Background

The use of radiolabeled peptides to target tumor cells overexpressing peptide receptors for imaging and therapy is a promising and rapidly expanding approach. The use of the radiolabeled somatostatin analog 68Ga-DOTATATE and 177Lu- DOTATATE, respectively, for imaging and treatment of neuroendocrine tumors, is a well-known example, but many other (neuro)peptide receptors are overexpressed in diverse tumors or subsets of tumors and support the development of this personalized medicine [1]. The neurotensin receptor-1 (NTS1) is a G-protein-coupled receptor under investigation in several cancers. In cell lines, neurotensin (NT) exerts numerous oncogenic effects involved in tumor growth and metastatic spread [2,3,4]. Also, overexpression of NTS1 has been documented in many cancers and has been associated with higher grade tumors and a poorer outcome [2, 5,6,7]. In breast cancer, the work of Sandra Dupouy and co-workers gave initial insight on the potential role of the NT/NTS1-axis [8]. High NTS1 expression in invasive ductal carcinoma (IDC) was associated with larger tumor size and Scarff-Bloom-Richardson (SBR) grade 3. Also, breast cancer patients with high NTS1 expression had a significantly lower ten-year survival rate [8]. The same team focused on the association between NT, NTS1, and human epidermal growth factor receptors (HERs). Under NT stimulation, EGFR, HER2, and HER3 are up-regulated. NTS1-positive samples significantly overexpressed HER3, while no association was found between NTS1 and HER2 [9]. All together, these data support a role of NT/NTS1 to breast cancer progression. Identification of the subsets of breast cancer patients in whom NTS1 is likely to be expressed can lead the way to clinical applications of radiolabeled neurotensin analogs for imaging and therapy. In this work, we examined the expression of NTS1 within the fully-annotated Tissue MicroArray (TMA) of primary breast tumors from Institut Bergonié and studied association with clinical, pathological, and biological parameters and outcomes.

Materials and methods

The check-list of the reporting Recommendations for Tumor Marker Prognostic Study (REMARK) [10] is provided in supplemental file 1. Tissue MicroArray (TMA). We investigated the fully-annotated TMA of invasive breast cancers of Institut Bergonié, Bordeaux, France. Details on this TMA have been previously reported [11]. Briefly, samples in quadruplicates from 1419 invasive breast tumors from primary-operated patients between 2000 and 2005 who did not receive neoadjuvant chemotherapy or hormone therapy were available to study NTS1 expression.

Validation of the primary antibody raised against NTS1

As for IHC studies of other G-protein-coupled receptors, validation of the NTS1 antibody is of primary importance [12]. We followed the methodology we previously described for gastrin-releasing peptide receptor (GRP-R) [11]. Briefly, these steps included western blotting of breast cancer cell line lysates, immunochemistry of these cell lines fixed and embedded with the same material than that used for the construction of the TMA, identification of positive and negative tumors within a dedicated TMA, and validation of the signal recorded in IHC by western blotting of corresponding frozen tumors.

Immunochemistry

For IHC staining on TMA, 4-µm-thick paraffin-embedded tissue sections were mounted on slides. The whole IHC procedure was performed using the Ventana Benchmark-ultra automated staining system and according to the manufacturer’s recommendations. Antibody was diluted within a DAKO diluent (S2022) and unmasking was performed using a CC1 protocol at pH8. Antibody was incubated during 52 min at 1:50. Signal was detected using ultraVIEW Universal DAB (760–500). The primary antibody was a mouse monoclonal IgM kappa light chain (ref sc-376958, Santa Cruz Biotechnology, clone B12) raised against an epitope mapping between amino acids 389–417 at the C terminus of NTS1 of human origin, also used in other recent studies [13]. NTS1 expression using IHC was assessed by light microscopy by a senior pathologist blinded to the study endpoint who scored the location, the staining, its intensity, and the percentage of tumor cells stained. The mean number of tumoral cells stained for NTS1 for each TMA (in quadruplicates) core was reported.

