Comparative Analysis of MYB Expression by Immunohistochemistry and RNA Sequencing in Clinical Gene Fusion Detection in Adenoid Cystic Carcinoma

Abstract

Purpose

MYB has been shown to play a central role in oncogenesis in a majority of adenoid cystic carcinomas (ACC). Testing for MYB expression via immunohistochemistry (IHC) or testing for the MYB gene fusion by next-generation sequencing (NGS) have become useful tools for the diagnosis of ACC. In addition, detection of MYB expression may have implications for patient management.

Methods

A cohort of 35 ACC cases was identified from the archival pathology files of the Massachusetts General Hospital. Cases were tested for MYB expression using a panel of 4 different commercially available MYB antibodies and scored using a modified Allred system. RNA-based NGS for MYB gene fusion detection was also performed.

Results

Among 4 different MYB antibodies, the sensitivity for MYB detection ranged from 26 to 97%. When a 30% threshold for determination of MYB immunohistochemical positivity was used, the AB_10900735 IHC clone showed the maximum sensitivity (97%). RNA sequencing revealed 19 (54%) cases positive for MYB fusions, and expression analysis derived from the sequencing data confirmed a significant association between MYB expression and fusion status (p = 0.036). Although less sensitive, the AB_778878 MYB clone showed a significant positive association between IHC staining and MYB RNA expression (R² = 0.15, p = 0.023).

Conclusion

The detection of MYB expression using immunohistochemistry varies significantly depending on the antibody used. Comparison with MYB fusion and transcription levels, as determined by NGS, reveals that MYB has a complex relationship between genetic alterations, transcript levels, and protein abundance.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12105-024-01719-1.

Introduction

Despite representing less than 10% of all salivary gland neoplasms, adenoid cystic carcinoma (ACC) is the second most common salivary gland carcinoma, accounting for approximately 22% of cases, and associated with 5- and 10-year survival rates of 62% and 48%, respectively [1–3]. There are 3 subtypes of ACC based upon histologic growth patterns which often occur in a combination of two or more patterns. More recently, variants with squamous differentiation, trabecular growth, and macrocystic architecture have been described as “metatypical” ACC [2, 4–7]. Histologically, ACC is a biphasic neoplasm composed of luminal epithelial cells, positive for CD117 by immunohistochemistry (IHC), and an outer layer of myoepithelial cells with IHC reactivity for myoepithelial markers (e.g., p63, S100, smooth muscle actin, calponin, and others) [8]. While ACC shares some histological features with other matrix-producing neoplasms of salivary origin, genetic studies have shown high expression of either the MYB or MYBL1 transcription factors in nearly all cases; more recently, next-generation sequencing (NGS) has become a routine approach for detecting ACC-associated translocation events involving MYB or MYBL1, which are present in approximately 60% of cases [7, 9–14]. These translocations preserve the MYB and MYBL1 DNA-binding and transactivation domains, while truncating each gene’s respective negative regulatory domain [15].

Surgical resection and adjuvant radiation therapy are the most frequent treatment modalities for ACC; however, given the aggressive biologic behavior of ACC, there is a need for additional novel therapeutic approaches [16]. There are currently multiple active clinical trials exploring various treatments for ACC, including immunotherapies, radiation therapy, novel small molecule inhibitors, monoclonal antibodies, and antibody-drug conjugates. Some of the therapeutic approaches being investigated have targeted the MYB pathway, supporting the necessity for appropriately determining the MYB status in ACC for inclusion in current and future clinical trials.

To investigate the expression of MYB in ACC, we present the performance metrics of a series of 4 commercially available MYB antibodies in a cohort of 35 ACCs. In addition, we use a clinical grade NGS assay designed for gene fusion detection to determine the MYB gene translocation status, and we compare the RNA MYB gene expression status with immunohistochemical and gene fusion data. The relevance of delineating MYB expression in ACC at different levels has the potential to significantly impact the diagnosis, trial options, and clinical research, ultimately aiming to improve patient outcomes.

