Background
Cancer treatment paradigms are evolving to exploit the sensitivity of tumors to inhibitors that target the products of genes carrying driver mutations [
1]. A number of genetic aberrations that drive and maintain tumorigenesis have recently been identified in non-small-cell lung cancer (NSCLC). These include fusion genes generated by chromosomal rearrangements between the rearranged during transfection (
RET) gene and other genes, most commonly kinesin family 5B (
KIF5B) and coiled coil domain containing-6 (
CCDC6) [
2-
12]. These fusions lead to overexpression of truncated RET proteins containing the RET kinase domain, which can induce transformation and occur in tumors that rarely harbor mutations in other common drivers, ie epidermal growth factor receptor (
EGFR),
KRAS, human epidermal growth factor receptor, and anaplastic lymphoma receptor (
ALK) genes.
RET rearrangement was first shown to be associated with papillary thyroid carcinoma (PTC), leading to the fusion oncoprotein (
RET/PTC) and constitutive activation of RET receptor tyrosine kinase in papillary cancer cells [
13]. In addition,
RET mutations are present in the germline of nearly all patients with hereditary forms of medullary thyroid cancer (MTC) [
14-
16] and approximately 50% of patients with sporadic MTC have somatic
RET mutations that are associated with a worse outcome [
17].
Vandetanib is a once-daily, oral anticancer agent that selectively inhibits vascular endothelial growth factor receptor (VEGFR), RET and EGFR signaling [
18,
19], and is licensed for the treatment of MTC in several geographical regions. Preclinical studies have demonstrated that vandetanib inhibits RET signaling arising from
RET mutations in a MTC cell line and inhibits growth of human PTC cell lines that carry spontaneous
RET/PTC rearrangements [
19]. In addition, vandetanib has been shown to inhibit the proliferation of cells expressing
RET-KIF5B [
2,
3] and a human lung adenocarcinoma cell line harboring an endogenous
RET-CCDC6 [
20].
Four randomized Phase III clinical trials have evaluated the efficacy of vandetanib in NSCLC in combination with docetaxel (NCT00312377; ZODIAC [
21]), in combination with pemetrexed (NCT00418886; ZEAL [
22]) or as a monotherapy (NCT00364351; ZEST [
23] and NCT00404924; ZEPHYR [
24]). These studies in unselected patient populations demonstrated antitumor activity of vandetanib, but there was no overall survival benefit when added to standard chemotherapy or as monotherapy versus erlotinib [
21-
24]. The aim of this study, a retrospective evaluation of tumor samples from the NSCLC Phase III program, was to determine the prevalence of
RET rearrangements and other potential RET biomarkers within this population and to investigate any association with outcome to vandetanib treatment.
Discussion
In this retrospective study, the overall prevalence of
RET rearrangements within the Phase III vandetanib NSCLC clinical program was determined as 0.7% among patients with a known
RET rearrangement status. We found consistent frequencies of
RET rearrangement in Asian (0.7%) and non-Asian patients (0.8%). In general,
RET rearrangement prevalence rates may be considered as low and may vary according to the proportions of smokers, racial origin and histological subtype in the study population. Prevalence rates of
RET rearrangement in Asian populations have been reported at 1–2% for NSCLC [
5,
11] and lung adenocarcinoma [
2,
3,
7,
12], and were estimated as high as approximately 6% in lung adenocarcinoma [
4]. Our own study contains a high proportion of non-Asian patients (67%, Additional file
1: Table S2) and smokers/ex-smokers (77%, Additional file
1: Table S2), in contrast to previous reports on
RET rearrangement prevalence rates [
3,
5,
7,
11,
12], in which study populations were either entirely or largely Asian and non-smokers.
RET rearrangements have previously been reported in squamous cell carcinoma [
5], lung neuroendocrine tumor [
5] and adenosquamous tumor [
11]; however, the majority occur in adenocarcinomas. This is consistent with our findings, which show a higher prevalence of
RET rearrangements in patients with adenocarcinoma (1.2%, 6/510) compared with those in other histology types (0.2%, 1/427). Our data are not in agreement with a number of studies that report a higher frequency of
RET rearrangements in non-smokers compared with smokers/ex-smokers; in our study, we have observed three and four
RET rearrangements, respectively (Table
3) [
2,
5,
11]. As in previous studies, we did not observe co-occurrence of
RET rearrangements with
EGFR and
KRAS mutations. Interestingly, one of the three
RET-KIF5B tumors reported by Go
et al. [
28] in lung adenocarcinomas negative for
KRAS and
EGFR mutations and
ALK rearrangements was from a smoker. However, all of these observations should be interpreted with caution given the small numbers.
The techniques used to identify
RET rearrangement genes in previously reported studies were sequencing [
2-
4] or reverse transcription-polymerase chain reaction followed by verification with FISH [
11], sequencing [
5,
12] or differential expression of 3′ and 5′
RET exons [
7,
9,
11]. Some of these techniques may underestimate the frequency of
RET rearrangement genes by failing to detect fusions to partner genes other than
KIF5B. We used a four-probe FISH assay to detect
RET rearrangements. This technique is highly sensitive in detecting chromosomal rearrangements and has the advantage of detecting other partners or isoforms, though it is not known whether all these rearrangements are functional. For example, in a study using a split FISH assay to evaluate 1528 lung cancers, 22 tumors were detected with split
RET signals, of which 12 were confirmed as fusions with
KIF5B and one with
CCDC6 [
10] and the remaining nine tumors showed little or no expression of the RET kinase domain.
