Patients with non–small cell lung cancer harboring ALK receptor tyrosine kinase (ALK), ROS proto-oncogene 1 (ROS1), and ret proto-oncogene (RET) gene rearrangements can benefit from specific kinase inhibitors. Detection of fusion genes is critical for determining the best treatment. Assessing rearrangements in non–small cell lung cancer remains challenging, particularly for lung cytology.
To examine the possible application of the multiplex, transcript-based NanoString system (NanoString Technologies, Seattle, Washington) in the evaluation of fusion genes in lung adenocarcinoma samples.
This study is a narrative literature review. Studies about NanoString, gene fusions, and lung adenocarcinoma were collected from PubMed (National Center for Biotechnology Information, Bethesda, Maryland). We found 7 articles about the application of the NanoString system to detect fusion genes on formalin-fixed, paraffin-embedded tumor tissues and one article evaluating the adequacy of lung cytologic specimens for NanoString gene expression analysis.
To maximize the yield of molecular tests on small lung biopsies, the NanoString nCounter system has been suggested to detect fusion genes. NanoString fusion gene assays have been successfully applied on formalin-fixed, paraffin-embedded tissues. Although there are only a few studies available, the application of NanoString assays may also be feasible in lung cytology. According to available data, the NanoString system could strengthen the routine molecular characterization of lung adenocarcinoma.
Lung cancer is the leading cause of cancer-related mortality worldwide, despite improvements in diagnosis and treatments.1 The most common histologic type is non–small cell lung cancer (NSCLC), which accounts for 85% of all lung cancer cases. In past years, the identification of specific genomic alterations has led to the development of new, targeted therapies in subgroups of patients with NSCLC. Mutations in the epidermal growth factor receptor (EGFR) were the first genetic lesions associated with a favorable response to tyrosine kinase inhibitors (TKIs) in patients with advanced NSCLC.2 In addition to EGFR mutations, NSCLC cases harboring ALK receptor tyrosine kinase (ALK) gene rearrangements respond well to specific kinase inhibitors, such as the US Food and Drug Administration (FDA)–approved crizotinib.3,4 The most common rearrangements occur between ALK and the echinoderm microtubule-associated protein-like 4 (EML4). The fusion gene is generated by an inversion of chromosome 2, which juxtaposes the 5′-end of EML4 with the 3′-kinase domain of ALK, permitting constitutive kinase activation and inducing oncogenic cell transformation, as observed in vivo and in vitro.4 Thirteen variants of the EML4-ALK fusion have been described, depending on the breakpoint of EML4 (from exon 2 to exon 20), and variants 1 and 3a/b recur the most often, accounting for 33% and 29% of cases, respectively. Furthermore, in addition to EML4, other ALK fusion partners have been reported, such as TRK-fused gene (TGF), kinesin family member 5B (KIF5B), and kinesin light chain 1 (KLC1).5–7 The incidence of ALK rearrangement in patients with NSCLC is approximately 3% to 5%, which is associated with younger age and light or no smoking history.8–10
Other gene fusions have been identified in patients with NSCLC as oncogenic drivers that can be targeted with existing TKIs. Recently, the FDA approved the use of the TKI crizotinib for patients with NSCLC harboring rearrangements of the ROS proto-oncogene 1 (ROS1).11–13 Further, reported molecular alterations involve the MET proto-oncogene (MET) and ret proto-oncogene (RET) genes.14–15 Therefore, detection of gene fusions, as well as EGFR mutations, is critical in the therapeutic management of patients with advanced NSCLC. However, assessing rearrangements in lung tumors remains a challenge because most available diagnostic methods separately investigate each fusion, requiring a lot of tissue, which is difficult to obtain from lung biopsies.
To date, fluorescence in situ hybridization (FISH) is the gold standard diagnostic method, but it is expensive, requires technical expertise that may not be readily available in a histopathology laboratory, and is not amenable to multiplexed analysis. Therefore, it may not be practical to screen for multiple gene fusions with FISH in every patient with NSCLC.16 Besides FISH, immunohistochemistry (IHC) assays have been validated and standardized for ALK and ROS1 rearrangements, which are relatively inexpensive, fast, and may be performed routinely in most laboratories, but they are usually performed on formalin-fixed, paraffin-embedded (FFPE) samples.17
Gene fusions can also be analyzed at the RNA level using reverse transcription–polymerase chain reaction tests. These tests provide the most-precise information about fusion patterns, but they require multiple reactions with specific primers, can miss unknown variants, and can risk false-negative results because of poor quality and quantity of RNA from FFPE and cytologic specimens.18,19 A sensitive, reverse transcription–polymerase chain reaction method was developed to detect the overexpression of 3′ regions of fusion transcripts for tumor genes that are usually repressed in lung tissues, such as ALK.20 That approach could be used for screening patients with ALK+ disease, but it is not adequate for ROS1, which is commonly expressed in healthy and hyperplastic lung tissues.11,21 In daily clinical practice, most patients with NSCLC at the time of diagnosis are at an advanced stage; therefore, surgery is seldom recommended. In those cases, the diagnostic tumor material comes from biopsy or cytologic specimens.22
Cytologic samples provide the only source of diagnostic material in more than 50% of patients with NSCLC, and the analysis of predictive biomarkers can be challenging because of the quantity, quality, and applicability of available diagnostic tests.5 Moreover, even when biopsy or cytologic material is assessed to contain an adequate representation of tumor cells and is processed correctly and managed carefully, the tissue may be insufficient to detect multiple gene fusions. In fact, multiple sections are necessary from small biopsy samples or cytology cell blocks beyond the sections for mutation analysis of EGFR. In the same way, multiple slides are required for molecular analysis of distinct gene fusions on cytologic smears.23,24
Therefore, there is an urgent need for a multiplex method that can screen a larger array of clinically informative targets with minimal tissue samples.
