IGFBP2 and IGFBP3 Protein Expressions in Human Breast Cancer: Association with Hormonal Factors and Obesity

Published OnlineFirst January 26, 2010; DOI: 10.1158/1078-0432.CCR-09-0957

Clinical Imaging, Diagnosis, Prognosis Cancer Research IGFBP2 and IGFBP3 Protein Expressions in Human Breast Cancer: Association with Hormonal Factors and Obesity

Nicole M. Probst-Hensch7, Julia H.B. Steiner1, Peter Schraml2, Zsuzsanna Varga2, Ursina Zürrer-Härdi2, Martina Storz2, Dimitri Korol1, Mathias K. Fehr3, Daniel Fink3, Bernhard C. Pestalozzi4, Urs M. Lütolf5, Jean-Philippe Theurillat2, and Holger Moch2,6

Abstract Purpose: The insulin-like growth factor (IGF) signaling system is involved in breast cancer initiation and progression. The prognostic relevance of tumor expression patterns of IGFI-related proteins remains poorly understood. This study associates the expression of selected IGF proteins with breast tumor and patient characteristics. Experimental Design: IGFI, IGFI receptor, IGF-binding protein (IGFBP)2, and IGFBP3 expression was measured in 855 primary breast carcinomas by immunohistochemistry using tissue microarrays. We investigated the association of tumor and nodal stage, grade, hormone receptor status, HER2 gene amplification, menopausal status, body mass index, and survival with IGF protein expression. Results: In contrast to IGFI, the expression of IGFI receptor, IGFBP2, and IGFBP3 was associated with estrogen receptor status. In addition, IGFBP3 was positively correlated with body mass index and premenopausal status. Importantly, IGFBP2 was an independent and positive predictor of overall survival (hazard ratio, 0.48; 95% confidence interval, 0.24-0.95; P = 0.04). There was a weak suggestion for IGFBP2 and overweight to modify each other's effect on survival. Conclusions: According to these results, which need confirmation in larger patient series, the prognostic relevance of IGFBP2 and IGFBP3 protein expressions in breast cancer may depend on the hormonal context and body weight. Clin Cancer Res; 16(3); 1025–32. ©2010 AACR.

Understanding the biology of breast cancer progres- gland development (3–5). They have mitogenic, antia- sion is an important aspect of identifying therapeutic poptotic, and proangiogenic effects and stimulate the targets. Both obesity at diagnosis and subsequent weight motility of breast cancer cells in vitro (3). The two gain are predictive of worse breast cancer survival (1). growth factors act through IGFI receptor (IGFIR) homo- – The obesity breast cancer association may in part be re- dimerization (3). Their bioavailability in circulation and lated to hyperinsulinemia, the concomitant decrease in tissues is regulated by at least six IGF-binding proteins. – insulin-like growth factor (IGF) binding proteins IGFBP3 and IGFBP2 are the most abundant IGFBPs in (IGFBP)1 and IGFBP2, and the resulting increase in bio- the blood. In addition to regulating IGF activity, IGFBPs active IGFI (2). IGFI and IGF2 play a role in mammary exert additional IGF-independent effects (6). Transgenic mice deficient in IGFI exhibit a decreased susceptibility to mammary carcinogens (7). Tumor inci- dence is increased in mice expressing IGFI or IGF2 in Authors' Affiliations: 1Chronic Disease Epidemiology/NICER, Institute of Social and Preventive Medicine, University of Zurich; 2Institute of Surgical the mammary gland (3). According to several meta-anal- Pathology, Department of Pathology, 3Department of Obstetrics and yses and systematic reviews, circulating concentrations of Gynecology, 4Clinic of Oncology, Department of Internal Medicine, and 5Clinic of Radio-Oncology, University Hospital of Zurich; 6Competence IGFI and less consistently of IGFBP3 were positively asso- Center of Systems Physiology and Metabolic Disease, Zurich, ciated with premenopausal breast cancer (8). Much less is Switzerland; and 7Institute of Social and Preventive Medicine at STI known about the prognostic relevance of autocrine and Associated Institute of the University of Basel, Basel, Switzerland. paracrine IGF signaling in tumor tissue. Several breast Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/). cancer studies reported on the correlation between es- N.M. Probst-Hensch and J.H.B. Steiner contributed equally to this trogen receptor (ER) status and the expression patterns work. of IGF-related proteins (9–12), but only few also evalu- Corresponding Author: Nicole M. Probst-Hensch, Chronic Disease ated the effect of IGF-related mRNA or protein expres- Epidemiology, Institute of Social and Preventive Medicine at STI Associ- sions in tumor tissue on survival (9, 13–21). ated Institute of the University of Basel, Steinengraben 49, 4051 Basel, Switzerland. Phone: 41-61-267-6066; Fax: 0041-61-2676190; E-mail: By analyzing a large number of breast cancers, we stud- [email protected]. ied the protein expression patterns of IGFI, IGFIR, IGFBP2, doi: 10.1158/1078-0432.CCR-09-0957 and IGFBP3 and their effect on clinical and pathologic ©2010 American Association for Cancer Research. parameters including prognosis.

