Expression profiling of receptor tyrosine kinases in high-grade neuroendocrine carcinoma of the lung: a comparative analysis with adenocarcinoma and squamous cell carcinoma

Large cell neuroendocrine carcinoma (LCNEC) is distinguished from small cell lung carcinoma (SCLC) based on its histological criteria, that is, a larger cell size, abundant cytoplasm, prominent nucleoli, vesicular nuclei or coarse chromatin, and a polygonal rather than a fusiform shape (Battafarano et al. 2005). Despite these differences, LCNEC and SCLC share many similarities in terms of not only immunohistochemistry, but also clinical characteristics (Gupta et al. 2004; Asamura et al. 2006; Fernandez and Battafarano 2006; Gollard et al. 2010; Nakachi et al. 2010; Dobashi et al. 2011; Li et al. 2012; Peifer et al. 2012; Rudin et al. 2012; Sun et al. 2012). Consequently, these lesions are often grouped together as high-grade neuroendocrine carcinoma (HGNEC). LCNEC also shares genetic alterations that are commonly seen in SCLCs, such as TP53, RB1, and EP300 (Jones et al. 2004; Peifer et al. 2012; Rudin et al. 2012; CLCGP-NGM 2013), suggesting a genetic similarity to SCLC. However, little is known about the differences in the protein expression profiles between these two histological types.

In addition, only fragmented information on therapeutically relevant gene alterations is available for HGNECs. Two reports regarding integrative genomic analyses of SCLC have shown that transcriptional deregulation (for example, via RB1, SOX2, and MYC family members and chromatin modifiers) might have a role in its biology.(Peifer et al. 2012; Rudin et al. 2012) To date, however, attempts to develop targeted therapies for these transcriptional deregulations have had limited success. Recently, we performed whole-exome sequencing of 51 Asian SCLC patients and demonstrated that the SCLC genome possessed distinguishable genetic features in the PI3K/AKT/mTOR pathway (Umemura et al. 2014). In this report, both gene mutations and copy number variations were analyzed, and genetic alterations in various targetable well-known receptor tyrosine kinase (RTK) genes were detected, but these variations were not correlated with the genetic changes in the PI3K/AKT/mTOR pathway, and their functional roles have remained unclear.

As already known, RTKs are the initial signaling gate on the cell membrane. Given their pivotal roles in tumor initiation and progression, RTKs have become one of the most prominent target families for drug development (IASLC 2009; Umemura et al. 2014). Therefore, in the present study, we analyzed the protein expressions of the major RTKs of the HGNEC tumors, which we examined using whole-exome sequencing, and compared them with those of adenocarcinoma (ADC) and squamous cell carcinoma (SQCC) to identify biologically distinctive alterations in HGNECs.

We performed an extensive RTK expression study for all four major histological types of lung cancer, focusing on HGNEC. Intriguingly, the overall RTK expression patterns of LCNEC and SCLC were similar, but they were quite different from those for ADC or SQCC. The similarity between LCNEC and SCLC has long been described since the concept of LCNEC was first proposed in the 1990s (Travis et al. 1991, 1998a, b). Based on the morphological features of LCNEC and SCLC, many reports have suggested a similarity in their clinical behaviors (Asamura et al. 2006; Kinoshita et al. 2013) and sensitivities to chemotherapy (Le Treut et al. 2013; Niho et al. 2013). Jones et al. (2004) also reported that they could not distinguish LCNEC from SCLC based on gene expression profiling. From this aspect, our analysis of the protein expression levels of major RTKs supports the past consensus.

In the present research, we analyzed the RTK expression levels in individual patients and plotted each RTK score for individual cases, since diversity in the expression levels was observed in individual cases even among those with the same histological type. This analysis revealed that about 20 % of LCNEC and SCLC tumors had some kind of strongly stained RTK, such as c-Kit, EGFR, IGF1R, KDR, or ERBB2. Interestingly, most of these strongly staining RTKs were mutually exclusive, evoking the oncogenic driver mutations (EGFR, KRAS, ALK, and ROS1) observed in lung adenocarcinoma. Additionally, we examined the relationship between the EGFR mutation status and RTK expression in ADC and showed that strongly positive non-EGFR–RTKs tended to be mutually exclusive with EGFR mutations. We anticipate that this exclusiveness might reflect the importance of these highly expressed RTKs for tumor proliferation, survival, or invasiveness.

In this study, one strongly positive ERBB2 tumor existed in the LCNEC group, but no strongly positive ERBB2 tumors were seen in the SCLC group. According to a recent report from The Clinical Lung Cancer Genome Project (CLCGP) and Network Genomic Medicine (NGM) (CLCGP-NGM 2013), a total of two (5 %) LCNEC tumors with ERBB2 amplification, but no SCLC tumors, have been reported, consistent with the present findings. The identification of LCNEC tumors with high ERBB2 expression levels suggests that a subset of these tumors might be sensitive to ERBB2 inhibitors, similar to ERBB2-positive gastric cancer and breast cancer (Asaoka et al. 2011; Stern 2012; Kumler et al. 2014).

