The Use of Quantitative Real-Time Reverse Transcriptase PCR for 5′ and 3′ Portions of ALK Transcripts to Detect ALK Rearrangements in Lung Cancers

ALK rearrangement represents a novel molecular target in a subset of non–small cell lung cancers (NSCLC; ref. 1). Fusion of ALK with the upstream partner, EML4, was found in NSCLC in 2007 (2). This occurred because of chromosomal inversion on chromosome 2p, resulting in formation of the EML4-ALK fusion oncogene. Subsequently, other fusion variants and fusion partners, such as TFG (3) and KIF5B (4), have been identified. Notably, ALK rearrangements occur almost exclusively in adenocarcinomas (ADC; ref. 5). More recently, crizotinib (ALK inhibitor) has been recommended as first-line systemic therapy for lung cancer patients with ALK rearrangements (6). However, the method for detecting the fusion gene has not been standardized.

Several methods, including FISH, reverse transcription PCR (RT-PCR) coupled with direct sequencing and immunohistochemistry (IHC) are currently used for detecting ALK rearrangements in cytology or tissue samples. However, each technique is associated with specific strengths and weaknesses for screening. Current clinical trials of crizotinib use FISH as the diagnostic method for detecting ALK rearrangements (7). Unfortunately, FISH assay is a low-throughput assay requiring specialized technical resources and expertise and is thus not appropriate for detecting ALK rearrangements in large sample sizes. RT-PCR can define both ALK fusion partner and fusion variant, but cannot detect unknown ALK rearrangements. In addition, because there are a number of EML4-ALK variants and non-EML4-ALK fusions in NSCLCs, RT-PCR is not the optimal detection method. Previously, IHC seemed not to be economical or efficient (4). Unfortunately, ALK immune reactivity is often weak and variable, making interpretation of results challenging.

Recently, Exon arrays examining expression levels of consecutive exons have been developed for detecting gene rearrangements (8). The breakpoint in ALK consistently occurs at exon 20 with the oncogenic effect of ALK rearrangement in NSCLC being due to EML4 or other fusion partners driving expression of ALK kinase domain (exon 20 to 29), with subsequent constitutive activation (5). In contrast, the remaining non-kinase region of ALK (exon 1 to 19) is not so expressed (8). Thus, we designed primer sets to amplify 5′ and 3′ portions of ALK mRNA. If unbalanced ALK mRNA expression was seen, ALK rearrangement was assumed to exist. Compared with other techniques, this assay seems to offer higher sensitivity, reduced complexity, and a new method for detection.

A total of 654 frozen NSCLC tissue samples and their corresponding paraffin-embedded tumor tissues, including 617 (94.3%) ADCs and 37 (5.7%) adenosquamous carcinomas (ASC), were collected from Oct. 2007 to Jun. 2011, with written informed consents from all patients. Tumor was obtained as previously described (9). The study was approved by The Committee for Ethical Review of Research at Fudan University Shanghai Cancer Center. Each case was reviewed and histologic diagnosis confirmed according to World Health Organization criteria by 2 pathologists (L.S and Y.L). Four hundred and seventy-nine (73.2%) samples harbored known mutant kinases, including 400 EGFR, 53 KRAS,16 HER2 and 10 BRAF, whereas 177 samples (26.8%) did not harbor any of the above mutations (“pan-negative” samples).

Exon array was done as previously reported (8). mRNA was isolated using RNeasy Mini Kit (Qiagen) as described by manufacturer's protocol. rRNA was first removed with RiboMinus Human Transcriptome Isolation (Invitrogen) and cDNA obtained by Whole Transcript Sense Target Labeling. cDNA was fragmented and biotin-labeled using Whole Transcript Terminal Labeling (Affymetrix). Biotinylated targets were hybridized onto Affymetrix HuEX-1.0 exon array analysis (Affymetrix) following manufacturer's protocol. Arrays were scanned on GeneChip scanner 3000 7G.

