Preoperative serum HER2 extracellular domain levels in primary invasive breast cancer

Human epidermal growth factor receptor-2 (HER2) overexpression or amplification in tumor tissue is reportedly observed in 20–30% of primary breast cancers. Tissue HER2 status is now routinely used as an important parameter for decision making regarding anti-HER2 therapy in the neoadjuvant [1], adjuvant [2], and metastatic [3] settings. The presence of HER2 in tumor tissues also adds predictive and prognostic information regarding shorter overall and disease-free survivals [4], preferential benefit from doxorubicin [5‐7], and possible resistance to tamoxifen [8‐10]. However, despite the continuous and varying degrees of tissue HER2 expression, tumors are arbitrarily dichotomized into HER2-positive and HER2-negative groups based on immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) [11]. As a result, the definition of tissue HER2 positivity remains a matter of controversy [12‐14]. In addition, there is recent evidence that some tissue HER2-negative patients appear to benefit from trastuzumab [12, 15].

Serum HER2 extracellular domain (ECD) can be detectable by the cleavage of transmembrane tissue HER2 protein which is probably mediated by a matrix metalloproteinase [16‐18]. Given the same origin of each HER2 subcomponent in tissue and blood, the technical problems inherent in current tissue HER2 testing methods have urged the exploration of clinical utilities of serum HER2 ECD [19]. The quantification of serum HER2 ECD in primary breast cancer could theoretically enhance the sensitivity of tissue HER2 testing in the minority population in which HER2 expression is significant, but is not sufficiently high to be considered as HER2-positive by the American Society of Clinical Oncology (ASCO) and College of American Pathologists (CAP) guidelines, possibly due to tumor heterogeneity [11, 20‐22]. In the metastatic setting, the findings of some studies further suggest that HER2 ECD levels and changes therein reflect the patients’ responses to antiestrogen therapy [23, 24], chemotherapy [25, 26], and trastuzumab [27, 28]. There have been a few positive results in the adjuvant setting, similar to the metastatic setting [29‐33], but most studies have addressed the relationship between abnormal HER2 ECD levels and tissue expression positivity with contradictory results. The current consensus is that although preoperative HER2 ECD appears to be correlated with tumor size and nodal involvement, it may not be related to tissue HER2 status, especially in primary breast cancer, and there is insufficient evidence to support incorporating measurement of this parameter into the routine clinical management of women with breast cancer as an independent prognostic factor [14, 34].

The aims of the present study were to establish the prevalence of elevated HER2 ECD levels, and determine whether there is an association between serum HER2 ECD levels and tissue HER2 overexpression in a large number of patients with primary breast cancer, using a single assay approved by the US Food and Drug Administration (FDA). Furthermore, we sought to elucidate differences in any such association between various subgroups of patients, as reported previously [35, 36], and to determine the prognostic usefulness of HER2 ECD and derive evidence for verifying previous equivocal reports [14, 34].

Patients who were diagnosed with and had surgery for stage I–III primary breast cancer at Asan Medical Center between January 2007 and December 2009 were enrolled for this study (n = 2,862). Patients with distant metastasis at the time of diagnosis, bilateral cancer, and the initial plan of neoadjuvant systemic therapy were excluded. All of the patients’ information and tumor characteristics were retrieved from our prospectively collected database. The size of the tumor, regional lymph node (LN) status, histologic grade, nuclear grade, presence of lymphovascular invasion (LVI), histological subtype, and immunohistochemical status of estrogen receptors, progesterone receptors, and HER2 were determined at our institute by pathological analysis on formalin-fixed paraffin-embedded tissue sections of the primary tumor excised at the time of definitive surgery. Tumor staging followed the tumor-node-metastasis classification of the 7th American Joint Committee on Cancer [37]. This study was reviewed and approved by the Institutional Review Board of Asan Medical Center (20141162).

The present study found that serum HER2 ECD level was correlated with a poor prognosis in primary breast cancer, and that patients with an elevated HER2 ECD level (>15.2 ng/ml) had a worse DMFS and BCSS than those with normal HER2 ECD levels. In particular, it was confirmed that its role as an independent prognostic factor was clinically robust at various tentative cutoff values in patients with positive tissue HER2 status.

There is still no general consensus as to a definitive dichotomizing HER2 ECD cutoff value for clinical use, so we adopted the most commonly used value of 15.2 ng/ml to define elevated HER2 ECD. HER2 ECD levels were previously reported to be elevated in 11.4% (range, 3.1–34.1%) of primary breast cancer patients and 36.5% (range, 23–62%) of metastatic breast cancer patients [34]. In the present study, elevated HER2 ECD levels were observed in 4.4% (126/2,862) of the primary breast cancer patients, which is a noteworthy finding because it was derived from a large number of patients at a single institute and using a single US FDA-approved assay. In addition, as shown in Table 1, significant differences in the prevalence of elevated HER2 ECD levels between the subgroups were confirmed, which is in accordance with previous results [30, 36]. And, other studies have suggested that setting a lower cutoff value such as 7.7 ng/ml and 10.2 ng/ml or analyzing early stage patients could come with a better outcome, however our additional study adjusted with these did not show a better result (Figure 7). However, it should be noticed that, our analysis did not address the issues on more optimal cutoff values for HER2 ECD raised by previous studies [31, 38, 39].