Statistical analysis

For association with clinical, pathological, and biological parameters, NTS1 expression was dichotomized according to a cut-off of 10% stained cells, that has been previously suggested [2, 14, 15]. We studied associations between NTS1 expression and various clinical and pathological parameters. Differences between categorized variables were assessed with the χ2 test or Fischer exact test. All P values were two-sided and a p value of less than 0.05 was considered statistically significant. Factors that are predictive of high NTS1 expression were investigated by univariate and multivariate analyses using a logistic regression. Odds ratios and their 95% confidence intervals (95%CI) were reported. The median follow-up was calculated by the reverse Kaplan–Meier method. Probabilities of overall survival (OS) and distant metastasis-free interval (DMFI) were calculated with the Kaplan–Meier method. The logrank test was used for comparison between NTS1 groups. For OS, all causes of death were considered as events. For DMFI, all distant metastatic events were considered and patients with no metastases were censored at the date of last news. Multivariate analyses for DMFI were calculated with a Cox model. Variables significant at the 0.05 level in the univariate analysis were introduced in the multivariate model, which used a stepwise descending maximum likelihood method. Hazard ratio (HR) were calculated with a Cox model; proportionality was tested with the residuals test. For lymph node status, patients who did not have lymph node dissection (Nx) were kept in all analyses as a separate category. Statistical analyses were performed using the SAS software (version 9.3; SAS Institute, Inc.)

Lymph node metastases

In order to avoid bias due to unavailability of tissue for analysis, only samples from patients with 4 invaded nodes or more were included in this study. Thus, 14 lymph node metastases (one lymph node per patient), whose primary breast tumors were positive for NTS1, were considered for IHC in order to assess whether NTS1 expression is maintained in metastatic tissue. Three cases with nuclear staining were not analyzable (2 cases did not present enough biological material). Therefore analysis of NTS1 expression using IHC was run on 11 lymph node metastases samples from 11 primary tumors with high cytoplasmic NTS1 expression.

Results

Validation of the NTS1 primary antibody

The study of several human breast cancer cell lines showed a good correlation in NTS1 expression between western blot and immunocytochemistry assessment. Labeling was cytoplasmic and granular on both, immunofluorescence and immunochemistry (not shown). Moreover, the study of 26 tumor samples Holland-Bouin fixed and paraffin-embedded showed that the monoclonal antibody can significantly discriminate tumors that are positive for NTS1 from negative ones (p = 0.017), using as reference standard western blot analysis on frozen samples from the same tumors. All results are summarized in Fig. 1.

Fig. 1
figure 1

Validation of the NTS1 antibody. A Representative NTS1 band at the expected molecular weight of 55 kDa on western blot of frozen human tumors. B Representative actin band on western blot of frozen tumors. C The new monoclonal antibody is able to discriminate NTS1-positive tumors and NTS1-negative tumors. The IHC cut-off was ≥ 10% of cells stained

Full size image

NTS1 expression in 1419 breast cancer samples, its cellular location, and association with pathological and biological parameters

A moderate or high NTS1 expression (i.e.,  > 10% of tumor cells stained) was present in 32.4% of all breast cancer (459/1419) and exhibited a granular staining with either cytoplasmic location (n = 304; 21.5% of all tumors) or nuclear location (n = 155; 11.0% of all tumors) in cancerous cells (Fig. 2A, B).

Fig. 2
figure 2

A Representative NTS1 immunohistochemistry of a primary breast tumor sample from the tissue microarray with numerous cancerous cells exhibiting a cytoplasmic staining of NTS1 (magnification 40X). B Representative NTS1 immunohistochemistry of a primary breast tumor sample from the tissue microarray with numerous cancerous cells exhibiting a nuclear staining of NTS1 (magnification 40X)

Full size image

Importantly, the nuclear vs cytoplasmic distribution appeared mutually exclusive. Only one case out of 459 NTS1-positive samples had both cytoplasmic and nuclear expression of NTS1. In the tumor-associated stromal tissue, mast cells were usually strongly positives for NTS1 [16]. Other cells of the tumor microenvironment showed no expression of NTS1.

In view of the dichotomy between cytoplasmic and nuclear location and because cytoplasmic location might be more easily amenable to targeting with radiopharmaceuticals, we carried out separate analysis of factors associated with NTS1 presence in breast cancer first at cytoplasmic level and then for nuclear localization.

Association between NTS1 positivity at cytoplasmic level and pathological and biological parameters in 1419 breast tumors

Cytoplasmic NTS1 staining was found in 304 tumors (21.5% of all tumors). In univariate analysis, factors associated with cytoplasmic NTS1 expression in tumors were a higher pT stage, a higher SBR grade, a node-positive status, a Ki67 score ≥ 20%, PR negativity, and positive HER2 status (Table 1). Age, histological subtype, and ER positivity level showed no significant association. The Low-Positive ER category in our TMA was defined prospectively as cancers with 1% to 9% of cells staining for ER; thus slightly different from the recent ASCO recommendation that put the upper limit for Low Positive as 10% [17]. In multivariate analysis, higher E&E grade, Ki67score ≥ 20%, and a higher pT stage remained significant predictors. On the other hand, breast cancer phenotype is strongly associated with cytoplasmic NTS1 expression (Table 1) but could not be introduced in the multivariate analysis because HER2, PR, and ER are part of the definition of the phenotypes. The luminal-A phenotype showed a lower probability of cytoplasmic NTS1 expression compared to the other phenotypes (17.3% for luminal A versus 32.9% for all other phenotypes combined; p < 0.0001).

