In the current study, we performed an extensive molecular analysis of several ICs, JAK2 and FoxO1, in a clinical series of BC. We found that most tumors overexpressed at least one IC. In fact, after analyzing the RQ data as a quantitative variable, CD276 (B7-H3), CD274 (PDL1), and PDCD1 (PD1) were overexpressed in the majority (95.2%). On the other hand, CTLA-4, PDCD1LG2 (PDL2), JAK2, and FoXO1 showed lower values, suggesting that these genes are frequently expressed close to normal. These statements were further corroborated by evaluating the RQ data as a two-category qualitative variable according to the cut-off point set at 1.5 by other investigators [51–54].
Our results partly agree with those of Fang et al. [55], who analyzed the gene expression of 50 different ICs in BC compared to healthy tissue and assessed their prognostic value. These authors observed that CD276 (B7-H3) and PDCD1 (PD1) were overexpressed, while PDCD1LG2 (PDL2) was underexpressed, which is consistent with our results. However, CTLA-4 was overexpressed, and CD274 (PDL1) was underexpressed. These discordances may be due, at least in part, to the lack of a reference or standard cut-off point to define the presence or absence of overexpression in the literature. Regarding JAK2 and FoXO1 expression, and in agreement with our results, Liu et al. [56] and Lallemand et al. [57] demonstrated in their BC clinical series that both genes were under-expressed. Thus, these results suggest that the gene expression profiles of both the ICs and the different signaling pathways analyzed are complex and highly variable in BC.
After analyzing gene expression in the different BC phenotypes, we observed that the TN/BL and luminal B/HER2-positive phenotypes had the highest RQ values for CTLA-4 and PDCD1 (PD1) receptors. At the same time, CD274 (PDL1) and PDCD1LG2 (PDL2) were overexpressed in the HER2-enriched subtype. On the other hand, CD276 (B7-H3) and JAK2 were elevated in the luminal B/HER2-negative phenotype, unlike FoXO1, which was overexpressed in luminal A tumors. In fact, after qualitative analysis, we confirmed that CTLA-4 and PDCD1 (PD1) were significantly associated with the phenotypes described above. Likewise, CD276 (B7-H3), JAK2, and FoXO1 correlated with the HER2-negative luminal A and B phenotypes. However, although CD274 (PDL1) and PDCD1LG2 (PDL2) were overexpressed in most HER2-enriched tumors, we found no significant differences compared to the other phenotypes.
These results are in line with what has been described in the literature. Several studies have demonstrated CTLA-4 expression in BC. One of the first was Contardi et al. [58], who detected the expression of this receptor in the membrane and cytoplasm of five invasive ductal carcinomas by immunohistochemistry. Subsequently, Mao et al. [59] and Kassardjian et al. [60] found in more extensive clinical series that it was elevated in neoplastic tissue at mRNA and protein levels.
The results of the current clinical series have been corroborated by our in vitro studies, showing intrinsic expression, especially in TN/BL cell lines [23]. Furthermore, some authors have suggested that CTLA-4 activation in neoplastic cells is directly related to cell adhesion, drug metabolism, and iron and amino acid transport, among other functions [61]. Considering all these data, we could suggest that CTLA-4 may be a critical IC in the aetiopathogenesis of TN/BL tumors.
Previous studies have also shown that PDCD1 (PD1) mRNA levels are elevated in neoplastic tissue relative to healthy tissue [62, 63]. However, although the intrinsic expression of this gene has not been studied in depth in BC, the few available data show that it remained stable in different BC cell lines [23] and tends to be overexpressed in HER2-enriched and TN/BL phenotypes [64]. Furthermore, contrary to our results, no correlation with ER expression has been observed [63]. These discrepancies may be due to the limited number of cases analyzed by these authors, as the methodology applied was the same. On the other hand, Muenst et al. [65] immunohistochemically evaluated the presence of PD1-positive TILs and observed a higher percentage in luminal B/HER2-positive tumors.
Fang et al. [66] observed a higher expression of this gene in samples with a higher percentage of PD1-positive TILs in luminal B/HER2-positive tumors. Following these observations, it has been described that the PDCD1 (PD1) in peripheral blood samples from patients with PR-positive tumors is higher than in HER2-positive tumors. These results could be related to the fact that hormones induce PD1 expression in various immune cells, including macrophages, dendritic cells, and B lymphocytes [67]. With this in mind, it may be plausible that our results are masked by immune cells in the TME.
