Abstract
In preclinical studies, celecoxib has been associated with reduced risk of breast cancer. In this study, the aim was to assess the biomodulatory effect of celecoxib on blood and benign breast tissue biomarkers in women at increased risk for breast cancer. Women at increased risk for breast cancer [5-year Gail risk score of >1.67%, history of atypical hyperplasia, lobular carcinoma in situ, or previous estrogen receptor (ER)-negative breast cancer] were treated with celecoxib at 400 mg orally twice daily for 6 months. Participants underwent random periareolar fine needle aspiration and blood draw at baseline and at 6 months for analysis of biomarkers: serum levels of insulin-like growth factor 1 (IGF-1), IGF-binding protein 1 (IGFBP-1), and IGFBP-3; tissue expression of Ki-67 and ER; as well as cytology. Forty-nine patients were eligible for analysis. Median IGFBP-1 levels increased significantly from 6.05 ng/mL at baseline to 6.93 ng/mL at 6 months (P = 0.04), and median IGFBP-3 levels decreased significantly from 3,593 ng/mL to 3,420 ng/mL (P = 0.01). We also detected favorable changes in cytology of 52% of tested sites after 6 months of celecoxib therapy. No changes in tissue Ki-67 and ER expression levels were observed. No grade 3 or 4 toxicity was recorded. Celecoxib was well tolerated and induced favorable changes in serum biomarkers as well as cytology in this pilot phase II trial. A phase IIb placebo-controlled study with celecoxib could be considered for women at increased risk for breast cancer.
Introduction
Selective estrogen receptor modulators (SERM) and aromatase inhibitors are FDA approved for reduction of breast cancer risk (1–8). However, these agents only reduce risk of estrogen receptor (ER)-positive breast cancer. Targeting the cyclooxygenase pathway with COX2 inhibitors holds promise as a preventive strategy. Indeed, over the last 2 decades, an increasing amount of data has shown the potential of NASIDs for the prevention of breast cancer (4, 5).
COX2 is an enzyme that converts arachidonic acid to prostaglandins, which are inflammatory markers that have been shown to have antiapoptotic, angiogenic, and proinvasive activities (9). To date, the most striking development in defining the relation of cancer and COX2 has been the finding of higher COX2 expression in colorectal cancer than in normal mucosa (10). Since then, COX2 inhibition has been considered a new avenue for chemoprevention and has been studied in lung, colon, and breast cancers as well. In a large clinical cohort of more than 34,000 women, prediagnostic use of COX2 inhibitors was associated with lower risk of breast cancer recurrence (11).
Large-scale prevention studies with cancer occurrence endpoint require a substantial amount of money and take from years to decades to reveal results. On the other hand, conducting short-term biomarker modulation studies is more cost effective, and is a practical method to understand mechanism of action of a given potential chemopreventive agent.
Here, we evaluated celecoxib's potential chemopreventive effect, by investigating the modulation of several candidate biomarkers that can serve as surrogate indicators of efficacy in phase I/II chemoprevention studies in breast cancer. Insulin-like growth factor 1 (IGF-1) is a polypeptide that promotes cell growth, proliferation, angiogenesis, and metastasis (12, 13). Its effects are regulated by the family of IGF-binding proteins, predominantly by IGFBP-3 (14, 15). IGF-1 and its binding proteins are associated with increased breast cancer risk. Increase in recombinant human IGF-1 was found to be an inducer of mitosis in mammary tissues of mouse models (16). In light of the functions of IGF-1 and its main regulator, IGFBP-3, it is possible that they could be prognostic biomarkers for breast cancer (17, 18). In addition, we also studied Ki-67, a proliferation index of neoplasm (19), as a prognostic biomarker.
In this study, we aimed to establish the effect of the COX2 inhibitor, celecoxib, in patients at increased risk for breast cancer by evaluating the modulation of a panel of potential biomarkers (IGF-1, IGFBP-1, IGFBP-3, Ki-67, and ER), as well as cytologic findings in breast tissue and serum samples. Our goal was to identify which of these biomarkers can serve as potential surrogate markers in a subsequent randomized, placebo-controlled phase II biomarker modulation trial of celecoxib as chemoprevention in women at high risk for breast cancer. We also assessed toxicity in this phase II pilot study.
Materials and Methods
Patient eligibility
Fifty-six pre- and postmenopausal women at increased risk for developing breast cancer were enrolled in this prospective, phase II breast cancer prevention trial using celecoxib. Patients were recruited from the High-Risk Breast Screening and Genetics Clinic at The University of Texas MD (UTMD) Anderson Cancer Center (Houston, TX). Increased risk was defined as having a 5-year Gail risk score of greater than 1.67%, a history of atypical hyperplasia, lobular carcinoma in situ (LCIS), or a previous history of ER-negative breast cancer. The study was reviewed and approved by the institutional review board, and all subjects provided written informed consent.
