NKX3.1 Expression in Salivary Gland “Intraductal” Papillary Mucinous Neoplasm: A Low-Grade Subtype of Salivary Gland Mucinous Adenocarcinoma

Abstract

Background

Salivary gland intraductal papillary mucinous neoplasm (SG IPMN) is a recently proposed entity characterized by a papillary-cystic proliferation of mucin-producing cells. Because of overlapping histologic features and a clonal AKT1 p.E17K variant, SG IPMN has been presumed to be a precursor or a low-grade subtype of mucinous adenocarcinoma. NKX3.1 is a tumor suppressor gene located on chromosome 8p and is a known immunohistochemical marker of prostate epithelium and mucinous acinar cells of the intraoral salivary glands.

Methods

We retrieved 12 SG IPMN cases, and performed histologic and genetic analysis. Given the association of SG IPMN with mucinous acinar cells, we also investigated the performance of NKX3.1 as a marker of this tumor entity.

Results

Diffuse and strong NKX3.1 expression was observed in all SG IPMN cases (12/12, 100%) as well as in normal mucinous acinar cells. In contrast, mucoepidermoid carcinoma and pancreatic IPMN cases as well as normal serous acinar cells were negative for NKX3.1. Genetically, 11 of 12 cases (92%) harbored an AKT1 p.E17K variant. A novel PTEN frameshift deletion (p.G36Dfs*18) was detected in the other single case. At least one of the histologic features implying malignant tumors, such as severe cellular atypia, brisk mitotic activity, high Ki-67 proliferating index, lymphovascular invasion, and lymph node metastasis, was detected in 6 SG IPMN cases (50%).

Conclusion

The findings suggest that SG IPMN is a low-grade subtype of mucinous adenocarcinoma which may be derived from mucinous acinar cells of the minor salivary gland.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12105-022-01471-4.

Introduction

Mucinous adenocarcinoma (MA) is defined as a primary salivary gland (SG) carcinoma that displays prominent intracellular and/or extracellular mucin [1]. MA is a histologically heterogeneous entity which include colloid carcinoma, papillary cystadenocarcinoma, and signet-ring cell carcinoma; however, a common AKT1 p.E17K variant was recently reported [2].

Low-grade papillary cystic proliferations of mucinous columnar cells were recently described as a distinct minor salivary gland tumor entity originally designated as intraductal papillary mucinous neoplasms (IPMN) [3, 4]. It is still controversial whether MA and SG IPMN are separate entities or a disease spectrum, but because both MA and SG IPMN are characterized by mucin production and the presence of AKT1 p.E17K variants, SG IPMN is currently considered a low-grade or non-invasive subtype of MA [2, 5, 6].

NKX3.1 is a multifaceted protein with roles in both prostate and minor salivary gland development [7, 8]. Concordantly in normal tissues, immunohistochemical expression of NKX3.1 has been observed in prostatic epithelium and in mucinous epithelium of minor salivary glands in the oral cavity [9]. Targeted disruption of the mouse Nkx3.1 gene resulted in defective branching morphogenesis of the prostate and palatine salivary glands [10]. Immunohistochemical staining for NKX3.1 is widely used as a marker to assess for metastatic prostatic adenocarcinoma and some subtypes of soft tissue sarcoma [9, 11]; however, in salivary tumors, NKX3.1 expression has been reported in some cases of salivary duct carcinoma and pleomorphic adenoma [9, 12, 13].

We have analyzed NKX3.1 expression in 12 SG IPMN cases and compared expression results with its histologic mimics. Further, we have described their histologic and genetic features, to concomitantly investigate their malignant potential. Because of the predominant occurrence of SG IPMN in the intraoral salivary glands and the morphologic similarity to mucinous acinar cells, we hypothesized that SG IPMN may be a neoplastic counterpart of mucinous acinar cells of intraoral minor salivary glands.

Materials and Methods

Patients and Histological Review

The present study was approved by the Institutional Review Board (IRB) of each collaborating institution, and the need to obtain informed consent was waived due to the retrospective nature of the analysis. We retrieved 12 SG IPMN cases from five academic institutions and hospitals. The surgical pathology slides of the tumors were reviewed by three head and neck pathologists (MN, WCF and PMS), and the diagnosis was confirmed based upon features described in prior publications [3, 4, 14]. Cases 4–9 in this study cohort were included in a previous study examining other features of SG IPMN [4]. We also selected and retrieved mucoepidermoid carcinoma and pancreatic IPMN cases to compare the NKX3.1 expression with its expression in SG IPMN. The patient characteristics are summarized in Supplementary Table S1.

