Differential expression of VEGFR2 protein in HER2 positive primary human breast cancer: potential relevance to anti-angiogenic therapies

The study population included a retrospective series of 186 female patients, including 171 Caucasian, 11 African American and 4 others with node-positive primary breast cancers (89 left, 90 right and 5 bilateral). Mean patient age was 59 years (range 33–88 years). Most of these patients received local radiation and chemotherapy, including Adriamycin, Cytoxan and 5-FU, as previously described [21]. Clinico-pathologic data were collated from the Yale Tumor Registry in accordance with the guidelines of the Yale Human Investigations Committee.

Mean primary breast cancer size was 3.4 cm (range 0.15–14.5 cm). Primary human breast carcinoma tissues (N = 186) were classified into invasive ductal, lobular, mixed ductal-lobular and mucinous carcinomas (N = 139, 22, 18, 7 respectively), based on the original pathologic evaluation at Yale University. Using the Nottingham Modification of the Scarff-Bloom-Richardson grading system, also known as the Nottingham Grading System (NGS) [22], the invasive carcinoma tissues (N = 186) were categorized into grade 1 (N = 8), grade 2 (N = 109), grade 3 (N = 69). Representative formalin-fixed, paraffin-embedded (FFPE) tumor tissue from each case was sampled as a single 0.6 mm core in a recipient tissue microarray block (Yale BRCA, YTMA 10), on a tissue-arraying instrument (Beecher Instruments, Silver Springs, MD). Sampling of human tissues in the Yale BRC TMA was based on the required institutional policies and approvals, including the patient consent to allow usage of tissue for research. Using the latest criteria proposed by the World Health Organization for Histologic Typing of Breast Tumors [23], all original pathologic diagnoses were confirmed on Hematoxylin & Eosin stained section of the Yale BRCA TMA (YTMA 10) by an experienced American Board-certified study pathologist (AN) with subspecialty expertise in breast pathology.

Five micron thick FFPE TMA sections were cut from the Yale Breast Cancer TMA above, stored in nitrogen chamber to prevent loss of antigenicity until immunostained for VEGFR2 and CD34, a sensitive IHC marker for tumor vasculature. For VEGFR2 protein we used a technically robust, sensitive, specific and selective immunohistochemical (IHC) assay developed and optimized by our laboratory (13) that had showed optimal performance on several different human tumor cohorts, including multiple human tissue and cell lines controls (14–17, 19). The IHC assay protocol, using one of the most specific commercially available monoclonal anti-VEGFR2 antibody (55B11) [24], optimization experiments and quality control procedures were previously described in detail [14]. For CD34, we used a technically validated IHC assay offered by a leading reference laboratory (Clarient, Aliso Viejo, CA, USA), including satisfactory positive and negative controls.

Unequivocal, crisp VEGFR2 immunoreactivity was demonstrated in the vascular endothelium but not in trophoblastic cells in the conventional sections of human placenta and also in the microvasculature of the invasive cervical squamous cell carcinoma favoring these tissues as optimal positive and negative tissue controls. Optimal reagent negative controls were run by replacing the primary antibody with control immunoglobulin.

As part of the analytical validation of the VEGFR2 IHC assay, the coefficients of variation (CVs) of immuno-pathological VEGFR2 scores for intra-run repeatability, inter-run reproducibility, and inter-observer reproducibility were less than 7% (data on file).

After immuno-pathologic review, each immuno-stained BRC TMA section was evaluated by the sub-specialty pathologist (AN), who was blinded to the BRC subtypes or other relevant clinico-pathologic or breast marker data. In each analyzable TMA core, the discrete VEGFR2+ and CD34+ tumor vessels were manually counted in the invasive tumor stroma. In order to qualify for a vascular structure, it had to have the histomorphologic appearance of a vessel with or without lumen. Scattered individual cells in the invasive cancer tissue stroma not conforming to the strict definition of a vessel above were excluded from VEGFR2+ and CD34+ vessel counts. Any suboptimal/inadequate TMA cores (cores with complete or major [>50%] tissue loss/fragmentation; those without well-preserved, viable tumor cells or those with any areas of tumor necrosis) were also excluded from scoring/analysis. After such exclusions, a total of 186 and 169 TMA cores were found to be adequate for manual assessment of CD34+ and VEGFR2+ vessels in the stromal component of the primary invasive carcinoma tissues sampled (Table 2).