Mutational status of EGFR (exons 18–22), KRAS (exons 2, 3), HER2 (exons 18–21), and BRAF (exons 11–15) were assessed with standard RT-PCR–coupled direct sequencing. For detection of EML4-ALK fusions, primers were designed to amplify all known fusion variants, using cDNA, as described previously (10).

mRNA was isolated by using RNeasy Mini Kit (Qiagen) as described by manufacturer's protocol. For mRNA analysis, an aliquot containing 2 μg of mRNA was transcribed reversely, using M-MLV reverse transcriptase and random primer (Invitrogen). Primers and conditions for qRT-PCR reactions are shown in Supplementary Table S1. qRT-PCR was done using SYBR Green PCR master mix (Applied Biosystems) for a total volume of 10 μL. Amplification protocol consisted of incubations at 95°C 10 minutes; 95°C 15 seconds, 60°C 1 minute, 40 cycles and detection by the ABI-Prism 7900HT Sequence Detection System (Applied Biosystems). Each sample was assayed in triplicate. Mean C_T values of 5′ or 3′ portion of ALK transcripts were calculated and the comparative C_T method used to determine the quantity of target sequences in 3′ portion of ALK transcripts relative to 5′ portion. Normalized expression value was calculated using the formula: 2^−ΔΔCT = ^{− [(CT (3}′^{portion of ALK) −CT (β- actin)) − (CT (5}′^{portion of ALK) −CT (β-actin))]}

ALK rearrangements were detected in formalin-fixed, paraffin-embedded specimens by FISH, using a commercially available break apart probe for the ALK gene (Vysis LSI ALK Dual Color; Abbott Molecular) in accordance with manufacturer's instructions. FISH was carried out on the same tissue area. The probe hybridizes to band 2p23 on either side of the ALK gene breakpoint. 5′ ALK signal was labeled with Spectrum Green (green), 3′ ALK signal with Spectrum Orange (orange). Criteria for probe signal interpretation in at least 100 interphase nuclei were as follows: (i) separated green and orange signals or single red signal identified in at least 15% of tumor cells analyzed indicated rearranged ALK; (ii) overlapping of red and green signals (yellowish) indicated cells in which ALK was not rearranged.

Samples having high expression in ALK kinase domain and positive for ALK FISH, but negative for EML4-ALK fusion genes, were subjected to 5′-RACE analysis. 5′-RACE was carried out using SMARTerTM RACE cDNA Amplification Kit (Clontech), as described by manufacturer's protocol. The first tailed cDNA strand was obtained from 1 μg of total RNA. cDNA was amplified by PCR with a universal primer and ALK-SP1 primer (5′-TCGTCCTGTTCAGAGCACACTTCAGG-3′) in ALK exon 22. Nested PCR amplification used the nested universal primer and ALK-SP2 primer (5′-GACACCTGGCCTTCAT ACACCTC-3′), which spans the ALK exon 21 region. Amplified PCR fragments were subjected to sequencing for identification of the fusion partner. KIF5B sequences were identified in one tumor (Case 93). RT-PCR was used to confirm the presence of KIF5B-ALK fusion transcripts in tumor and corresponding normal lung using the forward primer (5′-CAGAATTTCACAAGGAGGGAAG-3′) in exons 15 of KIF5B, and the reverse primer ALK-E20R (5′-TGCCAGCAAAGCAGTAGTTG-3′) in ALK exon 20. Amplified PCR fragments were sequenced for fusion partner identification.

Immunohistochemical staining was carried out on 4-μm thick, formalin-fixed, paraffin-embedded tissue. Heat-induced epitope retrieval was done with 1 mmol/L EDTA, pH 9.0 (Maixin). Slides were treated with 3% hydrogen peroxide for 15 minutes to block endogenous peroxidase activity. Mouse monoclonal anti-human CD246 (1:50 dilution, clone: ALK1; Dako) was applied and incubated for 120 minutes. Slides were washed in PBS and detected with horseradish peroxidase conjugated anti-mouse Envision + kit (Dako), followed by counterstaining with hematoxylin. ALK IHC scores were assigned as follows: 3+, strong staining intensity in more than 5% tumor cells; 2+, moderate staining intensity in more than 5% tumor cells; 1+, faint or weak staining intensity in more than 5% tumor cells; and 0, no staining. IHC scoring was carried out by 2 pathologists.