Recently, various transcript-based methods have been used to detect ALK, RET, and ROS1 fusion genes using FFPE samples from patients with NSCLC as well as with NSCLC cell lines. These assays are primarily based on the NanoString (NanoString Technologies, Seattle, Washing), Agena (Agena Bioscience, San Diego, California), and next-generation sequencing platforms.21,25–27
According to the data available in the literature, among the techniques cited above, NanoString is less sensitive to the preanalytic treatment of samples, such as fixation effects, because it allows direct counting of messenger RNA (mRNA) molecules. Furthermore, it is extremely easy to use and requires less hands-on time; it seems to be a rapid, sensitive method, which is a good candidate for use in clinical practice.25,26
The NanoString nCounter technology, based on a dual-probe system, provides a laboratory the ability to analyze up to 800 target genes per reaction through molecular fluorescent barcodes. The digital quantitation of nucleic acids is made by direct counting of target molecules without reverse transcription or amplification steps28 ; this aspect makes NanoString adequate for even degraded clinical samples, in which extracted nucleic acids are not easily amplified in polymerase chain reaction.29,30
NanoString assays include 3 primary steps: (1) the RNA is hybridized with probe pairs (reporter probe and capture probe), which are designed to be adjacent to one another along the target sequence of a transcript. The reporter probes are 5′ labeled with a 6-fluorophore sequence whose combination is unique for each target, and the capture probes are 3′ biotinylated. After the hybridization of capture and reporter probes with mRNA molecules in a 1:1 ratio, a stable tripartite structure is formed, and the excess probes are washed away with 2-step magnetic-bead purification; (2) the tripartite structure is bound to the surface of the sample cartridge that is coated with streptavidin, and reporters are aligned by an electric current and immobilized for data collection; (3) the sample cartridge is scanned by the digital analyzer, and each fluorescent barcode is counted and tabulated according to the gene identity (Figure 1).28–30
Graphical representation of nCounter gene expression system. Hybridization: the biotinylated capture probe and the barcoded reporter probe hybridize to a complementary target mRNA. Purification: once the tripartite complex is created, the excess probes are removed through a process of affinity-purification. Immobilization and alignment: purified complexes are immobilized to a streptavidin-coated cartridge by the biotinylated capture probe, and then they are aligned by applying an electric field. Digital counting: the aligned complexes carrying specific barcodes are digitally read. The level of expression is measured by counting the number of codes for each mRNA.
Graphical representation of nCounter gene expression system. Hybridization: the biotinylated capture probe and the barcoded reporter probe hybridize to a complementary target mRNA. Purification: once the tripartite complex is created, the excess probes are removed through a process of affinity-purification. Immobilization and alignment: purified complexes are immobilized to a streptavidin-coated cartridge by the biotinylated capture probe, and then they are aligned by applying an electric field. Digital counting: the aligned complexes carrying specific barcodes are digitally read. The level of expression is measured by counting the number of codes for each mRNA.
In this review, we will examine the possible application of the multiplex, transcript-based NanoString system in the evaluation of fusion genes in lung adenocarcinoma.
DATA SOURCES
To systematically collect studies of interest PubMed (National Center for Biotechnology Information, Bethesda, Maryland) was searched by separately using several combinations of the following key words: NanoString, lung adenocarcinoma, lung cancer, non–small cell lung cancer, gene fusions, ALK, ROS1, and RET; only English language publications were considered. We found 11 articles published from 2012 to 2017; 7 articles (64%) studied the application of NanoString to detect fusion genes in lung adenocarcinoma samples, 3 articles (27%) were about the analysis of deregulated lung adenocarcinoma genes and pathways, and 1 article (9%) evaluated the adequacy of lung cytologic specimens for NanoString gene-expression analysis.