Regarding CD274 (PDL1), several studies of both mRNA and protein have shown an association with HER2-enriched and TN/BL phenotypes [23, 68–72]. However, few studies correlate PDCD1LG2 (PDL2) expression with BC molecular subtypes. Baptista et al. [70] analyzed PDL2 protein expression in 192 cases. They found a significant association in tumors without ER and HER2 overexpression, similar to the present study. However, the association did not reach statistical significance. In our previous in vitro studies, the correlation between intrinsic PDCD1LG2 (PDL2) expression and BC phenotypes is unclear [23].
Previous molecular studies have shown that CD276 (B7-H3) is overexpressed in neoplastic tissue compared to healthy tissue and, to a lesser extent, in stromal cells [73]. The same results have been described by immunohistochemistry, finding expression in the membrane and cytoplasm of neoplastic cells [73–76]. Furthermore, this IC is also expressed in circulating neoplastic cells in the peripheral blood of BC patients, irrespective of disease stage [77]. Although consistent with our data, other investigators have found a correlation between HR expression and the absence of HER2. However, the association was not significant, probably because of the limited number of analyzed cases [73]. We have recently reported an in vitro intrinsic expression of CD276 (B7-H3) in luminal cell lines [23]. However, contrary to our results, Kim et al. [78] have described in a clinical series of 198 tumors a significant overexpression of B7-H3 in the TN/BL phenotype. These discrepancies suggest that mRNA and protein levels are probably not equivalent in the different phenotypes. Therefore, further studies are needed to better define the role of this gene among BC subtypes by expanding the number of cases in each group.
In the case of JAK2, our results agree with the recently published study by Liu et al. [56], which describes under-expression in BC relative to healthy tissue and other neoplasms. As in our clinical series, the published data are associated with luminal B/HER2-negative tumors [56, 79]. These results are congruent, as JAK2 signaling through the prolactin receptor (PLR) has been associated with an increased risk of developing luminal tumors [80–82]. Indeed, data in the literature suggest a relevant role of the PLR/JAK2/STAT5 pathway in the pathogenesis of some neoplasms in murine models [83]. In addition, it has also been observed that JAK2 is involved in the mechanisms of resistance to tamoxifen, a selective ER modulator commonly used in the treatment of luminal phenotypes [36]. All these data suggest that JAK2 could be essential in developing luminal tumors and acquiring resistance to hormone treatment.
Finally, FoXO1 expression in BC is quite variable, although it generally decreases compared to healthy breast tissue [57]. There are some contradictions as to its association with the different subtypes. According to our data, some studies show a significant association between HR expression and the absence of HER2 [42, 84], but there are contrary results [57]. Furthermore, it has been shown that FoXO1 is involved in HRhormone receptor signaling in neoplastic cells, depending on the cellular context. Therefore, it may promote or repress ER and PR functions in BC [85–88]. Recent studies have reported that FoXO1 expression is modulated by the transcription factor E2F1 [89], which the ER and its expression regulate, and its expression has been associated with tamoxifen resistance [90]. These data would support the complex role of FoXO1 in the pathogenesis of luminal phenotypes. Our study also evaluated the association between ICs and immunoregulatory pathways gene expression with clinicopathological features and patients’ outcomes. We observed that CTLA-4 was overexpressed in tumors with intermediate-high proliferation index, a high percentage of sTIL, and the absence of HR. These features are specifically consistent with the TN/BL phenotype.
Similarly, PDCD1 (PD1) overexpression was associated with tumors with intermediate proliferation and a high sTIL content. As for its ligands, CD274 (PDL1) and PDCD1LG2 (PDL2) were associated with a high percentage of sTIL and intermediate proliferation, respectively. All of the above supports that the expression of these genes is associated with unfavorable clinicopathological features in BC and that the expression of CTLA-4 and PDCD1 (PD1) receptors could be regulated by the presence of immune cells in the TME.
One of the first studies regarding CTLA-4 expression in BC described that high mRNA levels were associated with axillary metastases and more advanced clinical stages [59]. Similarly, a recent immunohistochemical study in a large clinical series (n = 229) demonstrated the association between CTLA-4 with lymph node involvement, histological grade 3, larger tumor size, and high proliferation index [91]. However, we could not confirm these results in our clinical series, except for the increase in Ki-67 expression. Nevertheless, there is some controversy in the relationship between the expression of this receptor and the proliferation of BC cells, as reported by Navarrete-Bernal et al. [61]. On the other hand, in recent literature, we found studies with similar results to ours, showing overexpression of CTLA-4 in HR-negative tumors [92] and the presence of TILs [93], mainly intratumoral Treg lymphocytes [94, 95]. The latter agrees that CTLA-4 expression is directly related to genes involved in leukocyte differentiation processes and regulation of T-lymphocyte activity [92]. Despite this, Catacchio et al. [96] found no correlation between CTLA-4 levels and the presence of TILs (stromal vs. intratumoral), reflecting the heterogeneity of this neoplasm.
