Wrong value of 1-specificity when drawing ROC curve when setting "legacy.axes = TRUE"

Question

I had drawn a ROC curve using the following codes:

roc_plot <- function(data, x, y){
    fit <- roc(response = data[[x]], predictor = data[[y]])
    plot(fit,
         print.auc = T,
         print.auc.x=0.2, print.auc.y= 0, 
         print.thres = T, 
         print.thres.cex = 1, 
         legacy.axes = T, 
         grid = T, 
         auc.polygon = T,
         lwd= 2, # ROC 線粗
         identity.lty =2, 
         smooth= T, 
         main="ROC Curve",col="#1c61b6")
}

roc_M2BP_plot <- roc_plot(sheet_4, "HCC", "M2BP")

At first, it looks normal

However, one day, it became like this

Note that the threshold, AUROC were all correct.

Actually, the plot is also correct if I set legacy.axes = F

What is the problem? I thought I didn't change the Excel data.