Although the prevalence of
RET rearrangements in NSCLC patients is low, RET inhibition may be efficacious within a subset of patients who carry these genetic aberrations. In this study, there were too few vandetanib-treated patients with
RET rearrangements to form conclusions regarding association with efficacy. A number of studies have reported increased expression of RET protein in NSCLC tumor cells, not necessarily associated with
RET rearrangements [
2,
3,
8,
11]. This led us to investigate both IHC and
RET copy number gain as possible predictive biomarkers for vandetanib response. No difference was observed in the ORRs between vandetanib and comparator arms for IHC and copy number analyses.
In our study, we observed
RET-KIF5B and other
RET rearrangements in samples that were negative for RET protein expression. This observation is in line with previous studies of NSCLC samples which have used a range of anti-RET antibodies (including the Epitomics EPR2871 antibody we have used here) and differing IHC techniques [
2,
3,
8,
28]. Sasaki
et al. [
8] reported three cases of RET translocation (from 371 NSCLC samples), all of which had weak positive cytoplasmic staining when evaluated with a 3F8 anti-RET mouse monoclonal antibody (Vector Laboratories, Peterborough, UK). In contrast, when using the EPR2871 antibody used in our study, weak, moderate and strong staining were reported for the three RET translocation positive samples; this suggests that apparent RET expression is dependent on both the antibody and the local IHC protocol used. Another study has reported weak to strong RET expression with IHC when using the 3F8 anti-RET antibody; however, only one of the RET IHC positive cases was also
RET-KIF5B positive [
3]. Using the EPR2871 antibody, Kohno
et al. reported 48/222 NSCLC cases to express RET in the absence of a
RET fusion; all cases of
RET-KIF5B were also RET positive with IHC [
2]. Go
et al. [
28] used three different anti-RET antibodies to screen 53 NSCLC cases for RET protein expression. RET IHC positive cases were defined as those with >30% of cells expressing cytoplasmic RET. Three samples that were
RET-fusion positive with whole transcript sequencing were negative for RET with IHC, whereas RET protein was identified in four samples, none of which harbored a
RET-KIF5B rearrangement. Taken together, these NSCLC studies, along with our results, suggest that not all cases of
RET-KIF5B or other
RET rearrangements express RET protein when evaluated with IHC. RET protein appears to be largely cytoplasmic; however, considerable inter-patient variation and heterogeneity among tumor cells within individual tumors is observed.
Investigation of RET inhibitors in NSCLC patients with a documented confirmation of a
RET rearrangement is an active area of research with three clinical studies currently ongoing (
NCT01639508,
NCT01823068 and
NCT01813734). Results from a study on the use of vandetanib in
RET-rearrangement-positive NSCLC patients (
NCT01823068) should provide further insight into the role of vandetanib in this patient population. Preliminary data from
NCT01639508, a prospective Phase II trial investigating the use of cabozantinib, a small-molecule inhibitor of MET, VEGFR2 and RET, has been published [
6]. For the first three patients treated with cabozantinib, two patients showed confirmed partial clinical responses and the third patient had prolonged stable disease approaching 8 months [
6]. A case study in a patient with poorly differentiated lung adenocarcinoma, positive for a
RET-KIF5B and refractory for previous chemotherapy, is also noteworthy. In this patient, 4 weeks of treatment with vandetanib 300 mg once daily produced a fluorodeoxyglucose-positron emission tomography/computed tomography response [
29]. In addition, in a preliminary study in which two heavily pretreated patients with confirmed
RET rearrangements were treated with vandetanib, stable disease was observed following treatment [
30].
Acknowledgments
We dedicate this article to the memory of our colleague Neil Gray, FIBMS.
The authors would like to acknowledge John Williams for facilitating the delivery of the contracts with external parties to time, quality and cost, Darren Hodgson for clinical data management, transmission, reporting and critical review of data analysis files, and Jennifer Bradford for assistance with analysis and interpretation.
This study was sponsored by AstraZeneca. Medical writing assistance was provided by Claire Routley from Mudskipper Business Ltd and funded by AstraZeneca.
Competing interests
G Bigley, A Dale, S Fan, H Fu, Q Ji, A Platt, J Read, X Su, V Williams, Q Ye, L Zheng and T Zhang are employed (other than primary affiliation; e.g., consulting) by AstraZeneca. L Blockley, E Donald, P Elvin, G Harrod, J Stevens, J Morten, C Cresswell, A Davies, A Gladwin and C Womack are employed by and own shares in AstraZeneca. J-S Lee has received research funding from AstraZeneca. R de Boer has received research funding and honoraria as a consultant from AstraZeneca. J Vasselli is employed by MedImmune. R Herbst has no potential conflicts of interest.
Authors’ contributions
AP, PE, JM, ED, AG, CW and QJ were involved in the conception and design of the study. XS, ED, VW, GB and CW developed the methodology for the study. All authors obtained data for the study. The analysis and interpretation of the data was performed by AP, JM, GB, LB, CC, AD, AD, NG, SF, HF, GH, JR, JS, VW, QY, TZ, QJ, XS, LZ, ED, CW and PE. All authors contributed to the writing and critical review of the manuscript. All authors read and approved the final manuscript.
Dr Womack was employed by AstraZeneca at the time of the study.