Quality Matters
Radiotherapy quality affects patient outcomes and trial outcomes. It's time we measured it, every trial, every time.
In cancer trials, radiation quality matters.
In a large share of drug-focused oncology trials today radiation is chosen for the primary local therapy; a crucial factor in ensuring the best possible outcome for a patient.
Despite this, only some trials that include radiation are evaluating the quality and the vast majority of those trials are asking a question about the efficacy of the radiation itself. Trials which do ask these questions assess the contour quality, the acceptable regimens, and delivery techniques and directly ask whether or not the radiation performed was acceptable across the board, on the site-level, and on the individual patient level.
The focus of this blog, where we believe quality especially matters, is on the trials which require radiation as part of standard treatment while testing something else, oftentimes a drug, but do not consider the quality of the radiation delivered to patients. Their primary question almost always echoes: does this new treatment approach provide a benefit to patients beyond the current standard of care?
Hundreds of millions of dollars (or more), years of work, and the hopes of patients depend on the answer. In a drug trial, theoretically, the drug should be the only variable and the definitive local therapy (radiotherapy or surgery) should be held to some consistency bar across patients with the drug's effect measured alongside it. Despite the heterogeneous nature of radiotherapy, it's often the case that quality requirements are not communicated whatsoever and even when they do exist they are infrequently validated.
Radiation planning is a human task: Two radiation oncologists may prescribe different regimens to the same patient, the same tumor can be outlined differently by different dosimetrists, and different physicists may choose different gantry angles which result in varying dosage to organs at risk. Every aspect has the potential to shift from one patient to the next.
That variation costs trials in two different ways:
- When the radiation requirements are the same in both arms it widens the spread, eats statistical power, and can bury a real drug effect that a quieter trial would have found.
- When the drug and the radiation interact, or when the drug's own toxicity changes how the radiation actually gets delivered, it stops being noise and starts shaping the comparison itself.
In either case, it's a significant determinant of an endpoint that is often overlooked. Despite this, it may be one of the best documented but least discussed problems in all of clinical oncology today.
Evidence & implications
Take one phase III trial in head and neck cancer, where a drug was added on top of chemoradiation. Patients whose radiation carried major deviations from the protocol had two-year overall survival of about 50 percent. Patients in the same trial whose radiation followed the protocol survived at about 70 percent (TROG 02.02, Journal of Clinical Oncology, 2010). A twenty point survival gap correlated with radiation adherence.
The obvious objection is that nobody randomized these patients to good or bad radiotherapy. The deviations were identified after the fact, and difficult tumors are both harder to plan compliantly and independently more lethal. It could be argued that some of that gap may be telling you which patients got deviations rather than what the deviations did to them. The association is real and it's been reproduced but a study has yet to be conducted to conclude whether the relationship is causal.
TROG's work is not the only example, though. One study pooled eight cooperative group trials and found patients whose radiotherapy deviated from protocol carried a 1.74 times higher risk of mortality, with a median of nearly a third of patients affected (Ohri et al., JNCI, 2013).
Perhaps the most concerning number is how rarely anyone asks the quality question at all, even in RT trials themselves. Across 42 phase III radiotherapy trials, 69 percent required any quality assurance, 45 percent required centers to be credentialed, and 50 percent reported protocol deviations when they occurred (Corrigan et al., Radiotherapy and Oncology, 2022). These are the trials which are actually most likely to assess quality, because the radiotherapy is the subject of the question that they're looking to answer. Widen the denominator to every registered interventional trial involving radiation and the share performing any radiotherapy quality assurance drops to roughly one percent.
None of this blames any clinician, center, or study. It describes a blind spot the field has lived with.
The haves and have nots of radiotherapy quality
There is a repeatable version of radiotherapy quality control that works today: Academic cooperative group trials, like the NCI's IROC and NRG, have spent decades building it: credentialing, phantom audits, plan review, deviation detection designed in from the start. These programs are real and trusted.
The biggest caveat is that these programs are voluntary and most often utilized for academic trials.
It's not entirely fair to say that commercial trials don't bother at all. Some run quality assurance comparable to a cooperative group: leveraging benchmark cases before a site opens, contour / plan reviews with tight SLAs, central sign-off on treatment plans before randomization. It does happen and there are examples of it worth looking at (which we intend to in a future entry of this series).
Industry is capable of RTQA, but no set of standards requires it nor describes it and it's difficult to distinguish which trials do and don't consider it from the outside looking in. Half of these trials never report deviations at all, and there's no radiotherapy-specific data standard to meet even if they wanted to. These same trials' results reach regulators, shape treatment guidelines, and ultimately decide which drugs patients get.
Why it matters now
Despite it being an uncomfortable reality, an uncontrolled variable inside a pivotal trial is an unpriced risk. A promising drug can look like a failure in a cohort of patients with risk-inducing RT protocol deviations. In extreme cases, these overlooked deviations can tip the scales and become very costly for sponsors. Assuring radiotherapy quality is relatively inexpensive and helps to minimize the likelihood of these situations.
There's a simpler reason too. Every point on a survival curve is a patient. When their radiation varied and nobody measured it, a trial can misjudge the root cause and the next generation of patients all have to inherit that.
Encouragingly, evidence has led to industry-wide standards before. Trial data formats were once inconsistent and hard to trust. An industry consortium built an open standard, CDISC, and it spread on its merits for a while. What made it the floor everyone meets was a regulator eventually requiring standardized data for submissions, an obvious win for clinical science.
What standard practice looks like
The goal is plain and achievable. Radiotherapy quality assurance should be a routine, expected part of any trial that includes radiation, as ordinary as the data standards drug sponsors already meet. The treatment plans and structures adhere to a standard, centers get credentialed before they contribute, and deviations get caught during planning instead of during trial readout or not at all.
Historically, collecting distributed imaging and treatment plans, standardizing them, running the checks, reviewing the results is a burden: slow, expensive, easier to skip. When the goalposts are not visible and the process is manual and, worst of all, the process can slow a trial timeline, of course it's easier to say 'no' than 'yes' to QA.
Ultimately, if we can, we should measure RT quality whenever it may have an impact on the patient outcome. Which is anytime it's the primary local treatment. Which is why quality matters.
---
Aitrium builds software for radiotherapy data collection and review in oncology trials, so we have an interest in a field that measures this. That being said, we believe the evidence above reads the same way with or without us. If you run radiotherapy quality assurance for trials and any of our ideas match or contradict what you believe, we'd like to hear from you.
