Introduction

Adenoid cystic carcinoma is a rare salivary gland-type tumor that arises in exocrine glands including the salivary, lacrimal, mammary, and bronchial glands [1]. The median age at diagnosis is 54.6 years and the female ratio is ~60% [1]. Although locoregional recurrence and distant metastasis are common, adenoid cystic carcinoma is an indolent cancer (5-year overall survival: 77%, 10-year: 60%, and 15-year: 45%) [1].

Histopathologically, the most characteristic architecture of adenoid cystic carcinoma is the cribriform pattern showing tumor nests with gland-like structures filled with a basophilic substance. The tubular pattern is comprised of glands containing inner epithelial and outer myoepithelial layers. In the solid pattern, tumor nests consist of monotonous proliferation of epithelial or myoepithelial tumor cells with occasional gland and gland-like structure formation. Depending on the case, these patterns are observed in a single tumor in different ratios. Other salivary-gland type tumors including basal cell adenoma, basal cell adenocarcinoma, pleomorphic adenoma, polymorphous low-grade adenocarcinoma, and epithelial–myoepithelial carcinoma may mimic these patterns of adenoid cystic carcinoma. Therefore, definitively diagnosing adenoid cystic carcinoma may be difficult, particularly when using small biopsy specimens.

In several salivary gland-type tumors, histopathological subtype-specific fusion genes have been identified, including PLAG1 and HMGA2 fusions in pleomorphic adenoma [2, 3], CRTC1-MAML2 in mucoepidermoid carcinoma [4, 5], ETV6-NTRK3 in secretory carcinoma [6], and EWSR1-ATF1 in clear cell carcinoma [7]. In adenoid cystic carcinoma, MYB-NFIB [8,9,10,11] and MYBL1-NFIB [12,13,14] were detected in ~60% of cases [15]. However, driver alterations in the remaining 40% of adenoid cystic carcinoma remain unclear. In the present study, we aimed to clarify the significance of MYB family genes in the pathogenesis of adenoid cystic carcinoma using histopathology, molecular biology, and bioinformatics for 100 adenoid cystic carcinoma cases.

Materials and methods

Case selection and histological review

Three pathologists histopathologically reviewed the slides of 123 patients with adenoid cystic carcinoma diagnosed and treated in the Cancer Institute Hospital (Tokyo, Japan) between January 2005 and October 2016. Twenty-three cases were excluded because a consensus diagnosis of adenoid cystic carcinoma was not obtained in 14 cases and archival samples were insufficient in 9 cases (Supplementary Table 1). This study was approved by the Institutional Review Board of Japanese Foundation for Cancer Research.

The three pathologists independently graded the remaining 100 cases based on the proportion of solid component; grade 1 was assigned when the solid component was <30% of the tumor area; grade 2, 30%≤ solid component <50%; and grade 3, 50%≤ solid component. When an area of histological transformation was identified, the suffix t was added to the grade. Each pathologist’s grade was defined as the consensus grade when it was agreed upon by 2 or 3 pathologists. Graded cases were grouped into low-grade (grades 1 and 2) and high-grade cases (grades 3, 1t, 2t, and 3t) for statistical analyses.

Fluorescence in situ hybridization (FISH)

Unstained formalin-fixed paraffin-embedded sections (4-µm-thick) were subjected to hybridization with bacterial artificial chromosome clone-derived probes for MYB, MYBL1, and NFIB. The clones used are shown in Supplementary Table 2 and genomic regions covered by the probes are illustrated in Supplementary Figure 1. For MYB, we developed two probe sets. One was designated as the wide-range MYB probe set covering 11.5 Mb upstream to 8.4 Mb downstream of MYB (Supplementary Figure 1A). The other was a conventionally designed probe set covering 386.7 kb upstream to 136.6 kb downstream of MYB, the region similar to those covered by previously reported and commercially available MYB probe sets (Supplementary Figure 1B) [11, 16,17,18,19,20]. The conventionally designed probe set was used for initial screening. Cases negative in both MYB and MYBL1 split FISH assays were again examined for MYB locus rearrangement using the wide-range probe set. Probe sets for MYBL1 and NFIB were designed to cover 681.1 kb and 631.8 kb, respectively (Supplementary Figure 1C, D). Hybridized slides were stained with DAPI and examined using a BX51 fluorescence microscope (Olympus, Tokyo, Japan).

