講演情報
[O21-2]既存のempirical modelを用いたBRCA1/2病的バリアントのリスク評価の有用性と限界
○二川 摩周1,2, 谷岡 真樹3, 浦川 優作2,4, 十川 麗美1, 加藤 芙美乃1, 山本 英喜1,2, 平沢 晃1,2 (1.岡山大学病院 臨床遺伝子診療科, 2.岡山大学大学院医歯薬学研究科 臨床遺伝子医療学, 3.岡山大学学術研究院 医歯薬学域 医療AI人材育成プログラム, 4.神戸市立医療センター中央市民病院 腫瘍内科)
Purpose.Empirical models are tools used for risk assessment of BRCA1/2 pathogenic variants (PVs), and the 2010 Myriad table and the 2018 HBOC consortium table (HBOC table) are often used in Japan. We aim to compare the accuracy of these empirical models using a prospective cohort study at Okayama University.Methods.We compared the accuracy of empirical models using 120 subjects, 107 of whom had the BRCA1/2 PVs and 13 of whom did not, who agreed to participate in a prospective cohort study. BRCA1/2 genetic testing (BRCA1/2 : 67 cases, BRCA1 or BRCA2 : 46 cases) and multi-gene panel testing (MGP) (7 cases) were analyzed separately. Variant retention probabilities identified from each table that were higher than the stepwise cutoff values (＜5%, ＜10%, and ＜15%) were judged positive.Results.The positive predictive values (PPV) for each cutoff were 50%, 30%, and 14% for the Myriad table and 98%, 85%, and 52% for the HBOC table in cases with BRCA1/2 genetic testing.The PPV of MGP cases were 75%, 50%, and 25% in the Myriad table and 100%, 100%, and 50% in the HBOC table. The HBOC table was superior to the Myriad table for PPV of the variants using any of the cutoffs, with the best values obtained at ＜5% in all cases.Discussion and Outlook.In Japan, the HBOC table with a cutoff value of ＜5% is a useful empirical model for risk assessment of BRCA1/2 PVs. On the other hand, about 51% (61/120) of the subjects performed BRACAnalysis, which may have caused selection bias. We plan to further minimize the selection bias and examine the usefulness of the empirical model.