The prognostic and predictive potential of Ki-67 in triple-negative breast cancer

As a cell proliferation biomarker, Ki-67 is principally used in ER+/HER2− breast cancer. However, the importance and the best cutoff point of Ki-67 in triple-negative breast cancer (TNBC) remains unclear and was evaluated in this study.A total of 1800 patients with early invasive TNBC between 2011 and 2016 at Fudan University Shanghai Cancer Center were consecutively recruited for this study. The optimal cutoff for Ki-67 was assessed by Cutoff Finder. Propensity score matching (PSM, ratio = 1:2) was performed to match the Ki-67low group with the Ki-67high group. Overall survival (OS) and disease-free survival (DFS) were compared between the two groups using the Kaplan-Meier method and Cox regression model. The most relevant cutoff value for Ki-67 for prognosis was 30% (p = 0.008). At the cutoff point of 30%, worse DFS and OS were observed in the Ki-67high group. In multivariate analyses, N-stage (p < 0.001), T-stage (p = 0.038), and Ki-67 at the 30% threshold (p = 0.020) were independently linked to OS. In subgroup analysis, Ki-67 cutoff at 30% had prognostic and predictive potential for DFS with either tumor size ≤2 cm (p = 0.008) or lymph node-negative (N−) (p = 0.038) and especially with T1N0M0 (stage I) TNBCs. For 945 N− TNBC patients, adjuvant chemotherapy (CT) was associated with better OS in the Ki-67high group (p = 0.017) than in the Ki-67low group (p = 0.875). For stage I/Ki-67low patients, adjuvant CT did not affect DFS (p = 0.248). Thus, Ki-67 cutoff at 30% had early independent prognostic and predictive potential for OS and DFS in TNBCs, and Ki-67 > 30% was significantly associated with worse prognosis, especially for stage I patients. For stage I/Ki-67low TNBC patients, the advantage of CT is unclear, providing the basis for future de-escalation therapy.

tenacious proliferation and poor prognosis. In a study conducted by Cheang 5 , the cutoff value for the immunohistochemically determined Ki-67 index to distinguish luminal B from luminal A BC was 13.25%, while some panels showed that a threshold of ≥20% was clearly indicative of 'high' Ki-67 status 6 . Currently, Ki-67 is principally used to evaluate prognosis, guide adjuvant treatment and predict the response to neoadjuvant treatment in ER+/ HER2− BC. The Ki-67 cutoff at 14% is an important parameter in subclassifying luminal BC into the luminal A subtype with good prognosis and the luminal B subtype with worse prognosis 6-8 . However, the importance and the best cutoff point of this marker as a prognostic and predictive factor in TNBC remains unclear. In the current study, we evaluated the prognostic and predictive potential of Ki-67 in TNBC.

patients. A total of 2465 female patients with TNBC between January 2011 and December 2016 at Fudan
University Shanghai Cancer Center (FUSCC) were consecutively recruited, as Ki-67 was not widely used in FUSCC as a routine test until 2011. After screening, only 1800 patients who were diagnosed with early invasive TNBC and underwent surgery were enrolled for this study. The other inclusion criteria for all participants were 18-70 years of age and BC as the only primary tumor. The exclusion criteria were neoadjuvant chemotherapy (CT) or radiotherapy (RT), deficiency of clinical data or lack of follow-up (Fig. 1).
Written permission was obtained for the collection of data from the FUSCC database, allowing the use of personal data for research purposes. The study was ethically approved by the Ethical Committee and Institutional Review Board of FUSCC, and all the methods were performed in accordance with the approved guidelines.
Pathology and immunohistochemistry. The ER, PR, and HER2 status and Ki-67 expression of paraffin-embedded tissue sections of breast surgical tumor specimens from the pathology center of FUSCC were evaluated according to standard procedures. All surgical specimens were fixed with 10% neutral phosphate-buffered formalin and paraffin-embedded 4 μm-thick slices of representative tumor blocks were stained with hematoxylin and eosin. The cutoff values for ER/PR positivity were 1% of positive tumor cells with nuclear staining by IHC. HER2 positivity was defined as 3+ by IHC using circumferential membrane-bound staining or as positive on fluorescence in situ hybridization (FISH), whereas HER2 negativity was defined as 0 to 1+ by IHC or 2+ by IHC without gene amplification confirmed by FISH detection 9 . TNBC was defined only when the tumors were ER−, PR− and HER2-negative.
The immunohistochemical analysis for Ki-67 value was performed on Benchmark XT system (Ventana, Tucson, AZ, USA), using MIB-1 antibody (dilution 1:100; Code M7240, Dako, Glostrup, Denmark). The reported Informed consent. Informed consent was obtained from all individual participants included in the study. Cutoff value for Ki-67 verified in the PSM cohort. To define the most relevant cutoff value for Ki-67 distinguishing prognosis, we used the online tool "Cutoff Finder". This method has been employed by several other studies for the prognostic determination of various tumors [15][16][17] . The optimal cutoff value of Ki-67 for distinguishing DFS was 30% (p = 0.020). We defined patients with Ki-67 > 30% as the Ki-67 high group, otherwise as the Ki-67 low group. The representative IHC images of the Ki-67 high group, the Ki-67 low group and Ki-67 at the 30% threshold were presented in Fig. 2. There were 464 (74.2%) patients with Ki-67 ≤ 30% and 1336 (74.2%) with Ki-67 > 30%. To exclude differences in other baseline levels that might affect the prognosis of the two groups, PSM (ratio = 1:2) was performed to match the patients in the Ki-67 low group with those in the Ki-67 high group. After PSM, all baseline characteristics (age, BMI, location, multifocality, T-stage, and N-stage) of the two groups were well balanced (Table S1). There were 464 patients with Ki-67 ≤ 30% and 928 patients with Ki-67 > 30%, with 1392 patients total in the PSM cohort. At the cutoff point of 30%, worse DFS and OS were observed in the Ki-67 high group of the PSM cohort (p = 0.008 and p = 0.043, respectively).

