Assessment of ALK Fusions in Uncommon Inflammatory Myofibroblastic Tumors With ALK IHC Positivity but FISH-Equivocal Findings by Targeted RNA Sequencing

Eighteen cases of IMT, including 12 consultation cases (cases 1–12) preliminarily diagnosed as uncommon IMTs with ALK IHC positivity but FISH negativity, plus 3 typical ALK-positive and 3 typical ALK-negative IMTs, were retrieved from the consultation files and surgical pathology files in the Department of Pathology, Fudan University Shanghai Cancer Center (Shanghai, China), during 2015–2019. Clinical information and pathologic features were obtained from the medical record, pathology report, and/or discharge summary. The follow-up information was taken from the referring hospitals or by direct telephone contact with the patients' relatives. Formalin-fixed, paraffin-embedded (FFPE) tissue blocks or unstained slides were reprocessed in our department for hematoxylin-eosin staining, IHC, FISH, and targeted RNAseq.

Immunohistochemical study was performed on 4-μm-thick unstained sections, using the following monoclonal antibodies: ALK1 (CD246, Dako), ALK (D5F3, Cell Signaling Technology), desmin (D33, Dako), myogenin (EP162, ZhongshanJinQiao), and myoD1 (5.8A, Dako). Staining was performed on a Ventana Benchmark XT autostainer (Ventana Medical Systems Inc, Tucson, Arizona) according to the manufacturer's instructions. Appropriate positive and negative controls were included.

FISH analysis was performed on 3-μm-thick unstained FFPE slides with break-apart FISH probes specific for ALK, ROS1, and FUS (Vysis, Abbott Molecular, Chicago, Illinois) according to the manufacturer's instructions on ThermoBrite Elite (Leica, Richmond, California). The fluorescence signals were examined with a fluorescence microscope (BX53 Olympus, Tokyo, Japan). Images were captured with the BioViewTM system (BioView Ltd, Tel Aviv, Israel). At least 100 tumor cells per each specimen were scored by 2 professional molecular pathologists. The rearrangement-positive cells were defined as those with split signals or isolated 3′ signal (3′ probe of ALK is orange, 3′ probes of ROS1 and FUS are green). There are no clear criteria for ALK FISH in IMT and the cutoff value of 15% is usually used for lung cancer according to the recommended criteria of the US Food and Drug Administration (FDA)–approved Vysis ALK break-apart FISH probe kit in non–small cell lung cancer (NSCLC),¹⁶  which was also used for the 12 consultation cases of IMT in primary hospitals. However, the criteria for ALK FISH in IMT in this study were as follows: When the ratio of positive cells to all tumor cells is 20% or greater, the specimen is considered to be typical positive; when the ratio is less than 10%, the specimen is considered to be typical negative; and when the proportion is between 10% and 20%, it is considered to be equivocal, borderline, or atypical positive. In this study, ALK rearrangement testing by FISH was repeated in all 18 cases, while testing for rearrangements of ROS1 and FUS by FISH was only performed in some cases to confirm corresponding fusion partners identified by targeted RNAseq.

The total RNA of the lesion was isolated from 5-μm-thick slices of FFPE samples by using RNeasy FFPE kit (Qiagen, Valencia, California). RNA concentration was evaluated by using 2 methods: spectrophotometric measurement of absorbance at 260/280-nm wavelength (NanoDrop 2000, Thermo Scientific, Wilmington, Delaware) and the fluorometric method based on binding of RNA-selective fluorescent dyes (Qubit 4.0 Fluorometer, Life Technologies, Invitrogen, Carlsbad, California). Complementary DNA (cDNA) was generated from 1 μg of total RNA with the First-Strand cDNA Synthesis System (Promega Corporation, Fitchburg, Wisconsin) and the Next Ultra II Non-Directional RNA Second Strand Synthesis Module (New England Biolabs, Beverly, Massachusetts). The purified cDNA was fragmented to a size range from 100 to 250 bp with Focused-Ultrasonicator (Covaris Inc, Woburn, Massachusetts). CaptureSeq libraries were prepared by the KAPA Hyper Prep Kit (for Illumina) (KAPA Biosystems, Wilmington, Massachusetts) and targeted custom capture of all coding sequences and intron/exon boundaries of coding exons from 630 genes (Integrated DNA Technologies, Coralville, Iowa); the genes included in this panel are listed in Supplemental Table 1 (see the supplemental digital content at https://meridian.allenpress.com/aplm in the October 2022 table of contents). The final library was quantified and analyzed with the Qubit 4.0 Fluorometer (Life Technologies) and Agilent 2100 Bioanalyzer assay (Agilent Technologies, Santa Clara, California), respectively. Paired-end 2 × 150-bp sequencing was performed on Illumina Next500 platform (Illumina, San Diego, California). The sequenced data were aligned to the GRCh37 reference genome and then analyzed by the STARFusion software.¹⁷  All fusion transcripts were manually reviewed through Integrative Genomics Viewer (IGV) software. The targeted RNAseq panel used in this study is only for investigational use.

