Introduction

Gastric cancer is among the most frequent cancers in the world. Although the incidence is decreasing, approximately 990,000 new cases and 738,000 deaths occurred in 2008 [1, 2]. In China, where the patients in this study were recruited, gastric cancer currently ranks the third most common cancer. Despite new treatment strategies, such as perioperative chemotherapy or adjuvant chemoradiation using external radiotherapy, the prognosis for patients with advanced gastric cancer remains poor, and new therapeutic approaches are needed. HER2 is an attractive target in exploring new strategies for anticancer therapy and is a member of the EGFR family of receptors and can, together with EGFR, HER3, and HER4, form dimers important for tumor growth and formation of metastases [3, 4]. Several therapy agents binding to HER2 are designed to interact with intracellular signaling and give therapy effects [5–7].

Trastuzumab (Herceptin™) is a humanized, recombinant monoclonal antibody targeting the HER2 extracellular domain and is well known for breast cancer therapy and is nowadays also considered for therapy of gastric cancer [8–11]. The ToGA trial (Trastuzumab for Gastric Cancer) was the first study evaluating the efficacy and safety of trastuzumab in combination with chemotherapy for patients with HER2-positive advanced gastric cancer. The ToGA trial showed survival benefit over chemotherapy alone [8, 10], and this proved that knowledge of the HER2 status is important for identifying gastric cancer patients who may benefit from HER2-targeted therapy. However, biological resistance to HER2-targeted therapies, due to mutations in intracellular signal pathways or other reasons [12–14], has been reported for breast cancer [15, 16], and such resistance seems to appear also in metastasized gastric cancer, and resistance against both the tyrosine kinase inhibitor lapatinib [17, 18] and the antibody trastuzumab [9, 19, 20] has been reported. It might therefore, as an alternative or complement, be beneficial to target the extracellular domain of HER2 in metastatic gastric cancer patients with radionuclides for HER2 imaging, dose planning, and therapy. Examples of suitable radionuclides for these applications are given in the “Discussion.”

The radionuclides can be delivered with, for example, antibodies, antibody fragments, affibody molecules, and peptides [21–24]. The application of radionuclide-labeled molecules for HER2-targeted therapy has so far, to the knowledge of the authors, not been clinically applied for metastatic gastric cancer. Radionuclides binding to the extracellular domain of HER2 give an absorbed dose that depends on HER2 expression but should not, or to a minor extent, be dependent on intracellular mutations, and we therefore decided to neither analyze mutations in signal pathways nor carry out analysis of gene amplifications.

Although HER2-targeted therapy is designed for HER2-expressing metastatic disease, the pathological HER2 testing is mainly carried out on the primary lesion. Metastases have often formed when gastric cancer is diagnosed and traditional therapies start, and information is needed on HER2 expression in these metastases. There are at least a few previous publications indicating rather good agreement when primary and metastatic lesions were compared considering HER2 expression [25–29] and gene amplification [26–29] with only a few discordant cases (in the range from zero up to about 10 %). Discordance could be due to genomic heterogeneity in the primary tumors and/or in the metastases giving nonrepresentative sampling [27]. In the present study, HER2 expression was investigated in a series of only 29 gastric cancers with synchronous samples from primary tumors and metastases from lymph nodes and liver. All samples for each patient were taken at the same occasion, and none were previously treated with radiotherapy, chemotherapy, or HER2-targeted therapy. To our knowledge, such samples have not been previously evaluated.

We analyze and discuss in this article whether HER2 is expressed so often and, with such good concordance between primary tumors and corresponding metastases that targeted radionuclide therapy, can be an alternative or complement to other modalities in the treatment of metastatic gastric cancers.

Material and methods

Patients

Gastric cancer patients with primary lesions paired with synchronous lymph node and liver metastases were diagnosed and treated in the Second Affiliated Hospital, Zhejiang University School of Medicine, in 2000–2007. The time span is wide since patients from which we could get material from primary tumors and both lymph node and liver metastases at the same day (within a few hours) were rare, and the collection took long time. The patient samples were not analyzed regarding HER2 expression until now because the documentation on resistance to HER2-targeting agents with “biological action” (trastuzumab and lapatinib) has only recently been published [9, 17–20]. The patients were included under the approval of the Institutional Review Board, and informed consent was obtained in accordance with the institutional guidelines. Patients who had received preoperative gastric radiotherapy or preoperative systemic chemotherapy were excluded as well as patients who had received HER2-targeted therapy. From a large number of patients (totally 1,982 gastric cancer patients were considered), only 30 could initially be considered for the study. However, in one case, there were no tumor cells in the sections supposed to be from a liver metastasis, giving 29 remaining cases with high-quality material of both primary tumors and the two types of metastases. All samples for each patient were taken by surgical resections at the same time, i.e., within a few hours. Hospital records were reviewed for clinical information; see Table 1. The ages of the patients scoring HER2 positivity and HER2 negativity were 64.2 ± 7.9 and 61.7 ± 10.1 years, respectively. Clinical pathologic staging, tumor node metastasis (TNM), of the primary tumors was determined according to the International Union Against Cancer [30], histologic tumor type according to Lauren’s classification [31], and differentiation grade according to the WHO classification [32].

