Targeted Resequencing Reveals ALK Fusions in Non-Small Cell Lung Carcinomas Detected by FISH, Immunohistochemistry, and Real-Time RT-PCR: A Comparison of Four Methods

Anaplastic lymphoma receptor tyrosine kinase (ALK) gene rearrangements occur in a subgroup of non-small cell lung carcinomas (NSCLCs). The identification of these rearrangements is important for guiding treatment decisions. The aim of our study was to screen ALK gene fusions in NSCLCs and to compare the results detected by targeted resequencing with results detected by commonly used methods, including fluorescence in situ hybridization (FISH), immunohistochemistry (IHC), and real-time reverse transcription-PCR (RT-PCR). Furthermore, we aimed to ascertain the potential of targeted resequencing in detection of ALK-rearranged lung carcinomas. We assessed ALK fusion status for 95 formalin-fixed paraffin-embedded tumor tissue specimens from 87 patients with NSCLC by FISH and real-time RT-PCR, for 57 specimens from 56 patients by targeted resequencing, and for 14 specimens from 14 patients by IHC. All methods were performed successfully on formalin-fixed paraffin-embedded tumor tissue material. We detected ALK fusion in 5.7% (5 out of 87) of patients examined. The results obtained from resequencing correlated significantly with those from FISH, real-time RT-PCR, and IHC. Targeted resequencing proved to be a promising method for ALK gene fusion detection in NSCLC. Means to reduce the material and turnaround time required for analysis are, however, needed.


Introduction
Chromosomal rearrangements involving anaplastic lymphoma receptor tyrosine kinase (ALK), a member of receptor tyrosine kinase family, have been detected in various cancers [1]. ALK rearrangements create an oncogenic ALK tyrosine kinase that activates many downstream signaling pathways resulting in increased cell proliferation and survival [1]. In non-small cell lung carcinoma (NSCLC), the most common ALK rearrangement is a fusion of ALK gene and echinoderm microtubule-associated protein-like 4 (EML4) gene, formed as a result of a small inversion within the short arm of chromosome 2, where the genes are located [2]. EML4-ALK fusion protein serves as a therapeutic target for an ALK tyrosine kinase inhibitor, which has showed promising results when used in treating NSCLC patients carrying ALK rearrangement [3][4][5][6]. In addition to EML4, other ALK fusion partners have also been reported in lung cancer, including TFG [7], KIF5B [8], and KLC1 [9].
ALK gene rearrangements can be screened by various methods, including commonly used �uorescence in situ hybridization (FISH), immunohistochemistry (IHC), and real-time reverse transcription-PCR (RT-PCR). Furthermore, next-generation sequencing technology holds a great potential as a new tool for cancer diagnostics [10]. Each of these methods has their own advantages and disadvantages. Hence, the optimal method to detect ALK fusions in NSCLC remains to be determined.
In this study, we screened ALK gene fusion status by FISH, IHC, real-time RT-PCR and targeted resequencing. Our aim was to compare the results obtained from these four methods in order to see how well they correlate with each other. We especially wanted to �nd out how well FISH, IHC, and real-time RT-PCR support the sequencing results. Moreover, we aimed to prove the suitability of targeted resequencing for testing of formalin-�xed paraffin-embedded (FFPE) tumor material. To our knowledge, this is the �rst paper comparing these four methods as means of ALK gene fusion identi�cation.

Patients.
Archived formalin-�xed paraffin-embedded tumor specimens were collected from 87 non-small cell lung carcinoma patients operated during 2005-2011 at the Hospital District of Helsinki and Uusimaa (HUS), Finland. Based on the WHO 2004 classi�cation, 80 of the patients were diagnosed with adenocarcinoma, four with large cell carcinoma, one with squamous cell carcinoma, and two with unclassi�ed NSCLC. Selection of the patients for the analyses was not random. e selected patients were mainly nonsmokers and those with adenocarcinomas. e EGFR gene mutation status was previously assessed for 57 patients, of which 9 carried EGFR mutation. e mean age of the patients was 63.7 years, ranging from 44 to 82 years. About half of the patients were male ( ) and half were female ( ). e tumor tissue percentage of the specimens was con�rmed by a pathologist to be >20%.
For ALK fusion testing, we applied dual-color, breakapart FISH, real-time RT-PCR, immunohistochemistry, and targeted resequencing. Interpretation of the results was done in double-blind manner without knowing the results by other methods.