There were 18 IMTs in this study, including 12 consultation cases preliminarily diagnosed as uncommon IMTs with ALK IHC positivity but FISH negativity, plus 3 typical ALK-positive (both IHC and FISH positive) and 3 typical ALK-negative (both IHC and FISH negative) IMTs. Of the 18 IMTs, 72.2% (13 of 18) were among females, with an average age of 44.4 years (range, 10–65 years). The IMTs ranged from 1.0 to 7.0 cm (average, 2.9 cm). Five cases of IMT were located in the lung, 4 in the bladder, 4 in the uterus, and 1 in the prostate, sigmoid colon, pelvic cavity, gallbladder, and chest wall, respectively. Most lesions were treated with excision. Six patients received crizotinib therapy after excision, and 2 also received chemotherapy. One patient (case 9) received radiotherapy after excision. Clinical follow-up was available for 16 of 18 patients and ranged from 18 to 61.2 months (mean, 38 months; median, 36 months). Of these patients, 14 (87.5%) were alive with no evidence of disease and 2 patients (12.5%) developed recurrence or metastasis within 2 years (cases 1 and 4), of which case 4 had metastasis within 1 year and recurrence within 2 years. The clinicopathologic characteristics of 18 IMTs are shown in the Table.

Expression of ALK protein was detected by IHC using both antibodies of ALK (D5F3) and ALK1 (CD246), and ALK gene rearrangement was performed by FISH in all cases. The results are shown in the Table. For the 12 consultation cases of uncommon IMT, immunohistochemical staining showed 100% (12 of 12) positivity by ALK-D5F3, but only 66.7% (8 of 12) positivity by ALK1 (Table); however, ALK gene rearrangement by FISH was all reported as negative (<15%) in primary hospitals, according to the recommended criteria of the FDA-approved Vysis ALK break-apart FISH probe kit in NSCLC.¹⁶  Interestingly, repeated analysis and re-analysis of FISH in our department indicated that most (83.3%, 10 of 12) of these uncommon IMTs showed equivocal ALK FISH signals, that is—the percentage of ALK-rearranged tumor cells was between 10% and 20% (eg, case 2 in Figure 1, A through C), and only 2 cases showed typical negative FISH signals (Table). On the other hand, ALK was positive in all of the 3 typical ALK-positive IMTs (eg, case 13 in Figure 1, D through F), while ALK was negative in all of the 3 typical ALK-negative IMTs by both IHC and FISH (eg, case 18 in Figure 1, G through I).

To further determine the authenticity of pathologic diagnosis and ALK status in these cases, especially for the 12 consultation cases of uncommon IMT, targeted RNAseq was performed, and the results are summarized in the Table.