Immunohistochemistry

MYB protein was detected on unstained 4-μm-thick formalin-fixed paraffin-embedded sections after deparaffinization with xylene using a rabbit monoclonal antibody to c-Myb (EP769Y) (Abcam, Cambridge, UK) at a dilution of 1:200.

Total RNA extraction and cDNA synthesis

For fresh frozen specimens, total RNA was extracted with RNeasy Mini Kit (QIAGEN, Hilden, Germany) and used to synthesize cDNA with SuperScript III Reverse Transcriptase (Thermo Fisher Scientific, Waltham, MA, USA) and random primers. For formalin-fixed paraffin-embedded specimens, regions of interest (tumor and normal regions) on 10-μm-thick sliced histopathological sections were manually dissected using disposable needles under a stereomicroscope and processed with RecoverAll Total Nucleic Acid Isolation (Thermo Fisher Scientific) to extract total RNA.

Multiplex RT-PCR for MYB-NFIB and MYBL1-NFIB fusions

To amplify MYB-NFIB and MYBL1-NFIB fusion cDNAs, multiplex reverse-transcription polymerase chain reaction (RT-PCR) analysis was conducted using AmpliTaq Gold DNA polymerase (Thermo Fisher Scientific) and the primers shown in Supplementary Table 3. The template cDNAs were incubated at 94 °C for 10 min and then subjected to 35 cycles of denaturation at 94 °C for 1 min, annealing at 64 °C for 1 min, and polymerization at 72 °C for 1 min. The PCR products were separated by agarose gel electrophoresis, followed by direct sequencing or sub-cloning, if needed.

3'-Rapid amplification of cDNA ends for MYB and M Y B L 1

For cases negative in the multiplex RT-PCR for MYB-NFIB and MYBL1-NFIB fusions, 3′-rapid amplification of cDNA ends was performed with one of the two forward primers designed for MYB and MYBL1 (Supplementary Table 3) under previously reported conditions [21]. Amplified products were subjected to direct sequencing.

Capture RNA sequencing

TruSeq RNA Access Library Prep Kit (Illumina, San Diego, CA, USA), which includes biotinylated capture probes targeting coding RNA, was used for coding RNA sequencing with an input of 20 ng total RNA from fresh frozen specimens. TruSight RNA Pan-Cancer Panel (Illumina) was used for RNA sequencing targeted to 1385 cancer-related genes with an input of 100 ng total RNA from formalin-fixed paraffin-embedded specimens. All samples were prepared according to the manufacturer’s instructions.

RT-PCR for MPZL1-NFIB, MPDZ-MYB, TRAM1-MYBL1, MYBL1-RAD51B, and RAD51B-AHI1

To confirm the fusion transcripts detected by coding RNA sequencing, RT-PCR was performed using Platinum Taq DNA Polymerase High Fidelity (Thermo Fisher Scientific) and the specific primer sets shown in Supplementary Table 3.

Sequencing analyses of rearrangements of MYB with downstream regions of N F I B

For the three cases in which fusion transcripts of MYB exon 14 to downstream regions of NFIB were detected by capture RNA sequencing (ACC005, ACC008, and ACC089), genomic PCR was performed with the primers shown in Supplementary Table 3 using LA Taq (TaKaRa Bio, Shiga, Japan). The PCR mixture was incubated at 94 °C for 1 min and then subjected to 30 cycles of denaturation at 98 °C for 10 s, annealing, and polymerization at 68 °C for 5 min. The PCR products were separated by agarose gel electrophoresis, followed by direct sequencing.

Sequence data analysis

Quality control was performed on fastq files using FASTX-Toolkit version 0.0.13 (http://hannonlab.cshl.edu/fastx_toolkit/index.html), PRINSEQ version 0.20.4 [22], and cmpfastq_pe (http://compbio.brc.iop.kcl.ac.uk/software/cmpfastq_pe.php) software. Sequenced reads containing <80% of bases with a quality score of Q20 or higher were deleted. Bases at read ends with N sequences or whose quality scores were less than Q20 were removed. Among the remaining reads, those shorter than 20 base pairs (bp) and the resulting non-paired reads were deleted. The qualified reads were mapped to hg19 with STAR version 2.4.2a [23]. Fusion gene analyses were performed with FusionCatcher [24] and in-house programs. Expression analyses for MYB and MYBL1 were performed using R version 3.3.2 [25] and HTSeq version 0.6.1 [26].