Table 1 Clinicopathological parameters and HER2 expression for the primary gastric tumors (n = 29)
Full size table

HER2 staining

The HER2 staining was made as previously described [33]. The tissues were fixed in 4 % buffered formalin, embedded in paraffin, and sectioned (4-μm thick). The sections were deparaffinized in xylene and hydrated through decreasing concentrations of ethanol, ending with distilled water. They were thereafter incubated 30 min in methanol and hydrogen peroxide for quenching of endogenous peroxidase. Antigen retrieval was done in water solution (98 °C, pH 6.0) for 40 min. The slides were then cooled at room temperature and washed in distilled water. Immunohistochemical staining was performed with the Elite ABC Kit (Vectastain, Vector Laboratories, Burlingame, CA, USA). A Blocking serum was applied for 15 min followed by incubation with rabbit antihuman c-erbB-2 oncoprotein (code no. A-0485, Dako, Denmark A/S, Glostrup, Denmark) diluted in 1:350. The sections were then incubated with the biotinylated secondary antibody and visualized with peroxidase substrate 3-amino-9-ethyl-carbazole (AEC) (Sigma A-5754, Sigma-Aldrich, Shanghai, China) as chromogen. Finally, the sections were counterstained with Mayer’s haematoxylin and mounted.

HER2 scores

A HER2 scoring system recommended for gastric cancer was used and has been described and reviewed by Hofmann et al. [34] and is well known by those considering HER2-targeted therapy of gastric cancer. The system considers basolateral, so-called U-shaped, HER2 expression as positive. HER2 score 0 corresponded to no staining, or membranous staining in <10 % of the tumor cells; 1+ was faint perceptible membranous staining in ≥10 % of the tumor cells, stained only in parts of their membranes; 2+ was weak to moderate complete or basolateral membranous staining in ≥10 % of the tumor cells; and 3+ was moderate to strong complete or basolateral membranous staining in ≥10 % of the tumor cells. Cytoplasmic staining was considered nonspecific and not included in the scoring. In-house positive control tissue sections and positive control sections supplied by DAKO (Denmark A/S, Glostrup, Denmark) were used. Normal tissues not expressing HER2 were negative controls as well as connective tissue in the same sections as the tumor cells. In lymph node metastases, lymphocytes and the surrounding capsule were negative controls. In the liver metastases, surrounding normal tissues were also used as negative controls. Samples that scored 2+ or 3+ were considered HER2 positive, while samples that scored 0 or 1+ were considered HER2 negative. Gene amplification analysis for 2+ and other cases was not made since receptor expression and not gene amplification is of no or minor importance for the binding to the receptors of the targeting agents that deliver the radionuclides.

Statistical analyses

Statistical analysis was performed using SPSS 17.0 (SPSS Inc, Chicago, IL, USA). Associations between HER2 overexpression and clinicopathological parameters, including gender and Lauren’s classification (histologic type), were assessed using Fisher’s exact test. The Kruskal-Wallis test was used for analysis of the relationship between HER2 expression and differentiation, disease stage, and number of metastastic lymph nodes. For the age parameter, which did not meet assumptions of normality and homoscedasticity (homogeneity of variance), the Mann-Whitney U test was applied. A two-tailed P value of <0.05 was considered as statistically significant. Data with non-normal distribution were evaluated as median and interquartile range (IQR).

Results

HER2 expression

HER2 positivity was seen in nearly 52 % of the primary tumors, and among these, 40 % scored 2+ and 60 % scored 3+ as can be calculated from the data in Tables 2 and 3. The detailed scorings (0, 1+, 2+, and 3+) for individual patients are shown in Table 2 with comparisons between primary tumors and lymph node metastases in the upper part and with liver metastases in the middle part. The detailed relation between lymph node and liver metastases is shown in the lower part. A summary of HER2 positive (2+ and 3+ together) and HER2 negative (0 and 1+ together) primary tumors and metastases is shown in Table 3.