For the 12 consultation cases of uncommon IMT, 11 cases were successfully analyzed by targeted RNAseq, with only 1 failure owing to poor RNA quality. Interestingly, ALK fusion transcripts were identified in all of the 9 cases with equivocal ALK FISH signals regardless of ALK1 expression, including 4 IMTs harboring FN1-ALK, 2 harboring SQSTM1-ALK, and 1 harboring TPM3-ALK, EML4-ALK, and DCTN1-ALK, respectively. In 2 cases with ALK IHC positivity but typical FISH negativity, RNAseq analysis showed IGFBP5-ALK fusion in case 11 (Figure 2, A through D) and FUS-TFCP2 in case 12 (Figure 3, A through D). To explore the reason for the FISH false-negative finding in case 11, re-analysis of RNAseq results showed that IGFBP5 and ALK both reside in chromosome 2 with the fusion presumed to result from intrachromosomal inversion. For case 12 with FUS-TFCP2 fusion, FUS gene rearrangement was further validated by FISH, showing typical positive split signals (Figure 3, E). Under a high-power microscope, the tumor cells were observed to be more atypical and mitotic (Figure 3, A). Further immunohistochemical staining showed positivity for desmin, myogenin, and myoD1 (Figure 3, F through H). Based on histomorphology and the results of IHC, FISH, and RNAseq, case 12 was misdiagnosed as IMT in the primary hospital rather than as spindle cell rhabdomyosarcoma. Similar observations of misdiagnosis of spindle cell rhabdomyosarcoma and other novel fusions were also reported in a recently study.⁷ 

For the 3 typical ALK-positive IMTs, as expected, ALK fusion transcripts were identified by RNAseq in all 3 cases, harboring TNS1-ALK, FN1-ALK, and EML4-ALK, respectively (Table). For the 3 typical ALK-negative IMTs, RNAseq results confirmed no ALK fusion, but additionally identified FN1-ROS1 fusions in 2 of 3 cases, which were further verified by FISH using ROS1 break-apart probes (Figure 4, A through E).

Overall, ALK fusion transcripts were successfully identified in 13 IMTs, including 9 equivocal, 3 typical positive, and 1 typical negative ALK FISH signals by RNAseq. As shown in the Table and Figure 5, FN1-ALK fusions were most common in ALK-positive IMTs (5 of 13), followed by EML4-ALK (2 of 13) and SQSTM1-ALK (2 of 13), while the partner genes of DCTN1, IGFBP5, TNS1, and TPM3 were identified in only 1 case, respectively. Further analysis showed that the breakpoint in the ALK gene most commonly occurred in exon 20 and exon 19, with a few occurrences in exons 18, 17, and 14. Although FN1 was the most common 5′ end partner of IMTs in this study, its exon breakpoint varied in 5 cases (exons 4, 6, 20, 24, and 29), while the 5-terminal exon breakpoint for EML4 and SQSTM1 was exon 2 and exon 8, respectively, for these 2 cases of IMTs.

In this study, 12 consultation cases of uncommon IMT with ALK IHC positivity but FISH equivocal finding, plus 3 typical ALK-positive and 3 typical ALK-negative IMTs, were analyzed by IHC (both ALK-D5F3 and ALK1), FISH, and targeted RNAseq. As a result, case 12 with FUS-TFCP2 fusion was misdiagnosed as IMT in the primary hospital, based on ALK positivity by IHC, but should be diagnosed as spindle cell rhabdomyosarcoma, based on histomorphology and the results of IHC, FISH, and RNAseq, suggesting the possibility of ALK immunohistochemical positivity not due to ALK gene rearrangement in some spindle cell tumors. In the other 11 uncommon IMTs, 90.9% (10 of 11) showed equivocal, borderline, or atypical positive ALK FISH signals by repeated analysis and re-analysis of FISH; all of these 10 cases were determined to harbor ALK fusion transcripts by targeted RNAseq except for 1 failure due to poor RNA quality, suggesting that using FISH to detect ALK gene rearrangement in IMT was prone to misdiagnosis, and a low threshold such as 10% for ALK FISH in IMTs should be proposed. However, RNAseq also identified IGFBP5-ALK fusion in 1 case with ALK IHC positivity but typical FISH negativity, suggesting the possibility of false-negative findings for ALK rearrangement by FISH. Immunohistochemically, ALK-D5F3 displayed positivity in all of these 11 cases of uncommon IMT, while 36.4% (4 of 11) displayed negativity for ALK1, indicating high sensitivity of ALK-D5F3 for detection of ALK protein in IMTs, and ALK1 alone was not recommended in the clinic. Furthermore, RNAseq identified FN1-ROS1 fusions in 2 of 3 cases of typical ALK-negative IMTs. Therefore, targeted RNAseq technology can provide more definitive fusion transcripts, which is of great value for the differential diagnosis and subsequent targeted therapy of IMTs, especially for those uncommon IMTs with ALK-negative, equivocal FISH signals or inconsistency between IHC and FISH.

