3.3 Determination of optimal cutoff values of EVI1 transcript for patient grouping
Firstly, we used the normal threshold 8.0% as the cutoff value. As shown in Fig. 1B, the grouped patients had similar 5-year RFS rates (P =0.48). Next, patients were grouped into quartiles according to EVI1 transcript levels (1st quartile to 4th quartile, from low to high levels). Although the 5-year RFS rates were similar among the four groups (P =0.37), there was a trend that 1st quartile patients had lower 5-year RFS rate compared to the other 3 groups (P =0.17, 0.40, and 0.15, Fig. 1C), and the other 3 groups had the similar RFS rate (P =0.75, Fig. 1C).
Then, the ROC curve analysis was performed to determine the optimal cutoff value of EVI1 transcript levels. Although the EVI1 transcript levels could not significantly differentiate patients in relapse (area under the curve 0.57, P =0.24), 0.10% had the maximal Youden index and was determined as the optimal cutoff values. As a result, 32 (18.2%) and 144 (81.8%) patients had EVI1 transcript levels <0.10% and ≥0.10% and were defined as low and high EVI1 expression in our cohort, respectively.