Few studies have also evaluated the expression of PDCD1 (PD1) in BC. One of these associated the expression of this IC with a lower clinical stage [63]; another barely detected the expression of this gene in neoplastic cells, thus finding no relationship with the clinicopathological characteristics analyzed [97], while the most recent one has only described an association with tumors of histological grade 2–3 [98]. In contrast, other researchers have demonstrated high mRNA expression in the peripheral blood of these patients [62, 66], reflecting an active systemic anti-neoplastic response. In this regard, most studies have focused on analyzing the expression of this receptor in TILs, showing an association with larger tumors, high grade, lymph node involvement, absence of ER, and elevated Ki67 expression [65]. Our results align with those recently described by Ren et al. [97], who observed a direct relationship between increased proliferation, the presence of TILs, and the absence of distant metastasis. The relationship between PD1 expression and a more significant number of TILs present in the TME, also described by other authors [96, 99, 100], was expected, as this gene is involved in the mechanisms of lymphocyte activation, mainly of T lymphocytes, and leukocyte migration [100]. Indeed, several investigators have demonstrated the association of PD1 with intratumoral and stromal CD8 + T cells, neutrophils, and fibroblasts [96, 100].
Regarding CD274 (PDL1), gene expression results in the literature are scarce and contradictory. Uhercik et al. [63] did not observe any association between mRNA levels and the clinicopathological features studied in a clinical series of 127 patients. On the other hand, and in agreement with our results, Schalper et al. [101] only found an association with a high percentage of TILs. In contrast, Sabatier et al. [68] detected overexpression in larger tumors, histological grade 3, HR-negative, and high Ki-67 expression. At the protein level, PDL1 expression in the tumor has been evaluated extensively in several clinical series, finding an association with younger patients, larger size, and high Ki-67 expression. In addition, several researchers have described the association with a high percentage of TILs [69, 71, 96, 102, 103]. However, there are some contradictions between studies regarding the association with histological grade, lymph node involvement, necrosis, and HR expression [69–72, 104]. These discrepancies may be due to subjective interpretation of the results or the use of different cut-off points established for statistical analyses, although technical problems cannot be excluded. It is also important to note that the anti-PDL1 antibody used in these studies (E1L3N - Cell Signaling Technology) is not FDA-approved for use in BC.
Currently, the only anti-PDL1 antibodies approved by the FDA as companion tests in metastatic TN/BL BC are SP142 (Ventana/Roche) and 22C3 (Dako/Agilent), which select patients as candidates for immunotherapy. These antibodies have been developed independently for different drugs (Atezolizumab and Pembrolizumab, respectively) and are based on different clones, protocols, staining platforms, and scoring algorithms. In addition, they have different cut-off points when interpreting the results. It should also be noted that each assay has a different sensitivity, depending on whether the test is aimed at detecting tumor cells (22C3) or immune cells (SP142), and that each clone recognizes a specific PDL1 epitope, thus presenting specific staining patterns. Different detection methods (platform and immunohistochemical reagents) may also influence the identification of PDL1-positive cells. Therefore, considering all this, there are questions about whether the tests can be used interchangeably when making treatment decisions for PD1 and PDL1 inhibitors [105], making quantifying gene expression a potential alternative. This requires the design of a more significant number of studies and clinical trials to support the evaluation of PDL1 with this molecular technology.
On the other hand, PDCD1LG2 (PDL2) has been analyzed very little in BC compared to CD274 (PDL1). We have only found one study in which the protein expression of this ligand was determined by immunohistochemistry in 191 tumors. These authors demonstrate an association between PDL2 and younger patients (< 50 years) with positive lymph node status and ER-negative tumors [70]. Unfortunately, we have not been able to confirm these findings in our clinical series. However, the methodology applied differs, so the results are not comparable.
Our results also showed that CD276 (B7-H3) overexpression was associated with tumors with low cell proliferation and sTIL content. These features align with the association described in our clinical series with luminal phenotypes. Similarly, Kim et al. [78] demonstrated an association with a lower density of sTIL, mainly CD8 + T lymphocytes. However, another study, including 82 patients, found that high gene expression was associated with larger tumor size, lymphovascular invasion, lymph node involvement, and more advanced clinical stages [73]. At the protein level, there are also discrepancies between studies regarding the association with different clinicopathological characteristics. Some authors describe the association of B7-H3 with axillary metastases and more advanced stages [106]; others observe only a correlation with smaller tumor size and low density of microvascularisation [74]; and finally, others do not find any significant association [75]. These discrepancies in the results could be explained by the existence of two isoforms of the protein, the possibility that these undergo post-translational modifications, and the ability to bind to different receptors, which, to date, remain unknown [107, 108]. Therefore, considering the results described above and all the available information, it could be said that the role of B7-H3 in BC may vary depending on the pathophysiological context. However, further studies are needed to test these hypotheses.