Some LCNEC and SCLC tumors have been reported to exhibit FGFR1 amplification (Peifer et al. 2012); however, no LCNEC or SCLC tumors that were strongly positive for FGFR1 were observed in our study. Although the amplification of FGFR1 is reportedly predominant in squamous cell carcinomas, the association with overexpression was inconclusive (Pros et al. 2013). Further examination of FGFR1 alterations is needed.

The histological type-specific findings in the present study were in accordance with previous reports of a high frequency of c-Kit expression in HGNEC (Pelosi et al. 2004a, b, Dy et al. 2005; Schneider et al. 2010; Lu et al. 2012), EGFR expression in SQCC (Mountzios et al. 2010; Pirker et al. 2012; Jiang et al. 2013), and ALK expression in a minority (three of 202 tumors, 1.5 %) of ADC tumors (Chen et al. 2012; Park et al. 2012; Nitta et al. 2013). Of note, the expression of c-Kit was considerably higher in HGNEC than in ADC or SQCC, suggesting its biological importance for tumorigenesis in HGNEC. High expression levels of c-Kit in LCNEC have also been previously reported (Araki et al. 2003; Rossi et al. 2003; Casali et al. 2004; Pelosi et al. 2004a, b, Rossi et al. 2005; Lopez-Martin et al. 2007; Schneider et al. 2010; Lu et al. 2012). Similar to our findings, Rossi et al. (2003) reported that c-KIT was frequently expressed in both SCLC and LCNEC, but not in ADC or SQCC. However, two phase II studies using imatinib, a c-Kit inhibitor, failed to demonstrate any clinical benefit even among selected SCLC patients harboring c-Kit-expression (Dy et al. 2005; Schneider et al. 2010). Unlike the situation for gastrointestinal stromal tumors (GIST), activating mutations in the KIT gene were minimally associated with the immunohistochemical expression of c-Kit in HGNEC. Actually, Rossi et al. reported that c-Kit was strongly expressed in 52 of the 83 LCNECs (62.7 %), but no activating mutation was detected in the KIT gene. We combined the current results with our previous data for whole-exon sequencing and a copy number analysis of SCLC samples (Umemura et al. 2014) and observed a similar tendency. Nevertheless, one SCLC case with strong c-Kit expression also had a mutation and amplification of the KIT gene. This rare case might be a candidate for targeted therapy, and IHC for c-Kit might be useful for patient screening. On the other hand, in imatinib-resistant GIST, the PI3K/AKT/mTOR pathway is a major contributor to proliferation and survival (Floris et al. 2013). This pathway has also been proposed as an actionable signaling cascade that is active in SCLC (Arriola et al. 2008; Ilic et al. 2011). This finding implies that a combined treatment with PI3K/AKT/mTOR inhibitor might be potentially effective for increasing the sensitivity to c-Kit inhibitors in HGNEC. In addition to tyrosine kinase inhibitors, antibodies are expected to be effective for tumor cells overexpressing target RTKs. A novel treatment using anti-c-kit antibody has been attempted in vitro (Yoshida et al. 2013b). Further studies will require the clarification of specific biological features and the development of c-Kit-targeted therapies.

As for the survival of patients with LCNEC and those with SCLC, the OS rates for both histological types were not significantly different, similar to the results of previous reports (Asamura et al. 2006; Kinoshita et al. 2013). Furthermore, the OS curve of HGNEC patients with RTK positivity was not significantly different from those without RTK positivity (Supplemental Figs. 2 and 3). The reasons for the similarity in OS curves were thought to include the small population of strongly positive RTKs, the short follow-up periods, the contributions of other RTK, and driver oncogenes that were not assessed in the present study. We did not find any significant effect of c-Kit expression on survival, similar to the results of previous reports (Rossi et al. 2005; Lopez-Martin et al. 2007). However, Casali et al. (2004) reported that c-Kit expression in LCNEC was a negative prognostic factor. Further investigation of possible correlations in larger studies is warranted.

In conclusion, LCNEC and SCLC are relatively similar, compared with ADC and SQCC, even at the protein expression level. Based on this background, the development of molecular-targeted agents for SCLC and LCNEC could be combined into the development of treatments for “HGNEC” (Pelosi et al. 2004b; Schneider et al. 2010; Sun et al. 2012; CLCGP-NGM 2013). Although RTK positivity cannot be used as the sole criterion for targeted therapies for HGNEC, strongly positive RTKs were observed in a mutually exclusive manner, suggesting their biological importance for tumorigenesis in HGNEC. In the future, we plan to perform both expression and genomic analyses of RTK in non-resectable advanced HGNEC tumors, in addition to surgically resected samples.