Materials and Methods

Study Design

The study was approved by the Institutional Review Board at the Massachusetts General Hospital (2013P001818, WCF). A retrospective search of the pathology archives for ACC over a 10-year period was performed at an academic, tertiary care center with a subspecialty pathology practice. The diagnoses were based on criteria outlined in the World Health Organization’s Classification of Head and Neck Tumours 5th edition [5], and for each tumor, the diagnosis was confirmed by at least 2 board certified pathologists with head and neck subspecialty expertise (ASF, WCF). For each patient, clinicopathological information was reviewed, including demographic information, tumor site, clinical course (i.e., primary vs. local recurrence vs. metastases), American Joint Committee on Cancer (AJCC) staging, histological features (i.e., growth pattern, solid component), and molecular findings.

Immunohistochemistry

Immunohistochemical studies were performed using 5 μm-thick, formalin-fixed paraffin-embedded (FFPE) sections of selected blocks from cohort cases of ACC. The full group of antibodies were selected based on preferred targeting of epitopes closer to the N-terminus of MYB to capture increases in both wildtype and chimeric MYB protein. Six antibodies were initially evaluated (Table 1) by 2 head and neck pathologists (ASF, WCF) for inclusion in the study, including testing and optimization of antibodies on positive and negative control tissues for appropriate staining. Only antibodies that were interpreted to show reliable sensitive and specific staining on control tissues and a sampling of tumors were included for further evaluation in the study.

Table 1.

Antibody characteristics

Once the antibody clones that met quality standards were included in the study, IHC slides were reviewed and nuclear staining was evaluated using a modified Allred-type scoring system [17]. Slides were scored semi-quantitatively by 2 head and neck pathologists (ASF, WCF) based upon the proportion of tumor cells staining (10–100% in 10% increments, 0% when no staining was seen, and 5% given to all cases with < 10% staining) and nuclear staining intensity (0 = negative staining, 1 = mild staining, 2 = intense staining). Cases for which scoring by the 2 pathologists differed by 1 for staining intensity or ≤ 30% for the proportion of tumor staining were averaged, and cases with a greater scoring difference were re-reviewed by the pathologists and a consensus for scoring was reached. A composite score was derived from the product of staining intensity and proportion of tumor cells staining.

Gene Fusion Detection by RNA-Based Next-Generation Sequencing

Genotyping at our institution was performed in a CLIA-certified molecular diagnostics laboratory using anchored multiplex PCR (AMP) technology for the identification of gene fusions [18]. Briefly, after extraction of nucleic acids from FFPE tissue specimens, samples were reversed transcribed and two hemi-nested PCR reactions were used to create sequencing libraries with custom designed FusionPlex Solid Tumor primers (ArcherDx Inc., Boulder, CO, USA). NextSeq 2 × 150 base paired-end sequencing results (Illumina, San Diego, CA, USA) were aligned to the hg19 human genome reference using bwa-mem. A laboratory-developed algorithm was used for fusion transcript detection and annotation. Assessment of RNA-based findings for MYB were queried using a panel of genes listed in Supplementary Table 1.

RNA-Based Next-Generation Sequencing-Derived Gene Expression Analysis

Relative gene expression of MYB on the exon level was calculated from read depth data. The sequencing platform covers 320 exons over 59 genes including MYB (exons 7–9, 11–16 of NM_001130173). Raw expression values were normalized using INSR, RAF1, and TFE3 as housekeeping genes. Briefly, we selected these genes based on an average coverage of at least 100 or more reads, over at least 10 exons, with median coefficients of variance between 5 and 12, median read counts of 500-1,500 (max median read range 10,000–25,000), and median standard deviation of 4,000–10,000. As typical in amplicon-based normalizations, for each case we calculated the normalized MYB expression level using the following equation:

Integration of MYB Expression at the Protein and RNA Levels

In addition to separating expression by exon, the specific coding region for the epitope of the AB_955796 MYB antibody (amino acids 281–294) was parsed. Cases were separated to those with and without gene fusion events, and the relative exon-level MYB expression was mapped against immunohistochemical scores. After sorting by protein expression level, exon-breakpoints as well as fusion partners and exons were plotted.