Table 2 HER2 scores for the primary gastric tumors and corresponding lymph node metastases (upper part) and liver metastases (middle part) and HER2 score relations between liver and lymph node metastases (lower part). n = 29
Full size table
Table 3 Summary of relations between positive (2+ and 3+) and negative (0 and 1+) expressions of HER2 in primary tumors and lymph node metastases (upper part) and with liver metastases (middle part). The relation between lymph node and liver metastases is shown in the lower part (n = 29)
Full size table

Concordant cases

The cases with HER2-positive primary tumors expressed HER2 also in their corresponding lymph node and liver metastases in 93 and 100 % of the cases, respectively, that can be calculated from the data in Table 3 upper and middle parts. The concordance between HER2 expression in lymph node and liver metastases was also high as seen in Table 3 lower part. However, the most important is the high co-expression of HER2 between the primary tumors and the corresponding metastases.

Discordant cases

There were only four cases with discordance between the primary tumors and corresponding lymph node metastases. Three of these were HER2 negative in the primary lesions but positive in lymph node metastases, i.e., ”positive conversions,” whereas only one showed “negative conversion” (Table 3, upper part). There were three cases with changes of HER2 status between primary tumors and corresponding liver metastases, and all three were “positive conversions” (Table 3, middle part). Five cases with discordance between the two sites of metastases were observed; three with positive liver metastases and negative lymph node metastases, and two changed the opposite way (Table 3, lower part).

Examples of stainings

Examples of HER2 stainings of primary tumors and corresponding lymph node and liver metastases from three patients, which scored 1+, 2+, and 3+, are shown in Fig. 1. Note that, for each patient, there was the same HER2 scoring for the primary tumors as for the two types of metastastic lesions.

Fig. 1
figure 1

Examples of immunohistochemical HER2 stainings from three patients with disseminated gastric cancer. Samples a–c are from one patient; samples d–f, from another patient; and g–i, from a third patient. a, d, g From the primary tumors. b, e, h From corresponding lymph node metastases. c, f, i From corresponding liver metastases. Samples a–c all scored 3+, samples d–f all scored 2+, and samples g–i all scored 1+. The bars correspond to 50 μm

Full size image

Statistical analysis

The statistical analyses revealed no significant association between HER2 expression and the clinicopathological features; age (P = 0.56), gender (P = 0.56), differentiation (P = 0.40), T stage (P = 0.51), N stage (P = 0.23), and histologic type (P = 0.36). All P values were too high to indicate significant associations.

Discussion

Half of the cases had positive HER2 receptors (approximately 52 %), and the positive cases had high expression frequencies in corresponding lymph node and liver metastases (93 and 100 %, respectively). The receptors are therefore suitable targets for substances that deliver radionuclides for receptor imaging, dose planning, and therapy. This is of special interest to consider when there is resistance to HER2-targeted therapies with tyrosine kinase inhibitors, e.g., lapatinib, and “naked” antibodies, e.g., trastuzumab, because of signal pathway mutations or other reasons [12–20], and this is in spite of existing HER2 expression. For radionuclide therapy, the extracellular domain of the HER2 receptors is of the main interest, and neither intracellular mutations in signal transduction nor gene amplifications are of major interest for binding of molecules that deliver radionuclides, and since we focused on receptor expression, we only did analyses with immunohistochemistry (IHC).

The IHC results give a first indication on which patients might benefit from HER2-targeted radionuclide therapy. However, samples analyzed with IHC might not be fully representative for the whole tumor and metastasis volumes, and there might also be variations in expression between metastases in the same patient. Thus, HER2 positivity due to IHC results only gives a crude estimate on which patients should be considered for radionuclide-based noninvasive and whole-body-based receptor imaging, as reported for other types of tumors [35–39]. However, the rather high frequencies of HER2 expression in our and other studies and also the high concordance between the primary tumors and corresponding metastases already now indicate that many patients with metastatic gastric cancers are of interest for targeting with substances that deliver radionuclides. There is low or no expression of HER2 in normal tissues [40, 41], and HER2 is therefore an interesting target for such therapy. Clinical imaging studies of HER2 has shown that normal tissues known to express low or no levels of HER2 most often give low or no signals. However, there are two exceptions seen in nuclear medicine analysis of HER2, that is, the liver and kidneys. These organs can have high HER2 signals, although they have no or very low HER2 expression [35, 37, 39].

