NEW YORK -- Researchers in Sweden have found that levels of the platelet-derived growth factor receptor beta (PDGFRb) protein in the tissue surrounding early-stage breast cancers could predict whether patients are likely to respond to radiation therapy.
The findings, published Wednesday in Clinical Cancer Research, are based on a retrospective analysis of patients with the earliest stage of breast cancer, ductal carcinoma in situ (DCIS), who after surgery were randomized to receive either radiation therapy or placebo. The clinical trial, dubbed SweDCIS, involved around 1,000 women with DCIS between 1987 and 1999, on whom 20 years of follow-up data were available.
Despite the generally positive prognosis for DCIS patients treated with standard-of-care breast conserving surgery and radiation, the authors of the Clinical Cancer Research paper, including Arne Östman of the Karolinska Institut, pointed out that roughly 10 percent of patients still experience recurrence within the first decade of treatment. Homing in on a biomarker to identify which patients are likely to experience recurrence — defined as an ipsilateral breast event — could help avoid expensive therapies unlikely to improve outcomes and, ultimately, develop combination treatment strategies to prevent relapse.
Fibroblasts as biomarkers
The PDGFRb protein that Östman and colleagues evaluated as a biomarker is expressed in the cells of tissue surrounding the breast tumor, or the stroma. Most research for predictive biomarkers tend to look primarily at their expression in the tumor cells themselves.
"My research focus has been on cancer-associated fibroblasts for a very long time, and this study exemplifies the potential of this cell type as a biomarker in general," Östman said. "This is maybe the first study that has used a randomized trial-devised tumor collection and obtained a very strong predictive signal for a fibroblast model in general."
To assess whether the fibroblast marker could predict responses to adjuvant radiation therapy in DCIS, Östman and colleagues retrospectively analyzed the samples from patients enrolled in the SweDCIS study, and considered their stromal PDGFRb expression alongside their clinical outcomes following treatment. The researchers performed immunohistochemistry (IHC) staining on samples from 590 patients, with a median follow-up of 17.4 years.
The researchers generated multiple IHC scores ranging from zero to three for each sample to determine an average score per patient. A score of zero denoted samples in which the average intensity of the PDGFRb stroma fraction was undetectable, while a score of three indicated a "strong" average intensity. They then took the median of all the samples scored by IHC, which turned out to be 1.65, and used it as the cutoff delineating PDGFRb-low and PDGFRb-high cancers.
Assessing the rates of recurrence alongside the IHC scores, Östman and colleagues found that among patients whose fibroblast cells had low PDGFRb expression, those who got adjuvant radiation had one-fifth of the recurrence risk as those who did not.
Among patients whose samples were PDGFRb-high, however, there was no significant added benefit to receiving radiation therapy. Further, when researchers looked at all patients who received adjuvant radiation, the risk of recurrence within 10 years was higher for the PDGFRb-high group versus the PDFGRb-low group.
Predictive versus prognostic
To truly assess whether PDGFRb expression in the DCIS stroma predicts for response to radiation therapy, and not just better or worse prognosis generally, Östman and colleagues compared the clinical outcomes among patients with PDGFRb-low versus -high samples in the group that did not receive adjuvant radiation. Here, there was no significant difference in recurrence rates.
The findings are in line with earlier research Östman has conducted. "The radiation therapy predictive signal was much stronger than the prognostic signal," he said. "[PDGFRb expression] is a biology that is affecting radiation sensitivity but not immediately the aggressiveness of the disease."
It will take further research into the mechanistic link between PDGFRb expression and radiation sensitivity to say for sure why the stromal protein expression functions as a predictive, rather than a prognostic, biomarker in DCIS, but Östman's research has led him to believe this mechanism hinges on the regulatory function of the protein.
"PDGFRb is a very important regulatory molecule for fibroblasts, and that is very well known from studies in developmental biology and tumor biology," he said, acknowledging though that at this stage, it's unclear which PDGFR-associated fibroblast activity is responsible for the predictive association.
In their paper, Östman and colleagues highlighted prior studies suggesting that the biomarker was associated with increased signaling activity and ligand production of TGF-beta, which is known to be a modulator of tumor radiosensitivity. "It will be important to perform experimental studies investigating how PDGFRb-signaling manipulations will affect sensitivity in in vivo models of adjuvant radiation therapy," they wrote.
The fact that stromal PDGFRb expression predicted for therapy response, instead of just being a prognostic marker, could ultimately give the biomarker an advantage over other scoring systems that physicians are using to guide DCIS treatment decisions. For example, Exact Sciences' OncotypeDx DCIS score, which analyzes a gene expression signature in tumor cells, is often used as a prognostic tool to predict recurrence risk.
The DCISionRT test from PreludeDx — a firm that provided guidance on the various statistical analysis conducted in the current study — uses an artificial intelligence- and machine learning-derived signature associated with disease progression pathways and clinical features to predict DCIS patients' response to radiation, but is also used to predict recurrence in general. Östman and colleagues hope to compare their biomarker to these scoring systems in further validation studies.
Next steps, treatment implications
Should future studies in larger DCIS patient cohorts validate PDGFRb expression as a predictive biomarker for adjuvant radiation, and in vivo studies clarify the biomarker's mechanistic role, it could inform future treatment strategies, according to Östman and colleagues. They wrote in their paper that "findings from such studies will eventually suggest novel radiosensitizing combination therapies."
Östman noted, for example, that existing drugs could be used to block PDGFRb, which in turn, could theoretically help patients with higher PDGFRb expression derive benefit from radiation plus one of these drugs.
"There are good [drug] candidates that have the capacity to block [these] receptors, so … there could be future experiments where we are testing … if blockade of PDGFRb receptors in fibroblasts would increase radiation sensitivity," he said. "There is an implication that this [research] is opening the door to new combination therapies to test."
According to Östman, there may be "good" drug candidates among already approved tyrosine kinase inhibitors such as imatinib (Novartis' Gleevec), sunitinib (Pfizer's Sutent), and sorafenib (Bayer's Nexavar), all of which have shown PDGF receptor inhibitory activity. "Eventually, it would be even more interesting if one had a tyrosine kinase inhibitor, which was more specific for the PDGFb receptor, for an even better PDGF inhibitory antibody."