Statistics analysis

Fisher’s exact test (two-tailed) was used to analyze the differences between MYB and MYBL1 locus rearrangement-positive head and neck adenoid cystic carcinoma (N = 78). For histological grade, Mann–Whitney U test was also used. To assess the statistical significance of variables affecting overall survival, the log-rank test was employed for univariate analysis. Multivariate analysis was performed on prognostic factors with significant difference in univariate analysis (P value < 0.05), using the Cox proportional hazard model. Univariate analysis was carried out with GraphPad Prism® Version 6.05 (GraphPad, Inc., La Jolla, CA, USA) and multivariate analysis was performed using R version 3.4.1 and survival version 2.38.

Results

Histopathological grading

In the 100 cases with adenoid cystic carcinoma, there were 71 female and 29 male patients. The median age of onset was 56.8 years (range: 14.9–84.7). The sites of origin included major salivary glands (N = 40), minor salivary glands (N = 37), and other sites (N = 23). The number of cases of head and neck (including four cases of orbital region) were 81 (Table 1; Supplementary Table 4). The 100 cases were histopathologically classified as follows: 65 cases as grade 1, 13 as grade 2, 15 as grade 3, five as grade 1t, one as grade 2t, and one as grade 3t, resulting in 78 low- and 22 high-grade cases. In 81 head-and-neck cases, 63 and 18 cases were graded as low and high, respectively (Table 1).

Table 1 Patient characteristics
Full size table

Expression of MYB and MYBL1 transcripts in the 27 cases with available frozen materials

We developed a multiplex RT-PCR system that could theoretically detect all possible fusion variants coding for the MYB-NFIB or MYBL1-NFIB fusion protein. Among the 27 cases for which frozen materials were available, 9 (33%) and 6 (22%) cases were found to express MYB-NFIB and MYBL1-NFIB fusion transcripts, respectively. In the remaining 12 (44%) cases, neither transcript was expressed (Table 2, Fig. 1, and Supplementary Figure 2).

Table 2 MYB/MYBL1 transcripts and FISH results of the 27 cases available for frozen materials and cases with atypical and negative FISH results
Full size table
Fig. 1
figure 1

Breakpoints in MYB-NFIB and MYBL1-NFIB fusion transcripts. Breakpoints in MYB and NFIB (a) and MYBL1 and NFIB (b) found to be fused to each other are shown in the same color. Arrows and triangles represent the breakpoints reported and those detected in the present study, respectively. Cases with asterisks showed alternative splicing. The primer locations for multiplex RT-PCR are also shown. Note that all fusion-positive cases retained MYB or MYBL1 DNA binding domain (DBD) and transactivation domain (TAD). NRD negative regulatory domain

Full size image

Next, we performed a conventional 3′-rapid amplification of cDNA ends assay and coding RNA sequencing using next-generation sequencing. Interestingly, all 27 cases were found to highly express either MYB (18 cases, 67%) or MYBL1 (9 cases, 33%) in a mutually exclusive manner, and three additional MYB (ACC005, ACC008, and ACC089) and one additional MYBL1 (ACC064) fusion transcripts were identified in the 12 multiplex RT-PCR-negative cases (Table 2). In the three additionally identified MYB transcripts, MYB exon 14 fused to 9p23 intergenic sequences, 207 kb (ACC005), 2.9 kb (ACC008), and 3.8 kb (ACC089) downstream of the 3′ untranslated region (UTR) end of NFIB (Supplementary Figure 3). The genomic breakpoints in MYB intron 14 were at 4.4 kb (ACC005), 0.7 kb (ACC008), and 4.2 kb (ACC089) downstream from the end of MYB exon 14 (Supplementary Figure 3). In ACC008, an ~0.9 Mb fragment of chromosome 9 was considered inserted in MYB intron 14 because the following two fusion transcripts were identified: MYB exon 14 fused to 3.6 kb downstream of the NFIB 3′UTR and MPDZ exon 24 fused to MYB exon 15 (Fig. 2a; Supplementary Figure 4A). In ACC064, MYBL1 exon 10-RAD51B intron 9 and TRAM1 exon 1-MYBL1 exon 11 fusion transcripts were expressed and confirmed by specific RT-PCR (Supplementary Figure 4B). MYBL1 and RAD51B were fused to each other in the opposite direction in this case (Supplementary Figure 5). The genomic breakpoint of MYBL1 was in intron 10, 87 bp downstream of exon 10, which was fused to intron 9 of RAD51B, 49.5 kb downstream of exon 9 (Supplementary Figure 5). As MYBL1, TRAM1, and RAD51B are located at 8q13.1, 8q13.3, and 14q24.1, the chromosomal rearrangement involving MYBL1 in ACC064 was considered complex.