Study design
After informed consent, patients underwent a baseline blood draw and bilateral random periareolar fine needle aspiration (RPFNA) for biomarker evaluation. Patients then started oral celecoxib at 400 mg twice daily for 6 months. Patients were given a pill diary and received a follow-up telephone call after 3 months for toxicity assessment. At 6 months, at the end of celecoxib therapy, patients returned to the clinic and underwent a second blood draw and repeat RPFNA and toxicity assessment.
Acquisition and processing of RPFNA samples
The RPFNA procedure was performed as described previously (20). Patients with a previous history of breast cancer underwent the procedure in their intact opposite breast. In all women, four RPFNA passes were done, two each at the 3 o'clock and 9 o'clock positions in each breast. Following injection of 2 mL of 1% lidocaine, the aspiration needle was moved in multiple directions to ensure sampling of most of the breast tissue, with emphasis on areas of dense breast tissue, where proliferative glandular tissue is often present. All the fine needle aspiration (FNA) samples were pooled in 5 mL of CytoLyte solution.
Cytologic specimens were prepared using the Thin Preparation (ThinPrep) Technique (Cytyc Corporation). Six to eight slides were prepared from each pooled FNA specimen. One slide was stained with Papanicolaou stain for cytologic diagnosis; the remaining slides were saved in the tissue bank for biomarker studies per the study protocol. Sample adequacy was defined as having 10 or more epithelial cells per slide. The categories used were nonproliferative epithelium (normal), hyperplasia without atypia (benign), atypical hyperplasia, and malignant lesion.
Cell numbers were quantified by directly counting epithelial cell clusters (groups containing >10 cells), single cells, and cells in small groups (groups containing nine or fewer cells). If more than 10 clusters were present, the cells within 10 randomly selected clusters were counted and averaged. The average number of cells was then multiplied by the total number of clusters present. Single cells and cells in small groups were counted by selecting a representative field from each quadrant of the sample.
Evaluation of tissue and serum biomarkers
IHC analysis was performed to determine expression of markers ER and Ki-67 in breast tissue. Apoptosis was determined using the terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling assay. Cytologic evaluation was performed by the cytopathologist using the Masood scoring system. Proteomic pattern analysis was performed on the Protein Biology System 2 SELDI-TOF Mass Spectrometer (Ciphergen Biosystems). Serum anti-glycan antibody analysis was performed using glycan array chips.
Blood samples were processed into serum and plasma fractions. The serum was frozen at −80°C for analysis of IGF-1 and IGF-binding proteins, IGFBP-1 and IGFBP-3. IGF-1, IGFBP-1, and IGFBP-3 were measured by ELISA using kits from Diagnostic Systems Laboratories according to the manufacturer's instructions. Baseline and 6-month samples were analyzed at the same time.
Data capture and entry
All participating patients were registered in UTMD Anderson's Patient Data Management System and in the Multidisciplinary Breast Cancer Database (Breast Cancer Management System), which are used to manage clinical trial programs and facilitate the collection of analysis of patient research data.
Statistical analysis
Patient characteristics were tabulated and compared between cytology classifications using the Kruskal–Wallis test or Fisher exact test. For markers measured on a continuous scale, the sample provided power of 0.87 to detect a change of 0.50 × SD from pretreatment to 6-month measurements (assuming a two-sided one-sample t test with significance level of 0.05).
Major analyses consisted of appropriate paired tests of pretreatment and 6-month marker measurements. These included the McNemar test to compare any differences in cytology before and after treatment, and one-sample t test for continuous variables. Nonparametric methods, such as the Wilcoxon test, were also employed to compare baseline markers and modulation of the markers between or among various characteristics groups, when appropriate.
Results
Patient characteristics
Of 56 enrolled patients, 49 were eligible and started the drug. The median age was 51 years (range: 39–73 years), and the majority of patients were Caucasian (82%). The risk factors for breast cancer were: 5-year Gail risk score of greater than 1.67% in 17 patients (35%), history of atypical hyperplasia in 14 patients (29%), history of LCIS in 13 patients (27%), and history of previous invasive ER-negative breast cancer with an intact contralateral breast in 5 patients (10%). Fifty-seven percent of the women were postmenopausal, and 43% were pre- or perimenopausal. The median duration of prior hormone replacement therapy (HRT) use was 60 months; 69% of the women had never used HRT (Table 1).