Interestingly, elevated serum HER2 ECD levels were detected in 22 (1.0%) of the 2,168 tissue HER2-negative patients (26.4 ± 35.6 ng/ml; range, 15.3–184.4 ng/ml). Possible explanations for this are chance false-positivity, normal elevation of serum HER2 ECD level as reported in healthy controls [40], a minority of HER2-positive cells being lower than the definition of tissue HER2 positivity as its source [41], and genetic differences between individuals with respect to matrix metalloproteinase activity, which is responsible for HER2 release into the serum [39]. Since metastasis had occurred in only 1 of the 22 patients in this subgroup, the clinical implication of unexpected serum ECD elevation could not be further investigated.

In this study we observed a significant association between elevated HER2 ECD level and the parameters of tumor aggressiveness (Table 2). More patients had undergone mastectomy among those with an elevated HER2 ECD levels than in those with a normal HER2 ECD level. This might be caused by the tumor characteristics being more aggressive in the former subgroup than in the latter, the cause of which was not fully established.

The association between serum HER2 ECD concentration and tissue HER2 status remains controversial in primary breast cancer, which may be attributable to the use of different cutoff values, small samples, and different patient populations in the various studies. For now it seems reasonable to suggest that serum HER2 ECD levels cannot substitute for tissue HER2 expression, but can provide additional information. The aim of the present study was to compare the concordances according to the subgroups and provide practical information when we suspected the inherent imperfection of tissue HER2 results in the case of elevated serum HER2 ECD levels. We applied two cutoff values: 15.2 ng/ml, which has been approved by the US FDA; and 10.2 ng/ml, which has been suggested as a more appropriate cutoff value for the Korean population [31]. Although a lower cutoff value could yield a higher sensitivity and negative predictive value (Table 3), it negatively affected the specificity and positive predictive value. In our opinion, in view of the adjunctive role of serum HER2 ECD level, the cutoff of 15.2 ng/ml may have greater clinical utility because higher specificity and positive predictive values are mandatory to reduce the likelihood of false suspicion of tissue HER2 status due to the abnormal serum HER2 ECD level. Unlike previous studies [35, 36], the present findings suggest that significant correlations between the expression of tissue HER2 and serum HER2 ECD level exist in primary breast tumor, regardless of whether the primary tumor is early or advanced (Table 3). However, the finding that AUCs increased as the indexes of tumor extent increased should not be overlooked (Figure 3).

Despite the relatively short follow-up, the inclusion of a large sample enabled us to demonstrate the poor prognostic role of HER2 ECD level. Moreover, we determined that tissue HER2 status had no prognostic impact on either DMFS or BCSS in patients with a normal serum HER2 ECD level. Thus, the serum HER2 ECD level can be a valuable prognostic factor in patients with primary breast cancer with tissue HER2 overexpression (Figure 6 and Table 4). The persistent adverse prognostic value of elevated HER2 ECD levels in stage I/II early breast cancer could reflect the presence of micrometastases or high rates of HER2 cleavage and shedding, with the production of truncated cell-associated fragments that contain the signaling kinase domain that is activated in the absence of the ECD. As a result, these tumors with a deregulated growth-promoting pathway could behave more aggressively [30].

To clarify the possibility of false-positive results in the present study in terms of prognosis according to the arbitrary cutoff value of 15.2 ng/ml, we validated its prognostic role among continuous subgroups classified relative to the quartile values of serum HER2 ECD levels in tissue HER2-positive tumors: the lowest-quartile, median, and highest-quartile values were 8.7, 10.6, and 13.1 ng/ml, respectively. The Kaplan-Meier curves (Figure 6C, D) and multivariate analysis revealed that the third- and highest-quartile subgroups independently had an ominous prognosis in terms of DMFS (third quartile: HR = 3.029, 95% CI = 1.003–9.152, P = 0.049; highest quartile: HR = 4.746, 95% CI = 1.591–14.160, P = 0.005; Table 4).

The possibility of an association between HER2 ECD level and trastuzumab response was assessed in 264 patients receiving trastuzumab and in 428 patients without trastuzumab administration. We found that HER2 ECD level was a poor prognostic indicator, irrespective of the treatment in terms of DMFS (Figure 4F). In terms of BCSS, there did not appear to be a significant prognostic role of elevated HER2 ECD level in the trastuzumab subgroup (Figure 5F). This lack of a significant finding may be attributable to the relatively short follow-up. Also, only 38.2% (264/692) of patients with HER2-positive tumors had been treated with trastuzumab in the cohort. It is partly related to the inclusion of patients with small tumors. Our study did not include patients receiving neoadjuvant systemic therapy. More importantly, the use of trastuzumab in the adjuvant setting was just covered for advanced breast cancer by the Korean national health insurance during the period of this study (2007 ~ 2009). In patients with tissue HER2 positive, T1 was 54.5% (377/692), and N0 was 54.0% (374/692), which needs to be taken into consideration. Due to the nature of a retrospective analysis without randomization, it was not possible to compare the responses or benefits of trastuzumab among the subgroups of 261 patients who received this drug with elevated (n = 53) and normal (n = 208) HER2 ECD levels.

While the present study is subject to the same limitations as other retrospective studies, it provides robust evidence supporting the valuable clinical utility of HER2 ECD level, drawn successfully because of the inclusion of a large sample. The presented results indicate that a preoperatively elevated HER2 ECD level reflects tumor extension, with significantly higher values being found in patients with larger tumors, with LN metastasis, or with LVI, and that patients with an elevated HER2 ECD level are more likely to develop distant metastasis irrespective of their disease status.

Given the prevalence of HER2 ECD elevation, preoperative serum HER2 ECD measurement can be clinically useful in patients with tissue HER2-positive primary breast cancer.