Sequence and Copy Number Variant Detection by DNA-Based Next-Generation Sequencing

Total nucleic acids were extracted from FFPE tissue specimens and tested using DNA-based AMP technology targeting 98 genes for the identification of single nucleotide variants (SNV), insertion-deletions (indel), and copy number variants (CNV). The samples were prepared and run as described above in the section of the methods describing gene fusion detection by RNA-based NGS. Gene targets (exon) for SNV, indel, and CNV detection are listed in Supplementary Table 2.

Statistical Analysis

To evaluate MYB gene expression in statistical analyses regardless of gene fusion status or genetic breakpoints, individual case expression was represented by a 5’ expression signature (MYB-5’) generated by calculating the mean expression of exons 7 and 8. The frequency distributions for staining intensity, proportion of positive tumor staining, and composite scores for all antibodies were plotted, and the cutoff MYB positivity by IHC was set to 30%. Simple linear regression (p < 0.05 for significance) was used to analyze the association between each IHC antibody’s performance and MYB-5’ RNA expression as well as between antibody performance and solid tumor architecture, while Mann-Whitney tests (p < 0.05 significance) were used for group comparisons. All statistical analyses and data visualizations were performed using GraphPad Prism version 10.2.3 for Windows, GraphPad Software, Boston, Massachusetts USA, www.graphpad.com.

Results

Patient Demographics and Tumor Histology

A summary of the characteristics of the 35 patients included in the study can be found in Table 2. The tumor specimens analyzed came from a cohort of 20 male (57.1%) and 15 female (42.9%) patients, with a median age of 56 years (range = 36–91 years). There were 25 primary ACC (71.4%), 7 metastatic ACC (20.0%), and 3 recurrent ACC (8.6%) cases in our cohort. Among primary and recurrent tumors (80%), the anatomic sites of origin included 10 from the major salivary glands (28.6%; 5 parotid, 4 submandibular, and 1 sublingual), 6 from the oral cavity (17.1%), 5 (14.3%) from the oropharynx and trachea, and 2 (5.7%) from the sinonasal tract. The cohort cases represented all 3 ACC subtypes, specifically cribriform, tubular, and solid (Fig. 1). A component of the solid growth pattern estimated to be over 30% and 50% of the total carcinoma was present in 10 and 9 cases, respectively. Most ACC in the cohort came from tumors in patients with stage IV disease by the AJCC staging guidelines, totaling 20 specimens that ranged from stage IVa-IVc. The remaining tumors were either stage I (n = 2), stage II (n = 6), or stage III (n = 4). Tumor staging data were not available for 3 cases (Table 2).

Table 2.

Summary characteristics of patient cohort (N=35)

Fig. 1.

Histological images of cohort adenoid cystic carcinoma cases demonstrating characteristic cribriform growth (A), tubular growth (B), solid growth (C), and perineural invasion (D)

RNA- and DNA-Based Next-Generation Sequencing

A total of 34 samples were successfully analyzed for gene fusions using RNA-based NGS. Nineteen (19) tumors were positive for MYB rearrangements with breakpoints that ranged from exon 8 to exon 14. Nearly all MYB fusions were with the gene fusion partner NFIB, which had breakpoints occurring between exon 6 and the 3’ untranslated region (UTR), and 1 tumor harbored a MYB::PKHD1 (exon 65) fusion. A total of 14 cases had no fusions identified; an additional tumor negative for NGS on in-house testing was subsequently found to have a MYBL1 translocation based on outside testing results. The final patient presented with a tumor that was known to harbor a MYBL1 fusion from earlier outside testing, and additional in-house testing was not pursued. Tumors from 19 patients had DNA-based molecular testing by NGS for sequence and copy number variants. The genes most frequently observed to harbor sequence variants were NOTCH1 and TP53, each altered in 3 tumors. Complete clinicopathologic details, including the results of molecular testing, are available in Table 3.