Characteristics of the 49 evaluable patients.
Compliance during the study was high, with 82% (40/49) of the patients completing 6 months of celecoxib therapy. Nine patients discontinued celecoxib early for the following reasons: adverse events (2 patients; at 2 and 6 weeks), refusal (1 patient; at 6 weeks), national clinical trials with COX2 inhibitors temporarily being held by the NCI (3 patients; at 1, 3, and 4 months), and inadequate tissue upon baseline FNA (2 patients; at 1 and 2 weeks). The compliance with FNA was very high; 39 of the 40 patients who continued celecoxib for 6 months underwent repeat FNA at the end of the study.
Toxicity
Celecoxib was well tolerated by the patients, as shown in Table 2. No grade 3 and 4 adverse events were reported. Grade 1–2 side effects include nausea, heartburn/indigestion, diarrhea, skin rash, headache, and fatigue. Two patients reported short-term chest pain, and cardiology evaluation was negative. One of these patients chose to discontinue celecoxib; however, the other patient resumed the study drug and completed the trial. Two patients experienced transient tachycardia; one elected to discontinue treatment, and one completed the trial.
Adverse events among the 49 evaluable patients.
Changes in serum biomarkers
The median IGFBP-1 levels were 6.05 ng/mL (range: 2.9–10.9) at baseline and 6.93 ng/mL (range: 0.4–30.07) at 6 months; this increase was statistically significant (P = 0.04). The median IGFBP-3 levels were 3,593 ng/mL (range: 1,780–5,662) at baseline and 3,420 ng/mL (range: 1,600–4,672) at 6 months; this decrease was statistically significant (P = 0.01; Fig. 1). The change in IGF-1 level from pre- to postcelecoxib therapy was not statistically significant (P = 0.27). As recent studies suggest serum IGF-1/IGFBP3 ratio correlates better with increased risk of cancer than serum IGFBP levels, we also looked at IGF1/IGFBP3 ratio. The median IGF-1/IGFBP3 ratios were 0.05 (ng/mL; range: 0.03–0.26) at baseline and 0.07 (ng/mL; range: 0.04 – 0.73) at 6 months (P = 0.0034; Fig. 2).
Change in IGFBP-1 (A) and IGFBP-3 (B) levels by patient in serum collected at baseline and after 6 months of celecoxib therapy.
Change in serum IGF-1/IGFBP-3 ratio collected at baseline and after 6 months of celecoxib therapy.
Changes in breast tissue: cytology and biomarkers
A total of 60 breast sites in 49 patients were fine needle aspirated before celecoxib therapy; 38 patients had FNA only in one site, and 11 patients had FNAs performed in two sites (one in each breast). Of 60 sites, four had insufficient FNA samples based on the above-mentioned adequacy criteria. Therefore, a total of 56 sites were available for analysis before celecoxib therapy. Of the 49 patients, 45 had pre- and postcelecoxib FNA samples (baseline and 6-month follow-up FNA in the same patient).
The cytologic findings are shown in Table 3. The rate of hyperplasia in baseline FNA samples was 22%, and the atypical hyperplasia rate was 17%. After 6 months of celecoxib therapy, the hyperplasia rate was 23%, and atypical hyperplasia rate was 10%. We found that 60% (6/10) of the sites with baseline atypical hyperplasia had regressed to hyperplasia and/or normal cytology. Moreover, regression from hyperplasia to normal cytology was also observed in 46% (6/13) of the tested sites. Therefore, in total, we observed favorable changes in cytology in 52% (12/23) of the tested sites after 6 months of celecoxib therapy (Table 3). In addition, levels of ER and Ki-67 expression were assessed via IHC analysis and did not significantly differ between pre- and postcelecoxib therapy (P = 0.32 and P = 0.6, respectively).
Changes in cytologic findings in tissue collected at baseline and after 6 months of celecoxib therapy.
Discussion
In this study, we found that 6 months' of celecoxib therapy caused significant alterations in the median IGFBP-1 and IGFBP-3 levels. Median IGFBP-1 levels increased significantly from 6.05 ng/mL at baseline to 6.93 ng/mL (P = 0.04), and median IGFBP-3 levels decreased significantly from 3,593 ng/mL to 3,420 ng/mL (P = 0.01). Furthermore, 60% (6/10) of the breast sites with baseline atypical hyperplasia regressed to hyperplasia and/or normal cytology. Regression from hyperplasia to normal cytology was also observed in 46% (6/13) of the tested sites. Therefore, in total, we observed favorable changes in cytology in 52% (12/23) of the tested sites after 6 months of celecoxib therapy.
