To the Editor,

I wish to congratulate Gallardo and colleagues for their article [1], in which they evaluated the prognostic significance of Ki67 and aimed to determine the most accurate prognostic cut-off. They found that the prognosis of luminal breast carcinoma can be predicted using Ki67 as a continuous variable and a standard cut-off value of 14%. It was reported that interlaboratory reproducibility of Ki67 was moderate. Although central staining demonstrated a better intraclass correlation coefficient of 0.7, in the setting of local staining, the intraclass correlation coefficient was low, i.e., 0.59 [2]. Furthermore, the National Comprehensive Cancer Network Clinical Practice Guidelines in Oncology do not recommend testing Ki67 routinely [3]. Therefore, Ki67 staining is not routinely carried out as a standard practice due to its inefficient reproducibility and insufficient recommendation other than for the purpose of clinical study.