In patients with clear cell renal cell carcinoma (ccRCC), radiogenomic analysis can reveal associations between specific features on CT imaging and messenger RNA-based tumor subtyping, authors of a preliminary analysis contend.
Among 177 patients with ccRCC for whom CT data and molecular subtyping information were available, well-defined vs. poorly-defined tumor margins were significantly associated with messenger RNA (mRNA) subtype m1 tumors, and a combination of margin definition and other qualitative tumor features was significantly associated with m3 subtype, reported Lan Bowen, MD, and Li Xiaojing, MD, from Huizhou (China) Central People’s Hospital.
“We demonstrated m1-subtype is positively associated with well-defined margin while m3-subtype is positively associated with ill-defined margin, renal vein invasion, and collecting-system invasion. These findings should be further investigated and validated in other cohorts,” they wrote. The report was published in Academic Radiology.
Radiogenomic analysis is a method for identifying associations between imaging features and gene expression profiles to provide information that can be used for clinical decision making.
The investigators used the technique to retrospectively explore associations between ccRCC mRNA-based subtyping and CT features.
They identified data on a total of 177 patients with ccRCC from The Cancer Genome Atlas for whom complete CT imaging, including contrast-enhanced images, and mRNA-based subtyping were available.
CT features studied included calcifications, margin definition, renal vein invasion, collecting-system invasion, multicystic tumors, nodular tumor enhancement, and intratumoral vasculature.
In univariate logistic regression analysis, well-defined tumor margins (vs. poorly-defined margins) were significant associated with m1 subtype (odd ratio [OR] 2.104, P = .041).
Similarly, a combination of well-defined tumor margins (OR 2.104, P = .013), no collecting system invasion (OR 0.421, P = .028), and renal vein invasion (OR 2.164, P = .026) were significantly associated with the m3 subtype.
They were no significant associations between CT imaging features and either the m2 or m4 subtypes, the authors found.
The authors did not report study funding sources or potential conflicts of interest.
SOURCE: Bowen L, Xiaojing L. Acad Radiol. doi: 10.1016/j.acra.2018.05.002.