www.aacrjournals.org 1025

Probst-Hensch et al.

measured using categorical variables. We thereby avoided Translational Relevance imprecision in assigning exact levels for the percent of positively stained cells. We used three categories for staining: The role of the insulin-like growth factor (IGF) sig- <10% stained cells (negative), weak or heterogenous stain- naling system in breast cancer development has been ing, and strong and homogenous staining. the focus of intense research. The prognostic relevance of local expression patterns of the IGF family proteins is still poorly understood. Our results from a large series of primary invasive breast cancers show that the Table 1. Tumor characteristics IGF protein expression patterns are correlated with estro- Characteristics n Mean/% gen receptor status (IGFIR, IGFBP2, and IGFBP3) as well as with menopausal status and body mass index Total no. 855 (IGFBP3). The findings of this study stress the need to Histotype study the prognostic relevance of IGF proteins by consid- Infiltrating ductal carcinoma 713 83.4 ering patient and tumor characteristics simultaneously. Infiltrating lobular carcinoma 98 11.5 Others 40 4.7 Missing 4 0.5 pT pT1 325 61.7 Materials and Methods pT2 386 45.6 pT3 55 6.5 Patients and tissues. Formalin-fixed, paraffin embedded pT4 81 9.6 primary breast cancer samples were retrieved from patients pN diagnosed with invasive breast cancer between 1991 and pN0 201 31.6 2004 at the Institute of Surgical Pathology of the Univer- pN1 343 54 sity Hospital Zürich. H&E-stained sections of all tumors pN2/pN3 91 14.3 were reevaluated by one pathologist (J.P.T.). All breast car- M cinomas were diagnosed according to the WHO tumor M1 43 classification (International Classification of Diseases in Grade Oncology). Histopathologic grading was done according G1 127 15.1 to the modified Bloom and Richardson system (22). Tis- G2 398 47.4 sue microarrays (TMA) construction was as previously de- G3 314 37.4 scribed (23). Overall survival data were provided by the ER Cancer Registry of the Canton of Zürich. Additional clini- Negative 162 17.8 cal information was obtained from the Department of Gy- Positive 693 82.2 necology, the Department of Oncology, and the PR Department of Radio-Oncology of the University Hospital Negative 243 28.8 Zürich. The project was approved by the local Commis- Positive 600 71.2 sion of Ethics (ref. no. StV 12-2005). HER2 Immunohistochemistry. TMA sections (2.5 μm) were Not amplified 287 77.8 transferred to glass slides using an adhesive-coated slide Amplified 82 32.2 system (Instrumedics, Inc.). Immunohistochemical stain- IGFI ing was done according to the Bond-maX automat proto- Negative 366 42.8 cols (Leica Microsystems, Medite). TMAs were incubated Weakly positive 328 38.4 with antibodies recognizing ER (1:1,000; Novocastra Strongly positive 161 18.8 Labs), progesterone receptor (PR; 1:1,000; Novocastra IGFIR Labs), IGFI (1:50; NeoMarkers), IGFIR (1:50;), IGFBP2 Negative 487 57 (1:25; Cell Signaling), and IGFBP3 (1:100; Calbiochem). Weakly positive 98 11.5 Diaminobenzidine was used as a chromogen. Tumors Strongly positive 270 31.6 were considered positive for ER and PR if unequivocal nu- IGFBP3 clear positivity was seen in at least 10% of tumor cells Negative 439 51.4 (24); tumors were considered positive for IGFI, IGFBP2, Weakly positive 360 42 and IGFBP3 if weak or strong cytoplasmic staining was ob- Strongly positive 56 6.6 served in at least 10% of tumor cells; and tumors were IGFBP2 considered positive for IGFIR if weak or strong membra- Negative 483 56.5 nous staining was seen in at least 10% of tumor cells. Be- Weakly positive 205 23.9 cause only one core with a diameter of 0.6 mm was Strongly positive 167 19.5 analyzed from each tumor, IGF protein expression was

1026 Clin Cancer Res; 16(3) February 1, 2010 Clinical Cancer Research

IGFBP2, IGFBP3, and Breast Cancer Prognosis

Table 2. Interrelationship between IGF family proteins within tumor samples

IGFI* IGFIR* IGFBP2* IGFBP3* ρ/P (Spearman)/P (γ) ρ/P (Spearman)/P (γ) ρ/P (Spearman)/P (γ) ρ/P (Spearman)/P (γ)

IGFI — 0.18/<0.001/<0.0001 0.05/0.12/>0.05 0.27/<0.001/<0.0001 IGFIR - ——0.07/0.03/<0.05 0.13/<0.001/<0.001 IGFBP2 ———0.09/<0.006/<0.05

*Based on ordinal expression levels.

