Incidence and patterns of ALK FISH abnormalities seen in a large unselected series of lung carcinomas

Background Anaplastic lymphoma receptor tyrosine kinase (ALK) gene rearrangements have been reported in 2-13% of patients with non-small cell lung cancer (NSCLC). Patients with ALK rearrangements do not respond to EGFR-specific tyrosine kinase inhibitors (TKIs); however, they do benefit from small molecule inhibitors targeting ALK. Results In this study, fluorescence in situ hybridization (FISH) using a break-apart probe for the ALK gene was performed on formalin fixed paraffin-embedded tissue to determine the incidence of ALK rearrangements and hybridization patterns in a large unselected cohort of 1387 patients with a referred diagnosis of non-small cell lung cancer (1011 of these patients had a histologic diagnosis of adenocarcinoma). The abnormal FISH signal patterns varied from a single split signal to complex patterns. Among 49 abnormal samples (49/1387, 3.5%), 32 had 1 to 3 split signals. Fifteen samples had deletions of the green 5′ end of the ALK signal, and 1 of these 15 samples showed amplification of the orange 3′ end of the ALK signal. Two patients showed a deletion of the 3′ALK signal. Thirty eight of these 49 samples (38/1011, 3.7%) were among the 1011 patients with confirmed adenocarcinoma. Five of 8 patients with ALK rearrangements detected by FISH were confirmed to have EML4-ALK fusions by multiplex RT-PCR. Among the 45 ALK-rearranged samples tested, only 1 EGFR mutation (T790M) was detected. Two KRAS mutations were detected among 24 ALK-rearranged samples tested. Conclusions In a large unselected series, the frequency of ALK gene rearrangement detected by FISH was approximately 3.5% of lung carcinoma, and 3.7% of patients with lung adenocarcinoma, with variant signal patterns frequently detected. Rare cases with coexisting KRAS and EGFR mutations were seen.


Background
Lung cancer is the leading cause of cancer-related deaths in the US with an estimated 160,000 deaths attributed to lung cancer in 2012 [1]. Histologically, the World Health Organization (WHO) classifies lung cancer into nonsmall cell lung cancer (NSCLC, 85%) and small-cell lung cancer (15%). NSCLC is further divided into three major subgroups: adenocarcinoma, squamous cell carcinoma, and a less well-characterized group of large cell carcinoma (40%, 30% and 15% of all lung cancer cases in the United States, respectively). Based on the patient's medical status and stage of disease, there are four general treatment strategies, including surgery, radiation therapy, chemotherapy, and/or targeted therapy. The use of targeted kinase inhibitors is based on the presence of specific genetic alterations, such as activating mutations in the epidermal growth factor receptor (EGFR) or activating chromosomal fusions involving the ALK kinase, and can be highly effective if the tumors harbor these associated molecular alterations [2]. Therefore, molecular tests are routinely performed to identify mutations in oncogenes in lung cancer, including EGFR and ALK; to identify those patients with a high likelihood of response to targeted therapy; and reduce unnecessary side effects of ineffective treatments. The mutation status of KRAS is usually also assessed since tumors that show such mutations do not respond to targeted kinase therapies.
ALK gene rearrangements have been reported in 2% to 13% of patients with NSCLC [2]. Most patients with ALK rearrangements typically have adenocarcinomas and are younger patients with minimal to no smoking history. ALK-rearranged tumors typically lack EGFR and KRAS mutations. Recent studies have demonstrated that lung cancers harboring ALK rearrangements do not respond to EGFR-specific tyrosine kinase inhibitors, but benefit from a small molecule inhibitor targeting the ALK kinase (ie, crizotinib [Xalkori W , Pfizer, Inc.]).
Fluorescence in situ hybridization (FISH) with a multicolor break-apart ALK probe is the currently FDA-approved method for screening for ALK gene rearrangements ( Figure 1A). Exon-scanning reverse-transcription PCR methods have also been developed. Inversion of the short arm of chromosome 2 is the most common rearrangement associated with the ALK gene in lung cancer. This rearrangement leads to fusion between the 5 0 end of the echinoderm microtubule-associated protein-like 4 (EML4) gene (located at 2p21) and the 3 0 end of the ALK gene (at 2p23): the EML4-ALK fusion. Alternate ALK FISH signal patterns which may indicate other activating rearrangements also occur and are studied here.
The reasons for the variability in the detection frequency of ALK gene rearrangements in NSCLC could be due to a variety of technical and biologic influences. These include biases due to small sample size, variations in the performance and interpretation of ALK FISH studies, and the relative proportions of smoking-related cases which usually lack both EGFR and ALK mutations. In a prior large screening study, ALK rearrangement was detected in 5.5% of 1500 cases of NSCLC [3]. In this study, we examined the frequency and patterns of ALK rearrangements in clinical specimens from a large cohort of lung cancer patients from both primary and tertiary settings across the United States referred for molecular profiling.