The clinical use of COX2 enzyme inhibition as a chemoprevention has long been investigated. Preclinical studies have shown remarkable reduction in the volume of breast tumor size in rats treated with celecoxib (21, 22). Other studies have shown higher COX2 overexpression by IHC analysis in patients with metastatic breast cancer (23). Moreover, Ristimaki and colleagues showed positive IHC staining for COX2 in 1,576 women with invasive breast carcinoma and found that COX2 overexpression was correlated with ER positivity, poor prognosis, tumor invasiveness, and reduced survival (24–27). Taken together, these findings constitute a reasonable rationale for the use of COX2 inhibitors as a risk reduction agent in women at increased risk for developing breast cancer.
The carcinogenic properties of COX2 are attributed to a molecule that COX2 produces: prostaglandin E2 (PgE2). PgE2 induces proliferation of breast cells (28) and increases angiogenesis (29) by a cascade related to the enzyme aromatase (30–32). It is hypothesized that while triggering the intracellular secondary messenger system, PgE2 causes transcription of the CYP19 gene and therefore induces concomitant production of aromatase. The increased aromatase activity eventually results in higher levels of estradiol, which is responsible for the genesis of new neoplastic cells, especially in ER-positive tumors (33). Hughes and colleagues also detected that COX2 enzyme induces aromatase activity by triggering the local production of PgE2. This explains why much more PgE2 is produced in ER-positive malignant breast tissue than in normal breast tissue (34, 35).
These findings are opposed to the observation that COX2 protein expression is also elevated in ER-negative breast cancer (36). Basudhar and colleagues (37) studied 248 human ER-negative breast cancers and identified a highly significant impact of NO synthase (NOS2) and COX2 interaction on patient survival. Patients whose tumors expressed high levels of both NOS2 and COX2 had dismal outcomes (HR > 21), with only one-third of patients were alive at 5 years compared with over 95% survival in patients with low NOS2 and COX2 expression.
Liu and colleagues have found another noteworthy mechanism in COX2-related tumorigenesis. Female transgenic mice with overexpression of COX2 in the mammary glands were shown to have increased expression of the antiapoptotic marker, BCL2, and reduced expression of proapoptotic proteins, Bax and BCLx(L) compared with nontransgenic mice (38).
In the last 2 decades, studies in a wide range of cancers, including breast, colorectal, and non–small cell lung cancer, have stressed the positive correlation between the COX2 enzyme and IGF-1 (9, 39). In vitro preclinical studies have shown that IGF is a natural growth hormone, which is essential for growth and survival. It is involved in inhibition of apoptosis, promotion of cell-cycle progression, angiogenesis, and metastatic activities in various cancers. The cellular responses to the IGFs are mediated primarily by the IGF-1 receptor, and IGF actions are regulated by IGFBP-1 and IGFBP-3 (16, 18, 39–45). In this study, we observed that 6 months of COX2 inhibitor therapy resulted in enhanced stimulation of IGFBP-1 and decreased levels of IGFBP-3, but no changes in IGF-1, Ki-67, or ER expression levels. We also detected favorable changes in the cytology of the breast tissue in 20% of our patient population.
It has been noted that women with breast cancer have higher levels of IGF-1 than those without disease (42). A collaborative analysis that evaluated the data of 12 prospective studies from 17 countries showed that IGF-1 is positively correlated with increased breast cancer risk (46). In this study, we did not see any significant changes in serum IGF-1 levels from baseline to posttreatment.
IGF-1 has six binding proteins; two of them, IGFBP-1 and IGFBP-3, have been well studied and found to be significantly relevant to cell growth and proliferation because their major role is to change the available amount of IGF in serum. Because they decrease IGF-1 levels by binding “free,” circulatory IGF-1, it is hypothesized that IGFBP-1 and IGFBP-3 may have the potential to inhibit proliferative effects of IGF-1 in breast cells (47, 48). These biomarkers have already been evaluated in chemoprevention trials with SERMs or aromatase inhibitors (49–52). A study, previously conducted by our group, showed a significant increase in IGFBP-1 in women who had 6 months of anastrozole therapy (53). In this prospective trial, we observed increased IGFBP-1 levels after 6 months of therapy with celecoxib, suggesting that it might have antiproliferative effects on breast cancer.
