Using a Tumor's Genomic Signature to Guide Radiation Therapy Doses

Researchers at the Moffitt Cancer Center have developed a genomics model called genomic-adjusted radiation dose (GARD) that can guide the radiation dose to match a specific tumor’s radiosensitivity. The model is expected to provide a framework for designing future clinical trials in radiation oncology.

The senior author on the study, Javier F. Torres-Roca, MD, has been a pioneer in the field of using genomic information to guide radiation therapy. A paper published in 2009 shared results of his group’s findings about a gene signature that could predict the intrinsic radiosensitivity of a tumor—a 10-gene signature led to the development of a radiosensitivity index (RSI), which was evaluated in head-and-neck cancer, rectal cancer, and esophageal cancer. With a greater RSI indicative of radioresistance, the researchers found that responders in all 3 cancers types had significantly lower RSI compared with nonresponders.

For their current study, which has been published in The Lancet Oncology, the research group used RSI to build a genomics model called the genomic-adjusted radiation dose (GARD). The hypothesis was that a high GARD value would indicate a high therapeutic effect of radiotherapy, which in turn can be correlated with therapeutic outcome. The retrospective study used data from the Total Cancer Care protocol to calculate GARD for primary tumors from 20 sites that were treated with radiation at the doses indicated for each disease type. Further, association studies were conducted to identify the relation between the GARD and clinical outcomes in 5 cohorts (Erasmus Breast Cancer Cohort, Karolinska Breast Cancer Cohort, Moffitt Lung Cancer Cohort, Moffitt Pancreas Cancer Cohort, and The Cancer Genome Atlas Glioblastoma Patient Cohort).

The researchers observed a wide range of GARD values (range 1·66—172·4) across the various participating cohorts, although the radiotherapy treatment doses matched what was recommended for that disease type.

“There is a high degree of variability among the GARD values for different patients within a single tumor type,” said Torres-Roca in a statement. “This suggests that different patients with the same type of tumor have different sensitivities to radiation therapy, further suggesting that the ‘one-size-fits-all’ approach to radiation therapy dose can be further optimized and personalized using tumor genomics.”

The authors report in their paper that despite the variance, GARD values independently predicted clinical outcomes in 4 cancer types: breast cancer, lung cancer, glioblastoma, and pancreatic cancer. The authors reported that in the Erasmus Breast Cancer Cohort, high GARD values were calculated in patients who had a longer 5-year distant metastasis-free survival, compared with those who had low GARD values (hazard ratio 2·11; 95% CI, 1·13-3·94; P = ·018).

These results highlight the potential of using the model to carefully carve out a precise dose of radiation for cancer patients based on their genomic signature. This can optimize the dose for maximal benefit, while avoiding the significant toxicity associated with radiotherapy.

Pointing out the value of using such an algorithm, Louis B. Harrison, MD, FASTRO, chair of Moffitt’s Radiation Oncology Department, said, “With multi-disciplinary care becoming standard for the majority of cancer patients, it is critical that precision medicine is expanded beyond drug therapy. The GARD model provides the first opportunity to genomically-inform radiation dose and is a safe and feasible approach to precision radiation oncology.”

Reference

Scott JG, Berglund A, Schell MJ, et al. A genome-based model for adjusting radiotherapy dose (GARD): a retrospective, cohort-based study [published online December 16, 2016]. Lancet Oncol. doi: http://dx.doi.org/10.1016/S1470-2045(16)30648-9.