Effects of tumor-specific CAP1 expression and body constitution on clinical outcomes in patients with early breast cancer

The MDCS enrolled participants living in Malmö, Sweden, between 1991 and 1996 with the objective to explore associations between dietary habits and subsequent cancer risk. This prospective population-based cohort included 17,035 women born 1923–1950, representing 42.6% of the eligible population [26, 27]. Exclusion criteria were limited to Swedish language insufficiency and mental disabilities impairing the respondent’s completion of study questionnaires. At baseline, the participants answered extensive questionnaires, underwent anthropometric measures including height, weight, waist and hip circumference, and bioelectrical impedance analysis of body fat percentage (BF%) obtained by trained study nurses, and blood samples were collected. Of the 17,035 study participants, 576 had a prevalent breast cancer diagnosis prior to baseline examination and were thus excluded. Information on incident breast cancer cases and vital status has been retrieved annually through record linkage to the Swedish Cancer Registry, the Southern Swedish Regional Tumor Registry, and the Swedish Cause of Death Registry [27]. Ethical approval was obtained from the Ethical Committee at Lund University (Dnr 427/2007) and all study participants signed informed consent at enrollment.

In total, 1016 women were diagnosed with primary breast cancer prior to January 1, 2011. Patients with carcinoma in situ only (n = 68), bilateral cancers (n = 17), distant metastasis at diagnosis (n = 14), neoadjuvant treatment (n = 4), breast cancer-related death within 0.3 years from diagnosis (n = 2), and one patient who declined treatment for 4 years prior to accepting surgery (n = 1) were excluded from the study population. The remaining study population consisted of 910 patients with incident invasive breast cancer of whom 718 had tumor tissue available in a prepared tissue microarray (TMA). A flowchart of the study population is shown in Fig. 1.

Information on tumor characteristics and breast cancer treatments were retrieved from medical records. For patients diagnosed prior to 2005, histological tumor type and grade were re-evaluated according to the WHO and Nottingham classifications [28] by one senior breast pathologist [29].

Information on tumor markers were retrieved from medical records and from immunohistochemistry (IHC) assessments of tumor tissue microarray (TMA) at the Center for Molecular Pathology, Malmö University Hospital, Malmö, Sweden, as previously described [29, 30]. In brief, estrogen receptor (ER) and progesterone receptor (PR) status were obtained from TMA data (1991–2004) and from medical records (2005 onward). Human epidermal growth factor receptor 2 (HER2) status was retrieved from TMA data, medical records, and patient registers (1991–2004) and medical records and patient registers (2005 onward). Ki67 proliferation index was obtained from TMA data (1991–2007) and from medical records (2008 onward).

In accordance with the Swedish clinical guidelines, ER and PR status were considered positive if > 10% of cancer cell nuclei were stained. Ki67 positivity was categorized into three groups (low, intermediate and high) based on tertile distribution within three assessment periods: 1991–2004, 2005–2007, and 2008–2014 [31]. HER2 was primary classified by in situ hybridization (ISH) and secondly by IHC. HER2 was considered positive if ISH showed HER2 amplification or if IHC was graded 3+. When ISH was negative for HER2 amplification or IHC was annotated 1+ or less, the tumor was regarded as HER2-negative. If no ISH data existed and IHC was graded 2+, HER2 status was regarded as missing.

Surrogate molecular subtypes were created based on ER, PR, and HER2 receptor status along with Ki67 positivity and histologic grade [31, 32]. Tumors were classified into luminal A-like [ER+ and HER2− tumors with (a) histologic grade I or (b) histologic grade II and low Ki67 or (c) histologic grade II, intermediate Ki67 and PR+], luminal B-like [ER+ and HER2− tumors with (a) histologic grade III or (b) histologic grade II and high Ki67 or (c) histologic grade II, intermediate Ki67 and PR−], HER2-positive [all HER2+ tumors], or triple-negative [all ER−, PR−, HER2− tumors].