Table 1 Univariate and multivariate analyses of predictors of high cytoplasmic NTS1 expression in 1419 primary breast cancers
Full size table

Association between NTS1 positivity at nuclear level and pathological and biological parameters in 1419 breast tumors

Nuclear NTS1 staining was found in 155 tumors (11.0% of all tumors). In univariate analysis, factors associated with nuclear NTS1 expression were a lower pT stage, a lower E&E grade, a Ki67 score < 20%, a higher ER expression (≥ 10% vs. < 10% [combining negative and Low-Positive (1%–9%) categories], PR positivity, and absence of HER2 overexpression (Table 2). Age, histological subtype, and lymph node status showed no significant association. In multivariate analysis, higher ER level, lower pT stage, and lower E&E grade remained significant predictors. Breast cancer phenotype was strongly associated with nuclear NTS1 expression, with luminal-A phenotype showing the highest positivity rate, followed by luminal-B subtypes, while none of the ER-negative tumors expressed NTS1 at nuclear level (p < 0.0001) (Table 2).

Table 2 Univariate and multivariate analyses of predictors of high nuclear NTS1 expression in 1419 primary breast cancers
Full size table

Correlates of nuclear versus cytoplasmic location of NTS1

After exclusion of one case with mixed location, we performed an analysis on the 459 patients with NTS1-positive tumors in order to elucidate factors associated with nuclear (n = 155) versus cytoplasmic (n = 304) location of NTS1 (Table 3). There was no significant difference in age distribution or histological subtypes (invasive ductal/lobular). The subgroup of patients with nuclear NTS1 had, as compared to the subgroup with cytoplasmic NTS1: a lower rate of tumors > 2 cm (9.6% vs. 31.6%; p < 0.001), a lower rate of E&E grade III tumors (10.3% vs. 38.5%; p < 0.001), less tumors with Ki67 ≥ 20% (10.3% vs. 34.2%; p < 0.001), less patients with pathological-positive lymph nodes (27.7% vs. 39.5%; p = 0.019), less ER-negative tumors (0% vs. 11.8%; p < 0.00002), less PR-negative tumors (7.1% vs. 23.4%; p < 0.001), and less tumors with a HER2-positive status (3.9% vs. 14.8%; p < 0.001). As a result, when considering tumor phenotypes, the subgroup of patients with nuclear NTS1, as opposed to the subgroup with cytoplasmic NTS1, had higher rates of luminal-A tumors (83.2% vs. 53.9%), lower rates of luminal-B tumors regardless of HER2 status (16.8% vs. 34.2%) and no ER-negative subtypes (0% vs. 11.8%) (Table 3). Most strikingly, all tumors with nuclear NTS1 expression were ER-positive, with only 1/155 being ER Low Positive.

Table 3 Factors associated with nuclear NTS1 location versus cytoplasmic NTS1 location in 459 NTS1-positive breast cancers
Full size table

Association between cytoplasmic NTS1 expression and outcomes and between nuclear NTS1 expression and outcomes

Median follow-up was 122 months. Figure 2 shows Kaplan–Meier curves for DMFI and OS. When considering cytoplasmic NTS1 expression, there was no difference in OS (Logrank P = 0.2018) and only a trend toward lower DMFI (Logrank P = 0.053) for patients with cytoplasmic NTS1 expression compared to those without cytoplasmic NTS1 staining. Ten-year DMFI were 85.7% (95%CI: 80.8%–89.4%) and 89.5% (95%CI: 87.3%–91.3%), respectively. Considering nuclear NTS1, 10-year DMFI were 92.8% (95%CI: 86.9%–96.1%) for patients with nuclear NTS1 and 88.2% (95%CI: 86.1%–90.0%) for patients without nuclear NTS1 expression (Logrank P = 0.1145). There was no difference in OS (Logrank P = 0.6175).