Stromal cells and epithelial tumor cells were distin- Results guished by experienced pathologists on the basis of their distinct histologic features. An unequivocal identification Tumor phenotype (tumor and nodal stage, grade), hor- was possible in all TMA cores even in the immunohisto- monal (ER and PR) and HER2 amplification status, as well chemically stained sections. Stromal cells (fibroblasts) are as the expression frequencies of IGFI, IGFIR, IGFBP2, and characteristically spindle-shaped cells with small nuclei, IGFBP3 are listed in Table 1. Examples of IGFI-, IGFIR-, whereas breast cancer cell complexes showed glandular IGFBP2-, and IGFBP3-positive breast cancers are shown in and tubular growth with hyperchromatic nuclei. Nuclear Supplementary Fig. S1. Of 855 primary breast carcinomas and cytologic features were also used to separate histio- analyzed, 83.4% belonged to the ductal and 11.5% be- cytes and lymphocytes from tumor cells. longed to the lobular subtype. Other tumor types (4.7%) Fluorescence in situ hybridization. TMA sections (2.5 μm) included invasive tubular, cribriforme, medullary, mucin- were also used for two-color HER2 fluorescence in situ hy- ous, apocrine, metaplastic, clear cell, micropapillary, and bridization. Fluorescence in situ hybridization was done papillary carcinomas. The tumor expression of IGFI was pos- as described in ref. (25), using the kits and protocols of Vysis itively correlated with the expression of IGFIR and IGFBP3. (Abbott Molecular, Inc.). The HER2 gene was considered In addition, IGFBP2 and IGFBP3 expressions were positively amplified if the ratio of gene/centromere was ≥2.0. correlated with each other and with IGFIR expression. Statistical analysis. The sample size of the study allowed Although the associations did reach statistical significance, us to categorize IGF protein expression into three levels as they were nevertheless relatively weak (Table 2). we lacked continuous expression measures (score 1, no ex- Patient characteristics including the distribution of follow-up pression; score 2, weak expression; and score 3, strong ex- time are described in Table 3. The mean age of the patients pression). The categories were chosen a priori based on the decision to (a) keep the patient group exhibiting no IGF protein separate and (b) achieve a balanced distribution of patients with positive staining for IGF proteins into two categories. Unless stated otherwise, results refer to the Table 3. Patient characteristics contrast of no versus any expression. In these cases, the use Characteristic n Mean/% of a variable with three categories did not provide addition- a al information in the analysis of ( ) correlation between Total number 855 b IGF protein expression, ( ) determinants of IGF protein ex- Age at diagnosis (y) 855 61.9 (SD 14.66) c pression, and ( ) associations with survival. The Spearman BMI at diagnosis (kg/m2) 364 26.0 (SD 5.3) rank correlation was used to correlate IGF marker protein Overweight 103 28.30% expressions; the statistical significance of the correlations Obese 72 19.80% γ χ2 was further confirmed by the statistic and the tests. Postmenopausal The association of patient and tumor characteristics with No 71 10.50% IGF protein expression was assessed by unconditional logis- Yes 602 89.50% tic regression. Overall and subgroup-specific survival was Hormone therapy estimated by the Kaplan-Meier method. Survival differences (postmenopausal women only) χ2 were assessed by the log-rank test ( test for dichotomous No 147 62.00% IGF protein expression). The independent effect of molecu- Yes 90 37.00% lar markers was determined by the Cox proportional hazard Follow-up (mo) model. The interaction between obesity and IGFBP2 pro- Mean 782 52.5 tein expression was assessed by the inclusion of an accord- median 782 43.8 P ing interaction term in the model. values of <0.05 were Range 782 0.5-167.5 considered significant in all statistical analyses, which were 5-y survival 782 76.90% done using the software package Stata (version StataSE 9).

www.aacrjournals.org Clin Cancer Res; 16(3) February 1, 2010 1027

Probst-Hensch et al.

Table 4. Association of tumor and patient characteristics with the expression of IGF family proteins in primary breast carcinomas

A: Adjusted association of tumor characteristics with the expression of IGF family proteins in primary breast carcinomas Determinant IGFI IGFIR IGFBP2 IGFBP3 OR/P OR/P OR/P OR/P n = 617 n = 617 n = 617 n = 617

pT (T1 vs >T1) 1.21/0.28 1.11/0.56 0.82/0.37 1.13/0.47 pN (N0 vs >N0) 0.83/0.31 0.90/0.57 1.03/0.88 1.08/0.68 Grade G1 —— — — G2 0.76/0.29 1.08/0.77 1.35/0.25 1.23/0.42 G3 0.84/0.52 1.46/0.17 1.66/0.06 1.76/0.04 ER (positive vs negative) 0.77/0.31 0.59/0.04 2.75/<0.001 0.49/0.007 PR (positive vs negative) 1.07/0.72 0.88/0.55 0.91/0.65 1.17/0.47 Additional her2 amplification n = 333 n = 333 n = 333 n = 333 Versus not amplified* 0.97/0.56 1.16/0.61 1.34/0.30 1.34/0.31

B: Adjusted association of patient characteristics with the expression of IGF family proteins in primary breast carcinomas

Determinant IGFI IGFIR IGFBP2 IGFBP3 OR/P OR/P OR/P OR/P n = 242 n = 242 n = 242 n = 242

Age (cont.) 0.99/0.43 0.98/0.10 1.00/0.73 1.01/0.69 BMI (cont.) 1.04/0.13 1.02/0.35 1.03/0.17 1.08/0.009 Postmenopausal: yes vs no 1.17/0.70 1.08/0.85 1.11/0.78 0.39/0.03