Ki-67 cutoff at 30% is an independent factor for OS and DFS.
For the 1392 patients, Ki-67 > 30%, larger tumor size and lymph node positivity were associated with shorter DFS and OS in the combined PSM and univariate Cox regression analysis, while the other clinical characteristics, such as age, multifocality and histological subtype, did not influence the prognosis. In the combined PSM and multivariate Cox regression analysis, N-stage (p < 0.001), T-stage (p = 0.038), and Ki-67 at the 30% threshold (p = 0.020) were independently linked to OS. However, only N-stage (p < 0.001) and Ki-67 at the 30% threshold (p = 0.010) were independently linked to DFS. The hazard ratios (HRs) of the high expression of Ki-67 were 1.947 (95% CI: 1.108-3.421) for OS and 1.604 (95% CI: 1.118-2.300) for DFS, with the low-level of Ki-67 used as a reference. All the results from the univariate analysis and final multivariate Cox regression model are presented in Tables 2 and S2.
Association between the expression of Ki-67 and adjuvant chemotherapy. For  www.nature.com/scientificreports www.nature.com/scientificreports/ Ki-67 high group (p = 0.016), while in the N−/Ki-67 low group, the interaction between adjuvant CT and prognosis was not statistically significant (p = 0.668). Kaplan-Meier survival analysis also confirmed that adjuvant CT was associated with better OS in the N−/Ki-67 high group (p = 0.017) than in the N−/Ki-67 low group (p = 0.875) (Fig. S1a,b). The HR for OS in the N−/Ki-67 high /without adjuvant CT group was 3.615 (p = 0.005, 95% CI: 1.108-3.421), which was higher than 2.408 in the N-/Ki-67 high group and 1.938 in the N−/without adjuvant CT group.  www.nature.com/scientificreports www.nature.com/scientificreports/ In the stage I/Ki-67 low group (n = 183), the median follow-up time was 39.5 months, among which five patients had recurrence without any death event. Further Kaplan-Meier analysis showed that adjuvant CT did not affect DFS (p = 0.248), while adjuvant CT played an important role in over stage I/Ki-67 low patients, both OS and DFS (Fig. 5a,b). For stage I patients who underwent adjuvant CT (n = 424), Ki-67 expression did not affect   www.nature.com/scientificreports www.nature.com/scientificreports/ OS (p = 0.189); for those without adjuvant CT (n = 99), the Ki-67 high group had a higher risk of recurrence than the Ki-67 low group (p = 0.049).