Fluorescence In Situ Hybridization.
We performed �uorescence in situ hybridization on 95 FFPE tumor specimens from 87 patients (two separate samples from 8 of the patients), for which the tumor tissue sections (2.5 m) were �xed on microscope slides. Deparaffinization of the sections, slide pretreatment, protease pretreatment, hybridization with Vysis LSI ALK Dual Color Break Apart FISH probes, slide washing, and counterstaining by DAPI were done following the manufacturer's protocol (Abbott Molecular Inc., Des Plaines, IL, USA). Paraffin melting and hybridization were performed using StatSpin ermoBrite Slide Processing System (Abbott Molecular Inc.). e results were viewed by �uorescence microscope. For each sample, the number of fused signals, single orange signals, and single green signals was counted in 50 nuclei. Cells containing at least one green and orange signal pair split apart by ≥2 signal diameters (pair-signal type fusion), or a single orange signal without corresponding green signal (single-signal type fusion), were considered to contain ALK gene rearrangement (positive). Fused or adjacent green and orange signals, or a single green signal (deletion of the corresponding orange signal), were an indication of a negative cell. Samples with ≥15% of positive cells were considered positive.
2.3. Immunohistochemistry. Immunohistochemical stainings were done to 14 specimens from 14 patients using mouse monoclonal ALK antibody (clone 5A4, Novocastra, Newcastle, UK), 1 : 100 with OptiView DAB detection Kit (Ventana, Tucson, Arizona, USA), and CC1 buffer in BenchMark XT (Ventana, Tucson, Arizona, USA). Due to long permission procedures, ALK immunostaining was performed only for FISH positive and some negative cases.

Real-Time RT-PCR.
We evaluated 95 FFPE tumour specimens from 87 patients for the presence of an EML4-ALK gene fusion using the AmoyDx EML4-ALK fusion gene detection kit (Amoy Diagnostics, Xiamen, China). Following xylene deparaffinization of a 16 M FFPE section, total RNA was extracted using the RNeasy FFPE Kit (Qiagen GmbH, Hilden, Germany) according to the manufacturer's protocol. RNA was quanti�ed using the Qubit �uorometric quantitation system (Life Technologies, Carlsbad, CA, USA). 100-500 ng of the extracted RNA was used for reverse transcription and real-time PCR in each of the four reactions of the EML4-ALK fusion gene detection kit according to the manufacturer's protocol. Reaction 1 ampli�es EML4-ALK variants 1, 2, 3a, and 3b, reaction 2 EML4-ALK variants 4 & 4′, reaction 3 EML4-ALK variants 5a, 5b, 5′, and 8, and reaction 4 the reference gene beta-actin. All assays were performed on an ABI7500 instrument (Applied Biosystems, Foster City, USA). Assay reactions achieving Ct values of ≤30 cycles were considered positive for one of the variants detected by that reaction mixture.

Next-Generation Sequencing
2.5.1. DNA Isolation. DNA was isolated from FFPE NSCLC tissue specimens using QIAamp DNA Mini Kit (Qiagen GmbH, Hilden, Germany). e isolation was performed according to the manufacturer's instructions for genomic DNA puri�cation from paraffin-embedded tissue material with following modi�cations: xylene was added twice and 100% ethanol was added three times for careful paraffin removal, samples were incubated at 90 ∘ C for 1 h aer overnight cell lysis, Buffer AL was added normally but an incubation at 70 ∘ C for 10 min was not performed, and in the last step of the protocol, QIAamp Mini spin column loaded with water was incubated at room temperature for 5 min before �rst centrifugation, and at 50 ∘ C for 5 min before the �nal centrifugation to increase the DNA yield. DNA was eluted in 50 L of puri�ed water. DNA concentration was measured by Qubit �uorometer (Life Technologies).

Target Capture.
We selected 192 genes associated with lung cancer for target enrichment. Baits were designed to capture all exons, 3′UTR, and 5′UTR regions of all genes, including ALK and EML4 genes. Additional baits were designed from intronic region between exons 19 and 20 of ALK gene to capture all breakpoint variants of ALK fusion gene. Baits were designed with e-array (Agilent Technologies, Inc., Santa Clara, CA, USA) and RNA baits obtained from Agilent. A total of around 1 MB region, including 2676 target regions, was captured.

Targeted
Sequencing. 2-3 g of DNA, isolated from 57 formalin-�xed para�n-embedded tumor specimens from 56 patients, was fragmented, ligated to adapters, and enriched for target regions using Agilent's SureSelect in-solution target capture and enrichment protocol. Paired-end sequencing of the target-enriched libraries was performed on Illumina HISeq2000 sequencer (Illumina, Inc., San Diego, CA, USA).