NOTE: Associations were mutually adjusted for all variables listed in the table. *Adding her2amp to the model does not alter associations between IGF proteins and tumor factors.

was 62 years (SD, 14.7; range, 23-97). Ninety percent of the As expected, grade, tumor size, lymph node involve- patients were postmenopausal. The median time of follow- ment, as well as ER and PR status were associated with up available for 782 subjects was 43.8 months. During the prognosis (Table 5). None of the IGF family members overall follow-up time of 167.5 months, 184 deaths oc- was associated with overall survival in the absence of ad- curred and 5-year overall survival was 76.9% (Supplemen- justment for established or hypothesized prognostic fac- tary Fig. S2: Kaplan-Meier survival curve in the total study tors and treatment (Supplementary Fig. S3A-D; Table 5). population). As only 25% of participants were followed-up Only in multivariate analyses adjusting for BMI as well for longer than 80 months, follow-up time was censored in as treatment was IGFBP2 tumor expression a positive pre- all survival analyses presented in this article. After censor- dictor of overall survival [hazard ratio (HR), 0.48; 95% ing, median follow-up time within 80 months of diagnoses confidence interval (95% CI), 0.25-0.95; P =0.04].The remained unchanged at 45.99 months; 157 of 782 subjects prognostic effect of ER status disappeared in this model. died within 80 months of diagnosis. There was a tendency for an association between over- The independent associations of the IGF protein expres- weight and poor survival (Supplementary Fig. S4; Table 5), sions with selected tumor and patient characteristics are yet this association did not reach statistical significance in presented in Table 4A and B. In models exclusively con- any model and disappeared upon adjustment for treat- taining tumor characteristics, IGFPB3 expression was asso- ment. However, there was a suggestion, albeit at a limit- ciated with high-grade tumors (Table 4A). ER-positive ed statistical power, that IGFBP2 and overweight may tumors were more likely to express IGFBP2, but less likely modify each other's effect on survival. In the absence of to express IGFIR and IGFBP3. If patient characteristics such adjustment for additional patient, tumor, and treatment as age, menopausal status, and body mass index (BMI) characteristics, overweight was related to poor overall sur- were added to the models, we observed that IGFBP3 expres- vival only in tumors expressing IGFBP2 (P =0.05ata sion was lower in postmenopausal when compared with power of 51%); vice versa, the lack of IGFBP2 expression premenopausal women and positively correlated with was related to poor survival only in nonoverweight patients BMI at diagnosis (Table 4B). IGFI was not associated (P = 0.06 at a power of 46%; Pinteraction = 0.10; Fig. 1A-D). with any tumor or patient characteristics. When studying the combined, unadjusted effect of

1028 Clin Cancer Res; 16(3) February 1, 2010 Clinical Cancer Research

IGFBP2, IGFBP3, and Breast Cancer Prognosis

IGFBP2 expression and overweight (Supplementary Fig. Extensive in vitro and in vivo data support the impor- S5A-B), survival was highest in the group of nonover- tant role of IGF-related signaling pathways in the growth weight patients with tumors staining positive for IGFBP2 regulation of breast cancer cells (26). But the relative im- when compared with all other patients (P = 0.04). Upon portance of locally produced versus systemically induced adjustment for additional patient, tumor, and treatment IGF-related proteins remains incompletely understood characteristics, the HR for the effect of overweight re- (26, 27). The correlations observed in this study between mained elevated, but not statistically significant among IGF protein expression and sex steroid hormone–related patients with tumors staining positive for IGFBP2 (HR, factors (ER status, menopausal status, and obesity) paral- 1.86; 95% CI, 0.57-6.08], and the HR for the lack of IGFBP2 lel the experimental data on an intense cross-talk be- expression was 4.00 (95% CI, 1.28-12.52) among nonover- tween the IGF system and sex steroids in breast cancer weight patients (Supplementary Table S1). cell lines and tissues (26, 28). The interactions observed between IGF and estrogen signaling pathways seem to relate to mitogenic and growth stimulatory effects. In addition, Discussion IGF signaling pathways may exhibit estrogen-independent nonmitogenic effects (26). In our large series of primary invasive breast cancers, the The correlation between IGF protein expression and the expression of IGFIR, IGFBP2, and IGFBP3, but not of IGFI clinicopathologic characteristics of breast cancer was the protein, was correlated with ER status. In addition, IGFBP3 focus of previous studies. Consistent with our results, expression was positively correlated with grade, BMI, and Koda et al. (29) reported an association between high premenopausal status. Despite these associations with fac- IGFIR and low ERα expressions measured immunohisto- tors predictive of breast cancer outcome, just IGFBP2 ex- chemically in breast tumors. pression was positively associated with overall survival, Several reports found high levels of IGFBP3 protein or and only upon adjustment for other prognostic factors in- mRNA in breast tumors associated with poor prognostic cluding BMI. factors including low levels of sex steroid receptors.