Antibody validation for IHC assessment of CAP1 was performed using established breast cancer cell models. ER-positive T47D and ER-negative MDA-MB-231 human breast cancer cell lines were purchased and validated by ATCC-LGC Standards. The cells were cultivated in Dulbecco’s modified Eagle’s medium (DMEM; with GlutaMAX™ and HEPES) supplemented with antibiotics (100 U/ml penicillin and 100 μg/ml streptomycin) and 10% fetal bovine serum. The cells were grown in a humidified 5% CO₂ atmosphere at 37 °C, routinely passaged once a week.

Knockdown of CAP1 in breast cancer cells was obtained using small interfering RNA (siRNA) with reverse transfection. Three different siRNA constructs (Silencer® Select s20547, s20548, s20579, ThermoFisher Scientific) were tested alone or in combination for optimal target knockdown. Briefly, the final siRNA transfection was performed as follows: 0.55 × 10⁶ of cells in 1.25 ml of antibiotics-free cell culture medium were added to a mixture of 25 nM CAP1 siRNA (T47D: s20547 and s20549; MDA-MB-231: s20549) and 10 μl Lipofectamine 2000 in 1.25 ml OptiMEM in 6-well plates. The Silencer® Select Negative Control No1 siRNA (ThermoFisher Scientific) was used as a non-targeting control. Following 72 h incubation, the cells were washed and collected for further analyses.

Total RNA was extracted using QIAGEN RNeasy (Qiagen, Mississauga, ON, Canada) according to the manufacturer’s instructions and quantified via Qubit Fluorometric system (Thermo Scientific, Waltham, MA, USA). cDNA was synthesized from 1 μg of total RNA using High Capacity cDNA Reverse Transcription kit (Thermo Scientific, Waltham, MA, USA). Quantitative reverse transcription PCR (qRT-PCR) was performed using TaqMan QuantiTect Probe kit (Qiagen, Mississauga, ON, Canada) with primers specific for CAP1 (Hs02860542_g1 ThermoFisher Scientific, Waltham, MA, USA). GAPDH (Hs99999905_m1 ThermoFisher Scientific, Waltham, MA, USA) was used as reference gene. All transcripts were measured in minimum duplicates and normalized to GAPDH. Relative CAP1 mRNA expression levels in CAP1 silenced cells compared with control were determined by 2−ΔΔCt method in three independent experiments (Additional file 1A).

Breast cancer cell lysates were prepared from CAP1 knockdown or control cells, and proteins were extracted using radioimmunoprecipitation assay buffer [RIPA; 10 mM Tris-HCl pH 7.4, 50 mM NaCl, 5 mM EDTA, 30 mM sodium pyrophosphate, 50 mM sodium fluoride, 100 μM sodium orthovanadate, 1% Triton X-100] supplemented with protease and phosphatase inhibitors. Protein quantifications were performed using Pierce BCA Protein Assay Kit, according to the manufacturer’s instructions. Proteins were separated by pre-cast SDS-PAGE (NuPAGE 10% Bis-Tris, Invitrogen) and transferred to nitrocellulose membrane. The membrane was blocked with 5% (w/v) non-fat dry milk in Tris-buffered saline with Tween-20 and incubated overnight at 4 °C with primary antibodies to CAP1 (Abcam; ab133655, 1:10000) or GAPDH (Merck; MAB374, 1:1000). The blots were subsequently incubated with horseradish peroxidase-conjugated secondary antibodies (CAP1, 1:2000; GAPDH, 1:10000) for 1 h and proteins visualized using SuperSignal West Dura Extended Duration Substrate (ThermoFisher Scientific) and LI-COR Biosciences Odyssey Imaging System. Relative protein levels were quantified by densitometry using ImageJ software (NIH) and normalized against the GAPDH housekeeping protein.

Cell pellets of T47D and MDA-MB-231 breast cancer cells collected following siRNA exposure were fixed in 4% formalin overnight, stained with hematoxylin, dehydrated, and paraffin embedded. A cell microarray was constructed with multiple 1.0 mm cores from each cell preparation using a semi-automatic Tissue Arrayer (Pathology Devices, MD, USA). For immunocytochemistry analysis, 3 μm sections were automatically pretreated using a pressure cooker (Histolab Products AB) and stained for CAP1 (Abcam; ab133655, 1:10,000) using Autostainer Plus (Agilent, DK).