Data were analyzed using the Statistical Package for Social Sciences Version 16.0 Software (SPSS Inc.). Heatmap was generated using Microsoft Office Excel 2007.

Four of 81 (4.9%) ADCs harbored ALK fusions by FISH. To confirm the chromosomal translocations leading to upregulation of 3′ portion of ALK transcripts, samples were profiled on Affymetrix human Exon 1.0 arrays. Compared with expression in 5′ portion of ALK exons, 4 ALK-rearranged tumors showed higher expression level in 3′ portion (kinase domain). There were no significant differences in expression levels of ALK between 5′ and 3′ exons in 77 ALK wild-type samples (Supplementary Fig. S1). On the basis of these findings, qRT-PCR was applied to detect ALK rearrangements. Nine sets of primers were designed in 5′ portion of ALK mRNA (exons 4–18), and 5 sets in 3′ portion (exon 22–27), with qRT-PCR done in 81 ADCs. Four sets of primers in 5′ portion and 3 in 3′ portion were eligible for detecting ALK expression. Results showed unbalanced ALK mRNA expression in only ALK fusion samples, which had a significantly higher expression in the 3′ than in the 5′ portion (Fig. 1).

According to previous reports, EGFR, KRAS, HER2, BRAF, and ALK mutations are almost always mutually exclusive. To validate feasibility in a larger sample size, we included 177 “pan-negative” cases for screening of ALK rearrangements by qRT-PCR. Patient characteristics are summarized in Supplementary Table S2. Of the 177 tumors, 40 (22.6%) showed unbalanced expression in 5′ and 3′ portions of ALK mRNA and balanced expression in 137 of 177 (77.4%; Fig. 2). The normalized 3′ portion of ALK expression value ranged from 32.2 to 1573.7 (median value: 248.9) in samples with unbalanced and 0.45 to 11.8 (median value: 2.24) with balanced ALK expression. To confirm ALK rearrangements, all cases were subjected to ALK FISH and RT-PCR. Samples with unbalanced expression were confirmed positive for ALK FISH, whereas 137 tumors with balanced expression were negative (Fig. 3). ALK fusions existed in 36 of 617 (5.8%) ADCs and 10.8% (4 of 37) of ASCs. Two cases with unbalanced ALK expression were originally misdiagnosed and classified as ALK FISH-negative (cases 81 and 93). Because tumor content in case 81 was lower than 10%, the FISH result was misinterpreted, the original FISH analysis rereviewed and was incorrect. Case 93 was an ASC. The cause of misdiagnosis was due to only the ADC component being positive for ALK FISH.

According to the relative expression in 3′ portion of ALK mRNA, NSCLCs could be divided into 2 subtypes: unbalanced ALK expression (relative expression in 3′ portion higher than 32) and balanced expression (relative expression in 3′ portion lower than 12). Even using 3′ portion of ALK expression value of 12 as a diagnostic criteria, qRT-PCR showed 100% sensitivity and specificity for detecting ALK-rearranged tumors. Diagnostic accuracy then was reevaluated using 2 sets of primers (amplified 5′ and 3′ portions, respectively). High sensitivity and specificity for detecting ALK-rearranged tumors was achieved setting diagnostic criteria at 32 (Supplementary Table S3). Therefore, criteria for detecting ALK-rearranged tumors by qRT-PCR was felt best set at 12 when multiple sets of primers were applied and 32 when only 2 sets were used in the validation set. There were also 4 cases found to be negative by RT-PCR but positive for FISH and qRT-PCR. Sensitivity of RT-PCR was 90%. ALK protein expression was detected in 90% of 40 ALK positive samples, which included scores of 0 (n = 4), 1+ (n = 6), 2+ (n = 9), and 3+ (n = 21) (Supplementary Fig. S2). Relative expression levels in 3′ portion of ALK mRNA were significantly higher in samples with strong immunohistochemical staining than weak and moderate staining (665 ± 106 vs. 203.1 ± 67.7, P = 0.002; Table 1).