Photomicrographic images representing the differential levels of immunohistochemical expression of VEGFR2 in tumor stromal vasculature in HR+, HER2+ and triple-negative BRCs were captured from high-resolution digital scans of the stained TMA slides (Scanscope XT; Aperio Technologies, Vista, CA).

Based on the breast marker IHC panel results (ER, PR and HER2) from the contributing institution (Yale University, New Haven, CT, USA), as used in the standard management of breast cancer patients, each case was grouped into one of the three BRC subsets, i.e., hormone receptor+ (HR+), HER2+ and triple-negative (TNBC).

Counts of VEGFR2+ tumor vessels from 3 breast cancer subtypes (HR+, HER2+, and TNBC) were compared for all pairs using Tukey–Kramer HSD procedure (JMP 12.1.0, SAS Institute Inc.). Because the distribution of the data was heavy-tailed to the right, various transformations were tried to normalize the data. However, none of the transformations normalized the data satisfactorily. Hence, zero-inflated Poisson regression was used to account for excessive zeroes in the data [25, 26]. Counts of VEGFR2+ tumor vessels from 2 aggregated breast cancer subtypes (HER2+ vs. HER2−) were compared using t test and zero-inflated Poisson regression.

The observed CD34+ and VEGFR2+ tumor vascular counts in individual breast cancer cases were heterogeneous. Overall, the BRC cases analyzed had larger numbers of CD34+ tumor stromal vessels per TMA core (mean 11; range 0–45) as compared to VEGFR2+ tumor stromal vessels per TMA core (mean 3.4; range 0–20). Also, 89 of 186 evaluable TMA cores exhibited more than 10 CD34+ vessels/core, while only 8 of 169 evaluable TMA cores had more than 10 VEGFR2+ vessels/core (Table 2), implying that only a proportion of CD34+ tumor stromal vessels co-expressed VEGFR2 protein in their endothelial lining.

Overall, the levels of vascular expression of VEGFR2 were relatively low in histologically characterized breast cancer tissues. Of 164 breast cancer cases including all histologic types, 127 (77.4%) had a few (<5) or no VEGFR2 positive tumor vessels in the tissue sampled in the TMA cores (Table 3), while 37 (22.4%) showed intermediate or high vascular expression of VEGFR2. among the various molecular BRC subtypes, however, VEGFR2+ tumor stromal vessel counts were significantly higher in HER2+ (mean 6.1 [sd 5.5], median 6) as compared to HR+ (mean 3.2 [3.3], median 3, p = 0.04) and triple negative BRCs (mean 3.0 [3.6], median 2, p = 0.02) tissues (Figs. 1, 2, 3, 4, 5). There was no significant difference between VEGFR2+ tumor stromal vessel counts between HR+ and triple negative BRCs (p = 0.69). As compared to HER2+ breast cancer cases illustrated in Fig. 4, in which many of the tumor stromal vessels localized by CD34 immunoreactivity (a–d; right panels) were also VEGFR2+ (a–d; left panels), the HR+ breast cancers illustrated in Fig. 3, despite showing frequent localization of CD34+ tumor stromal vessels (a–d; right panels), only showed an occasional VEGFR2+ vessel in the tumor stroma (a–d; left panels, black arrows). Similarly, compared to HER2+ breast cancer cases illustrated in Fig. 4, in which many of the tumor stromal vessels localized by CD34 immunoreactivity (a–d; right panels) were also VEGFR2+ (a–d; left panels), the TNBCs illustrated in Fig. 5, despite showing frequent localization of CD34+ tumor stromal vessels (a, b; right panels), only showed an occasional VEGFR2+ vessel in the tumor stroma (a, b; left panels, black arrows).

Based on CD34+ and VEGFR2+ vascular counts in various human breast cancer subtypes, cases were ranked from negative to low (1–5) to intermediate (6–10) to high (>10) VEGFR2+ vessel counts (Fig. 2). Compared to HR+ BRCs and TNBCs, a greater proportion of HER2+ BRC cores had higher numbers of VEGFR2+ tumor vessels (Fig. 2a–c). Also, compared to HER2-negative BRCs, HER2+ BRCs had significantly higher VEGFR2+ tumor vessels count (p = 0.007). In mucinous carcinoma, weak VEGFR2 staining was found in an occasional tumor stromal vessel.

Thirteen of 169 (8%) cases also showed tumor cell (cytoplasmic and/or membrane) expression of VEGFR2 protein.