This retrospective study investigated the value of Ki67 as a predictive factor in relation to neoadjuvant chemotherapy and possible effects on prognosis. Ki67 was found to be an independent predictor for pathological complete responses and for the prognosis in all patients across all subtypes. Looking at Ki67 values in different molecular subtypes, it seemed that patients with triple-negative or hormone receptor-positive, HER2-negative breast cancer had a more favorable prognosis when a pCR was achieved, although these patient groups had a higher Ki67 proliferation rate. These results might suggest that the Ki67 cut-off values in patients undergoing chemotherapy may need to be set at a higher level in these subgroups to allow prediction of the chemotherapy response with a translation to the prognosis.
In the present study, the pCR rate in patients with triple-negative tumors (47.3%) was in the range reported in previously published studies (22-58%) [8, 16–19]. The pCR rates in HER2-positive carcinomas (28% without and 52% with neoadjuvant trastuzumab treatment) were higher than in the NOAH study (19% and 38%) , GeparQuattro study (31.7% for patients treated with trastuzumab) , and TECHNO study (42% for patients treated with trastuzumab) , but the sample sizes in the present study were much smaller. The pCR rate in the HER2-negative, hormone receptor-positive group (5.7%) was also consistent with that in other published studies . The present study also confirms previous reports that pCR is associated with a more favorable prognosis in some molecular subgroups [8, 24].
Previously published parameters such as age, BMI, tumor stage, histological type, hormone receptor and HER2 status correlated with pCR, as in other previously published studies [23, 25]. Ki67 also had a strong correlation with pCR and added independently to the predictive value of a logistic regression model. This effect was present with regard to the total group of patients and with regard to the molecular subtypes of breast cancer as defined by hormone receptor and HER2 status. It reached statistical significance in the hormone and HER2 receptor-positive groups, with a cut-off at 13%, but not with regard to the triple negative group. Cut-off calculations within the molecular subgroups showed that much better differentiation between treatment response groups could be achieved with much higher Ki67 cut-off values for the hormone receptor-positive group (between 36% and 40%) and the triple-negative group (between 30% and 40%). For the HER2-positive group, the cut-off value was between 17% and 20%. However, this group is difficult to interpret, as it included patients with and without neoadjuvant trastuzumab treatment.
Triple-negative tumors generally have a much higher proportion of Ki67-positive cells, and differentiation between responsiveness groups could thus be expected at a higher level. However, in the hormone receptor-positive group, previous molecular analysis determined a cut-off at 13% to differentiate between luminal A and luminal B tumors . This cut-off does not appear to be the best for predicting the chemotherapy response.
Patients with tumors that have a very high level of proliferation might possibly have a better prognosis than those with lower Ki67 values as a result of a successful therapy response. In the present study, this was shown indirectly for the triple-negative and hormone receptor--positive subgroup. Patients with a pCR had a better prognosis and a higher mean Ki67 value, whereas patients with a lower Ki67 value had a more unfavorable prognosis. This might explain some of the inconsistencies in reports concerning the prognostic value of Ki67 .
One aspect of the present study involves both advantages and disadvantages. On the one hand, the methods of Ki67 staining and evaluation used are part of routine clinical practice. The staining and assessment of whole sections may be a strength, as most published studies use tissue microarrays and are unable to account for heterogeneously expressed Ki67 in a whole slide section. In addition, the fixation and staining procedures were carried out directly after the fixation and embedding of the core biopsies into paraffin. This may have reduced the potential for variability in studies using paraffin blocks of different ages, ranging up to decades. On the other hand, routine clinical assessment means that different batches of chemicals and antibodies are used, and also that there are different observers involved. Another problematic issue might concern the arbitrary molecular classification of tumors used . Approximately 30% of luminal B tumors are HER2-positive . Patients with these tumors were included in the HER2-positive group in the present study. However, further subcategorization was not possible due to the small sample sizes in the subgroups, and not all of the HER2-positive patients were treated with trastuzumab.
It would have been desirable to validate other cut-off values for Ki67 in a sample set of patients treated with chemotherapy, but the sample size in the present study appeared to be too small to pursue this aim.