Primary
Analysis. Data obtained from sequencing was processed with a variant-calling pipeline (VCP) developed at the Finnish Institute of Molecular Medicine (FIMM) [11]. Sequence reads were �ltered for quality, paired-end reads aligned to the reference genome with the Burrows-Wheeler Aligner (BWA) [12], and duplicate fragments removed by rmdup algorithm. For variant calling, VCP utilized SAMtools' pileup [13]. Single nucleotide variation (SNV) calling and read end anomaly (REA) calling were based on FIMM's own developed algorithm. Detection of indels was performed with the Pindel [14], and Circos [15] was used for visualization of anomalously mapping paired-end (PE) reads. Results were visualized in the Integrative Genomics Viewer (IGV) [16].
For identifying ALK fusions from data, all exons and intronic region of ALK (chr2:29,446,208-29,448,431 in build GRCh37 of the human reference genome) were checked for anomalously mapped paired-end reads within this region.

Results
FISH and real-time RT-PCR were performed on all the 87 NSCLC cases, targeted resequencing on 56 cases, and IHC on 14 cases. Successful results were obtained from all specimens by each method. Table 1 lists all ALK-positive cases detected by FISH, IHC, real-time RT-PCR, and targeted resequencing.
We detected ALK rearrangement by FISH in �ve patients and one replicate sample. Two of the positive cases had pairsignal type fusion and three of them had single-signal type fusion. We also detected one ALK fusion-positive borderline case with 20% of positive cells observed by �rst reader. Aer evaluations by two other readers, it was, however, considered negative (<15% of positive cells). No ALK fusion was observed in this case by real-time RT-PCR, targeted resequencing, or IHC.
Using real-time RT-PCR, we identi�ed EML4-ALK fusion variants 1, 2, or 3a/b (without distinguishing between them) in �ve patients and one replicate sample.
Targeted resequencing revealed four patients and one replicate sample harboring EML4-ALK fusion (the �h positive patient was not studied by targeted resequencing). All positive cases by targeted resequencing had a breakpoint in intronic region between exons 19 and 20 of ALK, and in  Table 1 shows detailed breakpoints of the fusions in ALK gene. By IHC, we detected ALK staining in �ve cases. e results of FISH, real-time RT-PCR, targeted resequencing, and IHC corresponded to each other, and no discrepant results were observed. Figures 1 and 2 show a detection of ALK fusion-positive result by targeted resequencing, and Figure 3 represents examples of ALK fusion-positive results detected by FISH, real-time RT-PCR, and IHC.
In total, we detected �ve (5.7%) ALK-positive cases in 87 NSCLC patients studied altogether. ree of these �ve EML4-ALK fusion containing tumors were adenocarcinomas and two were large cell carcinomas. Of all the studied cases, four men and one woman carried ALK fusion, while 41 men and 41 women were considered ALK negative. All the patients with ALK gene rearrangement were nonsmokers, and their age ranged between 44 and 63 years (mean age = 50.4 years). e age range among ALK fusion negative patients was 45-82 years (mean age = 64.5 years). None of the patients with EGFR mutation harbored ALK gene fusion. All replicate samples of the 8 patients included in our study showed concordant results in each method used.