Immunohistochemistry

Immunohistochemical (IHC) studies were performed on 5-μm-thick formalin-fixed paraffin-embedded sections. NKX3.1 IHC was conducted using a monoclonal antibody specific to the NKX3.1 (pre-diluted, clone EP356; Cell Marque, Rocklin, CA). Heat-induced antigen retrieval was performed using an autoclave step (105 °C, 5 min) in pH 9 antigen retrieval buffer. The results of NKX3.1 IHC were evaluated by three expert head and neck pathologists (MN, WCF and PMS). Normal prostate gland was used as a positive control.

Variant Analysis

Variant analysis was performed via two different methods using total nucleic acids extracted from formalin-fixed paraffin embedded tumor tissue. In 2 cases (Cases 1, 2), an Anchored Multiplex PCR assay was performed for detecting single nucleotide variants, gene rearrangements, insertions, deletions, and copy number changes [15]. The single nucleotide variants targeted by this test include AKT1, HRAS, PIK3R1, PTEN, TP53 and ninety-five additional target genes described previously [16]. In 10 SG IPMN cases (Cases 3–12), polymerase chain reaction and subsequent Sanger sequencing were performed. The protocol was the same as that described previously [17]. Variant analysis for AKT1 (exon 2) and HRAS (exons 2 and 3) was performed for all cases, and analysis of BRAF, PIK3CA, KRAS, and GNAS was performed for 5 cases. The primer sequences are listed in Supplementary Table S2.

Results

Clinical Findings

Patient characteristics of our SG IPMN cohort are summarized in Table 1. The patients’ ages ranged from 61 to 84 years (mean 72.3 years), with a female:male ratio of 7:5. Tumors ranged in size from 5 to 85 mm (mean 23.7 mm). Except for 2 sublingual cases, the other 10 SG IPMN cases arose in the intraoral minor salivary glands. The most common anatomical site was buccal mucosa (4 cases, 33%), followed by the upper lip, oral floor, and sublingual gland (2 cases each, 17%). Surgical excision or excisional biopsy were performed for all cases and 1 case received post-operative radiation therapy. In 8 cases with clinical follow-up data, all patients were alive without recurrence or metastasis with follow-up periods of 2 to 96 months (median, 20 months; excluding recent cases).

Table 1.

Clinical and genetic characteristics of salivary gland intraductal papillary mucinous neoplasm

Case no	Age (years)	Sex	Anatomical site	Size (mm)	Genetic alteration	Follow-up (mo)
1	61	M	Oral floor	85	PTEN p.G36Dfs*18	PORT, NED (18)
2	78	F	Base of tongue	10	AKT1 p.E17K, PIK3R1 p.M582Dfs*20	No data
3	82	F	Sublingual gland	49	AKT1 p.E17K, homozygous	NED (16)
4	76	F	Buccal mucosa	22	AKT1 p.E17K	NED (96)
5	66	M	Buccal mucosa	20	AKT1 p.E17K HRAS p.Q61R	NED (48)
6	82	M	Buccal mucosa	20	AKT1 p.E17K	NED (3)
7	76	F	Sublingual gland	18	AKT1 p.E17K	NED (2)
8	63	M	Gingiva	7	AKT1 p.E17K	NED (22)
9	63	F	Oral floor	25	AKT1 p.E17K HRAS p.Q61R	NED (60)
10	66	F	Upper lip	5	AKT1 p.E17K	Recent case
11	84	M	Upper lip	10	AKT1 p.E17K	Recent case
12	70	F	Buccal mucosa	13	AKT1 p.E17K	Recent case

NED no evidence of disease, PORT post-operative radiation therapy

Histologic Findings

The histologic characteristics of our SG IPMN cohort are summarized in Table 2. Six cases (50%) were well-demarcated and circumscribed tumor (Fig. 1A) and the other 6 cases (50%) had irregular peripheries. The lesions were comprised exclusively of a papillary or tubular growth of mucin-producing epithelial cells. Cytologic atypia varied among cases and among areas within the same tumor. Four cases (33%) were composed of tumor cells with only minimal atypia. The monotonous bland round nuclei were basally located and the cytoplasm contained apical mucin (Fig. 1B). In tumor cells with moderate atypia (4 cases, 33%), the nuclear size was slightly enlarged and the papillary architecture became more crowded and complex (Fig. 1C). In the other 4 cases (33%), tumor cells exhibited severe atypia. In these tumors, the nuclear polarity was partially lost and brisk mitotic activity was observed. The cytoplasmic mucin was less prominent than tumor cells with mild and moderate atypia (Fig. 1D).

Table 2.