Table 5. Univariate and multivariate analyses of the effect of pT and pN category, histologic grade, age, BMI (where indicated), and IGF family protein expressions on overall survival (80 mo follow-up)

Determinant Univariate Multivariate Multivariate* Multivariate† HR (95% CI)‡ HR (95% CI)‡ HR (95% CI)‡ HR (95% CI)‡ without BMI adj. with BMI adj. with BMI adj.

Grade G1 —— — — G2 2.29 (1.14-4.60) 1.38 (0.54-3.53) 1.99 (0.49-8.91) 2.14 (0.47-9.73) G3 3.68 (1.84-7.34) 2.24 (0.88,5.75) 5.36 (1.22-23.56) 5.70 (1.26-25.75) Age (cont.) 1.03 (1.01-1.04) 1.02 (1.01-1.03) 1.03 (1.01-1.05) 1.03 (1.01-1.05) pT (>T1 vs T1) 3.52 (2.28-5.44) 2.91 (1.55-5.47) 2.78 (1.13-6.85) 2.61 (1.05-6.49) pN (>N0 vs N0) 2.10 (1.26-3.48) 2.18 (1.29-3.70) 1.88 (0.88-4.02) 2.14 (0.97-4.73) IGFI 1.15 (0.83-1.61) 1.04 (0.67-1.60) 0.96 (0.52-1.79) 1.16 (0.60-2.23) IGFIR 1.02 (0.74-1.40) 0.79 (0.53-1.20) 0.70 (0.38-1.31) 0.70 (0.37-1.33) IGFBP2 0.95 (0.69-1.31) 0.73 (0.49,1.12) 0.48 (0.25-0.93) 0.48 (0.24-0.95) IGFBP3 1.12 (0.81-1.54) 1.13 (0.73-1.75) 1.21 (0.60-2.47) 1.21 (0.60-2.47) BMI ≤25 kg/m2 —— — — >25 kg/m2 1.39 (0.79-2.42) — 0.77 (0.40-1.53) 0.72 (0.35-1.46) >30 kg/m2 1.25 (0.66-2.37) — 1.32 (0.63-2.78) 1.04 (0.47—2.30) ER (positive. vs negative) 0.43 (0.31-0.60) 0.52 (0.30-0.91) 0.85 (0.37-1.68) 0.68 (0.29-1.62) PR (positive. vs negative) 0.50 (0.36-0.69) 0.75 (0.45-1.27) 0.78 (0.39-1.53) 0.80 (0.36-1.46) Her2am amplification vs not amplified§ 1.52 (0.71-3.22) 1.00 (0.41,2.48) 1.73 (0.61-4.86) 2.55 (0.83-7.81)

*Model contains all variables except her2 amplification (n = 254 observations in the model). †Additional adjustment for surgery, radiotherapy, chemotherapy, and tamoxifen therapy. ‡Associations statistically significant at P < 0.05 are bold. §Additional adjustment for her2 amplification (amplified vs not amplified) did not materially alter the HRs for the other factors reported above.

www.aacrjournals.org Clin Cancer Res; 16(3) February 1, 2010 1029

Probst-Hensch et al.

Fig. 1. A–D, association of BMI and IGFBP2, respectively with overall survival during 80 mo of follow_up in patients (A) positive for IGFBP2 expression and (B) negative for IGFBP2 expression and (C) with BMI ≤25 kg/m2 and (D) with BMI >25 kg/m2.

Despite that, the prognostic role of IGFBP3 itself remains nant cells and tissues were predictive of poor prognosis in elusive. This reflects the complexity of biological IGFBP3 many malignancies including leukemia, brain tumors, as activity in tumor cells (10, 30–33) and possibly the fact well as cancer of the colon, ovaries, lung, and prostate that IGFBP3, due to its expression in endothelial cells, is (9, 39–43). Studies on IGFBP2 expression in breast tumor increased in highly vascular tumors (34, 35). Furthermore, tissue are still limited in number. The reports regarding the IGFBP3 concentration in tumor tissue depends on the the correlation between IGFBP2 and ER expression are patients' attributes. In the Shanghai Breast Cancer Study, contradictory (9–12). Few studies investigated the prog- IGFBP3 mRNA expression in breast cancer tissue was high- nostic effect of IGFBP2. Wang et al. (11) observed an in- er in premenopausal as opposed to postmenopausal verse correlation between IGFBP2 and ER and reported women (36). Yu et al. (31) found the level of IGFBP3 pro- IGFBP2 tumor expression to predict metastases in early- tein in breast cancer tissue to decrease with age. Guo et al. stage breast cancer. The largest study on the prognostic (37) compared gene expressions in breast epithelial cells role of IGFBP2 included tumor specimens from 4,186 from women with different reproductive history and re- breast cancer patients (9). IGFBP2 expression was higher ported an IGFBP3 overexpression for premenopausal par- in ER-positive tumors, but not associated with overall sur- ous women. In women with nonmetastatic breast cancer, vival. We could not confirm the trend for an inverse effect high levels of circulating IGFBP3 predicted distant recur- ofIGFBP2expressiononsurvivalinpatientswithhor- rence exclusively in postmenopausal women with ER- mone receptor–negative tumors observed in the study by positive tumors (38). So and colleagues (9). Instead, we found that IGFBP2 ex- IGFBP2 is the second most abundant IGFBP in the cir- pression increased women's overall survival after adjust- culation. High IGFBP2 concentrations in blood or malig- ment for other breast cancer prognostic factors including