Fig. 2
figure 2

Estimated genomic rearrangements in MYB split FISH-negative but MYB-NFIB fusion FISH-positive cases. a Transcripts of MYB exon 14-NFIB 3′ downstream and MPDZ exon 24-MYB exon 15 were detected in ACC008, suggesting an ~0.9 Mb insertion of the NFIB locus into MYB intron 14. b In ACC101, a transcript of NFIB exon 4-LINC01235 exon 2 was identified. Therefore, a part of the NFIB locus, which consisted of the NFIB sequence from intron 4, intergenic region (gray) between NFIB and LINC01235, and LINC01235 intron 1, was likely to juxtapose the MYB locus. Not to scale

Full size image

In the remaining eight cases, full-length MYB (ACC024, ACC098, ACC102, ACC110, ACC120, and ACC122) and truncated MYBL1 (ACC003 and ACC051) were highly expressed. In a case with the full-length MYB transcript (ACC098), an MPZL1 (1q24.2) exon 4-NFIB exon 4 fusion transcript was detected (Table 2; Supplementary Figure 6). In confirmatory RT-PCR, a transcript with an insertion of 69bases from 6q25.3 between MPZL1 exon 4 and NFIB exon 4 was also identified (Supplementary Figure 4C; Supplementary Figure 6).

Taken together, the following results were obtained in the analyses of MYB and MYBL1 transcripts in the 27 cases. Either MYB or MYBL1 transcripts were expressed in all cases in a mutually exclusive manner, regardless of its form (full-length, truncation, or fusion). In the fusion forms, both in-frame and out-of-frame fusions were identified. In all MYB and MYBL1 transcripts, the DNA-binding domain and trans-activation domain were retained (Table 2; Fig. 1).

FISH for MYB and MYBL1 loci in the 27 cases

We subsequently performed FISH assays using formalin-fixed paraffin-embedded specimens to determine whether genomic rearrangements occurred around MYB and MYBL1 in these 27 cases. In adenoid cystic carcinoma, recent studies reported the fusion transcripts of NFIB and genes around MYB (6q23.3), such as NKAIN2 (6q22.31) and AIG1 (6q24.2) [13]. These data prompted us to design probes that covered the ~20 Mb region around MYB (Supplementary Figure 1A), which was much broader than those covered by previously reported and commercially available FISH probes (Supplementary Figure 1B) [11, 16,17,18,19,20]. Among the 27 cases, 17 and 8 were positive in MYB and MYBL1 split FISH assays, respectively (Table 2; Supplementary Figure 7). One of the remaining two cases (ACC008) was negative for MYB split signals, but was positive in the MYB-NFIB fusion FISH assay. The other case (ACC037) was negative for MYBL1 split signals, but was positive in the MYBL1-NFIB fusion FISH assay.

Collectively, genomic rearrangements around the corresponding highly-expressed gene were observed in all 27 examined adenoid cystic carcinoma cases (18 cases for MYB and 9 cases for MYBL1), and the NFIB locus (9p23-22.3) was involved in 16 of 18 MYB and 7 of 9 MYBL1 locus rearrangements (Table 2). In addition, it was shown that MYB and MYBL1 were the target genes for overexpression in MYB and MYBL1 locus rearrangements, but not the other adjacent genes around MYB (ALDH8A1, HBS1L, AHI1, and LINC00271) and MYBL1 (TRIM55, ADHFE1, C8orf46, and C8orf44-SGK3), respectively (Supplementary Table 5; Supplementary Figure 8).