Discussion
Non-small cell lung carcinoma patients harboring ALK gene rearrangement have been reported to bene�t from tyrosine kinase inhibitor treatment [3][4][5][6], which makes the reliable screening of ALK fusions important. In addition to the commonly used FISH, IHC, and RT-PCR, targeted resequencing holds great potential in the detection of ALK rearrangements.
We studied NSCLCs using FISH, IHC, real-time RT-PCR, and targeted resequencing to detect ALK gene rearrangements. e results obtained from all four methods were   interpretable and no discrepancies between the methods were observed. Table 2 shows the comparison of the methods used for ALK fusion screening in this study. FISH is a relatively rapid, well-standardized, rather expensive method which only requires a small amount of tumor biopsy. e FISH assay performed in this study is the FDA-approved companion diagnostic for the NSCLC crizotinib, which uses probes that target to ALK locus on chromosome 2p23, thus recognizing all ALK gene rearrangements including non-EML4-ALK fusion partners. FISH could not, however, identify the fusion partner nor can it distinguish between different fusion variants. e interpretation of FISH results is rather subjective, and a considerable variability in interpretation of FISH results between readers have been reported [17]. In our study, all cases were evaluated by two experienced viewers, who read the signals for most of the cases fast (10 to 30 minutes) and reliably. e borderline ALK-positive case (20% of positive cells) observed by �rst FISH reading was evaluated by two other readers, who both detected <15% of positive cells. As the mean value of the three analyses was less than 15% of positive cells, the sample was considered, according to the provider of the FISH kit, negative, underlining the relevance of reevaluation of the borderline positive cases. e results from real-time RT-PCR, targeted resequencing, and IHC support the interpretation, since no ALK fusion was detected in this case by these methods.
Immunohistochemistry is a rapid and relatively inexpensive method for diagnosing ALK-rearranged NSCLCs that require only a small biopsy sample. A great advantage of immunohistochemistry is that tissue morphology is retained during analysis, and thus the detection of ALK fusions can be focused on the tumor areas. We performed immunohistochemistry by using ALK antibody clone 5A4, the use of which has showed good sensitivity in ALK fusion detection in lung carcinomas [18][19][20]. In the present study, the results obtained from IHC were readily interpreted, and all ALK positive cases showed a strong positive staining.
Real-time RT-PCR is a fast and sensitive method for detection of expressed EML4-ALK gene fusions in NSCLC. It recognizes known EML4-ALK fusion variants for which speci�c primers have been designed, not other ALK gene rearrangements. e real-time RT-PCR method used in this study was able to identify 10 EML4-ALK fusion variants of which variants 1, 2, or 3a/3b were detected. Only a small amount of RNA (100-500 ng) was needed for the test as a starting material. Although two samples gave only 84 ng and 90 ng of input RNA, the b-actin Ct values were considered acceptable (a Ct value <30). All in all, the assay worked 100% in all runs and the results were reproducible.
Targeted resequencing is a promising method for detection of ALK gene rearrangements. In the present study, the baits for sequencing were designed to capture all breakpoint variants of ALK fusion gene. We observed 73-96% of target 20-fold coverage in our samples, which shows a great efficiency of the target enrichment. All patients with EML4-ALK fusion detected by resequencing had EML4-ALK fusion variant 1, which is the most common variant observed in 33% of NSCLC patients [21]. e results obtained by targeted resequencing were interpretable and in signi�cant correlation with results obtained by the commonly used methods, indicating the reliability and sensitivity of targeted resequencing in EML4-ALK gene fusion identi�cation. However, an accurate sensitivity analysis by dilution series could not be performed using the material available for our study. Regarding the probability of false positive results, targeted resequencing did not detect any anomalously paired reads from ALK region in any of 51 ALK fusion negative (FISH, real-time RT-PCR, IHC) specimens, proving the speci�city of targeted resequencing in ALK fusion detection. In addition to the detection of known ALK gene fusions, targeted resequencing holds a great advantage over other methods by enabling the simultaneous screening of novel ALK fusion partners as well as other lung cancer-related gene mutations, fusions and copy number alterations, saving time and material in the long run. In this study, we selected 192 lung cancerrelated and miRNA genes for resequencing, although we only focused on the ALK fusions here. Targeted resequencing also provides a huge amount of novel information, for instance, increasing our knowledge of resistance of some NSCLCs to tyrosine kinase inhibitor treatment. However, the method still faces many challenges before it is applicable to an average laboratory of pathology. For example, expertise is needed for analysis and interpretation of the sequencing results, as well as means to lower the cost and speed up the turnaround time. Furthermore, sequencing requires a relatively large amount of DNA, which might be difficult to isolate from small FFPE biopsies. In the present study, the amount of DNA in two analyzed specimens was less than the optimal 2-3 g (1.6 g and 1.7 g, resp.) but interpretable results were obtained.
Similarly, as targeted resequencing can be used for detection of ALK gene rearrangements and other gene fusions in lung carcinomas, it can be applied on detection of fusions in other diseases, such as in hematologic malignancies and sarcomas, in which multiple chromosomal translocations have been characterized, some of them with clinical importance [22,23].
Formalin-�xed, paraffin-embedded tumor specimens are oen the only tumor tissue material available for diagnostic purposes. Hence, the feasibility of FFPE to the detection methods is essential for reliable ALK fusion screening. In this study, we showed that FFPE material is applicable for all the four methods used for ALK gene fusion detection. However, it is important to note the low prevalence and limited selection of ALK gene fusions within the patient population, which is a common issue for the development, validation, and comparison of ALK gene fusion assays, as well

Conclusions
Our study proved that FISH, IHC, real-time RT-PCR, and targeted resequencing can be performed successfully on FFPE tumor specimen material. We found signi�cant concordance among the four methods studied. Our results support the use and further development of targeted resequencing in detection of ALK gene fusions.

Con�ict of �nterests
Lab21 is a distributor of the AmoyDx EML4-ALK fusion gene detection kit and Abbott the Vysis LSI ALK Dual Color Break Apart FISH kit. Sakari Knuutila, Kaisa Salmenkivi, and Aija Knuuttila attended advisory board meetings for P�zer and were sponsored by P�zer to attend a medical congress. Sakari Knuutila attended advisory board meeting for Abbott Molecular Inc.