1030 Clin Cancer Res; 16(3) February 1, 2010 Clinical Cancer Research

IGFBP2, IGFBP3, and Breast Cancer Prognosis

BMI and treatment. Data on the independent effect of evidence point to a role of IGFBP2 in adipocyte biology IGFBP2, adjusted for other prognostic factors, is not avail- (47). Circulating IGFBP2 tends to be reduced in the pres- able from the article by So et al. (9) for comparison with ence of obesity (48–50). our findings. Frommer and colleagues (44) reported the In conclusion, none of the protein expressions were proapoptotic effects of IGFBP2 in the breast cancer cell strong independent predictors of overall survival in the line Hs578T, a result in line with the positive prognostic study population. But our study suggests that the ex- effect of IGFBP2 in our study. Furthermore, IGFBP2 was pression of IGF axis proteins is dependent on sex steroid found to interact with integrins to negatively modulate hormone–related tumor and patient characteristics. If IGFI-mediated migration and tumor growth (45). These the suggested interaction between IGFBP2 and obesity is results contrast with abundant experimental evidence confirmed in larger studies, the two factors would need showing IGFBP2 to promote cell proliferation and growth; to be considered in combination for prediction of patient in fact, Zhang et al. (46) recently identified IGFBP2 as a prognosis. potent stimulant of hematopoietic stem cell proliferation. Given the discordant effects of IGFBP2 on cell prolifer- Disclosure of Potential Conflicts of Interest ation and tumorigenesis (39), our results stress the relevance of studying the effect of IGFBP2 expression and D. Fink has received honoraria from Essex Oncology and is a consultant patient or tumor characteristics on patient survival in com- for Essex Oncology, Novartis, GSX, and MSD. bination. The interaction between overweight and IGFBP2 expression suggested in this study has not been reported Acknowledgments before. Clearly, the current study was underpowered for assessing effect modifications. This is evidenced by wide We thank Silvia Behnke and Martina Storz for their skillful technical CIs associated with HR estimates for the combined effect assistance, Norbert Wey for his help in preparing the photographic illustrations, M. Aerne for supporting database programming, and Dr. Ch. of overweight and IGFBP2 expression. Lack of power is al- Schindler for the statistical support. so a likely explanation for the wide overlap in CIs for the hazard of dying between subgroups of patients not be- Grant Support longing to the group of IGFBP2-positive/nonoverweigth women. The results are nevertheless presented to stimulate The study was supported by UBS AG (made possible by an anonymous confirmation in larger patient series, as this result is of donor). potential clinical interest. It suggests that prognostic pre- The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked dictions for obese women with breast cancer may be advertisement in accordance with 18 U.S.C. Section 1734 solely to improved by considering IGFBP2 expression. The ob- indicate this fact. served interaction between IGFBP2 and BMI is further- Received 4/15/09; revised 10/27/09; accepted 11/19/09; published more biologically plausible, given that several lines of OnlineFirst 1/26/10.

References 1. Majed B, Moreau T, Senouci K, Salmon RJ, Fourquet A, Asselain B. 9. So AI, Levitt RJ, Eigl B, et al. Insulin-like growth factor binding pro- Is obesity an independent prognosis factor in woman breast cancer? tein-2 is a novel therapeutic target associated with breast cancer. Breast Cancer Res Treat 2008;111:329–42. Clin Cancer Res 2008;14:6944–54. 2. Renehan AG, Frystyk J, Flyvbjerg A. Obesity and cancer risk: the 10. Yee D, Sharma J, Hilsenbeck SG. Prognostic significance of insulin- role of the insulin- IGF axis. Trends Endocrinol Metab 2006a;17: like growth factor-binding protein expression in axillary lymph node- 328–36. negative breast cancer. J Natl Cancer Inst 1994;86:1785–9. 3. Sachdev D, Yee D. Inhibitors of insulin-like growth factor signaling: a 11. Wang H, Arun BK, Wang H, et al. IGFBP-2 and IGFBP-5 overexpres- therapeutic approach for breast cancer. J Mammary Gland Biol Neo- sion correlate with the lymph node metastasis in T1 breast carcino- plasia 2006;11:27–39. mas. Breast J 2008;14:261–7. 4. Bonnette SG, Hadsell DL. Targeted disruption of the IGF receptor 12. Clemmons DR, Camacho-Hubner C, Coronado E, Osborne CK. In- gene decreases cellular proliferation in mammary terminal end buds. sulin-like growth factor binding protein secretion by breast carcino- Endocrinology 2001;142:4937–45. ma cell lines: correlation with estrogen receptor status. 5. Hovey RC, Harris J, Hadsell DL, Lee AV, Ormandy CJ, Vonderhaar Endocrinology 1990;127:2679–86. BK. Local I nsulin-like growth factor-II mediates prolactin-induced 13. Haffner MC, Petridou B, Peyrat JP, et al. Favorable prognostic value mammary gland development. Mol Endocrinol 2003;17:460–71. of SOCS2 and IGF-I in breast cancer. BMC Cancer 2007;7:136. 6. Firth SM, Baxter RC. Cellular actions of the insulin-like growth factor 14. Shin A, Ren Z, Shu X-O, Cai Q, Gao Y-T, Zheng W. Expression pat- binding proteins. Endocr Rev 2002;23:824–54. terns of insulin- like growth factor 1 (IGF-I) and its receptor in mam- 7. Wu Y, Cui K, Miyoshi K, et al. Reduced circulating insulin-like growth mary tissues and their associations with breast cancer survival. factor I levels delay the onset of chemically and genetically induced Breast Cancer Res Treat 2007a;105:55–61. mammary tumors. Cancer Res 2003;63:4384–8. 15. Chong YM, Williams SL, Elkak A, Sharma AK, Mokbel K. Insulin-like 8. Patel AV, Cheng I, Canzian F, et al. IGF-1, IGFBP-1, and IGFBP-3 growth factor 1 (IGF-1) and its receptor mRNA levels in breast can- polymorphisms predict circulating IGF levels but not breast cancer cer and adjacent non-neoplastic tissue. Anticancer Res 2006;26: risk: findings from the Breast and Prostate Cancer Cohort Consor- 167–73. tium (BPC3). PloS One 2008;3:e2578. 16. Happerfield LC, Miles DW, Barnes DM, Thomsen LL, Smith P, Hanby