Validation of MYB and MYBL1 rearrangements using 73 additional cases

On the basis of the fact that all of the examined 27 adenoid cystic carcinoma cases harbored genomic rearrangements in the MYB or MYBL1 locus, we further performed FISH analyses using formalin-fixed paraffin-embedded specimens of 73 additional adenoid cystic carcinoma cases. MYB split signals and MYBL1 split signals were identified in 52 and 15 cases, respectively. In ACC062 and ACC101, MYB-NFIB fusion signals were positive, although MYB split signals were negative as in ACC008. In ACC055, MYBL1-NFIB fusion signals were positive, although MYBL1 split signals were negative as in ACC037 (Fig. 3).

Fig. 3
figure 3

Clinicopathological and genetic landscape of 100 cases with adenoid cystic carcinoma. At the top of the figure, MYB or MYBL1 expression level is shown using fragments per kilobase of exon per million mapped fragments (FPKM) (pink bar: MYB, green bar: MYBL1)

Full size image

Among the 100 cases examined, 97 showed either of the following signal patterns: both MYB split and MYB-NFIB fusion signals in 57, only MYB split signals in 12, only MYB-NFIB fusion signals in 3, both MYBL1 split and MYBL1-NFIB fusion signals in 18, only MYBL1 split signals in 5, and only MYBL1-NFIB fusion signals in 2. The remaining three cases (ACC020, ACC114, and ACC119) were negative for any of these split/fusion FISH signals (Fig. 3).

Further analyses of genomic rearrangements in cases with atypical FISH patterns

Thirteen cases with atypical FISH patterns were further analyzed by capture RNA sequencing (Table 2). In the five cases negative for MYB split with the conventional probe set but positive with the wide-range set (ACC024, ACC056, ACC093, ACC102, and ACC112), full-length MYB expression was detected except for ACC093 and ACC112, which were not evaluable due to the low quality of RNA.

MYB split was negative but MYB-NFIB fusion was positive in three cases (ACC008, ACC062, and ACC101). In ACC008, the discrepant FISH results support our above-mentioned estimation that a ~900 kb fragment of chromosome 9p23-p22.3 (3.6 kb downstream from NFIB 3′UTR to MPDZ exon 24) was probably inserted in MYB intron 14 (Fig. 2a), as the insertion size is far below the resolution limit of the split FISH assay on formalin-fixed paraffin-embedded slices. In ACC062, an MYB exon 8-NFIB exon 12 fusion transcript was identified, indicating that a small fragment containing NFIB exon 12 and after was inserted in MYB intron 8 through a similar mechanism as in ACC008. In ACC101, full-length MYB was highly expressed and an NFIB exon 4-LINC01235 exon 2 fusion transcript was identified. LINC01235 is located 650 kb downstream from NFIB, and therefore the region from NFIB intron 4 to LINC01235 intron 1 may be juxtaposed to MYB (Fig. 2b).

MYBL1 split was negative while MYBL1-NFIB fusion was positive in two cases (ACC037 and ACC055). A MYBL1 exon 14-NFIB exon 9 fusion transcript was expressed in ACC037. Therefore, as in the three cases (ACC008, ACC062, and ACC101), a small fragment containing NFIB exon 9 and after was likely inserted into MYBL1 intron 14. Capture RNA sequencing was not successful in ACC055 for unknown reasons.

In the three cases negative in all FISH analyses (ACC020, ACC114, and ACC119), the expression of full-length MYB mRNA and MYB protein was detected (Supplementary Figure 9). Any causative genomic rearrangements for the expression of MYB were not observed in the present FISH assays or capture RNA sequencing. Interestingly, all three FISH-negative cases were of external ear canal origin, and the remaining external ear canal case was also atypical in FISH (ACC024), which was negative for MYB split with the conventional probe set but positive with the wide-range set.

