The studies that stick with me after a conference are the ones that answer a question clinicians are already asking, or surface a problem the field hasn't quite named yet. By that measure, this year's ASCO had a lot to offer.
Across ASCO 2026, academic researchers presented 18+ studies powered by Flatiron’s real-world data (RWD) that spans treatment selection, biomarker adoption, end-of-life care, and historically underrepresented patient populations. What connected these studies was more fundamental than subject matter—each one answered a question that clinical trials, by design, aren't built to ask. Trials, which are constructed around specific hypotheses, in defined patient populations, with endpoints that close at a fixed point in time, are oftentimes not reflective of day-to-day clinical care. Real-world data, drawn from actual clinical settings, longitudinally, and at scale, fills in what sits beyond that frame.
Here are four studies that stood out.
An NSCLC treatment comparison oncologists have been waiting for
Sotorasib and adagrasib are both approved for patients with KRAS-G12C mutated metastatic NSCLC who have received at least one prior systemic therapy. However, similar to many competing approved treatments, no head-to-head trial has ever compared them. While previous indirect comparisons attempted to bridge the two trials statistically, those trials enrolled selected patients with limited representation of the complex real-world cases clinicians actually see.
A team at Penn Medicine's Abramson Cancer Center used the Flatiron Health NSCLC Panoramic database, comprising over 340,000 patients with non-small cell lung cancer, to conduct the first large-scale, multi-institutional real-world comparison across 1,133 patients. The headline: no meaningful difference in overall survival or progression-free survival between these two agents. But sotorasib showed longer time to treatment discontinuation (TTD)—3.8 versus 3.3 months (adjusted HR 0.80; p = 0.006)—which may suggest better tolerability. Adagrasib was more commonly used in patients with CNS metastases; sotorasib in patients with poorer ECOG performance status.
This is exactly the type of study real-world data is built for. It reveals the nuances of cancer care, how oncologists are already individualizing care between these agents, and confirms that both deliver similar survival outcomes. The TTD difference is an intriguing signal, and further work into understanding the potential tolerability advantage would be highly informative. For a patient weighing two options with similar survival outcomes, tolerability is often the deciding factor. For someone who will spend months on this therapy, often while continuing to work, care for family, or pursue the things that matter most to them, tolerability is the difference between treatment that fits their life and treatment that doesn't.
A new biomarker, and the testing gap that followed
When zolbetuximab received FDA approval for CLDN18.2-positive advanced gastric and GEJ cancer in October 2024, it made CLDN18.2 an actionable biomarker overnight. But approval is only the starting line, not the finish line.
While clinical trials demonstrate that a therapy works in patients who harbor an actionable biomarker, they aren't designed to measure what happens next in clinical practice: how quickly testing adoption spreads, where gaps persist, or what delays mean for patients who might otherwise benefit. Those questions only become visible when you can track real patients, longitudinally, from the point of approval forward.
A team from Yale School of Medicine, led by Dr. Yifei Zhang, used Flatiron's real-world data on over 22,000 patients with gastric cancer to track testing adoption from before the SPOTLIGHT trial through post-FDA approval. The trajectory is encouraging: testing rose from 1.1% pre-SPOTLIGHT to nearly 40% post-approval. But nearly one-third of patients still underwent sequential rather than upfront comprehensive biomarker testing, with a median gap of 91 days between tests.
In an advanced cancer where treatment timing matters greatly, a 91-day gap in comprehensive biomarker testing means some patients may exhaust certain therapeutic options before others are recognized. This is a systems problem that can be addressed with practical solutions, and now there is evidence to support addressing it.
A distinct population on the rise
Lung cancer in patients under age 50 is rising, but this population remains uncharacterized in clinical research. Young-onset lung cancer is rare enough that even high-volume cancer centers see relatively few of these patients each year. Building a cohort large enough to draw clinically meaningful conclusions, and compare it meaningfully against an older-onset population, requires data at a scale no single institution can generate alone. It's the kind of research question that becomes answerable only when investigators have access to multi-institutional, population-scale data.
SPARK-Lung, presented by Teja Voruganti, a fellow at Penn Medicine's Abramson Cancer Center, used the Flatiron Health NSCLC Panoramic database, comprising over 340,000 patients with non-small cell lung cancer, to create one of the most comprehensive portraits of young-onset NSCLC to date.
Across nearly 3,800 young-onset and 120,000 older-onset patients, the differences are meaningful. Younger patients are more often female, more racially and ethnically diverse, less likely to have a smoking history, and more likely to carry actionable driver mutations — particularly ALK fusions (9.3% vs. 1.2%). They more often present with metastatic disease (70% vs. 49%), yet survive longer when treated.
This reinforces the case for aggressive molecular profiling and early access to targeted therapies in this population, and a recognition that biology here is genuinely different.
A necessary look at end-of-life care
Led by Dr. Umang Swami at the Huntsman Cancer Institute, this study examined end-of-life treatment patterns across six common advanced cancers in the U.S. The team drew from Flatiron’s Panoramic data on over 1.9 million patients with non-small cell lung, breast, prostate, colorectal, pancreatic, and bladder cancers to better understand the real-world treatment patterns of over 260,000 patients with advanced or metastatic cancers in relation to their time of death.
In terms of population-level relevance, this study is hard to overstate. Care patterns in a patient's final weeks, across multiple cancer types simultaneously and at this scale, sit outside what controlled research is designed to observe. A study of this kind—population-level, longitudinal, capturing what actually happens to patients rather than trial participants— is only possible with real-world data.
The findings are striking: up to one-third of patients received systemic treatment within one month of death. Extend the window to three months, and the numbers are even more sobering — across all six cancer types, a majority of patients were still receiving treatment, ranging from 60% in metastatic prostate cancer to nearly 80% in metastatic pancreatic cancer.
These numbers surface a question our field needs to sit with: what is driving systemic therapy use this close to death, and what would it take to change it? This study doesn't answer that directly — it wasn't designed to measure palliative care utilization or the timing of goals-of-care conversations. But by quantifying the scale of late treatment across six major cancer types, it provides the kind of evidence base that can inform those interventions. Understanding the pattern is the necessary first step toward changing it.
Empowering the next wave of research
What connects these four studies is the possibility they represent for providers, patients and researchers. These studies consistently demonstrate that real-world data answers questions that clinical trials alone cannot—questions oncologists are actually asking, about the patients they're treating, in the settings where care is delivered.
Working with Flatiron RWD gives investigators access to research-ready data on over 6 million oncology patients, analytic and visualization tools, rapid study feasibility assessment, and — when relevant — letters of support for external grant funding. This model supports rapid hypothesis and insight generation, stronger grant applications, and democratization of high-impact research opportunities for trainees, ultimately accelerating researchers and broadening the institution's research footprint.
It is also worth highlighting that the sotorasib vs. adagrasib study was led by a resident and the SPARK-Lung study by a fellow. When researchers at every career stage have access to research-ready data at population-scale, it strengthens their ability to produce practice-changing science—advancing both their careers and the field simultaneously.
If you're an academic researcher or health system leader looking to do more with real-world oncology data, from powering high-impact research to enabling your bench of rising research talent, we'd love to connect. Reach out to us here