Figure 3 shows DMFI and OS curves restricted to patients whose tumors expressed NTS1 and comparing outcomes of those with cytoplasmic location of NTS1 versus those with nuclear location of NTS1. There were no differences in OS (Logrank P = 0.2748). However a lower DMFI was found for patients with cytoplasmic NTS1 staining compared to those with nuclear NTS1 expression (Logrank P = 0.033). Patients with cytoplasmic expression of NTS1 had a twofold higher metastatic risk compared to patients with nuclear NTS1 staining (HR = 2.03; 95%CI: 1.04–3.94, p = 0.037). Differences in DMFI can be explained, at least in part, by the associations described above between cytoplasmic NTS1 expression and clinical and pathological characteristics. The number of events did not allow for performing a multivariable-adjusted Cox models. Further studies are needed to better discriminate any specific role of NTS1 on prognosis.

Fig. 3
figure 3

Distant metastasis-free interval according to the presence or absence of cytoplasmic A, or nuclear C NTS1 expression in primary breast tumors, or opposing patients with nuclear NTS1 expression to patients with cytoplasmic NTS1 expression (E). Association of cytoplasmic B, nuclear D, or nuclear vs cytoplasmic F NTS1 expression in primary tumors with OS

Full size image

NTS1 expression in lymph node metastases from patients with NTS1-positive breast tumor

Invaded lymph nodes were available for the current study from 14 patients with NTS1-positive primary tumors. In three of these patients there was not enough residual material for IHC analysis. In all other 11 patients, NTS1 expression could be assessed on tumoral lymph node tissue. All 11 primary tumors had cytoplasmic NTS1 expression. In 8 (72.7%) of these cases, lymph node tumoral tissue also expressed NTS1 with cytoplasmic staining of NTS1 as shown in Fig. 4.

Fig.4
figure 4

A Representative NTS1 immunohistochemistry of a metastatic lymph node from a NTS1-expressing primary breast tumor at 5X magnification. B Corresponding HES staining. C NTS1 staining was cytoplasmic in the lymph node metastasis as it was in the primary tumor. Image was taken at 20X magnification

Full size image

Discussion

The overexpression of neuropeptide receptors in breast cancer raises hope for the selective addressing of radiolabeled molecules [1, 18] or drugs using suitable cargos [19]. In this work we aimed to identify breast cancer patients who might benefit from NTS1-based imaging or targeted therapy.

We found NTS1 to be expressed in ~ 32% of breast tumors from patients who received primary surgery. In two-thirds of NTS1-positive cases, NTS1 showed cytoplasmic distribution, while in one-third of cases it was located in the nucleus. Interestingly, cytoplasmic versus nuclear location appeared mutually exclusive, with only one tumor showing mixed staining. Another important finding is that all 156 tumors with nuclear NTS1 staining were ER positive. Therefore, the sub-cellular location of NTS1 is strongly associated with estrogen receptor expression. Again, 92.9% of tumors with a nuclear expression of NTS1 expressed PR (Table 3). To further understand factors driving the sub-cellular location of NTS1, we performed an analysis on the 459 patients with NTS1-positive tumors. When compared to the subgroup with cytoplasmic NTS1 (n = 304), the subgroup of patients with nuclear NTS1 (n = 155) had lower rates of tumors > 2 cm, E&E grade III tumors, tumors with Ki-67 ≥ 20%, as well as less ER-negative tumors, PR-negative tumors ,and HER2-overexpressing tumors (Table 3).

It has been previously raised that neurotensin is expressed and up-regulated by estrogen in normal epithelial breast cells [8]. Also, women receiving high doses of steroids have increased immunoreactivity of neurotensin and NTS1 in human uterine myometrium and uterine leiomyomas [20]. On the other hand, neurotensin regulates the expression of NTS1 in different cell compartments [20,21,22]. Under activation by neurotensin, NTS1 is internalized to the perinuclear region [21] or addressed to the nuclear envelope and the nucleus soma [22] due to the strong nuclear-targeting signal contained in the NTS1 receptor [23]. The NTS2 receptor subtype may also participate in the intracellular processing of NTS1. It has been suggested that heterodimerization of NTS1 with NTS2 may cause the retention of NTS1 in sub-cellular compartments and interfere with its proper recruitment to the plasma membrane [24]. These mechanisms of autocrine/paracrine activating loops deserve further investigations but may be tumor specific as, we and others, did not find nuclear localization of NTS1 in primary prostate tumors [15, 25], whereas a nuclear localization of NTS1 was found in gastro-intestinal stromal tumors [26]. Nuclear localization of other G-protein-coupled receptors has been described [27]. Whether the nuclear trafficking of NTS1 receptor would allow transporting NTS1-targeting agents to the nucleus deserves investigation [28,29,30].