Secondary MYBL1 locus rearrangement as a passenger event

In the 97 cases showing rearrangements, those in the MYB and MYBL1 loci were fundamentally mutually exclusive (Fig. 3), except for one case (ACC018). The sample from the primary lesion (lacrimal gland) was positive in MYB split FISH and negative in MYB-NFIB fusion FISH and MYBL1split FISH. However, MYBL1 split signals were also detected in samples from relapsed lesions. We examined all four lesions (primary and relapsed at 2, 4, and 6 years after) by capture RNA sequencing with TruSight RNA Pan-Cancer Panel using RNA extracted from formalin-fixed paraffin-embedded tissues. Full-length MYB mRNA expression was constantly detected, while MYBL1 expression remained very low (Supplementary Figure 10). Therefore, the MYBL1 locus rearrangement in this MYB-rearranged case was likely a secondary and passenger event. A fusion transcript of RAD51B exon 7 (14q24.1) fused to AHI1 exon 23 (114 kb downstream from MYB) was detected, which was confirmed by RT-PCR (Supplementary Figure 4D), supporting the FISH results (positive for MYB split and negative for MYB-NFIB fusion).

Clinicopathological analysis of 78 head and neck cases with confirmed MYB/MYBL1 locus rearrangement

Prognostic analyses were performed on 78 head and neck cases (including four cases in the ocular regions) with confirmed MYB or MYBL1 rearrangement. Three variables, primary tumor size (30 mm<, P = 0.0028), primary lymph node metastasis (positive, P < 0.0001), and histological grade (high, P = 0.0022) were significantly associated with poorer overall survival in univariate analysis (Supplementary Table 6; Fig. 4). In multivariate analysis of these three variables, primary tumor size (P = 0.003), primary lymph node metastasis (P = 0.002), and histological grade (P < 0.001) were independent prognostic factors.

Fig. 4
figure 4

Overall survival curves in the 78 MYB/MYBL1 locus rearrangement-positive cases whose primary sites were the head and neck (including orbital region). Cases with clinicopathological factors: a >30 mm primary tumor size (N = 38) (P = 0.0028), b primary lymph node metastasis (N = 5) (P < 0.0001), and c high histological grade (N = 18) (P = 0.0022)

Full size image

Any clinical features tested including overall survival did not significantly differ between the cases with MYB and MYBL1 alterations in the present cohort. However, histological grade was generally higher in cases with MYB alterations (P = 0.10, Fisher’s exact test; P = 0.023, Mann–Whitney U test) (Supplementary Table 7). Given that histological grade was found to be an independent prognostic factor, a large-scale study with detailed examination of MYB/MYBL1 alterations would reveal the prognostic significance of the alteration.

Discussion

In the present study, we found that 97 of the 100 adenoid cystic carcinoma cases were associated with genomic rearrangement either in the MYB locus (73 cases) or in the MYBL1 locus (24 cases). In addition, all 27 cases examined expressed MYB (18/27, 67%) or MYBL1 (9/27, 33%) transcripts in a mutually exclusive manner. There were various forms of the transcripts including full-length, truncated, and fusion forms. In fusion forms, both in-frame and out-of-frame fusion transcripts were detected. In all transcripts, the DNA-binding domain and trans-activation domain were retained. Taken together, these observations suggest that the fundamental alteration in the development of adenoid cystic carcinoma is dysregulated overexpression of the two domains in MYB or MYBL1 led by translocation of super-enhancers (mostly those of NFIB) in the proximity of MYB or MYBL1.

Given its high frequency, MYB/MYBL1 locus rearrangement can be a sensitive diagnostic marker of adenoid cystic carcinoma. However, the diversity of the rearrangements was also revealed and thus the method of detection should be designed judiciously. Although the frequency of MYB-NFIB fusion transcript in adenoid cystic carcinoma in the initial report was 100% (11/11) [8], the frequency obtained by RT-PCR was relatively low considering subsequent reports (51.1%, 162/317) [8,9,10, 12, 16, 18, 19, 27,28,29,30,31]. Aside from technical issues (primer design, formalin-fixed paraffin-embedded vs. frozen materials, etc.), this was explained by the presence of other types of MYB expression including full-length MYB, truncated MYB, and MYB fusion transcript with other partner genes (PDCD1LG2 at 9p24.1, EFR3A at 8q24.22, TGFBR3 at 1p22.1, and RAD51B at 14q24.1) [28, 32]. In addition, the MYBL1-NFIB fusion was reported in 2015. The reported frequency of the fusion varies depending on methodology used (10.9%, 16/147, transcript-based; 11.9%, 16/135, MYBL1-NFIB fusion FISH), and other types of MYBL1 expression including truncated MYBL1 and MYBL1 fusion with other partner genes have been reported [12,13,14, 33]. Although MYB/MYBL1 alterations are highly characteristic to adenoid cystic carcinoma, this diversity in MYB/MYBL1 transcripts dampens the efficacy of RT-PCR-based approaches for confirmatory diagnosis of adenoid cystic carcinoma.