Histologic characteristics of salivary gland intraductal papillary mucinous neoplasm

Case no	Cellular atypia	Mitotic counts/2 mm2	Ki-67 PI	Necrosis	Boundary	Lymphovascular invasion	LN metastasis	Protrusion from Orifice	Mucin extravasation	P63 rimming
1	Moderate	3	8%	(−)	Irregular	( +)	( +)	(−)	(−)	(−)
2	Mild	1	Low	(−)	Irregular	(−)	(−)	(−)	(−)	(−)
3	Severe	18	20%	(−)	Irregular	(−)	( +)	(−)	(−)	(−)
4	Mild	0	2%	(−)	Regular	(−)	(−)	(−)	(−)	(−)
5	Moderate	6	70%	(−)	Regular	( +)	(−)	(−)	(−)	(−)
6	Severe	15	30%	(−)	Regular	(−)	(−)	(−)	(−)	(−)
7	Mild	0	1%	(−)	Regular	(−)	(−)	( +)	(−)	(−)
8	Mild	0	2%	(−)	Regular	(−)	(−)	(−)	(−)	(−)
9	Moderate	2	5%	(−)	Regular	(−)	(−)	(−)	(−)	(−)
10	Moderate	0	1%	(−)	Irregular	(−)	(−)	(−)	(−)	Partially ( +)
11	Severe	13	40%	(−)	Irregular	( +)	(−)	(−)	( +)	(−)
12	Severe	25	35%	(−)	Irregular	(−)	(−)	( +)	(−)	(−)

Ki-67 PI Ki-67 proliferation index, LN lymph node

Fig. 1.

Histologic features of salivary gland intraductal papillary mucinous neoplasm (SG IPMN). A A well-demarcated and circumscribed tumor is observed. B SG IPMN with mild cellular atypia. Tumor cells have monotonous basally-located nuclei and apical mucin. C SG IPMN with moderate cellular atypia. The papillary pattern becomes more complex and hypercellular than SG IPMN with mild cellular atypia. D SG IPMN with severe cellular atypia. Brisk mitotic activity is observed. The nuclear polarity is partially lost and cytoplasmic mucin is less remarkable than in SG IPMN with mild or moderate atypia. E Lymph node metastasis is detected. F The papillary growth protrudes from the orifice. G Invasion into surrounding parenchymal tissue with mucin extravasation is observed. H Non-neoplastic mucinous acinar cells are present in the neoplastic papillary growth

Mitotic counts ranged from 0 to 25/2 mm². Cases with severe atypia tended to have higher mitotic counts. Necrosis was absent in any cases. Four cases (33%) exhibited lymphovascular invasion or metastasis to lymph nodes (Fig. 1E). Protrusion of the tumor from the orifice was observed in 2 cases (Fig. 1F). Mucin extravasation and invasion into the surrounding parenchyma was detected in 1 case (Fig. 1G), and a mixture of non-neoplastic mucinous cells was present in 1 case (Fig. 1H).

Immunohistochemical Findings

Results of NKX3.1 IHC are summarized in Table 3. All SG IPMN cases exhibited diffuse and strong NKX3.1 expression (Fig. 2A, B). In contrast, none of the mucoepidermoid carcinomas (Fig. 2C) and pancreatic IPMN (Fig. 2D) were positive for NKX3.1. In normal minor salivary gland parenchyma, mucinous acinar cells were positive for NKX3.1 in 13 cases (93%) (Fig. 2E), but serous acinar cells and striated duct cells were negative (Fig. 2F).

Table 3.

Results of NKX3.1 immunohistochemistry in normal salivary gland, salivary gland intraductal papillary mucinous neoplasm, and other histologic mimics

	NKX3.1 positive cases (%)
SG IPMN	12/12 (100)
IPMN mimics
Mucoepidermoid carcinoma	0/12 (0)
Pancreatic IPMN	0/11 (0)
Normal cells
Mucinous acinar cells	13/14 (93)
Serious acinar cells	0/6 (0)
Striated duct cells	0/14 (0)

SG IPMN salivary gland intraductal papillary mucinous neoplasm

Fig. 2.

NKX3.1 immunohistochemistry in SG IPMN, its mimics, and normal salivary gland. H&E stain of SG IPMN (A) and corresponding NKX3.1 immunohistochemistry (IHC) (B). Tumor cells show diffuse and strong NKX3.1 expression. The expression is limited to the nucleus (inset). C Mucoepidermoid carcinoma and its NKX3.1 IHC (inset). All tumor cells including mucous cells are negative for NKX3.1, D Pancreatic IPMN morphologically resembles SG IPMN, but completely lacks NKX3.1 expression. E Normal mucinous acinar cells are positive for NKX3.1, while striated duct cells are negative. F Serous acinar cells are negative for NKX3.1

The Ki-67 proliferation index ranged from 1 to 70%. The index varied remarkably between tumors and even between areas in the same tumor (Table 2; Fig. 3A, B). 92% of SG IPMN cases (11cases) lacked p63-positive myoepithelial/basal cells, while 1 case had partial rimming by p63-positive cells (Fig. 3C, D).

Fig. 3.