Results
ALK FISH was performed on an unselected series that included 1387 samples from patients with NSCLC. Overall, 49 of the 1387 tumor samples (3.5%) had altered ALK signals. Thirty eight of these 49 samples (38/1011, 3.7%) were from patients with confirmed adenocarcinomas (Table 1). In 32 cases, there was the expected pattern of one split orange and green signal indicative of an ALK gene rearrangement and one single fusion signal from the intact ALK gene ( Figure 1B). Among variant signals, the most commonly seen (15 cases) was deletion of the 5 0 ALK (green signal, Figure 1C). More complex FISH patterns included normal split patterns along with loss of the 5 0 signal; this was often seen in tumors that showed gains of intact ALK signals consistent with polysomy ( Figure 1D).
One of the 15 patients with deletion of the 5 0 ALK, also showed 5-25 copies of the 3 0 ALK (high level amplification, Figure 1E). Two samples showed a deletion of the 3 0 ALK (orange signal, Figure 1F). According to the new FDA-approved Vysis FISH ALK break-apart probe kit guidelines, patients with deletion of the 3 0 ALK split signal should be considered negative for rearrangements, because the tyrosine kinase domain of ALK is located in the 3 0 region of the ALK gene.
To investigate the significance of variant signal patterns, we also performed exon scanning multiplex RT-PCR to identify EML4-ALK fusion transcripts in a subset of both typical and variant cases. Five of these samples (3 with break apart and 2 with loss of 5 0 ALK) were confirmed positive for EML4-ALK fusion. Among these, one was positive for fusion variant 1, one for variant 3a, and two for both variants 3a and 3b. These are the most common ALK fusion variants reported by other investigators [2]. In addition, new transcription variants were identified, derived from a fusion of EML4 exon 17 to the 3 0 end portion of ALK exon 20 with two different small insertions (Figure 2 for gene structures and putative proteins, and Additional file 1 for nucleotide sequences and amino acid sequences). The 3 samples that did not show EML4-ALK fusion by PCR had FISH patterns with deletion of the 5 0 ALK probe.
Among the 45 ALK-rearranged samples tested for mutations in exons 18-21 of the EGFR gene, only one mutation, T790M, was detected in a patient with a typical positive break-apart FISH signal pattern of the ALK gene. Two KRAS mutations were detected among 24 ALK-positive samples examined, G12F in a patient with a deletion of the green 5 0 end of the ALK signal and G13C in the sample that showed amplification of the 3 0 ALK region ( Figure 1E).