www.aacrjournals.org Clin Cancer Res; 16(3) February 1, 2010 1031

Probst-Hensch et al.

A. The localization of the insulin-like growth factor receptor 1 (IGFR-1) D. Correlation of insulin-like growth factor-binding protein expres- in benign and malignant breast tissue. J Pathol 1997;183:412–7. sion in primary breast cancer tissues: detection of higher levels in 17. Mizukami Y, Nonomura A, Yamada T, et al. Immunohistochemical tumors with poor prognostic features. J Natl Cancer Inst 1996;88: demonstration of growth factors, TGF-α, TGF-β, IGF-I and neu on- 601–6. cogene product in benign and malignant breast tissues. Anticancer 34. Shivakumar S, Prabhakar BT, Jahashree K, Rajan MGR, Salimath Res 1990;10:1115–26. BP. Evaluation of serum vascular endothelial growth factor (VEGF) 18. Nardon E, Buda I, Stanta G, Buratti E, Fonda M, Cattin L. Insulin-like and microvessel density (MVD) as prognostic indicators in breast growth factor system gene expression in women with type 2 diabe- carcinoma. J Cancer Res Clin Oncol 2009;135:627–36. tes and breast cancer. J Clin Pathol 2003;56:599–604. 35. Schmid MC, Bisoffi M, Wetterwald A, et al. Insulin-like growth factor 19. Papa V, Gliozzo B, Clark GM, et al. Insulin-like growth factor-I recep- binding protein-3 is overexpressed in endothelial cells of mouse tors are overexpressed and predict a low risk in human breast can- breast tumor vessels. Int J Cancer 2003;103:577–86. cer. Cancer Res 1993;53:3736–40. 36. Ren Z, Shin A, Cai Q, Shu X-O, Gao Y-T, Zheng W. IGFBP3 mRNA 20. Schnarr B, Strunz K, Ohsam J, Benner A, Wacker J, Mayer D. Down- expression in benign and malignant breast tumors. Breast Cancer regulation of insulin-like g growth factor-I receptor and insulin recep- Res 2007;9:R2. tor substrate-1 expression in advancer human breast cancer. Int J 37. Guo S, Russo IH, Russo J. Difference in gene expression profile in Cancer 2000;89:506–13. breast epithelial cells from women with different reproductive history. 21. Voskuil DW, Bosma A, Vrieling A, Rookus MA, van't Veer LJ. Insulin- Int J Oncol 2003;23:933–41. like growth factor (IGF)-system mRNA quantities in normal and tumor 38. Goodwin PJ, Ennis M, Pritchard KI, et al. Insulin- like growth factor breast tissue of women with sporadic and familial breast cancer risk. binding protein 1 and 3 and breast cancer outcomes. Breast Cancer Breast Cancer Res Treat 2004;84:225–33. Res Treat 2002;74:65–76. 22. Ellis IO, Schnitt SJ, Sastre-Garau X, et al. Pathology and Genetics 39. Hoeflich A, Reisinger R, Lahm H, et al. Insulin-like growth factor- Tumours of the Breast and Female Genital Organs. IARC WHO Clas- binding protein 2 in tumorigenesis: protector or promoter? Cancer sification of Tumours, No 4 Edition. IARC Press; 2003. Res 2001;61:8601–10. 23. Kononen J, Bubendorf L, Kallioniemi A, et al. Tissue microarrays for 40. Renehan AG, Jones J, Potten CS, Shalet SM, O'Dwyer ST. Elevated high-throughput molecular profiling of tumor specimens. Nat Med serum insulin- like growth factor (IGF)-II and IGF-binding protein-2 in 1998;4:844–7. patients with colorectal cancer. Br J Cancer 2000;83:1344–50. 24. Torhorst J, Bucher C, Kononen J, et al. Tissue microarrays for rapid 41. Baron-Hay S, Boyle F, Ferrier A, Scott C. Elevated serum insulin-like linking of molecular changes to clinical endpoints. Am J Pathol 2001; growth factor binding protein-2 as a prognostic marker in patients 159:2249–56. with ovarian cancer. Clin Cancer Res 2004;10:1796–806. 25. Theurillat J-P, Zürrer-Härdi U, Varga Z, et al. NY-BR-1 protein ex- 42. Lin Y, Jiang T, Zhou K, et al. Plasma IGFBP-2 levels predict clinical pression in breast carcinoma: a mammary gland differentiation anti- outcomes of patients with high-grade gliomas. Neuro-oncol 2009;11: gen as target for cancer immunotherapy. Cancer Immunol 468–76. Immunother 2007;56:1723–31. 