Following antibody validation, immunohistochemical staining of tumor-specific CAP1 was performed using a prepared TMA with duplicate 1 mm tissue cores from individual tumors of the 718 patients in the study population with available tumor tissue (Fig. 1). TMA sections (4 μm) were automatically deparaffinized and pretreated for antigen retrieval using a pressure cooker (Histolab Products AB). IHC was performed and stained for CAP1 (Abcam; ab133655 at 1:10,000 for 30 min) using EnVision FLEX, high pH (Agilent K801021-2) and Autostainer Plus (Agilent, Denmark) with subsequent hematoxylin counterstaining (Agilent S2020).

Evaluation of the CAP1 staining intensity was done twice by one observer (MB), blinded to patient information and tumor characteristics, using light microscope Olympus BX53. The overall CAP1 protein positivity rate was > 75% in tumor cells, whereby only differences in CAP1 staining intensity was annotated. Cytoplasmic CAP1 staining intensity was assessed in five categories: negative (−), weak (1+), moderate (2+), strong (3+), and intense (4+), with representative images of staining intensities shown in Fig. 1. Conflicting assessments were low (4% of the cases), and in cases of discrepancy between the readings, a third evaluation was made. For remaining inconsistency, a second observer (AR) was consulted until consensus. In the event of intra- or inter-duplicate tumor core heterogeneity, the highest score was applied for tumors borderline between two scores. Based on the distribution of CAP1 staining intensities and survival estimates in relation to the initial five CAP1 categories (Additional file 2), the CAP1 tumor expression were subsequently compiled into three groups according to the cytoplasmic intensity scores: low [negative (−)/weak (1+)], moderate [moderate (2+)], and high [strong (3+)/intense (4+)].

The difference in distribution by CAP1 expression across patient characteristics or clinicopathological parameters was analyzed by linear-by-linear association chi-square test for trend for the categorical data and Jonckheere-Terpstra for the comparisons of medians. Categorical variables are presented as number (n) and frequency (%) of patients, continuous variables as medians with interquartile range (IQR). In survival analyses, Kaplan-Meier estimates and LogRank-trend test were used to assess the association between CAP1 expression and time-to-event defined as breast cancer-specific survival (BCSS) or overall survival (OS). The follow-up time was defined as date of diagnosis to date of death, emigration, or end of follow-up up until December 31, 2016. Univariable and multivariable Cox regression models were used to calculate crude and adjusted hazard ratios (HRs) with 95% confidence intervals (CI) for association between CAP1 tumor expression and breast cancer outcome. Models were adjusted for potential confounding factors. Model 1 was unadjusted (crude), and model 2 was adjusted for age at diagnosis (continuous), tumor size (≤ 20 mm or > 20 mm) and any lymph node involvement (positive or negative). Model 3 was additionally adjusted for histologic grade (1–2 or 3), Ki67 (low, intermediate or high, HER2 (normal or overexpression), and ER (positive or negative). Student’s t test was applied to test statistical difference between two groups for experimental in vitro data. Statistical analyses were performed using IBM SPSS Statistics for Windows v 25.0 (Armonk, NY: IBM Corp.) for clinical data and GraphPad Prism v 7.03 for experimental data. All tests were two-tailed, and the P value was considered as strength of evidence against the null hypothesis. The study adheres to the REporting recommendations for tumor MARKer prognostic studies (REMARK) guidelines, to ensure methodological quality [33].

To ensure target specificity and functional application validity, a thorough antibody validation was performed prior to IHC staining. The specificity of the antibody was determined by applying a genetic strategy of siRNA-mediated CAP1 gene knockdown. qRT-PCR analyses confirmed an efficient CAP1 knockdown with 6 and 2% detectable mRNA expression in T47D and MDA-MB-231 cells, respectively, compared with the non-silencing controls (all P values < 0.001, Additional file 1A). The specificity and functional application of the monoclonal Abcam CAP1 antibody was further determined by Western immunoblotting and immunocytochemistry of constructed cell microarray. Equivalent to the mRNA results, approximately 90% and > 99% reduction of CAP1 protein expression at the expected molecular size (52 kDa) was detected by the Abcam antibody in the CAP1 silenced T47D and MDA-MB-231 cells, respectively (all P values < 0.001, Additional file 1B). Similar findings were obtained in the immunocytochemistry analyses (Additional file 1C), thereby confirming the validity of the antibody. An independent antibody validation was performed using an affinity isolated polyclonal CAP1 Prestige antibody (Sigma Aldrich; HPA030124), with equivalent results (data not shown).

