Breast cancer risk in papilloma patients: Osteopontin splice variants indicate prognosis

This study investigated biopsies from a total of 114 women with premalignant breast lesions and follow-up information over 5–11 years, comprising 49 patients from Wroclaw, Poland, and 65 patients from Cincinnati, USA. In addition, 7 tissues from Wroclaw were stained but either had no follow-up information or encountered death from other causes and were therefore excluded from evaluation. These cases comprise all papillomas with the necessary follow-up information that were accessible to the authors from both study sites. To assure uniformity of the analysis, the tissues were stained and read by two pathologists in the same institute, and the initial diagnoses were confirmed by them. All the evaluations were made based on international standards and classifications (specifically, College of American Pathologists [CAP] for USA and the standards of the Polish Society for Pathologists, which were also based on CAP outlines). The research project was approved by the ethics committees at Wroclaw Medical University, Poland, and at the University of Cincinnati, USA.

Previous immunohistochemistry studies of Osteopontin-c, reported by us and others [17‐23], were conducted with the polyclonal chicken IgY that had been generated by us and made available to other researchers through Gallus Immunotech, later Exalpha Biologicals. The polyclonal rabbit antibody LF161 (obtained from Larry Fisher, NIH) served as a marker for Osteopontin exon 4. Here, we use monoclonal antibodies. Exon 4 is recognized by the mouse monoclonal antibody MAB193P (obtained from BBI Solutions). Using the immunizing peptide ac-SEEKQ|NAVS (the vertical line marks the Osteopontin-c splice junction) coupled to KLH via a C-terminal cysteine, we generated 3 clones of rabbit monoclonal anti-Osteopontin-c. All of the monoclonal antibodies displayed strong and selective reactivity with their target antigen, and they showed neutralizing activity in soft agar colony formation. Dissimilar to the polyclonal antibodies, the monoclonal OPN-c stain located predominantly to the cytoplasm, while the exon 4 stain was nuclear and cytoplasmic (Fig. 1).

For each antibody, formalin-fixed and paraffin-embedded biopsy specimens from papillomatous lesions were cut on a microtome. The antibodies used in this study, after blocking endogenous peroxidase for 10 min with DAKO Peroxidase Blocking Solution and non-specific proteins for 15 min with DAKO Protein-Block Serum-Free, were monoclonal anti-OPN-c rabbit Ig clone F2 (Georg F. Weber recognizes the Osteopontin-c splice junction), and murine monoclonal anti-OPN-exon-4 Ig MAB193P (BBI Solutions recognizes OPN-a and OPN-b, but in the breast is indicative of OPN-a because OPN-b is absent). The primary antibodies were applied at 1.25 μg/ml (F2) or 2.2 μ/ml (MAB193P) dilution for 30 min followed by DAKO EnVision + System HRP-labeled polymer for 30 min. Color development was achieved with DAB + chromogen from DAKO for 5 min and consecutive counterstaining with hematoxylin for 1 min. For each antibody, the tissues were scored according to staining intensity (0, 1, 2, or 3) and percent positivity (0, 1, 2, or 3). In addition to analyzing the indicators in their original scale, we dichotomized the immunohistochemical biomarkers into low (0–1) or high (2–3). We have previously found this dichotomizing to strengthen the power of the analysis. Further, it enables the development of a clear-cut risk score [17, 20]. In those predecessor studies, a cutoff between 0 and 1 was found to place many low-risk patients into the high group, whereas a cutoff between 2 and 3 would have missed a non-trivial portion of patients, who later developed cancer. All microscopic slides were independently evaluated by two pathologists, and in the rare cases of discrepant initial scores, a final score was agreed on after discussion.

The pathology scores assess staining intensity and percent positivity. The predictors were each categorical (pathology scores 0–3) or dichotomized (pathology scores low versus high). Statistical analysis was done with the t test and calculation of odds ratios with confidence intervals.

The papillary breast lesions encompass a spectrum of masses, which present as fronds attached to the inner mammary duct wall by a fibrovascular core with both epithelial and myoepithelial cells; although not malignant, papillary disease is associated with a risk of progression to invasive breast cancer [1, 24, 25]. From the patients seen at the two locations involved, 114 paraffin blocks and follow-up information were available (Table 1). These women grouped into those who did not experience cancer over at least 5 years, those who progressed to cancerous lesions in the same breast (12 patients or 10.5%, slightly higher than most literature estimates for progression risk), and those who later encountered cancer in the contralateral breast (6 patients or 5.3%).

All lesions had two sections prepared for staining with monoclonal antibodies to either OPN-c or OPN-exon-4 (Fig. 2). The signal for Osteopontin-c was cytoplasmic, whereas Osteopontin exon 4 stained both cytoplasm and nucleus. In each case, only the papillomatous lesion was scored for intensity and percent positivity.

The pathology scores covered a range, with none of the patients who later experienced cancer in either breast having intensity or percent positivity of 0. The OPN-c intensity scores were significantly higher in women who experienced cancer at a later time than in women who remained free of progression. This was the case for a two-tailed t test (assuming equal variance) (p = 0.041) and for a one-tailed t test assuming that in progression the direction of change is an increase in the pathology score (p = 0.020). The exon 4 staining contributed very little, with the staining intensity not differing significantly among the patient groups. Only the percent positivity was marginally significantly elevated (p = 0.024, two-tailed t test, but not significant in the one-tailed t test) for patients with later cancer in the other breast (Fig. 3).

Previously [17], we found it helpful to develop a risk score, in which we added the pathology scores for intensity and percent positivity followed by ranking into low, intermediate, or high for each marker. We could then generate a combined score for several markers (specifically OPN-c and OPN-exon-4) using the analogous process of addition followed by categorization. About 75% of women who later developed cancer in the same breast had high OPN-c staining at the time of papilloma biopsy; for later cancer in the other breast this percentage was over 80. OPN-exon-4 immunohistochemistry was only marginally discriminating for cancer in the same breast, but may have a substantially reduced fraction of low-risk scores when cancer is later experienced in the contralateral breast (Fig. 4, for various cutoffs, see Additional file 1: Figure S1). For low, intermediate, or high additive OPN-c pathology scores, the cancer risk increased 3.2%, 5.7%, 18.8% for the same breast and 3%, 0%, 10% for the contralateral breast. A risk progression was less clear for the staining with anti-OPN-exon-4 antibody or when merging the two markers into a combined risk score (Table 2). The relative risk for progression in the ipsilateral (RR 4.043, CI 95% 1.159–14.109) or contralateral (RR 7.143, CI 95% 0.866–58.946) breast was significantly elevated in the high versus low comparison for OPN-c intensity (Table 3; the relative risk by pathology score is shown in Additional file 1: Table S1). The overall incidence of progression for papilloma is about 10.5% (and for cancer in the other breast about 5.3%).

In multi-year follow-up, patients with mammary papilloma have sometimes displayed cancerous lesions in the contralateral breast [10]. In this study, highly staining lesions were predictors for contralateral breast malignancy (Fig. 4, Table 2), although over a longer follow-up (8 ± 2 years, mean ± range). This may suggest that there are genetic factors, which predispose to elevated Osteopontin splice variant expression. The resulting abundance of Osteopontin variants increases the progression risk, when early transformed lesions occur in either breast.