High HER2 expression can be detected in a wide spectrum of malignancies, e.g., breast, prostate, colorectal, and urinary bladder cancer [42]. There are nowadays strong indications also on a role of HER2 in aggressive growth and formation of metastases in gastric cancer [8, 10, 11, 25–29, 43], and the HER2 expression has been associated with prediction of therapy results for individual patients but, so far, not clearly for prognosis of survival on the population level [43–46]. Jorgensen et al. [46] has summarized the HER2 expression in primary gastric tumors, covering 39 reports dating until 2011, and in combination with results from other reports [26, 29, 43, 44], there were large variations in HER2 positivity (range 4–53 %). The variations could potentially be attributed to differences in inclusion criteria (e.g., metastatic spreading), geographic distributions (etiology), and methodology of the IHC procedures and scoring system and even due to sampling errors. The HER2 expression in our study, nearly 52 %, was higher than in most previous studies. This seems reasonable since our cohort consists only of patients with advanced gastric cancer having both lymph node and liver metastases and none of them being previously treated with radiotherapy, chemotherapy, or HER2-targeted therapy. The high frequency of HER2 expression usually correlates with high T stage, aggressive course, and poor survival, and Janjigian et al. [43] and Mizutani et al. [47] found higher rates of HER2 positivity in patients with liver metastasis than those with no diagnosed liver metastases. Another reason for our high HER2 positivity could be that we used the HER2 scoring criteria for gastric cancer by Hofmann et al. [34] which, as mentioned in “Material and methods,” considers basolateral, so-called U shape, HER2 expression as positive.

In our samples, there was “downregulation” of HER2 from the primary tumors to the corresponding lymph node metastases in only one case and in liver metastases in none of the cases. Our discordance frequencies are similar to those in previous reports [25–29]. Kim et al. [27] suggested that discordant cases might be due to intratumoral heterogeneity of HER2 expression and gene amplification in the primary tumors. Liver metastasis is associated with recurrence and bad prognosis of gastric cancer [43, 48], and in our study, good HER2 concordance was observed between primary lesions and the corresponding liver metastases. Previously published studies on distant metastases [26, 27] have also indicated a reasonably good such concordance, although the types of distant metastases were not analyzed separately but often included, e.g., the liver, abdominal wall, and intestine. Bozzetti et al. [26] reported the HER2 concordance to be as high as 95 % after analysis of distant metastases with IHC.

Consistent with findings of Kim et al. [27] our results gave more positive conversions (negative primary lesions and positive metastases) than negative conversions: six positive conversions and only one negative. We could not find an association of HER2 positivity with clinical characteristics, probably due to the low number of cases in our study. However, the study by Marx et al. [29] (166 cases) could neither find such a relationship using IHC-based analysis, but there was a relation between HER2 gene amplification analyzed with FISH and histology type [29].

Antibodies, antibody fragments, affibody molecules, and peptides are suitable for the delivery of radionuclides [21–24]. Radionuclides suitable for therapy are β−-emitters (e.g., 67Cu, 90Y, 131I, 177Lu, 186Re, 188Re) and α-emitters (e.g., 211At, 212Bi, 213Bi, 225Ac, 227Th), and for receptor imaging and/or dose planning, 99mTc, 111In, and 123I are suitable for SPECT cameras and 18 F, 64Cu, 68Ga, 76Br, 86Y, 89Zr, 110In, and 124I for PET cameras [21–24, 42, 49]. Suitable radiolabeling methods that do not severely disturb receptor binding are in most cases available.

The heterogeneity problem is a concern in targeted radionuclide therapy. The radionuclides with short range, e.g., 177Lu emitting β−-particles with a mean range of 200–300 μm and 211At emitting α-particles with a range of approximately 100 μm, might not be good enough if areas with HER2-positive tumor cells also contain areas with HER2-negative tumor cells. In those cases, a long range β−-emitter like 90Y (mean range of the β−-particles is approximately 4 mm) might be preferable or used as a complement. The long range β−-emitters give rather homogeneous local absorbed dose distributions since the radiation from the targeted cells give absorbed dose also to neighbor cells (cross-fire radiation) [24].

As an alternative to radionuclides and as the therapy active components in a targeting conjugate, there are interesting toxins [50, 51], but these are not discussed in this article.

Few targeted radionuclide therapy methods are routinely used on a large scale. However, there is promising clinical and preclinical research results published during some decades. One example is therapy of lymphomas, which normally are radiosensitive, by using radiolabeled anti-CD20 antibodies such as 90Y-Zevalin [52, 53], and another example is therapy of neuroendocrine tumors using somatostatin analogs such as 177Lu-Octreotate [54, 55]. Note that there are promising results with radionuclide therapy of lymphomas [52, 53], although their tumor cells are localized in large body volumes. This indicates that targeted radionuclide therapy probably also can be of value for therapy of disseminated solid tumors that also can have tumor cells spread into large body volumes. There is also potential for clinical studies of various types of tumors since preclinical research and early clinical studies with new targeting agents with good pharmacokinetics and biodistributions have been tried [23, 35–38, 56–62].

In conclusion, HER2 positivity was seen in about half of the primary gastric tumors, and there was good concordance with the corresponding metastases in lymph nodes and the liver. Thus, it is of interest to target HER2 for radionuclide-based therapy of disseminated gastric cancers and thereby decrease the influence of resistance to other forms of therapy. This can hopefully allow more patients to be treated with curative instead of palliative intention.