Accumulated evidence indicates that ALK gene rearrangement and protein expression are more specific diagnostic markers and subsequent therapeutic targets for IMTs. However, when we used FISH to detect ALK gene rearrangement in IMTs, equivocal, borderline, or atypical FISH signals were often observed. According to the recommended criteria of the FDA-approved Vysis ALK break-apart FISH probe kit in NSCLC,¹⁸  these cases were often misdiagnosed as ALK negative, just like the 12 consultation cases in primary hospitals. Actually, IMTs with this pattern of equivocal ALK FISH signals were all determined to be positive for ALK fusions by targeted RNAseq, suggesting that a low threshold for ALK FISH in IMTs might be proposed, and more attention should be paid to the equivocal (around 15%) ALK FISH signal cases. Consistent with previous publications, Tan et al¹⁹  found that the positive and negative predictive values of the Vysis ALK break-apart FISH probes were discordant in different ALK-related hematolymphoid neoplasms and nonhematolymphoid lesions, including IMTs, and proposed the term conditional threshold positivity to encourage the adoption of different cutoff values for making positive calls in lesions of different origin. Thus, a lower threshold, such as 10%, should be adopted when using FISH to detect ALK rearrangement in IMTs, and this threshold should be adjusted in different laboratories. Actually, in our clinical practice, we have validated the ALK gene rearrangements in non-IMT spindle cell tumors, including leiomyomas, leiomyosarcomas, fibrosarcomas, nodular fasciitis, and others, using ALK break-apart probes. As a result, none of the cases with a ratio of ALK-positive cells to all tumor cells of 10% or greater was observed (data not shown). Thus, a low threshold such as 10% for ALK FISH in IMTs does not increase the possibility of false positives. In addition, to avoid occasional false-positive cases, IHC (ALK-D5F3) or RNAseq is recommended for the cases with equivocal ALK FISH signals.

IGFBP5-ALK was identified by RNAseq in 1 case of uterine IMT with ALK IHC positivity but typical FISH negativity, suggesting the possibility of false negatives for ALK rearrangement by FISH. The IGFBP5-ALK fusion that had been reported recurrently occurred in uterine IMTs, and coincidentally, in 1 of the 3 uterine IMTs with IGFBP5-ALK fusion showing ALK-negative FISH signals.²⁰  Although IGFBP5 and ALK were both located on chromosome 2, the genomic distance (∼188 M) was far enough for FISH to be able to detect the inversion. Thus, the negative FISH result is very hard to explain. Assuming the fusion is caused by intrachromosomal inversion, it is possible that a false-negative FISH result was rendered because the spatial separation between 5′ and 3′ fluorescent signals in a split-apart FISH assay was subtle and not appreciated, similar to some instances of EML4-ALK fusion in NSCLC.^21,22  Thus, it is very valuable to analyze ALK and other gene fusions by RNAseq in such problematic cases.

Immunohistochemically, ALK-D5F3 and ALK1 are 2 clones commonly used for ALK protein detection in clinical practice. Consistent with previous studies,^10,23  our results demonstrated that ALK-D5F3 had high sensitivity and superior overall performance in terms of intensity and extent of staining in comparison with ALK1 antibody in IMTs. Up to 36.4% (4 of 11) of uncommon IMTs were negative for ALK1, suggesting that detection of ALK protein expression in IMTs by ALK1 clone alone might result in false-negative findings. On the other hand, we should also be cautious in making a misdiagnosis of IMT based on positive expression of ALK-D5F3 clone, since ALK protein expression not caused by ALK gene rearrangement may be present in several spindle cell tumors that mimic IMTs morphologically, including spindle cell rhabdomyosarcoma,²⁴  just as case 12 with ALK IHC positivity but typical FISH negativity was identified as harboring FUS-TFCP2 fusion by RNAseq in this study.

