Key Points The genomic-adjusted radiation dose (GARD) is the first opportunity to use genomic information to optimize radiotherapy doses. GARD was significantly associated with survival and recurrence, whereas the physical radiation dose was not.

Using the predicted radiation dose effect derived from an individual patient's genomic profile may result in a better outcome according to a new study appearing in Lancet Oncology (2021;22:1221-1229. doi:10.1016/S1470-2045(21)00347-8). The study focused on a novel algorithm that a team of researchers developed on the basis of a personalized GARD model using individual tumor genomic parameters rather than the commonly administered “one-size-fits-all” dosages based on the cancer diagnosis. The researchers of this new study noted an incentive for the study: “Patients we treat uniformly do not have a uniform response.”

The development of GARD was spurred by an understanding that the researchers had uncovered from previous large-scale classification studies showing that the clinical heterogeneity of radiation response was influenced by individual-level differences in genetic variables in the tumors themselves. Using a gene expression–based radiosensitivity index that the study authors had developed earlier, they followed with the GARD model, which combined the radiosensitivity index and radiation dosages to pinpoint the biological effect on specific patients.

Study Details

In this study, researchers from the Cleveland Clinic, Case Western Reserve University, and the Moffitt Cancer Center conducted a pooled analysis of data from 11 previously published clinical cohorts of patients with varying outcomes for 7 different types of cancer: breast cancer (including triple-negative breast cancer), head and neck cancer, non–small cell lung cancer, pancreatic cancer, endometrial cancer, melanoma, and glioma. They used tumor genomic data from these studies and the clinical information about each patient's treatment to calculate individualized GARD values, and they performed a stratified Cox regression analysis to test for associations of GARD and physical radiation dose with 2 clinical outcomes: time to first recurrence and overall survival. Finally, the researchers estimated the impact that varying GARD would have on 3-year survival for each of the 7 cancer types.

With the exception of 3 cohorts, both time to first recurrence (local, regional, and distant metastasis) and overall survival were calculated from the end of treatment. These patients' outcomes were calculated from the date of pathological diagnosis for the glioma and endometrial cancer cohorts and from the date of randomization for the head and neck cancer cohort.

There were 1615 patients included overall. Among those patients evaluated for time to first recurrence, 982 received radiotherapy, and 316 did not receive radiotherapy. Among those patients evaluated for overall survival, 424 received radiotherapy, and 253 did not.

Study Results

The researchers found that although the range of physical radiation doses in the cohorts was limited to values near those of the standard of care and the doses were delivered in standard fraction sizes, the GARD values showed a wide range of predicted biological effects.

In analyses of the 7 cancer types combined, the researchers found that GARD was associated with time to first recurrence (hazard ratio [HR] per unit change in GARD, 0.98; 95% CI, 0.97-0.99; P = .0017) and overall survival (HR, 0.97; 95% CI, 0.95-0.99; P = .0007). Although the effects per unit change in GARD were small, patients who had similar or identical physical radiation doses often had GARD values differing by 20 or more units, so differences of this magnitude had clinically meaningful implications for patient outcomes.

The GARD values for patients who did not receive radiotherapy (designated as “sham-GARD”) were calculated on the basis of usual doses of physical radiation and were intended as controls. These sham-GARD values were not significantly associated with either time to first recurrence (HR, 1.00; 95% CI, 0.97-1.03) or overall survival (HR, 1.00; 95% CI, 0.98-1.02), and this supports the conclusion that the association of GARD with clinical outcomes is truly related to patients having received radiotherapy.

“We believe our study suggests that GARD is a new unit of measure—and it is the unit we should be using to dose patients.”—Jacob Scott, MD, DPhil

Patients' physical doses of radiation were not significantly associated with either outcome, with HRs of 0.99 (95% CI, 0.97-1.01) and 1.00 (95% CI, 0.96-1.04) for time to first recurrence and overall survival, respectively.

In analyses of all cancers together that included cancer type as a covariate, associations of GARD and 3-year survival were statistically and clinically significant, even though patients with each type of cancer received similar physical radiation doses.

Study Interpretation

“We think our study breaks new ground,” says study author Javier F. Torres-Roca, MD, senior member in the Department of Radiation Oncology at the Moffitt Cancer Center in Tampa, Florida. He adds that GARD is the first opportunity to use genomic information to optimize radiotherapy doses. “It tells you more about the outcome of a patient than the actual RT dose delivered. Importantly, we demonstrate that GARD is not associated with outcome in patients treated without RT and we show that the interaction between GARD and RT is significant, showing that it predicts RT treatment benefit.”

Lead study author Jacob Scott, MD, DPhil, associate professor and staff physician-scientist at Cleveland Clinic and Case Western Reserve University School of Medicine, adds, “We believe our study suggests that GARD is a new unit of measure—and it is the unit we should be using to dose patients.”

Three experts in the field wrote an accompanying comment article in Lancet Oncology (2021;22:1200-1201. doi:10.1016/S1470-2045(21)00411-3): Orit Kaidar-Person, MD, PhD (Sheba Medical Center, Ramat Gan, Israel); Philip Poortmans, MD, PhD (Iridium Network and University of Antwerp, Antwerp, Belgium); and Roberto Salgado, MD, PhD (Department of Pathology, GZA-ZNA Hospitals, Antwerp, Belgium). They wrote that the efforts of Dr. Scott and his colleagues need to be applauded worldwide “because radiotherapy is considerably lagging compared with the enormous progress done in the field of personalized medicine that currently mainly applies to decisions on the use of systemic therapy or targeted agents.”

They did have lingering questions about the research. Among their questions is gaining an understanding of the biological basis of GARD. In response, Dr. Scott says that this is largely a phenomenological signature. “Other academics always want to look at the list of genes and see ones that they know about, and then be able to say something about the biology,” he says. “This is human nature. However, due to the network-based method used, that approach doesn't work here.”

Dr. Scott says, “It isn't so much that the genes themselves need to be mechanically causative, but instead what this signature is capturing is a signal from the larger network—each of these genes is more like a canary in the coal mine rather than a specific, causative thing. As a physicist, I'm fine with this. It's just a list of genes that pick up signals that can be used to predict the state of a complex thing.”

The comment authors wrote that although the “clinical utility of genomic assays in daily practice remains to be demonstrated,” they agree with the researchers that a GARD-based framework for radiotherapy should be adopted as the new paradigm for trial design. They recommended that the assay be developed and integrated into clinical trials according to a set of criteria set forth by the National Cancer Institute. “By supporting such projects, including those aiming to identify genetic variants associated with susceptibility to radiotoxicity, we can move toward a more personalized radiotherapy approach for our patients,” they wrote.