The metabolic reprogramming of cancer cells and its implications for tumor growth and dissemination has gained increasing prominence and could contribute to a better understanding of NAC response. Some proteins like glucose tranporters and monocarboxilate transporters are essential for metabolic control and have been characterized as predictors of response and prognostic factors. Thus, this study evaluated the expression of MCT1, MCT4, CD147, GLUT1 and CAIX in locally advanced BC submitted to NAC and their relationship with pCR, DFS, DSS and OS. Unexepectedly, CAIX expression has been showed as predictor of pCR and was associated with higher DFS and DSS in patients with locally advanced breast cancer treated by NAC using AC-T.
The present study evaluated a cohort of patients with breast cancer at stages IIb and III treated with NAC, whose tumor size was greater than 5.0 cm in most of the cases. Moreover, there was a long follow-up time with a small number of missed patients. In this population, the expression of MCT1, MCT4, and CD147 was lower than that observed by Pinheiro et al. (19.4%, 7.3% and 11.0%, respectively) [20]. GLUT1 and CAIX expressions were also lower than the frequencies of 46.0% and 18.0% seen in the study by Pinheiro et al. [17] and 28.5% and 12.5% in the study of Vleugel et al. [25]. It should be considered that in Pinheiro et al studies [17, 20] and Vleugel et al study [25], the percentage of the population with tumors larger than 5 cm ranged from 9.9% to 17.6%, while in the present study, tumor size was greater than 5.0 cm in 90.3% of the cases. In addition, the antibodies and the positivity criteria used by Vleugel et al. are different from those used in the present study [25].
In accordance with previous studies [17, 18, 20, 26], the expression of the metabolism-related proteins was associated with worse prognostic factors. For instance, tumor characteristics related to loss of differentiation and higher growth and probability of dissemination, like histological grade of Nottingham III, mitotic score 3 and nuclear grade G3 were associated with MCT1, GLUT1 and CAIX. In addition, presence of necrosis was associated with MCT4, GLUT1 and CAIX, while lymph node involvement was associated with MCT4, CD147 and GLUT1 expressions. Finally, the lack of ER and PR expression was associated with MCT1, CD147, CAIX and GLUT1. The hyperglycolytic and acid-resistant phenotype in undifferentiated cells is responsible for the acidification of the extracellular environment, which, in turn, stimulates tumor progression and dissemination [15, 27-30]. Also, rapid growth, partly maintained by the hyperglycolytic phenotype, leads to hypoxia and increased necrosis, which also contributes to the metabolic reprogramming towards an hyperglycolytic metabolism, thus creating a cyclic process to stimulate tumor growth and dissemination [15, 27-30]. Therefore, there would be a process of natural selection where tumor cells with characteristics of greater aggressiveness, when manifesting the hyperglycolytic phenotype, would have adaptive advantages for greater proliferation and dissemination.
The percentage of pCR observed (16.3%) is consistent with data seen in prospective phase II and III clinical trials, ranging from 15 to 30% and using sequential use of docetaxel to chemotherapy [31, 32] or weekly paclitaxel [33]. However, pCR is often related to higher survivals and is more frequently associated with aggressive tumors [7-12, 34-36]. This behaviour has been referred to as the "triple negative paradox phenomenon" [37]. It may be related to the expression of proto-oncogenes and immune response regulatory genes, as well as the lack of an additional therapeutic option (eg hormone therapy), which would allow the rapid evolution of the disease in those cases that do not reach pCR with NAC [37, 38]. In this study, pCR was also associated to aggressive tumors, occurring in 19.4% of triple negative compared to 9.1% in luminal A. Our results is in agreement with previous report describing pCR rates ranging from 20.0 to 34.0% in triple negative, and 0.0 to 7.5% in luminal A tumors [12]. Additionally, associations were observed between pCR and age, absence of ER expression, HER2 overexpression, mitotic score, as well as GLUT1 and CAIX expression. In multivariate analysis, only regional lymph nodes staging (TNM - N), mitotic score and CAIX expression were independent predictors of pCR.