Table 3.

Clinical, pathological, and molecular characteristics of patient cohort (N=

Gene Expression Analysis

The 34 specimens with RNA-based NGS for gene fusion detection were analyzed for MYB gene expression at the exon level, and the MYB-5’ gene expression signature was calculated for consistent comparison across tumors. Nearly all tumors (n = 32, 94.1%) showed maximal expression in MYB-5’ relative to downstream exons present on the fusion panel. The remaining 2 cases included a fusion-positive tumor (patient 3) with its second highest expression within MYB-5’ (exon 7), and a fusion-negative tumor (patient 34) with maximal expression observed outside MYB-5’ (Supplementary Table 3). An examination of the relationship between MYB-5’ expression and fusion status revealed a significant association between increased MYB-5’ expression and the presence of MYB gene fusions (p = 0.0362) (Fig. 2).

Fig. 2.

Comparison of MYB expression, represented by MYB-5’, in tumors harboring MYB gene fusions and tumors negative for MYB gene fusions (Mann-Whitney test, p < 0.05 for significance)

Immunohistochemistry

Of the 6 antibody clones evaluated, 4 clones were successfully optimized and met quality standards for appropriate staining on positive and negative controls for inclusion in the study, including AB_10900735, AB_10912656, AB_955796, and AB_778878. The LS-B11880 and AB_1078540 clones targeting MYB and MYBL1, respectively, did not meet these standards and were excluded from further investigation. Following performance of IHC and scoring of all 4 antibodies (Table 4), the highest percentage of tumor cells stained (64.8% ± 23.2%) and highest composite staining score (109.5 ± 55.2) were observed using AB_10900735 (Fig. 3), while the highest mean intensity was observed for AB_778878 (1.9 ± 0.4) (Fig. 4). The AB_10900735 clone also showed the best performance based upon its sensitivity for detecting MYB expression in over 97% of cases when using a 30% tumor cell staining threshold. The other antibodies showed diagnostic sensitivities of 71% (AB_955796), 26% (AB_10912656), and 66% (AB_778878) (Supplementary Figures S1-S2). No significant association was seen between solid component by histology and antibody staining.

Table 4.

Summary statistics of antibody scoring

Fig. 3.

Immunohistochemical analysis of MYB expression using the AB_10900735 clone. These images demonstrate the range of MYB immunohistochemistry performance seen, as measured by composite score, with examples of strong, diffuse staining (A and C) and poor staining characteristics (B and D) regardless of MYB fusion status

Fig. 4.

Immunohistochemical analysis of MYB expression using the AB_778878 clone. These images demonstrate the range of MYB immunohistochemistry performance seen, as measured by composite score, with examples of strong, diffuse staining (A and C) and poor staining (B and D) regardless of MYB fusion status

Integration of Molecular and Immunohistochemical Findings

An expression analysis comparing MYB protein using all 4 IHC clones to MYB-5’ expression by NGS showed a significant positive trend for only the AB_778878 antibody (R² = 0.1512; p = 0.0230). The positive trend seen for the AB_955796 antibody was not significant (R² = 0.1036; p = 0.0634), and no significant association was seen for AB_10900735 or AB_10912656 (Fig. 5). Only the AB_955796 antibody had a target epitope with a coding region covered by RNA-based gene fusion NGS, and no significant correlation was seen between AB_955796 staining and RNA expression for this region. No significant association was found between each antibody’s performance and the MYB gene fusion status.

Fig. 5.