43. Mehrian-Shai R, Chen CD, Shi T, et al. Insulin growth factor-binding 26. Helle SI. The insulin-like growth factor system in advanced protein 2 is a candidate biomarker for PTEN status and PI3K/Akt breast cancer. Best Pract Clin Endocrinol Metab 2004;18:67– pathway activation in glioblastoma and prostate cancer. Proc Natl 79, Review. Acad Sci U S A 2007;104:5563–8. 27. Eppler E, Zapf J, Bailer N, Falkmer UG, Falkmer S, Reinecke M. IGF-I 44. Frommer KW, Reichenmiller K, Schutt BS, et al. IGF-independent ef- in human breast cancer: low differentiation stage is associated wtih fects of IGFBP-2 on the human breast cancer cell line Hs578T. J Mol decreased IGF-I content. Eur J Endocrinol 2002;146:813–21. Endocrinol 2006;37:13–23. 28. Hamelers ICH, Steenbergh PH. Interactions between estrogen and 45. Pereira JJ, Meyer T, Docherty SE, et al. Biomolecular interaction of insulin-like growth factor signalling pathways in human breast tumor insulin-like growth factor (IGF) binding protein-2 with ανβ3 negatively cells. Endocr Relat Cancer 2003;10:331–45. modulates IGF-I mediated migration and tumor growth. Cancer Res 29. Koda M, Przystup W, Jarzabek K, et al. Expression of insulin-like 2004;64:977–84. growth factor-I-receptor, estrogen receptor α, Bcl- 2 and Bax pro- 46. Zhang CC, Kama M, Iizuka S, Huynh HD, Lodish HF. Angiopoietin- teins in human breast cancer. Oncol Rep 2005;14:93–8. like 5 adn IGFBP2 stimulate ex vivo expansion of human cord blood 30. Vestey SB, Perks CM, Sen C, Calder CJ, Holly JMP, Winters ZE. Im- haematopoietic stem cells as assayed by NOD/SCID transplantation. munohistochemical expression of insulin-like growth factor binding Blood 2008;111:3415–23. protein-3 in invasive breast cancer and ductal carcinoma in situ: im- 47. Wheatcroft SB, Kearney MT, Shah AM, et al. IGF- binding protein-2 plications for clinicopathology and patient outcome. Breast Cancer protects against the development of obesity and insulin resistance. Res 2005;7:R119–29. Diabetes 2007;56:285–94. 31. Yu H, Levesque A, Khosravi MJ, Papanastasiou-Diamandi A, Clark 48. Carmichael AR. Obesity and prognosis of breast cancer. Obes Rev GM, Diamandis EP. Insulin-like growth factor-binding protein-3 and 2006;7:333–40. breast cancer survival. Int J Cancer 1998;79:624–8. 49. Ahmed RL, Thomas W, Schmitz KH. Interactions between insulin, 32. Yu H, Levesque MA, Khosravi MJ, Papanastasiou-Diamandi A, Clark body fat, and insulin-like growth factor axis proteins. Cancer Epide- GM, Diamandis EP. Associations between insulin-like growth factors miol Biomarkers Prev 2007;16:593–7. and their binding proteins and other prognostic indicators in breast 50. Wheatcroft SB, Kerney MT. IGF-dependent and IGF-independent ac- cancer. Br J Cancer 1996;74:1242–7. tions of IGF- binding protein-1 and -2: impications for metabolic ho- 33. Rocha RL, Hilsenbeck SG, Jackson JG, Lee AV, Figueroa JA, Yee meostasis. Trends Endocrinol Metab 2009;20:153–62.

1032 Clin Cancer Res; 16(3) February 1, 2010 Clinical Cancer Research

IGFBP2 and IGFBP3 Protein Expressions in Human Breast Cancer: Association with Hormonal Factors and Obesity

Nicole M. Probst-Hensch, Julia H.B. Steiner, Peter Schraml, et al.

Clin Cancer Res 2010;16:1025-1032. Published OnlineFirst January 26, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-09-0957

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2010/01/26/1078-0432.CCR-09-0957.DC1

Cited articles This article cites 49 articles, 15 of which you can access for free at: http://clincancerres.aacrjournals.org/content/16/3/1025.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/16/3/1025.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/16/3/1025. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.