NanoString fusion gene assays have never, to our knowledge, been evaluated on cytologic specimens, which are often barely sufficient for FISH or IHC analysis, so it is difficult to evaluate other molecular tests. Here, we review the current literature on detection of lung cancer fusion genes by NanoString and describe our own experience in detecting ALK, ROS1, and RET fusions using FFPE and cytologic lung adenocarcinoma specimens.
NANOSTRING AND LUNG ADENOCARCINOMA FUSION GENES
In 2012, Suehara and colleagues31 used a NanoString assay that was specific for the transcripts of 90 tyrosine kinases to evaluate imbalances between the 3′ kinase domain and 5′ exons for the first time. Indeed, most gene fusions lead to the formation of a chimeric fusion protein and cause differences in RNA expression levels of the exons that are 5′ or 3′ to the fusion point.31 First, they validated the performance of NanoString assay on 75 lung adenocarcinoma samples (6 [8%] from fresh-frozen tissues; 69 [92%] from FFPE blocks) of which 24 (32%) were ALK+, and 51 (68) produced negative results by FISH. NanoString assay correctly determined ALK status in 74 of the 75 cases (99%). Furthermore, by examining serial dilutions of RNA from cell lines harboring known fusion genes, they found that samples with at least 25% ALK-fusion–positive tumor cells could be detected with this platform. Then, they analyzed 69 additional lung adenocarcinoma samples that had negative results for the most-frequent mutations in the KRAS proto-oncogene, GTPase (KRAS), EGFR, B-Raf proto-oncogene, serine/threonine kinase (BRAF), mitogen-activated protein kinase 1 (MAP2K1), erb-b2 receptor tyrosine kinase 2 (ERBB2), and for ALK fusions. They identified 2 samples with ROS1 and RET fusions. A deeper molecular investigation revealed that the 2 samples harbored novel gene rearrangements (GOPC-ROS1 and KIF5B-RET), which had never been described in lung adenocarcinoma.31
Two advantages of the NanoString system were clear: (1) the low amount of RNA used in the study (100–250 ng of total RNA), and (2) the detection of gene rearrangements by evaluating the balance or imbalance between 3′ and 5′ regions of mRNA, which does not require a prior knowledge of fusion variants.
The interrogation of imbalanced 3′:5′ expression is ideal for ALK and RET, whose endogenous transcript levels are low in lung tissue in the absence of rearrangements. In contrast, it may not be sensitive enough for ROS1, whose endogenous transcript may obscure fusion 3′ overexpression11,20,21 ; hence, the direct detection of fusion transcripts might be helpful.21,32
Considering that point, Lira and colleagues32 first designed a multiplex NanoString assay specific for ALK fusions, and in a second step, they improved and extended the technology to allow for the simultaneous analysis of ALK, ROS1, and RET rearrangements.21
Generally, in the Lira et al32 NanoString assays, the detection of fusion transcripts depended on a dual strategy: (1) the use of reporter and capturing breakpoint-specific probes for the most common ALK, ROS1, and RET rearrangements and (2) the use of pairs of probes that targeted wild-type 3′ and 5′ gene regions of mRNA to evaluate their balances or imbalances without an upfront knowledge of fusion variants. NanoString panels always include probes for 6 spike-in–positive controls and for 8 spike-in–negative controls to determine the technical reliability of the analyses and probes for housekeeping genes, to normalize samples, and to evaluate the quality and quantity of the RNA. Data analysis usually includes a 2-step normalization procedure, which is based on spike-in–positive controls and housekeeping genes, a background removal based on the spike-in–negative controls, and fusion prediction that is based on both the 3′:5′ ratio (positive, if higher than a prespecified threshold) and fusion probe expressions. However, only a high 3′:5′ ratio may be evident, indicative of a fusion transcript from a novel or a rare variant, which specific fusion probes would miss.21,32 It may also happen that a sample is positive for fusion-specific probes but negative for the 3′:5′ imbalance, a situation likely to be a technical artifact that usually requires further investigation.25
In their first study, Lira et al32 validated the NanoString assay specific for the ALK gene, which included 8 pairs of probes for 3′ and 5′ ALK regions, 7 pairs of probes for ALK known fusion variants, and 4 pairs of probes for housekeeping genes (GAPDH, AZIN1, POLR2A, GUSB). By testing 66 archival FFPE NSCLC samples, of which 34 ALK samples had positive results by FISH, and using 500 ng of total RNA, the authors obtained a concordance with either FISH or IHC equal to 93%, and they also managed to detect low-abundant ALK fusion transcripts when the tumor cell content was as low as 10%.