Discussion
In this study, the frequency of ALK gene rearrangements detected using a dual-color break-apart FISH probe was 3.7% of adenocarcinoma cases. This is somewhat lower B. Split ALK signal A. ALK FISH probe  than earlier studies that generally had smaller number of patients, but close to 3.9% which was observed in two cohorts of 720 and 1121 lung adenocarcinomas from two series in Japan [4,5]. Our results are consistent with previous studies where ALK rearrangements have been largely (but not completely) restricted to adenocarcinomas that lack EGFR or KRAS mutations.
Here, we show that the ALK FISH signal patterns may vary from a single split signal to very complex signal patterns. Gains of intact ALK signals as well as typical and  Figure 2 Characterization of novel EML4-ALK transcription variants in a patient with lung adenocarcinoma. This patient had a typical positive break-apart FISH signal pattern of the ALK gene. A) Two transcription variants were detected, derived from a fusion of EML4 exon 17 to the 3 0 end portion of ALK exon 20, with 46 base pairs deleted from the 5 0 end of ALK exon 20. These two fusion transcripts were from the same gene rearrangement with either a 23-nucleotide or a 6-nucleotide insertion from EML4 intron 17. B) Putative proteins from EML4-ALK fusion transcripts. The transcript with an insertion of 23 nucleotides produced a truncated EML4 protein and no EML4-ALK fusion protein, as a result of an early stop codon. However, the transcript with an insertion of 6 nucleotides produced an EML4-ALK fusion protein, containing the N-terminal region of EML4 and the C-terminal region of ALK with the protein tyrosine kinase domain (PTKc). variant hybridization patterns for ALK gene rearrangements were observed simultaneously in some samples likely due to intratumoral heterogeneity. Although polysomy of chromosome 2 did occur frequently in NSCLC (and was not regarded as evidence of an ALK-specific rearrangement), high-level amplification of the ALK fusion signal was a rare event.
The current study and a previous report using the same PCR method [6] showed correlation between PCR and FISH for detection of EML4-ALK fusions even when loss of 5 0 green signals was seen. In our study, 5 samples (3 with break apart and 2 with loss of 5 0 ALK) were positive for an EML4-ALK fusion by multiplex RT-PCR. This is similar to a previous study that an EML4-ALK fusion was detected by RT-PCR in 22 out of 31 patients with FISH-positive ALK rearrangements [3]. However, FISH also detected variant ALK signal patterns in three cases that were negative for EML4-ALK fusion by PCR; this may be indicative of translocations involving ALK and genes other than EML4 [3], or other mechanisms of ALK activation. The general correlation between the two methods supports using loss of the 5 0 ALK signal as presumptive evidence of an ALK gene rearrangement.
EML4-ALK and other 2p23/ALK gene rearrangements lead to a constitutively activated ALK kinase, which confers sensitivity of NSCLC tumors to ALK inhibitors [3]. Lung cancer patients with ALK rearrangements have been shown to have significant reductions in tumor burden in response to treatment with the ALK inhibitor crizotinib, which led to its accelerated FDA approval in the United States in 2011 [3,7,8]. Crizotinib has also been shown to have activity against the 1-2% of patients with NSCLC that show rearrangements of another receptor tyrosine kinase gene, ROS1, located at 6q22 [5,9].
Patients who harbor ALK rearrangements do not benefit from treatment with the EGFR-specific tyrosine kinase inhibitors showing that the EGFR inhibition is bypassed. The clinical behavior of NSCLC cases with variant ALK signal patterns is less clear. Amplification of the ALK fusion has been seen in a patient undergoing crizotinib therapy [10] and may be a sign of developing resistance. But the response of untreated NSCLC with ALK gene amplification is not yet clear. Given the large number of signaling pathways that are influenced by ALK fusion products [7], the behavior and appropriate treatment of NSCLC cases that have variant ALK signal patterns also require further study.

Conclusions
Use of a break-apart ALK FISH probe is an effective method for assessing ALK gene rearrangement status in routinely submitted formalin fixed paraffin-embedded NSCLC tumor samples from a wide variety of tissue sources and clinical settings. Variant ALK FISH signals usually, but not always, represent EML4 -ALK fusions.

Patient samples
Samples from 1387 lung cancer patients (1011 with a submitting diagnosis of adenocarcinoma) were included in this study. Most were small transbronchial biopsies, or cell blocks of lymph node aspirates or malignant pleural or pericardial effusions. These samples were consecutively submitted for ALK rearrangement testing. The presence of lung cancer cells was verified by pathological examination of a hematoxylin-eosin-stained section adjacent to the slide used for FISH analysis. Mutation studies for EGFR and KRAS were also performed if sufficient tumor was available, as was EML4-ALK RT-PCR.

ALK FISH
FISH using a dual-color break-apart probe for ALK was performed on paraffin-embedded tissue sections according to the Vysis ALK Break Apart FISH Probe Kit protocol (Abbott Molecular, Des Plaines, IL), using the BX51/BX52 Olympus fluorescence microscope (Olympus, Richardson, TX). The SenSys W CCD camera (Photometrics, Tucson, AZ) was used to capture selected images. The occurrence of an ALK rearrangement (ALK positive) was concluded if >15% of tumor cells showed split orange and green signals and/or deletion of green signals; otherwise, the specimen was classified as negative for ALK rearrangement.

EGFR and KRAS mutations
Sanger sequencing of dissected lung cancer samples was performed to detect mutations in EGFR and KRAS. DNA was PCR amplified for exons 18-21 of EGFR and exons 1 and 2 of KRAS. The PCR products were purified and sequenced on a DNA sequencer (ABI; Carlsbad, CA).
Exon scanning RT-PCR for EML4-ALK fusion transcripts RNA was extracted from formalin fixed paraffinembedded sections, reverse transcribed and PCR amplified using one-step RT-PCR kit and products detected using fragment analysis on ABI 3730 genetic analyzer, as previously described [6]. The assay utilized 22 EML4 exon primers and a single ALK reverse primer to detect a variety of inversion variants. Fusion types were identified based on expected sizes of the PCR products, with the new variant identified by standard dideoxy chaintermination DNA sequencing of the abnormally-sized cDNA PCR product, as described [6].