Tumor content of snap-frozen specimens and relative expression in 3′ portion of ALK mRNA are shown in Supplementary Fig. S3. There was a trend for tumor content more than 50% having a higher relative expression in 3′ portion of ALK mRNA (328.7 ± 374 vs. 447.7 ± 421.6), but without statistical significance (P = 0.389). Of 7 ALK-positive samples with only 10% tumor content, the relative expression in 3′ portion of ALK mRNA was 51.6–685 (median: 101.1).

Previously unidentified were 6 ALK fusion variants, including one KIF5B-ALK and 5 EML4-ALK. Case 93 was positive for qRT-PCR and FISH but negative for RT-PCR, using primers spanning between exon 20 of ALK and reported EML4-ALK variants. Therefore, 5′RACE and sequencing were carried out. A novel KIF5B variant and a fusion consisting of an in-frame fusion between exon 17 of KIF5B and exon 20 of ALK were found, confirmed by RT-PCR (Fig. 4A).

A previously reported breakpoint of ALK gene occurred in exon 20 (2, 4), one novel ALK breakpoint (case 1523) being identified, an in-frame fusion between exon 6a of EML4 and exon 19 of ALK. This fusion point was different from those present in KIF5B-ALK and EML4-ALK previously identified (Fig. 4B).

A novel isoform of EML4-ALK transcript with 53 bp deletion in an original exon 10 was observed (case 988), mRNA splicing being referred to as exon 10b and the original exon 10 as exon 10a (Fig. 4C). Also identified were 3 novel EML4-ALK variants with 18 bp (case 1494), 39 bp (case 981), and 65 bp (case 737) insertions between EML4 and ALK exon 20. Extra base pairs (cases 1494 and 981) were mapped to ALK intron 19. cDNA would thus be expected to encode a fusion protein with an insertion of 6, 10, or 33 amino acids between EML4 and ALK sequences of the protein, respectively. Another novel isoform of EML4-ALK transcript with 58 bp deletion in an original exon 17 was observed (case 981), with mRNA splicing being referred to as exon 17b (31 bp) and the original exon 17 as exon 17a (69 bp; Supplementary Fig. S4).

We carried out this study to evaluate the potential role of qRT-PCR as a diagnostic method for ALK-rearranged lung cancers. Its importance in NSCLCs relates to the apparently high predictive value of the gene translocation with regard to tumor sensitivity to ALK inhibitors. Fusion gene frequency is approximately 3% to 7% (1), suggesting that more than 90% of patients will not benefit from ALK inhibitors. Therefore, accurate and timely identification of such patients have important therapeutic implications.

We present a novel method for the diagnosis of lung cancers harboring ALK rearrangements. By using a qRT-PCR assay to detect expression in 5′ and 3′ portions of ALK transcripts, a cohort of 177 “pan-negative” lung cancers were screened, of which 40 (22.6%) samples harboring ALK arrangements were found. High sensitivity and specificity were achieved when using multiple sets of primers, selecting the relative 3′ portion of ALK expression value more than 12 as diagnostic. Therefore, qRT-PCR, alone, seems sufficient for diagnosis, pending finding new ALK fusion variants.

qRT-PCR offers high sensitivity, suggesting that the greater relative expression in 3′ portion of ALK mRNA has a high probability of representing an abnormal ALK gene. Currently, FISH, RT-PCR, and IHC are being evaluated for detecting ALK rearrangements. Although FISH is used for screening in clinical trials, it has limitations. As EML4-ALK results from a chromosomal inversion on chromosome 2p and because the 2 genes are normally only approximately 12 Mb apart, split signals may be subtle and easily overlooked (11). Previously, sensitivity varied between 93% and 98.6% (12–14). ALK rearrangements are infrequent and thus considered low-throughput so that FISH is not suitable for detecting ALK rearrangements in large sample sizes. Because FISH requires specialized technical resources (such as fluorescence microscope) and expertise, it is not readily available. IHC is a readily available diagnostic tool, with 3 specified antibodies for ALK (ALK1 (15), D5F3 (12), 5A4 (16). Initial attempts to use ALK antibodies detecting ALK-rearranged tumors were disappointing (17). However, more sensitive ALK antibodies and techniques enhanced immunohistochemical signals so that immunohistochemical sensitivity has greatly improved. For example, the tyramide amplification method (using antibody ALK1) improves the ability to detect from 40% to 80% of specimens defined by FISH (12). By using IHC scoring, one report had a sensitivity of 90%, suggesting IHC can be used for screening in surgically resected specimens (15). We also used an ALK1 antibody (Dako) in our initial screening attempts, but some ALK rearranged tumors only displayed faint staining and a few wild-type tumors were positive (data not shown). These results showed that IHC may be promising but not optimal for screening.