Then, the same group validated the NanoString assay for ALK, ROS1, and RET fusions on 295 surgically resected lung adenocarcinoma samples (211 FFPE [72%] and 84 fresh-frozen specimens [28%]) enriched for gene fusion–positive samples (104 ALK+, 7 ROS1+, and 14 RET+ sample results by FISH). In that case, the nCounter CodeSet included 24 probe pairs targeting wild-type 3′ and 5′ regions of ALK, ROS1, and RET, 27 probe pairs specific for the most-common gene fusions, and 4 probes for the same housekeeping genes as before. NanoString system achieved 100% concordance with FISH for all genes, whereas the concordance between IHC and NanoString for ALK detection was 97.8%. Moreover, this approach allowed the researchers to identify a RET fusion involving a novel partner, further characterized as the cutlike homeobox 1 gene (CUX1). In that validation setting, the total amount of RNA input was 500 ng for FFPE and 250 ng for fresh-frozen tissues. However, the authors proved that 50 to 100 ng of total RNA would be sufficient to predict the presence of fusion transcripts with good accuracy.21 In 2016, the NanoString panel for ALK, ROS1, and RET designed by Lira et al32 was used to evaluate the presence of gene fusions on 214 lung squamous cell carcinomas with an extremely low incidence. Authors used 500 ng of total RNA, and none of the screened tumor tissues harbored gene rearrangements, thus confirming the rare prevalence of ALK, ROS1, and RET fusions in lung squamous cell carcinomas.33
In the past 2 years, 4 additional research groups used NanoString assays for fusion gene detection in lung adenocarcinoma following the Lira et al32 approach for both assay design and data analysis.
Reguart et al25 evaluated an nCounter CodeSet, including 23 probe pairs designed to bind to 27 specific fusion transcripts; 24 probe pairs targeting wild-type ALK, ROS1, and RET 3′ and 5′ regions; and 4 probe pairs for the same housekeeping genes used by Lira.32 Reguart et al25 tested 108 FFPE tissue samples from patients with advanced NSCLC, 98 samples (91%) were successfully analyzed by nCounter, and concordance rates with FISH were 87.5% and 85.9% for ALK and ROS1, respectively. Using FFPE blocks derived from cell lines, they found that a tumor cell content of 10% and 25 ng of total RNA were sufficient for the successful detection of all fusion transcripts; however, 200 ng was established as the optimum amount.25
Lindquist and colleagues34 investigated the clinical potential of NanoString multiplexed gene-fusion analysis on 169 FFPE lung cancer tissues from a Swedish cohort. The NanoString success rate was 80%; 5 ALK+, 3 RET+, and 2 ROS1+ case results were identified in complete agreement with FISH, and no FISH-positive case was missed by the NanoString assay. The total input of RNA used in that study was 100 to 250 ng, and by dilution experiments, the authors found that the assay could detect fusion genes in a sample with 5% or more tumor cells.34
The NanoString system is not the only multiplex method under investigation for the analysis of fusion genes. Recently, Rogers et al26 compared 3 different transcriptome-based methods: (1) a NanoString assay, (2) an Agena LungFusion panel (Agena Bioscience, San Diego, California), and (3) a next-generation sequencing fusion panel (Thermo Fisher Scientific, Waltham, Massachusetts) to detect clinically relevant fusion genes in 51 surgically resected NSCLC samples.26 The NanoString assay used in that study included 29 probe sets, based on the Lira et al32 procedure, targeting several known fusion variants, and the same 4 housekeeping genes for content normalization of samples. Total input of RNA was from 100 to 500 ng; 3 (6%) of the samples failed the NanoString analysis, and the overall agreement with FISH was 96%; the overall agreement between FISH and Agena and between FISH and next-generation sequencing fusion panel was 94% and 86%, respectively.26
Li and colleagues35 characterized oncogenic drivers in lung adenocarcinoma in an Asian population and profiled 271 tumors for gene-fusion analysis, using the Lira et al32 assay implemented with a hybridization probe spanning MET gene exons 13 and 14. The innovation of this study consisted of the use of a single NanoString assay to detect not only fusion genes but also the presence of MET exon 14 skipping. They found 7.4%, 2.2%, 2.2%, and 1.8% of ALK-, ROS1-, RET-, and MET-altered tumors, respectively.
From the studies described, we assess nCounter multiplex assays can detect ALK, ROS1, and RET fusion transcripts, representing a cost-effective strategy for predictive medicine in lung cancer. Indeed, NanoString assays work on minimal amount of FFPE RNA, have high sensitivity, and save time. Regardless of the nCounter CodeSet design and considering all the gene fusions, the reported concordance with the gold standard FISH was always greater than 85%. Unfortunately, few data are available about clinical response to TKI of discordant NanoString and FISH cases. In this context, Reguart et al,25 who found the lowest concordance between NanoString and FISH, collected clinical data from 25 fusion-positive patients who achieved a good response to TKI: 24 (96%) were positive by the nCounter assay and 22 (88%) by FISH. In detail, 3 patients (12%) who were ALK+ by nCounter but had negative FISH results responded well to TKI. Those results are in agreement with articles reporting TKI response in some patients who were ALK− by FISH but had positive IHC results.36 Of course, further clinical investigation is needed to establish whether NanoString may substitute for FISH in identifying fusion-positive patients.