A simple linear regression shows a significant association (R² = 0.152; p = 0.0230) between MYB-5’ expression and the AB_778878 antibody (black squares). The AB_955796 antibody (blue triangles) showed a non-significant positive trend with MYB-5’ (R² = 0.1036; p = 0.0634), and the AB_10900735 (red circles) as well as AB_10912656 (magenta circles) antibodies showed no significant correlation

Discussion

We present a cohort of 35 patients with ACC, for which a combined assessment including IHC and RNA-based NGS for gene fusion detection was performed. A total of 4 commercially available IHC assays were evaluated for MYB staining properties, and our results show that when using a 30% staining threshold, the AB_10900735 antibody has optimal sensitivity (97%). This can be useful in the setting of diagnostic surgical pathology practice as well as in the identification of a MYB driver for MYB-targeted clinical trial inclusion. We selected antibodies to enrich for target epitopes in the N-terminus, intending to capture increased expression of both wildtype and translocated MYB. The evaluated antibody clones have shown successful staining by immunohistochemistry in prior investigations, including AB_10912656 in the evaluation of MYB in rhabdomyosarcoma [19], AB_955796 in castration-resistant prostate cancer [20], AB_10900735 in studies focusing on renal cell carcinoma [21], colorectal carcinoma [22], and glioblastoma [23], and AB_778878 in other studies focusing on adenoid cystic carcinoma [24, 25]. The AB_778878 clone was by far the most used in published data according to literature searches for each antibody. The final 2 antibody clones were excluded from the study for failing to meet quality standards during optimization and testing on control tissues. These included LS-B11880, which has not been mentioned in any publications, and AB_1078540, which has had some success in MYBL1 protein assessment by immunohistochemistry in the literature [9].

Data from RNA-based NGS intended for gene fusion detection was further analyzed for gene expression, providing insight into gene fusions involving MYB that could be found in DNA, as well as the expression status of MYB in each tumor. These analyses identified 19 tumors with MYB gene fusions and showed that increased MYB gene expression, represented by MYB-5’ RNA, was significantly associated with the presence of a MYB gene fusion. Our study also showed that AB_778878 antibody staining is associated with MYB expression by RNA sequencing. DNA-based NGS also revealed a variety of genetic alterations, most frequently in NOTCH1 and TP53, which showed no specificity for MYB fusion status, as previously described [7, 26, 27].

The correlation between MYB expression in RNA and protein, measured by NGS and IHC, respectively, could be expected as a corollary of translating abundant RNA to protein. However, the results from our study suggest a more complex process, given a significant association between MYB RNA and protein levels was seen using only one antibody clone (AB_778878), notably the only monoclonal antibody tested in the study, which was distinct from the antibody that showed the best sensitivity in the ACC cohort (AB_10900735), while 2 antibodies showed much lower performance metrics, and 2 clones failed to be included in the study. The biological cause for the variable IHC findings is unclear and is likely multifactorial, especially when considering the overlapping epitopes for multiple antibodies. A search through the literature shows that the AB_778878 is the only one that has been previously studied in the setting of ACC, while the other antibodies investigated in this study have been used in up to several investigations each, all in other tumor types. Also, despite a significant association between MYB gene fusion status and MYB-5’ expression level, there was no significant association between any antibody clone and gene fusion status. Taken together, these findings suggest that the dependence on MYB signaling for oncogenesis in ACC is mediated through a complex interplay between DNA-level changes (i.e., gene fusions), MYB transcription levels, and MYB protein abundance. This is consistent with previous data showing MYB gene expression measured by RNA is found in approximately 90% of ACC tumors despite translocations only being found in a subset of those tumors, raising the question of what factors contribute to variability in MYB transcription and translation [10, 12, 28]. With regards to MYB::NFIB translocations, the most frequently observed genetic alteration in ACC, it has been shown that H3K27 acetylation-rich enhancer hijacking drives increased MYB expression [29], while other studies have identified an alternative promoter (TSS2) for MYB expression that leads to increased expression in metastatic ACC relative to primary tumors [30].