Cytoplasmic NTS1 location facilitates targeting with radiopharmaceuticals. It is of interest that cytoplasmic overexpression of NTS1 was found more frequently in tumors with poorer prognostic factors (E&E grade II/III, Ki67 ≥ 20%, HER2 positive status, ER negativity, and PR negativity). Similar results regarding cytoplasmic expression of NTS1 in breast cancer were found in earlier studies [8, 9]. As regarding breast cancer phenotypes, our study further shows that the probability of cytoplasmic NTS1 expression is higher in higher-risk phenotypes. Cytoplasmic expression of NTS1 is found in 17.3% in luminal-A tumors and in ~ 30% of the other phenotypes combined (p < 0.0001). Another important finding for therapeutic purposes is that NTS1 expression is most often preserved in nodal metastases from NTS1-positive primary tumors (8/11: 72.7%) (Fig. 3), but this observation requires confirmation, and studies of NTS1 expression in distant metastases are needed. Our results support recent findings on the genomic data of NTS1 mRNA in primary breast tumors with axillary lymph node metastasis [31]. Therefore, in N+ breast cancer patients with luminal-B phenotype, or HER2-enriched phenotype, or triple-negative phenotype, the development of NTS1-targeted agents would be of interest for imaging and/or therapy of these subsets of patients.

Dupouy and co-workers found that breast cancer patients with cytoplasmic overexpression of NTS1 had a higher relative risk of dying [8]. In our series of mostly low risk (primary operable breast cancer) we found no difference in OS, but a lower DMFI (Logrank P = 0.033) for patients with cytoplasmic NTS1 expression compared to those with nuclear NTS1 staining. 10-year DMFI were 85.7% and 92.8%, respectively.

A pilot study was performed in 6 patients with metastatic pancreatic adenocarcinoma, patients using low doses (1.5 GBq) of the radiolabeled NTS1 antagonist [177Lu]Lu-3BP-227 for imaging, followed by higher therapeutic activities in patients with adequate tumor uptake on imaging [32]. This pilot study shows the current interest in NTS1 targeting and many more NTS1-targeting radiopharmaceuticals for imaging and therapy is being developed [33,34,35]. In breast cancer, NTS1 imaging using Positron Emission Tomography (PET) or Single Photon Emission Computed Tomography (SPECT) would be beneficial in specific phenotypes. In triple-negative breast cancer, identification of NTS1 as a potentially targetable receptor paves the way of an “image and treat” strategy using 68 Ga- and 177Lu-labeled radiopharmaceuticals. Interestingly, Dupouy and colleague, reported that neurotensin/NTS1 promotes activation of EGFR, HER2, and HER3 rendering breast tumors aggressive yet highly responsive to lapatinib and metformin in mice [9] and NTS1 imaging might also be interesting in this regard. Finally, exploring tumor heterogeneity by both [18F]-FDG PET [36] and NTS1-PET might offer complementary staging and prognostic information and would be beneficial to better predict response to targeted radionuclide therapy as today performed in neuroendocrine tumors (using [68Ga]Ga-DOTATATE/[18F]-FDG) [37] and prostate cancer (using [68Ga]Ga-PSMA/[18F]-FDG) [38].

Our study has limitations inherent to the use of a TMA. We could only assess the NTS1 expression in the peritumoral environment of the TMA cores which may reflect only partially the expression of NTS1 by normal breast cells. Moreover, the low number of samples is too small to draw solid conclusion as compared to the NTS1 expression in related metastatic lymph nodes. Finally, future studies of NTS1 expression in distant metastases will also be needed to consolidate the “theranostic” potential of this new target in breast cancer.

As perspectives and before considering future NTS1-targeted therapeutic applications, assessing the effect of previous treatments (hormone therapy or chemotherapy) on NTS1 expression is a necessary step. This work is scheduled in our consortium.

Conclusion

Our data from this large tissue bank documents the overexpression of NTS1 in about one-third of primary breast tumors with a dichotomic intracellular distribution (cytoplasmic vs. nuclear). Cytoplasmic overexpression of NTS1 was found in 21.5% of breast tumors from patients undergoing primary surgery, with higher frequency in aggressive subtypes. These findings encourage the development of radiopharmaceuticals for imaging and therapy. On the other hand, nuclear NTS1 location, found in 11% of patient samples, and its strong association with ER positivity deserve investigation of the mechanisms involved. We hope that the expression profile we described in this study would stimulate new developments for imaging and therapy of breast cancer patients using NTS1-targeted agents.