In addition, NanoString was reported to be highly compatible with low-quantity and low-quality RNA, probably because it is based on hybridization without enzymatic reactions. It seems that the best total input of RNA from FFPE is about 200 to 300 ng to ensure a good analytical performance, but 25 to 50 ng might be enough.21,25 Although quality mRNA parameters for NanoString analysis have yet to be assessed, the screening failure was acceptable in most studies. Indeed, Reguart and colleagues25 did not verify the quality of the RNA before NanoString analysis, and they had a 6% screening failure. Rogers et al26 assessed RNA yield and integrity before transcript-based analysis using a Bioanalyzer assay (Agilent, Santa Clara, California). As expected with FFPE-derived material, most of the samples had an RNA integrity number less than 3, so the percentage of RNA fragments greater than 300 bases was considered to estimate RNA integrity, and they had a screening failure similar to that one reported by Reguart et al25 (3 of 51 samples [6%]). Lindquist et al34 obtained the highest NanoString failure rate, equal to 20%, which could be due to RNA degradation in the FFPE blocks caused by the fixation process. Interestingly, the authors found that the proportion of inconclusive NanoString cases was equivalent to the 17% of cases with an inconclusive ALK status by FISH or IHC in the analyzed clinical cohort. However, they did not find a significant association between an inconclusive ALK FISH/IHC and an inconclusive NanoString analysis; it is likely that different degradation processes specifically interfere with each technique. NanoString is more practical and advantageous for fusion-genes analysis than current standard diagnostic methods. In comparison to FISH and IHC, some NanoString assays allowed not only the ability to determine the presence of gene rearrangements but also to identify specific, known variants. The ability to identify known gene rearrangements depends on how reporter and capture probes are designed. For instance, regarding determination of ALK-known fusions, Lira and colleagues21 designed reporter probes targeting exon 20 and capture probes specific for the most-common fusion partners. Because the same ALK exon 20 reporter probes are used for more than one variant, it is not possible to discriminate among all of them. Although the discrimination among fusion partners is reliable using the Reguart et al25 approach, which is based on the use of a universal capture probe that is specific for ALK exon 20 and reporter probes with different barcodes specific for each fusion partner. The characterization of fusion variants may have an important role in clinical practice, considering that recent evidence showed that fusion partners influence TKI efficacy.25 In addition, the use of appropriate analysis software leads to an objective determination of gene rearrangements.21,25
Finally, new multiplex transcript-based diagnostic devices for fusion assays have been validated for in vitro diagnostic use under the 98/79/EC directive, such as the RealQuant Lung Fusion Genes kit (Diatech Pharmacogenetics, Jesi, Italy) based on the NanoString technology.37 That device was validated on FFPE lung adenocarcinoma samples in a multicenter clinical study created from the European Union's Horizon 2020 research and innovation program and approved by competent ethical committees. That study involved our Unit of Pathological Anatomy at the University Hospital of Pisa (Pisa, Italy) and the Biosciences Laboratory of the Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori (Meldola, Italy). The RealQuant lung fusion genes kit includes reporter and capture breakpoint-specific probe pairs for 21 ALK, ROS1, and RET known rearrangements, 24 probe pairs targeting wild-type ALK, ROS1, and RET 3′ and 5′ regions of mRNA, 21 probe pairs for housekeeping genes, and 6 probe pairs for tumor-associated genes that are differentially expressed in tumors and healthy lung tissues to ascertain that the percentage of tumor cells is sufficient. Along with the RealQuant lung fusion genes kit, the iGenetics RealQuant analysis software (Diatech) is provided to evaluate gene rearrangements. As reported by the manufacturers on the Diatech Web site, the concordance of their NanoString-based device with FISH was 97.7%, 100%, and 100%, for ALK, ROS1, and RET, respectively.37 Although one of the main advantages of the nCounter gene expression system is the opportunity to customize gene panels according to the routine necessity of individual laboratories, each custom panel has to be validated before its use in clinical practice. The availability of validated assays for in vitro diagnostic use could be advantageous, particularly for clinical laboratories that do not perform research activities. For that reason, we decided to participate in the validation of the RealQuant lung fusion genes kit, which provides the same advantages reported for the NanoString probe CodeSets.