Among the 718 breast cancer patients with tumor tissue included in the TMA, CAP1 expression was assessable in 669 tumors (Fig. 1), of which 106 tumors (15.8%) displayed a low, 273 (40.8%) moderate, and 290 (43.3%) high CAP1 expression. Baseline patient characteristics according to distribution of CAP1 intensities are presented in Table 1. Patients with tumors of low to moderate CAP1 expression were more likely to be older at baseline (P < 0.001), have a higher BMI (P = 0.009), larger waist (P = 0.008), wider hip (P = 0.014), and higher BF% (P = 0.006), compared to patients with tumors of higher CAP1 expressions. No further associations were found between CAP1 expression and anthropometric measures.

Patients with low to moderate CAP1-expressing tumors were older at breast cancer diagnosis (P < 0.001) compared to patients with tumors of high CAP1 expressions (Table 2). Further, patients with low CAP1-expressing tumors were more likely to present with high proliferative (defined as Ki67 high; P < 0.001) and histological grade III (P < 0.001) tumors and had a higher percentage of ER-negative tumors than patients with tumors of higher CAP1 intensities (P = 0.014). Similarly, low CAP1 expression was more frequent among patients with HER2-positive and triple-negative (P < 0.001) molecular subtypes. Patients with low CAP1 tumor expression tended to be more likely to received adjuvant endocrine treatment compared with patients with moderate to high CAP1 tumor expression (P = 0.002; Additional file 3). Other breast cancer treatment modalities were not related to CAP1 tumor expression.

Patients were followed for up to 25 years, with a median follow-up time of 10.9 years. The median time between baseline examination and date of breast cancer diagnosis were 9.6 years. Of the patients included in the survival analyses, 239 died during follow-up, and 117 of these died from breast cancer-related causes. Three patients had emigrated, and 476 were still alive at end of follow-up.

Patients with low CAP1 tumor expression had worse long-term survival, both in terms of BCSS (LogRank P_trend= 0.002) and OS (LogRank P_trend < 0.001; Fig. 2), compared to the patients with tumors of moderate to high CAP1 expression. When stratified by ER status, the same trend was observed for patients with ER-positive breast cancer (BCSS: LogRank P_trend = 0.020 and OS: LogRank P_trend = 0.002; Fig. 2), whereas a non-linear association was found among patients with ER-negative breast cancer (Fig. 2).

In Kaplan-Meier estimates when patients were stratified on BF%, BMI, waist circumference, or WHR, CAP1 tumor expression was associated with worse BCSS among patients with low adiposity status across all four anthropometric measures, BF% (LogRank P_trend = 0.002; Fig. 3a), BMI (LogRank P_trend = 0.003; Fig. 3b), waist circumference (LogRank P_trend = 0.001; Fig. 3c), and WHR (LogRank P_trend < 0.001; Fig. 3d), compared with patients with higher CAP1 expression. No association between CAP1 expression and survival was observed for patients with the highest adiposity status. Similar results across anthropometric measures were found for the association between CAP1 expression and body fatness regarding OS (Additional file 4).

Univariable Cox analyses demonstrated that low CAP1 tumor expression was a prognostic indicator for inferior BCSS (HR = 0.46, 95% CI 0.28–0.77; Fig. 4) as well as poor OS (HR = 0.54, 95% CI 0.38–0.78; Fig. 4) among all patients. CAP1 remained a marker of poorer BCSS and OS after adjustments for age at diagnosis, tumor size, and any axillary lymph node involvement. However, the association did not remain after adjustments for histological grade, ER status, HER2, and Ki67. Among patients with ER-negative tumors and patients with low adiposity status, low CAP expression remained associated with poor BCSS after adjustments for age at diagnosis, tumor size, and any axillary lymph node involvement (Figs. 2 and 3). Further adjustment for BMI in the multivariable analyses did not significantly alter the results (data not shown).