Given the importance of MYB expression as a potential marker to aid in the diagnosis of ACC, a recently described RNA in situ hybridization (ISH) assay has been shown to be effective. Rooper et al. reported a 92% sensitivity of the RNA ISH assay for ACC, including in 97% of tumors with MYB rearrangement and 83% of tumors without rearrangements [31]. The greater sensitivity for MYB-targeted RNA ISH in MYB-translocated ACC is consistent with the observation in our cohort that MYB expression by RNA sequencing is increased for MYB-translocated cases when compared to fusion-negative tumors. The presence of high MYB expression in some fusion-negative tumors may be, at least in part, mediated by peri-MYB translocations, as recently demonstrated by Ueda et al. [32] RNA ISH showed an 89% specificity, with MYB overexpression in basaloid and sinonasal carcinomas thought to drive the lower specificity relative to sensitivity. MYB IHC in that study showed a marginally higher sensitivity for ACC (94%) than RNA ISH, which is consistent with the 97% MYB IHC sensitivity observed in our cohort. The lower specificity reported in that study using a 5% threshold for determination of positive IHC staining might be mitigated in future use with a higher threshold for positivity, such as the 30% threshold used in the present cohort, without significant loss of sensitivity [31].

The standard therapeutic approach continues to rely primarily on complete surgical excision with negative margins and adjuvant radiotherapy. ACC has still proven to be a challenging tumor to treat, with its characteristic perineural invasion, delayed onset of recurrences, and distant metastases. From the molecular perspective one can speculate that the latter may be, at least in part, related to the complex nature of MYB signaling in ACC, which is supported by the enrichment for TSS2 activation in ACC metastases and the activated biological programs seen in poor survival-associated gene expression signatures [33, 34]. Despite transcription factors being considered generally poor candidates for targeted therapy, MYB has stood apart, particularly in early studies of leukemia models. Specifically, a large number of recent efforts have focused on chemical compounds that disrupt the interaction with the KIX domain of p300 via MYB’s transactivation domain, a necessary step in mediating MYB signaling [15]. Others have focused on targeting the ATR and BUB1 proteins in ACC cell lines and patient-derived xenografts with successful results [35, 36].

There were some limitations to the present study, such as the use of a targeted clinical NGS assay that includes a subset of MYB exons as opposed to the whole gene, the effects of which were mitigated using the MYB-5’ signature. Another limitation is the lack of tumors in the differential diagnoses for ACC, which was inherent to this investigation of gene-transcript-protein interactions as well as sensitivity optimization for MYB identification via IHC in ACC. In this study we have presented a review of multiple MYB antibodies with potential diagnostic value and identified a single clone (AB_10900735) with optimal sensitivity. Furthermore, through the integration of RNA sequencing data, we were able to show the MYB fusion status, the increased expression of MYB in the setting of MYB gene fusions, as well as the association between a single MYB antibody (AB_778878) and MYB expression. Based on these findings, it is evident that the biology of ACC related to MYB is complex, necessitating future studies to develop better diagnostic, prognostic, and therapeutic biomarkers.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Author Contributions

A.S.F. and J.K.L. wrote the main manuscript text, A.S.F. and A.A.F. prepared the figures and tables, and all authors reviewed the manuscript.

Funding

This study was funded by a pilot grant from the Adenoid Cystic Carcinoma Research Foundation.

Data Availability

No datasets were generated or analysed during the current study.

Declarations

Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. For this type of study formal consent is not required. This study has obtained IRB approval from the Institutional Review Board at the Massachusetts General Hospital (2013P001818, WCF).

Consent to Participate

This study has obtained IRB approval from the Institutional Review Board at the Massachusetts General Hospital (2013P001818, WCF) and the need for informed consent was waived.

Consent for Publication

For this type of study consent for publication is not required.

Competing Interests

The authors declare no competing interests.