designated Comprehensive Cancer Centers and found error rates of 25%, 30%, and 50% across imaging time points.3 After those centers moved onto a comprehensive, cloud-based platform that enforced protocol-specific criteria at the point of the read, error rates fell below 3%.4
Second, labor: In current manual workflows, study teams spend up to eight hours managing imaging assessments, data transcription, and audit preparation for every hour of radiologist time.3 That ratio captures the argument in one number. Ironically, the expensive radiologist is the smallest part of the cost. The largest imaging expenses involve study staff burden that few budget lines fully capture, and one incremental hire can only bring incremental relief. Scaling capacity by standardizing methods, modernizing technology, and reimagining workflows have exponentially greater payoffs.
Centers are already making this trade
Sites are not waiting for permission. The 2023 ACRP site challenges survey found staffing and retention to be the top concern at research sites, cited by 63% of respondents. The same survey found that sites list complexity as a reason to forego opening specific trials.5
Read that against a flat budget. Centers are declining protocols they are scientifically capable of running, because the operational math does not work. Sponsors willing to pay what is necessary to run their trial at these sites still run into the same capacity wall. Progress slows. That is the cost of a people-based model that has run out of people.
Focus on simple things, not fancy AI things
A long, futuristic conversation about AI’s potential to improve differential diagnosis, trial subject matching, and real-world data initiatives is tempting. Those may well prove out. They drive demand for services more than they increase capacity to provide them.
The payoffs available now are simpler, less glamorous, and much more near-term. Sites and sponsors can standardize their ways of working, leverage technology to elevate people’s work, and connect directly to source data systems with elegant human-in-the-loop workflows to remove manual, error-prone steps.
Trial protocol imaging response assessment criteria are increasingly complex and vary by trial. It is impossible for site readers to manage this compliantly across all the trials at their site, and even more so across sites in multi-site trials, without a comprehensive protocol-specific imaging network solution. What, then, is the justification for having no eClinical solution to drive these complex imaging workflows efficiently both within and across sites? None of this requires a research breakthrough to fix. It requires first realizing that the current way is a choice rather than a constraint, avoiding the temptation to staff a problem, and adopting technology to realize the desired gains.
Good clinical practice is outrunning sites’ capabilities
There is a compliance dividend to solving the capacity problem with technology, not more people. ICH E6(R3)6 is now the new FDA GCP standard and will be enforced more strictly going forward. Anecdotally, FDA reviewers increasingly ask sites to trace imaging endpoint values through the entire chain of custody back to the source images and the decisions that underlie the results. A true tech-forward workflow that produces that link as a natural byproduct is considerably more reliable and cheaper to defend than a manual process that must painstakingly reconstruct it after the fact, if it can at all.
Doing more with more (technology)
There is a better framing available. Instead of doing more with less money, we could all do more with better technology.
We are all going to have to spend money and watch costs differently as cancer therapies advance. More treatments, more trials, more complexity, more patients, more sites, more data elements, more queries, and more monitoring will affect not just cancer centers, but sites, CROs, and sponsors alike. The alternative is to keep waiting for funding to reach the level the old model required, and no one can say when (or if) that might happen.
Cancer centers spent a year defending the research enterprise and won. Next year will almost certainly be a bit more of the same, about what each enterprise can do to get by with what it already has. Thankfully, each organization has an unprecedented opportunity to shape its own future with technology, a future where technology investments demonstrate a cost-effective return, save money through efficiency gains, and positively impact everyone.
Jeff Sorenson is CEO and co-founder of Yunu, which provides cloud-based imaging workflow and data management infrastructure for clinical trials. Yunu supports more than 8,500 active trials and 400 pharmaceutical sponsors, including at 20 of the 58 NCI-designated Comprehensive Cancer Centers7. For more information, visit yunu.io.
References
- Consolidated appropriations for fiscal year 2026, Labor, Health and Human Services division, enacted February 3, 2026. NIH funded at $47.2 billion and the National Cancer Institute at $7.352 billion, an increase of $128 million over fiscal year 2025. Summarized by the American Cancer Society Cancer Action Network, February 2026. fightcancer.org/releases/congress-demonstrates-strong-commitment-cancer-prevention-and-cures-fy26-appropriations
- Thiel P, with Masters B. Zero to One: Notes on Startups, or How to Build the Future. Crown Business, 2014.
- Alsumidaie M. Dangers and costs of clinical trial imaging errors. The Clinical Trial Vanguard, June 24, 2024. https://www.clinicaltrialvanguard.com/article/urgent-warning-dangers-and-costs-of-clinical-trial-imaging-errors/
- Cruz A, Lankhorst B, McDaniels H, Weihe E, Correa E, Nacamuli D, Somarouthu B, Harris GJ. The complete workflow solution for quantitative imaging assessment of tumor response for oncology clinical trials. Presented at the AACI Clinical Research Innovation Meeting, Chicago, IL, 2024.
- Association of Clinical Research Professionals. Top site challenges of 2023: data and insights on site burden and trial efficiency. August 2023. acrpnet.org/2023/08/02/top-site-challenges-of-2023-data-and-insights-on-site-burden-and-trial-efficiency-2
- International Council for Harmonisation. Guideline for Good Clinical Practice E6(R3). ich.org
- Yunu company data, August 2026.