For IGFBP-3, there was less concrete evidence regarding biomarker properties. Theoretically, IGFBP-3 is expected to be a marker as reflective as IGFBP-1 in terms of reducing antiapoptotic and proliferative properties of IGF-1. Levitt at al observed 20% growth inhibition when they treated human mammary epithelial cells with 500 ng/mL IGFBP-3; they attributed this change to IGFBP-3 potentially blocking the actions of IGF-1 and subsequently attenuating the growth of breast epithelial cells (39). Supporting this finding, recent studies have associated the increase in IGF-I/IGFBP-3 ratio in combination with an increased risk of cancer (42, 54, 55). However, there are contradictory reports that outline controversy regarding IGFBP-3 levels (56). For instance, Helle and colleagues showed an overall decrease in IGFBP-3 levels in subjects who had a clinical response to tamoxifen treatment, but an increase in IGFBP-3 levels in patients with progressive disease (56). They interpreted this finding as indicating that IGFBP-3 may have a close correlation with tumor burden. Some other reports have confirmed the role of IGFBP-3 in predicting poor prognostic features. Rocha and colleagues observed 3 times higher levels of IGFBP-3 in tumors that were relatively larger, had a high rate of cell-cycle synthesis, were negative for ER and/or progesterone receptor, and were aneuploid (57, 58). Moreover, Goodwin and colleagues found that a high level of IGFBP-3 was related to distant metastasis and recurrence in postmenopausal women with ER-positive breast tumors (59). Amenable with these results, our study revealed not only significantly lower levels of IGFBP-3 after 6 months of celecoxib therapy, but also the IGF-I/IGFBP-3 ratio was significantly higher (P < 0.034). Taking all these findings into account, it can be postulated that IGFBP-3 could be more associated with tumorigenesis than previously thought. Given our results, we believe that IGFBP-3 is a biomarker with high potential to predict poor prognosis, and reducing its levels could be the aim of further chemoprevention trials.
Cytomorphology is a potential surrogate endpoint for breast cancer prevention trials. However, some chemoprevention studies have not shown any change in cytology after treatment with tamoxifen, letrozole, and alpha-DFMO (50, 60, 61). In our study as well, we did not observe any significant changes in cytology after 6 months of treatment with celecoxib (in the majority of patients?), but we observed favorable changes in 20% of women, such as regression from atypical hyperplasia to hyperplasia, or from hyperplasia to normal cytology.
Our results should be interpreted with caution and considered only hypothesis generating, given two limitations. First is the small number of patients enrolled, as this was a single-institution study. However, prospective accrual to prevention trials that require paired breast tissue sampling is challenging. Second, the influence of other uncontrolled factors, including the follicular phase of the menstrual cycle, could have affected Ki-67 results in FNA samples.
Importantly, this is the first clinical trial that has investigated whether COX2 inhibition can prevent IGF-1–related proliferation in women at increased risk of primary or secondary breast cancer. The acceptable side effect profile and significant modulation of the proliferation markers support continued evaluation of celecoxib as a breast cancer prevention agent. Our results validate IGFBP-1 and IGFBP-3 as surrogate biomarkers to identify treatment efficacy in clinical studies of the preventive effects of COX2 inhibitors. Therefore, this short-term biomarker modulation study justifies the potential of COX2 inhibitors as risk-reducing therapy, which could be investigated in a prospective randomized phase II placebo-controlled study to validate the biomarkers. Such a trial, in turn, could lead to larger phase III prevention trials in patients at increased risk for ER-negative breast cancer, such as BRCA germline mutation carriers or patients with a previous history of ER-negative breast cancer, for contralateral secondary breast cancer prevention.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
S. Bayraktar: Data curation, writing-original draft, writing-review and editing. S. Baghaki: Data curation, writing-review and editing. J. Wu: Formal analysis, methodology, writing-review and editing. D.D. Liu: Formal analysis, writing-review and editing. A.M. Gutierrez-Barrera: Investigation, methodology, writing-review and editing. T.B. Bevers: Writing-review and editing. V. Valero: Data curation, investigation, writing-review and editing. N. Sneige: Formal analysis, investigation, writing-review and editing. B.K. Arun: Conceptualization, resources, data curation, formal analysis, supervision, investigation, methodology, writing-original draft, project administration, writing-review and editing.
Acknowledgments
Editorial assistance was provided by Sunita Patterson, Scientific Publications, Research Medical Library, University of Texas MD Anderson Cancer Center. This study was supported, in part, by the NIH/NCI under award number P30CA016672 (to B.K. Arun).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Footnotes
Cancer Prev Res 2020;13:795–802
- Received February 27, 2020.
- Revision received April 1, 2020.
- Accepted June 2, 2020.
- Published first June 8, 2020.
- ©2020 American Association for Cancer Research.
References
Volume 13, Issue 9