In the current series, we also detected JAK2 overexpression in tumors of the luminal phenotype and those with intermediate proliferation. Little information in the literature regarding its association with clinicopathological features in BC exists. Liu et al. [56] describe that high mRNA levels correlate with favorable characteristics such as age < 50, smaller size, early clinical stage, and PR expression. On the other hand, immunohistochemistry has confirmed its association with a higher density of TILs, mainly T and B lymphocytes, macrophages, neutrophils, and dendritic cells [109, 110]. Since our data have been only partially corroborated and there is limited published data, further studies are needed.
Finally, we observed FoXO1 overexpression in smaller tumors with histological grade 2, no lymphovascular invasion, a low-to-intermediate proliferative index, and luminal phenotype, all good prognostic features. Other authors have also described the association with favorable clinicopathological features, such as smaller tumor size, negative nodes, and hormone receptor and HER2-negative expression [42, 84]. In contrast, other investigators find an association only with negative PR, HER2 expression, and elevated Ki-67 percentage [57]. Due to the scarce data and the existing discordance, it is necessary to continue studying this transcription factor in BC better to define its role in the pathogenesis of luminal tumors.
Regarding patients' outcomes, we found none of our genes of interest were significantly associated with OS. We only observed that patients with tumors overexpressing FoXO1 had higher OS, but only as a trend, in line with its correlation with more favorable clinicopathological features. However, several studies have shown that FoXO1 expression may be modulating progression and metastasis in BC [111, 112], although this has not been confirmed by Lallemand et al. [57], who show that this gene does not appear to be influencing metastasis-free survival in BC patients.
Concerning SLE, we observed that patients developed fewer recurrences when their tumors overexpressed CTLA-4. Furthermore, we showed that it could act as an independent prognostic factor in these patients. Our results support a recent similar molecular study [55]. However, opposite results have been published [91, 113, 114]. These discrepancies can be attributed, at least partially, to the different methodology used (mRNA vs. protein expression) or the interpretation of the data. Specifically, the work by Yu et al. [113] evaluated CTLA-4 protein expression in tumor cells and TILs. They showed a correlation between high CTLA-4 expression in TILs and a better prognosis, but unfavorable if the positivity was in tumor cells. Therefore, these data suggest that our results could be influenced by the presence of CTLA-4-positive TILs in TME.
Analysis of OS and DFS depending on CTLA-4 expression levels and stratification by phenotype also showed an association with good prognosis in patients with HER2-enriched tumors. However, as discussed above, the data described in the literature are contradictory. On the one hand, overexpression of this receptor, both at mRNA and protein levels, has been associated with poorer OS and DFS [113, 115]. On the other hand, it has been associated with improved survival, especially in TN/BL tumors [55, 92]. Again, these data suggest a relevant role in the presence of CTLA-4 + TILs in this tumor subtype. However, these results should be corroborated in further independent studies.
Studies in animal models have shown that anti-CTLA-4 treatment promotes T-lymphocyte activation and neoplastic cell killing [116, 117]. Based on this, the use of two humanized monoclonal antibodies targeting CTLA-4 (ipilimumab and tremelimumab) has been investigated in different clinical trials. A phase I trial [118] evaluated the anti-neoplastic activity of tremelimumab in combination with exemestane (aromatase inhibitor) in 26 patients with advanced luminal BC. The results showed that 42% (11/26) had stable disease for at least 12 weeks. They also observed increased levels of CD4 + and CD8 + T-lymphocytes and decreased presence of Treg lymphocytes at the peripheral level [118]. Similar results were seen in a pilot study combining ipilimumab with cryoablation in 19 patients with early BC [119]. On the other hand, the NCT02536794 phase II clinical trial [120] determined the efficacy of the combined treatment of durvalumab (anti-PDL1) and tremelimumab in 18 patients with luminal (n = 11) or TN/BL (n = 7) metastatic BC, where preliminary results showed clinical benefit in 71% of patients with TN/BL tumors. However, this study did not proceed, as the objective response rate (ORR) did not meet the required criteria. Several clinical trials are currently underway in patients with TN/BL tumors studying the combination of nivolumab (anti-PD1) and ipilimumab with cryoablation (NCT03546686; phase II), radiotherapy (NCT03818685; phase II) or oncolytic viruses (NCT04185311; phase I), as well as durvalumab and tremelimumab alone (NCT02658214; phase I / NCT03982173; phase II) or in combination with nab-paclitaxel and neoantigen vaccines (NCT03606967; phase II). However, their preliminary results are not yet published.