An optimized microRNA signature algorithm to predict clear cell renal cell carcinoma metastasis: A 265-case validation study using paraffin specimens.

Abstract

612 Background: Clear cell renal cell carcinoma (ccRCC) metastasis is associated with a dismal prognosis. Early identification of patients at high or low risk of metastasis may help to guide appropriate treatment. There are currently no reliable clinico-pathologic and laboratory modalities for this purpose in a routine clinical setting. We previously performed a microarray study using frozen ccRCC specimens and identified dysregulation of 4 microRNAs (miRNAs), miR-10b, miR-139-5p, miR-130b and miR-199b-5p, to be highly associated ccRCC metastasis and prognosis. Methods: We established a training cohort including formalin-fixed paraffin-embedded (FFPE) specimens of localized (n = 13; pT1) and metastatic (n = 15; M1) ccRCCs. We measured the expression of the 4 miRNAs of the training cohort samples by quantitative RT-PCR, normalized with miR-24, one of the most stably expressed miRNAs in human renal tissue as we previously reported. Using the risk score method, we rebuilt a mathematical formula with cutoff points to predict a patient to be at high, low or equivocal risk of metastasis. We validated the signature in an FFPE test cohort of 265-case primary ccRCCs. We examined the signature performance and its correlation with the cancer-specific survival. Results: Multivariate logistic regression analysis showed the signature to be highly associated with ccRCC metastasis (OR 7.67, 95%Cl 2.80-21.09, p < 0.0001), with the overall sensitivity and specificity of 80% and 76%. The sensitivities and specificities were 72% and 78%, 70% and 67%, and 78% and 75% for stage I, II or III patients, respectively. The signature predicted metastasis of small ccRCCs ( ≤ 4.0 cm), with the sensitivity of 60% and specificity of 77%. The overall positive and negative predictive values (PPV and NPV) were 61% and 89%. The area under the curve of ROC was 0.8. The signature was also found to be well correlated with worse cancer-specific survival (HR 3.06, 95%Cl 1.49-6.93, p < 0.01). Conclusions: This optimized miRNA signature can reliably predict ccRCC metastasis and prognosis and may help to stratify patients for more appropriate therapies and suitable clinical trials.