NANOSTRING AND LUNG ADENOCARCINOMA FUSION GENES ON CYTOLOGY
The NanoString system have been demonstrated to be adequate for FFPE specimens because it does not require high-quality materials and works on a small amount of RNA. Because of those same characteristics, the system could be applied to detect fusion genes in cytologic samples, which are the only source of material in many lung adenocarcinomas. Indeed, NSCLC specimens are typically obtained by endobronchial ultrasound fine-needle aspiration (FNA), transthoracic FNA, bronchial secretions or brushes, bronchoalveolar lavage, pleural effusions, or FNA from other metastatic sites. Processing cytologic specimens for FFPE cell blocks is an elective method because it allows the samples to be treated the same as histologic samples.38,39 However, cell blocks sometimes have an insufficient number of cancer cells for molecular analysis, and discriminating tumor cells from reactive cells is more difficult than in conventional cytology.40 By contrast, conventional cytology (smears, cytospins, or liquid-based preparations) is advantageous, allowing the selection of an optimal cytologic slide, and DNA and RNA quality in air-dried or alcohol-fixed cytologic specimens is better than after formaldehyde fixation, which causes cross-linking and chemical modification of nucleotides.41 A FISH analysis of fusion genes is applicable to almost all types of cytologic samples.42 However, each FISH test requires a cytologic slide, and often, it is impossible to have the number of slides equal to the number of genes to be tested, considering that cytologic specimens should be archived for potential reevaluation and legal purposes.
Although IHC can be a valuable method for analyzing the expression of fusion proteins on cell blocks,43,44 its application can be challenging for cytologic smears because it may be greatly influenced by preanalytic factors. Moreover, additional studies are needed to validate various antibody clones and revelation systems for the latter sample type.
There is limited literature on the feasibility of the NanoString system to analyze RNA from lung cytologic smears, except for a recent study by Sgariglia and colleagues.45 In that study, RNA was extracted from 12 archival NSCLC stained (Papanicolaou or Diff-Quik), cytologic smears with 30% or more neoplastic cells and 100 or more cells. The quality and quantity of the RNA were determined with the 4200 Tape Station (Agilent); RNA concentration ranged from 3.24 to 12.96 ng/μL, and the RNA integrity number was greater than 3 for 11 of 12 cases (92%). A total of 15 ng of RNA was used for each sample to perform a 48 gene-panel nCounter analysis. NanoString data were analyzed with the nSolver software (NanoString) and 11 of 12 samples (92%) were adequate, fulfilling the quality-check parameter. Although preliminary, that study showed that the NanoString system deserves further investigation on lung cytologic smears.
We performed a preliminarily evaluation of whether the RealQuant lung fusion genes kit could be applied on samples beyond FFPE, such as cytologic specimens, and its concordance with FISH. We retrospectively selected as many FISH ALK+, ROS1+, and RET+ samples as possible from our archival materials, considering the low incidence of those rearrangements. Finally, we compared NanoString and FISH results of 17 FFPE tissues (13 whole tissue sections from surgical procedure [76%], 2 preoperative small biopsies [12%], and 2 cell blocks [12%] from pleural effusions) and 11 cytologic lung adenocarcinoma samples (exfoliative in 3 cases [27%] and FNA in 8 cases [73%]), obtaining a high degree of concordance in both groups (Table 1).
Molecular Status of Tumors: Results of Fluorescence In Situ Hybridization (FISH), MassARRAY, and NanoString for all the Patients Analyzed
The ALK, ROS1, and RET status was determined by FISH, and ALK FFPE-positive samples were also analyzed by IHC, as previously reported.46 In detail, 15 FFPE samples were positive by FISH: 6 (40%) for ALK (example in Figure 2), 7 (47%) for ROS1 (example in Figure 3, A and B), and 2 (18%) for RET (example in Figure 3, C and D); 2 FFPE samples were triple negative for the gene fusions mentioned. For the cytologic smears, 6 of 11 samples (55%) had positive results by FISH, 5 for ALK (example in Figure 4, A and B) and 1 for ROS1 (Figure 4, C and D), whereas 5 (45%) had triple-negative FISH results.
![Figure 2. A, Surgical specimen of lung adenocarcinoma. B, Immunohistochemical staining for the ALK receptor tyrosine kinase (ALK) protein revealing strong, granular cytoplasmic expression. C, Fluorescence in situ hybridization analysis of the same sample revealing cells with a characteristic ALK translocation (hematoxylin-eosin, original magnification ×20 [A]; original magnification ×20 [B]; original magnification ×60 [C]).](https://allen.silverchair-cdn.com/allen/content_public/journal/aplm/142/4/10.5858_arpa.2017-0135-ra/3/m_i1543-2165-142-4-480-f02.png?Expires=1750243977&Signature=slc~YxP27TsM4F738fqliU2kZy7l0fH83PwYzSWyZrdty1EacmgSqzctz81CUKesijo1YYZECOjUzvbCCpyUL2P2fUjs8Cs~bfZBPFUu3cblmqjgg7lvKDqPQnWsJ5exDh7FUBCY75r-VDSqwTJpOvz3XS0YZStTz0KoYIKaMEL6TW5dwLEDKA65P4piRSS7yV~-hymudVH449IZzjoEuhrhbnCUxjpBjWtUsMwrZGZzImQLKGisPiddwEVNlZyUruMPYRB97VlHDHl6duORc0V0kVoR4DHYORbI7UPxSC1DRv0WnivlHr-4YA4l-48AHv204plkkZq7zD7ywgP4YA__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
A, Surgical specimen of lung adenocarcinoma. B, Immunohistochemical staining for the ALK receptor tyrosine kinase (ALK) protein revealing strong, granular cytoplasmic expression. C, Fluorescence in situ hybridization analysis of the same sample revealing cells with a characteristic ALK translocation (hematoxylin-eosin, original magnification ×20 [A]; original magnification ×20 [B]; original magnification ×60 [C]).