To the best of our knowledge, CAIX expression has not been previously described as an independent predictor of pCR. Aomatsu et al. observed that CAIX expression is related to lower pCR rate and considered this protein a chemoresistance marker [39]. In that study, CAIX expression frequency was 46.0% [39], whereas in the present study it was only 7.4%. Another difference between the two studies is the frequency of pCR seen in 29.0% of patients in Aomatsu study versus 16.3% in the present one [39]. However, the differences in samples’ characteristics should be emphasized; while in the present study the sample was comprised of patients with locally advanced tumor treated with AC-T, the Aomatsu et al. study sample consisted of 102 patients with early-stage breast cancer treated with 5-fluorouracil, epirubicin, and cyclophosphamide [39].
Other explanations related to the phenotypic manifestation could explain the unprecedented result of the present study. In a recent study, Euceda et al. [40] evaluated, through magnetic resonance spectroscopy, the metabolic behavior of breast cancer of 122 patients treated with NAC and randomized to sequential use of bevacizumab. Good responders presented an initial metabolic profile related to greater aggressiveness and elevated levels of lactate were observed, which progressively increased throughout the treatment. The authors suggested that patients with tumors with a metabolic profile associated with increased aggressiveness are more likely to benefit from this treatment in terms of reduced tumor size, possibly due to a change in their phenotype - becoming metabolically non-glycolytic - or related to morphological changes that would block lactate excretion [40]. This would likely alter the tumor microenvironment, reducing extracellular acidity, which would improve the efficacy of chemotherapeutics, classified as weak bases that ionize under low pH conditions [41]. This context is very similar to that observed in the present study, especially with regard to the greater CAIX expression in pre-treatment tumors from patients who reached pCR after NAC. Even with the expression of a protein responsible for pH control and promoter of an appropriate microenvironment to tumor growth and proliferation, the expected aggressive phenotype was not able to manifest in the group of patients evaluated in this study, which allowed higher rates of pCR, contrary to the initial expectations.
In line with the association with pCR, CAIX expression was also associated with higher DFS and DSS. These findings were also not previously described, and go against previous studies showing CAIX as a poor prognostic factor [17, 39, 42, 43]. Generali et al. demonstrated women with breast cancer treated with epirubicin and tamoxifen had lower DFS and OS when expressing CAIX [42]. Similarly, Pinheiro et al. observed that CAIX expression was associated with an increased risk of relapse [17]. In the study by Aomatsu et al., in which CAIX expression was evaluated in breast cancer tumor samples before and after NAC, the presence of the protein was prognostic of lower DFS in both situations [39]. As a counterpoint, it is important to cite two studies. In the first one, Ivanova et al. evaluated breast cancer samples of 3,455 patients and observed high expression of CAIX mRNA was associated with lower DFS in basal-like and triple negative subtypes and lower OS in luminal B, but not in luminal A and HER2 + [43]. On the other, Chen et al evaluated the expression of CAIX and CAXII mRNA, enzymes with the same catalytic function, but with related different prognostics predictions (CAIX related to worse and CAXII to good prognosis) [44, 45]. Chen et al. observed high expression of CAIX mRNA was associated with increased survival in the luminal subtype while CAXII mRNA expression was linked to reduced survival in basal and HER2 positive breast cancer [44]. Furthermore, they suggest that CA enzymes could have their functions regulated by changes in the pH of the tumoral microenvironment [44]. Thus, we can assume that in our study, the conditions of the tumor microenvironment (related to the large tumor size and NAC based on AC-T) may have determined CAIX functional alterations and, consequently, may have been associated with pCR and higher survival. Moreover, in the samples evaluated in our study, the low CAIX expression could be compensated by a higher CAXII expression, unfortunately not evaluated by us. It should be noted the lack of correlation of triple-negative cases with pCR rates in the multivariate analysis. We consider, however, that this finding is strictly related to statistical power. Due to the number of included variables, the final sample size in this analysis was substantially reduced, probably determining this lack of correlation. In addition, among the triple-negative cases that demonstrated pCR, only one of them had CAIX expression. Given these data, we can state that there is no strong correlation between CAIX and pCR expression between triple-negative tumors, even with the result found in the multivariate analysis..
Since the biological material used in TMA construction is dated from 2005 to 2011, its quality should be considered as a limitation of this study. Although all the samples come from the same service, differences in the techniques of fixing and preserving the material should be considered, which could contribute to the reduction of antigenicity, decrease in the sensitivity of the IHC reaction and, of course, lower detection of protein expression [46-48]. It is also worth noting that the TMA blocks used in the present study were composed of single samples from each patient and, as already mentioned, there were a considerable number of cases excluded by the lack of tumor representativeness.