The targeted RNAseq-based next-generation sequencing approach has become more and more advantageous in the detection of fusion genes.¹⁵ ALK fusion transcripts were detected in all 13 IMTs including 9 equivocal ALK FISH signals, 3 typical positive ALK FISH signals, and 1 typical negative ALK FISH signal, except for 1 failure due to poor RNA quality. In addition, RNAseq identified FN1-ROS1 fusion in 2 of 3 ALK-negative IMTs. Previous research has indicated that ROS1 occurs in 5.6% of IMTs²⁵  and in 10% of IMTs presenting in children.¹⁴  Patients carrying TFG-ROS1 fusion have partial response to ceritinib.²⁶  Except for ROS1 and ALK, therapeutically targetable fusions involving other kinases such as RET,²⁷ NTRK3,²⁸  and PDGFB¹ were reported in IMTs. Thus, RNAseq can simultaneously detect multiple gene fusions (without considering the break location of the intron region) and provide more information on the diagnosis and treatments of IMTs.

By the RNAseq approach, in addition to the known partners, such as TMP3,²⁹ EML4,^14,30  and FN1,³¹  we also found several other oncogene fusions that only recently/rarely have been identified. SQSTM1-ALK was found in lung and prostate IMTs for the first time in this study, although it had been reported in large B-cell lymphoma previously³²  and was recently discovered in 1 case of head and neck IMTs.¹⁸  In our 2 IMT cases, the breakpoint of EML4-ALK (exon 2 of EML4) is different from the common breakpoints in NSCLC (exon 6 or 13 of EML4). TNS1-ALK fusion with a new exon breakpoint was also identified in a pelvic IMT and had been reported with the same partner but different breakpoint in a uterine IMT originally diagnosed as leiomyosarcoma.³³  These findings highlight the dual role that genomic profiling plays in aiding with classifying diagnostically challenging cases and enabling patients to benefit from targeted therapy. DCTN1-ALK had been reported in several cases, not just in IMT,³⁴  including lung cancer³⁵  and pancreatic ductal adenocarcinoma.³⁶ 

From the results of RNAseq, a significant observation is that most of the ALK fusion–positive IMTs fused the 5′ partners (IGFBP5, FN1, TNS1) to the start of exon 17, 18, or 19 of ALK, different from the typical exon 20 of ALK in NSCLC. Moreover, we even found a new breakpoint located between exon 18 of TNS1 and exon 14 of ALK. In addition, in the recurrently mutant FN1-ALK, the exon of 5′ FN1 varied in 5 cases (exons 6, 24, 43, 20, and 29), fused to exon 19 of ALK. Different fusion partners and exon breakpoints will determine the subcellular localization and the intrinsic properties of the fusion oncoprotein and will condition the protein-protein interactions and modulate oncogenic signaling and molecular consequences. The different fusion transcripts might impact response to ALK inhibition. In vitro studies showed consistent differences (5- to 10-fold) in drug sensitivity across the 7 ALK fusions.³⁷  And even the same EML4-ALK fusion with different exon breakpoints appears to exhibit differential sensitivity to crizotinib.³⁸  Future studies are needed to investigate the potential biological and therapeutic significance of these fusions.

In summary, by re-analyzing 12 uncommon IMTs plus 3 typical ALK-positive and 3 typical ALK-negative IMTs by IHC, FISH, and RNAseq, some important conclusions were achieved. Firstly, using FISH to detect ALK gene rearrangement in IMT was prone to misdiagnosis; a low threshold for ALK FISH in IMT might be proposed and more attention should be paid to the cases with equivocal FISH signals. Secondly, ALK-D5F3 showed high sensitivity and superior staining characteristics compared with ALK1 antibody, but extreme caution should be paid to the possibility of non–rearrangement-induced ALK protein expression. Last but most importantly, RNAseq could simultaneously detect multiple gene fusions and provide more information on fusion forms and breakpoints, which was of great value for the differential diagnosis and subsequent targeted therapy of IMTs, especially for those uncommon IMTs with ALK-negative, equivocal FISH signals or inconsistency between IHC and FISH.