A, Surgical specimen of lung adenocarcinoma. B, Immunohistochemical staining for the ALK receptor tyrosine kinase (ALK) protein revealing strong, granular cytoplasmic expression. C, Fluorescence in situ hybridization analysis of the same sample revealing cells with a characteristic ALK translocation (hematoxylin-eosin, original magnification ×20 [A]; original magnification ×20 [B]; original magnification ×60 [C]).
![Figure 3. A and C, Two surgical samples of a lung adenocarcinoma that were positive for ROS proto-oncogene 1 (ROS1) gene rearrangement (B) and for ret proto-oncogene (RET) gene rearrangement (D) on fluorescence in situ hybridization (hematoxylin-eosin, original magnification ×20 [A and C]; original magnification ×60 [B and D]).](https://allen.silverchair-cdn.com/allen/content_public/journal/aplm/142/4/10.5858_arpa.2017-0135-ra/3/m_i1543-2165-142-4-480-f03.png?Expires=1750243977&Signature=UJIM2rlM8HI6ERYdbf5NHm80mdw5zK6fMUR0HWGFj4OhHxKDmsPwUWpur5I6v9X10AVaeJEnJJARZLsnjV29JpkRn0I70BIOkeXXOdcg2JGUS8T8Nqy8AiLAu5Qo4acfTOv-k6vE~zSwbVhxo2VeeCch7W41r1hm0gKAi0VKi9ZN56oHg5a27InXs2J7gXJtFLSQXyEX8grl3ksX1Z3g~prNgD0K6bmjKbBDpGnuE33rvjVWe1GKL2t5VpoaRYn4pTrk~VzDLm0ZfifLq~3FZcjw~6nRcsdwIbMifcJTjdueA5Wc1536v5IrrKNGwj58Ys2kGe4dcrzhjybS1ivkrQ__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
A and C, Two surgical samples of a lung adenocarcinoma that were positive for ROS proto-oncogene 1 (ROS1) gene rearrangement (B) and for ret proto-oncogene (RET) gene rearrangement (D) on fluorescence in situ hybridization (hematoxylin-eosin, original magnification ×20 [A and C]; original magnification ×60 [B and D]).
A and C, Two surgical samples of a lung adenocarcinoma that were positive for ROS proto-oncogene 1 (ROS1) gene rearrangement (B) and for ret proto-oncogene (RET) gene rearrangement (D) on fluorescence in situ hybridization (hematoxylin-eosin, original magnification ×20 [A and C]; original magnification ×60 [B and D]).
![Figure 4. A and C, Two cytological samples of lung adenocarcinoma that were positive for ALK receptor tyrosine kinase (ALK) gene translocation [B] and for ROS proto-oncogene 1 (ROS1) gene rearrangement [D]) on fluorescence in situ hybridization (Papanicolaou, original magnification ×20 [A and C]; original magnification ×100 [B and D]).](https://allen.silverchair-cdn.com/allen/content_public/journal/aplm/142/4/10.5858_arpa.2017-0135-ra/3/m_i1543-2165-142-4-480-f04.png?Expires=1750243977&Signature=oVyV8nVYNL3Y5P9S3rV7gj~Cqa9JifHoyIynGCOdZsQDTd8N36VwF~7N6NaJOirio5RNCdvlj5hNggUJuYJWJzEMFw9eGoTtsFqLmolDXSb6IHcuv4UtZvZJQXlhyTsFQ80~g4xGqCAkh8iGHQuEqEbqsfDfHQ5Cri1Qx09413Gls3M~kkW10IJbuS0FPWcxBHPNqBlti9noHAspAH~R4oYRSfPm80GBOvRd4Nj7j8RT~Tj9DOAN~0PLDj2a~VRIrFp5M66QzjVXaBbn3eU84238a1USzhTqPvXDyAaPHZB81FGH41t~GZUVAAGGvbUFIjP~sntbysxGUlAm6LqjBw__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
A and C, Two cytological samples of lung adenocarcinoma that were positive for ALK receptor tyrosine kinase (ALK) gene translocation [B] and for ROS proto-oncogene 1 (ROS1) gene rearrangement [D]) on fluorescence in situ hybridization (Papanicolaou, original magnification ×20 [A and C]; original magnification ×100 [B and D]).
A and C, Two cytological samples of lung adenocarcinoma that were positive for ALK receptor tyrosine kinase (ALK) gene translocation [B] and for ROS proto-oncogene 1 (ROS1) gene rearrangement [D]) on fluorescence in situ hybridization (Papanicolaou, original magnification ×20 [A and C]; original magnification ×100 [B and D]).