Discussion
Because of the poor prognosis of TNBC and the lack of effective treatment so far, the direction of future research involves the search for effective biomarkers to guide treatment and predict prognosis 18 . In this study, we consecutively enrolled 1800 patients to further explore the prognostic and predictive potential of Ki-67 expression in TNBC. As treated patients had more unfavorable baseline characteristics, the effect of Ki-67 expression could have been underestimated. We therefore conducted PSM in our analyses, and Cox regression analysis verified significant differences between the Ki-67 high group and the Ki-67 low group in the PSM cohort. All patients were recruited from a single clinical center, ensuring the stability of the quality of pathological marker testing, clinical diagnosis and evaluation, and treatment decisions.
Currently, since testing for Ki-67 is more convenient and economical, the assessment of the level of Ki-67 expression has been proposed in routine practice to be a measure for the quantification of cell proliferation in www.nature.com/scientificreports www.nature.com/scientificreports/ BC samples 19 , particularly to define prognostic subgroups of HR+ tumors. However, among most cohorts of BC studies, the number of cases of TNBC and HER2+ BC was quite small 20 , which might limit the ability of Ki-67 to predict the prognosis and identify clinically distinct subclasses. In addition to the shortcoming of a small number of samples in previous studies, a clear and uniform cutoff in TNBC has still not been established. Since baseline Ki-67 values for TNBC are much higher than those for luminal tumors 20 , the definitions of Ki-67 cutoff values in TNBC have been diverse and controversial, ranging from 10% to 61%, with one study in South Korea defining the cutoff as 10% 21 and one study in Italy defining the cutoff as 35% 22 . In our study, we found that the optimal cutoff value of Ki-67 in TNBC patients for distinguishing prognosis was 30% in the combined PSM and multivariate Cox regression analysis. When Ki-67 expression ranged between 10 and 25%, the risk of misclassification was 37%; when Ki-67 expression was <10% or ≥25%, the risk of misclassification was only 11% 4 . We defined Ki-67 as a classification variable at a 30% threshold, which might stabilize the variability between observers and across laboratories and reduce the influence of manual interpretation errors on prognostic evaluation. Standardized cut-off values for Ki-67 have not been established, and laboratory-specific values should be used.
With Ki-67 at the 30% threshold, N-stage and T-stage were independently linked to prognosis, and those patients with Ki-67 > 30% were associated with shorter DFS and OS. However, this cutoff value had the greatest prognostic and predictive potential for prognosis in patients with a tumor size ≤ 2 cm or those who were lymph node-negative, and especially those with stage I disease. Therefore, we speculated that the Ki-67 cutoff at 30% was an early independent predictor in early invasive TNBC, but its use was limited in TNBCs over stage I. In previous studies, Ki-67 > 30% might be attributed to the early recurrence pattern of TNBC within the first three years of follow-up 23,24 . In this study, 82.4% (75/91) of relapses occurred in the first three years following the initial pathological diagnosis, and among these patients, 81.33% (61/75) had Ki-67 > 30%. Therefore, TNBC with its high proliferative potential should be followed-up more frequently in the first three years after diagnosis.
As for treatment, many studies have shown that the expression level of Ki-67 is related to the subsequent treatment selection and efficacy [25][26][27] . Previous studies have demonstrated that Ki-67 is one of the markers of chemosensitivity for BC, but such an association has rarely been found in TNBC. Interestingly, in this study, we found that N−/Ki-67 high TNBC patients could benefit from adjuvant CT; however, the advantage of CT was not clearly observed in stage I/Ki-67 low TNBC patients. One of the reasons behind this phenomenon might be that the low expression of Ki-67 is insensitive to CT, and another reason could be that these stage I/Ki-67 low patients themselves had a favorable prognosis and might be exempted from CT, which provides a basis for future de-escalation therapy. Further studies on the correlation between Ki-67 and chemosensitivity in TNBC might be necessary.
One of the limitations of our study was that there were few TNBC patients in T 1b N 0 M 0 , and whether these patients need adjuvant CT after surgery is still controversial [28][29][30] ; thus, we were unable to carry out in-depth survival analysis and prognosis prediction for the T 1b N 0 M 0 group, nor can we guide clinical practice. Besides, neoadjuvant therapy is becoming an option for more and more TNBC patients, which was not included in our study. Therefore, the application of this conclusion to neoadjuvant therapy in TNBCs is limited and deserves further study. Furthermore, Ki-67 might be even more powerful if incorporated in the design of trials of risk-adapted adjuvant therapies in the future.

conclusions
For TNBC patients, Ki-67 cutoff at 30% had early independent prognostic and predictive potential for OS and DFS, and Ki-67 > 30% was significantly associated with worse prognosis, especially for stage I patients. For N−/ Ki-67 high TNBC patients, adjuvant CT may be necessary; however, for stage I/Ki-67 low TNBC patients, the advantage of CT is unclear, providing the basis for future de-escalation therapy. Ki-67 cutoff at 30% can be used for further classification of TNBC into two subtypes with different responses and prognoses.

Data availability
The data that support the findings of this study are available from the corresponding author, but restrictions apply to the availability of these data, which were used under license for the present study, and so are not publicly available. Data are, however, available from the authors upon reasonable request and with permission of the corresponding author.