qRT-PCR is rapid, simple, relatively high-throughput, and inexpensive for detecting ALK rearrangement. This assay can be completed in 4 hours, with an easy-to-use protocol. Cost per patient is less than 10 dollars using SYBR Green PCR master mix (Applied Biosystems). ALK FISH assays, however, require 2 days and at least $100 for each sample with the commercially available ALK probe (Vysis LSI ALK Dual Color; Abbott Molecular) and other FISH reagents. Furthermore, qRT-PCR can detect 16 samples with the 96-Well Detection System and 64 samples with the 384-Well Detection System by using 2 sets of primers each time. This relatively high-throughput assay is thus suitable for clinical screening.

Several factors, including loss of 5′ materials shown by FISH, polysomy, and tumor content can influence assay performance. Although expression level is very low or absent for wild-type ALK, possibly loss of the 5′ ALK portion can lead to higher relative expression in the 3′ portion. In this study, 4 samples (cases 654,781, 563, and 636) had loss of 5′ material by FISH and showed high expression in the 3′ portion. Expression levels in 3′ portions of ALK mRNA were 78.4- (case 654), 103.2- (case 781), 879.2- (case 563), and 300-fold (case 636) higher than the 5′ portion, respectively. Also, polysomy may influence performance of this method. In one study, a few lung cancers harbored multiple copies of rearranged chromosome (14). Thus, rearranged ALK mRNA expression level would be higher and more easily detected. However, we found no such cases. In addition, we showed that real-time RT-PCR is less prone to interference from nontumor cells, with 10% tumor content being sufficient. Low tumor content samples, such as malignant pleural effusion and small biopsy samples, may be efficient and sensitive for detection.

Although our study shows the promise of real-time PCR for detecting ALK fusion, the inherently demanding nature of RNA may hamper its wide clinical application. Thus, careful sample processing for obtaining high quality RNA is important. Because reagents and instruments for real-time PCR may be different, each laboratory trying to use this method should establish its own cut-off values. The enzyme might also sometimes decrease by using OligodT primers during reverse transcription, so that the 5′ end is absent. Therefore, for obtaining more accurate results, random primers were used.

In conclusion, with wider clinical use of the ALK inhibitor crizotinib timely and accurate diagnosis of ALK arrangement NSCLCs are essential to identify those patients suitable for this target therapy. We believe that qRT-PCR assay is a preferred method for identifying such tumors in clinical practice.

No potential conflicts of interest were disclosed.

Conception and design: R. Wang, Y. Pan, X. Luo, Y. Sun, H. Chen

Development of methodology: R. Wang, Y. Pan, X. Luo, Y. Sun, H. Chen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Wang, C. Li, Y. Zhang, X. Luo, Z. Fang, H. Ji, Y. Sun, H. Chen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Wang, H. Li, X. Luo, Z. Fang, H. Ji, Y. Sun, H. Chen

Writing, review, and/or revision of the manuscript: R. Wang, C. Li, H. Li, X. Luo, J. Zhang, Z. Fang, D.H. Garfield, Y. Sun, H. Chen

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Wang, Y. Pan, C. Li, H. Hu, X. Luo, Y. Li, L. Shen, H. Ji, Y. Sun, H. Chen

Study supervision: X. Luo, J. Zhang, H. Chen

The authors thank Qiong Lu, Shilei Liu, and Shihua Yao for their technical support.

This work was supported by the funds from National Natural Science Foundation of China (81101761, 81172218), key construction program of the National “985” Project (985-YFX0102), grants no. WJP1101 from Wu Jieping Foundation, and Grant from Shanghai Municipal Health Bureau.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.