Moreover, except for one case that was positive for the p.G12C mutation in the KRAS gene, all samples were wild type for the most-common lung adenocarcinoma driver mutations analyzed by the Sequenom (San Diego, California) MassARRAY using the Myriapod Lung Status Kit (Diatech) according to the manufacturer's protocol (Table 1).
All the samples analyzed by NanoString assay had more than 40% tumor cells and minimal contamination from benign cells. RNA was considered adequate for gene-expression analysis whenever its concentration, determined by a Xpose spectrophotometer (Trinean, Gentbrugge, Belgium), was 30 to 90 ng/μL, and its quality was acceptable if the ratio between the value of absorbance (A) at 260 nm and that at 280 nm was ≥ 1.9 and the ratio between the value of absorbance at 260 nm and that at 230 nm was 2 or more. Up to 450 ng of total RNA was used for each sample. According to manufacturers' instruction, the optimal input of total RNA ranges from 150 to 450 ng for FFPE samples; the optimal range has not yet been clearly defined for cytologic smears, so the same protocol was followed as that for FFPE samples.
The NanoString output data were analyzed with the iGenetics RealQuant software.
All FFPE samples were adequate and suitable for NanoString analysis. Regarding cytologic specimens, one sample gave an RNA yield that was too low to perform the NanoString test and another failed the content normalization step. Technical characteristics of the samples are reported in Table 2.
We found a concordance between NanoString and FISH of 100% for FFPE samples, whereas among the cytologic smears, NanoString results were concordant in 8 of 9 cases (89%) that passed the normalization step, and 1 ALK+ case was not detected. For that discordant case, a biopsy sample was available, which was ALK+ by both FISH and IHC. Therefore, that sample was likely a NanoString false-negative; it would have been interesting to evaluate the clinical response to TKI, but the patient underwent surgery and was not eligible for treatment.
Moreover, one patient only had positive results for ALK 3′/5′ imbalance; consequently, no variant could be identified in the patient (Table 2). Probably the fusion transcript was rare or has not yet been described.
During preanalytic testing, it was essential that all phases were checked before performing the NanoString fusion genes assay, with attention to determining the percentage of tumor cells and the sample integrity. According to our experience, for NanoString analysis, it is recommended that samples have at least 40% or more tumor cells and a total input of RNA equal to 150 ng or more, and the RNA quality should be acceptable if the ratio between absorbance 260 nm and 280 nm is 1.9 or more and between absorbance at 260 nm and at 230 nm is 2 or more. Moreover, it is advisable to perform NanoString analysis as soon as possible because the older the sample is, the worse the test performs.47 Reassuringly, we obtained a good yield and quality of RNA for almost all cytologic samples. The evaluation of RNA using a Trinean spectrophotometer was satisfactory for identifying adequate samples for NanoString analysis, as previously reported.47 Of note, a deeper evaluation of RNA methods used, which could also determine the degree of RNA fragmentation, might improve the preanalytic selection of samples.
Considering the importance of cytology in diagnosing lung cancer, our findings, together with those of Sgariglia et al,45 merit further validation.
CONCLUSIONS
The therapeutic approach to lung cancer has changed dramatically in recent years because of the widespread use of molecular testing, which has optimized the ability to identify molecular targets for biologic drugs.48 The presence of FDA-approved drugs for treating ALK and ROS1 fusion-positive NSCLC, beyond EGFR TKI, increased the importance of testing fusion genes to define patient therapy.9,11 Because targeted therapies are mostly applied in patients with advanced-stage NSCLC, there is an increasing clinical need to investigate molecular aberrations in small biopsies and cytologic specimens.
To maximize the yield of molecular tests on small lung biopsies, NanoString transcript-based assays have recently been suggested to detect gene fusions. According to the data reported, NanoString offers several advantages, including sensitivity, technical reproducibility, and robustness, and it does not suffer during the preanalytic treatment of samples, such as fixation, because it allows direct counting of mRNA molecules without any amplification steps. Furthermore, the system is easy to use and reduces testing time; it seems to be a good candidate for use in clinical practices.25,26,29
Although NanoString assays have been successfully applied to determine fusion genes on FFPE tissues, there are few data available on lung cytology. The explanation for such a lack of data may be the difficult management of cytologic samples, which require a more-accurate preanalytic manipulation of RNA to preserve its quality.
However, according to results by Sgariglia et al45 and our own experience, applying NanoString fusion gene assays may be feasible on cytologic specimens. In most patients with lung cancer, multiple genes are difficult to analyze because of the limited availability of tumors; therefore, a multitarget approach is preferable. The efficacy of the NanoString system has been proven for histologic specimens, but studies on cytology material are rare, and further investigation is warranted to better explore its applicability in the routine practice of lung cancer.