In the present study, we detected 10 cases (10%) whose MYB/MYBL1 locus rearrangements could not be detected using the conventionally designed split FISH probe set (Table 2). In five cases, genomic rearrangements in the MYB locus turned out to be visible with the wide-range probe set. The probe set covered the ~20 Mb region around MYB (Supplementary Figure 1A), which was extremely broad compared with those covered by previously reported and commercially available FISH probes (Supplementary Figure 1B) [11, 16,17,18,19,20]. Thus, the breakpoints in the MYB locus are far distant from MYB (~10 Mb) in these cases. In the other five cases, a small insertion (<1 Mb) that could not be recognized by split FISH assay was considered to occur within MYB (ACC008 and ACC062), close to MYB (ACC101), within MYBL1 (ACC037), and within or close to MYBL1 (ACC055). The super-enhancer elements in NFIB, TGFBR3, or RAD51B loci translocated to the proximity of MYB (regardless of whether it was upstream or downstream) caused overexpression of MYB protein, which binds the elements and sustains MYB expression in a positive feedback manner [32]. On the basis of previous results and those of the present study, MYB can be upregulated even if a genomic region containing the super-enhancer elements is translocated “far distant” from MYB (but the elements should physically interact with MYB promoter region) and/or if the genomic region is too small to be observed by conventional karyotyping and FISH assays.

FISH assay is a powerful tool for detecting genomic rearrangements. Particularly, split FISH assay is suitable for screening for rearrangement of genes of interest, as it is available regardless of the rearranged partners. For example, the split FISH assay for ALK has been widely used as a companion diagnostic for ALK inhibitors in ALK-positive lung cancers, in which ALK is rearranged with various potential fusion partner genes including EML4 [34], KIF5B [35], KLC1 [21], and others. However, the presence of above-mentioned atypical patterns of MYB/MYBL1 locus rearrangement in adenoid cystic carcinoma may limit FISH as a confirmatory diagnostic assay. Using conventionally designed split FISH probe sets, the frequency of MYB/MYBL1 locus rearrangement in our cohort was 87% (87/100), which is comparable to that in a study in which both MYB and MYBL1 were examined by FISH (88%, 29/33) [33]. Therefore, to maintain the sensitivity of FISH for detecting MYB/MYBL1 locus rearrangement in adenoid cystic carcinoma, combined analysis using the split FISH assay with a wide-range probe set and fusion FISH assay using probes for known fusion partners (at least NFIB) is necessary. For example, one of the practical ways is following; 3-color FISH using a mixture of a conventionally designed MYB probe set (e.g., set A in Supplementary Table 2) and NFIB (e.g., set C) is performed in a case. If the result is negative, a mixture of a MYBL1 probe set (e.g., set B) and NFIB (e.g., set C) is performed. If the result is negative, FISH using wide-range MYB probe set is performed. The first two or all three FISH tests can be done at the same time.

Among the 14 excluded cases without a consensus diagnosis of adenoid cystic carcinoma, 11 were evaluable and negative for MYB/MYBL1 rearrangement (data not shown). Capture RNA sequencing was performed in 4 cases. MYBL1 expression was not detected in the 4 cases, but the expression of full-length MYB transcript was detected in 3 cases. This means that the expression-based assays that cannot distinguish MYB forms (full length, fusion, and truncation), such as anti-MYB immunohistochemistry, are less useful in differentiating adenoid cystic carcinoma from its mimics in diagnostic settings.