Immunohistochemical features of SG IPMN. The Ki-67 proliferation index is low in SG IPMN with mild atypia (A), while SG IPMN with severe atypia show a high Ki-67 proliferation index (B). Most SG IPMN cases lack p63-positive myoepithelial rimming (C, left portion) in contrast to normal mucinous salivary gland myoepithelial cells (C, right portion). A partial myoepithelial cell lining is observed in one SG IPMN case (D)

Genetic Findings

The results of genetic analysis are summarized in Table 1. Eleven of 12 cases (92%) harbored an AKT1 p.E17K variant. Among them, 1 case (Case 3) had a homozygous p.E17K variant (Fig. 4). In the single case without an AKT1 variant (Case 1), a PTEN frameshift deletion variant (c.107delG) which causes functional loss of the protein was detected. Two cases showed an HRAS p.Q61R variant, and NGS analysis revealed an additional PIK3R1 p.M582Dfs*20 variant in 1 case (Case2).

Fig. 4.

Chromatogram of DNA sequencing of AKT1 (case 3). While surrounding normal salivary tissue is AKT1 wild-type, tumor cells exhibited a c.49G > A (p.E17K) peak

Discussion

NKX3.1 plays a crucial role in the development of oral minor salivary glands [8]. Concordantly, normal mucinous acinar cells are immunohistochemically positive for NKX3.1 [9]. One study has investigated NKX3.1 expression in various salivary gland tumors including adenoid cystic carcinoma, mucoepidermoid carcinoma, acinic cell carcinoma, adenocarcinoma NOS, and pleomorphic adenoma; however, none of the 38 tumors investigated were positive for NKX3.1 [9]. In the present study, diffuse and strong nuclear NKX3.1 expression was observed in all SG IPMN cases, while mucoepidermoid carcinomas were negative for NKX3.1 despite the presence of neoplastic mucinous cells.

SG IPMN was named as such due to the morphologic resemblance to pancreatic IPMN [3]. Despite their low-power histologic similarity, immunohistochemistry shows these minor salivary gland tumors are not intraductal (p63 largely absent). The genetic features of these two entities also differ: SG IPMN are associated with AKT1 variants while pancreatic IPMN are characterized by variants in GNAS [3, 4, 18]. Due to both discrete molecular changes and pancreatic IPMN lacking NKX3.1 expression, their immunohistochemical and molecular features indicate that pancreatic and SG IPMN are distinct neoplasms, and the latter, in fact, are not even intraductal. Overall, NKX3.1 appears to be a useful immunohistochemical marker distinguishing SG IPMN from its local histologic mimics, primarily mucoepidermoid carcinoma (Table 3).

Because NKX3.1 acts as a co-factor of androgen receptor (AR), two studies have investigated NKX3.1 in salivary duct carcinoma (SDC), a primary high-grade salivary gland carcinoma characterized by AR expression [12, 13]. Immunoreactivity for NKX3.1 was detected in 12–36% of SDC cases. From these data, NKX3.1 expression does not appear to be entirely specific to distinguish SG IPMN from all salivary gland tumors, but specifically, AR would be co-expressed in SDC that might also manifest more focal NKX3.1 expression than SG IPMN [12, 13, 19]. Further, mucin production is rare in SDC, and differentiating SDC from SG IPMN is typically morphologically straightforward.

The characteristic and common gene alteration of both SG IPMN and MA is an AKT1 p.E17K variant. Until now, including this study, 24 out of 26 reported cases (92%) of SG IPMN have been found to harbor an AKT1 p.E17K variant [2–4]. Among 2 AKT1 variant-negative cases, one case in the present study contained a frameshift deletion of PTEN (p.G36Dfs*18) resulting in functional loss of PTEN, a phosphatase of the PI3K/AKT pathway that is frequently involved in oncogenesis [20]. Because both AKT1 and PTEN are key components of the PI3K/AKT pathway, the frameshift deletion of PTEN in our cohort SG IPMN case is presumed to be a driver variant of SG IPMN [21]. Interestingly, PTEN loss has not been reported previously in SG IPMN or MA. Another SG IPMN case in our cohort exhibited only an AKT1 c.49G > A peak (p.E17K) in the chromatogram (Fig. 4). As the surrounding normal salivary gland tissue was AKT1 wild-type, this result could be interpreted as a homozygous variant or one variant allele with deletion of the other allele. In other 3 cohort cases, an additional HRAS or PIK3R1 variant together with an AKT1 variant was detected as previously described [1, 4]. Recently, frequent TP53 variants (88%) were reported in MA [1]; however, previous analyses of 3 SG IPMN cases and 2 cases in the current study revealed no TP53 variant [3]. Frankly invasive MA including colloid carcinoma and signet-ring cell carcinoma, namely non-IPMN MA, may tend to harbor TP53 variants in addition to AKT1 variants; however, more SG IPMN and MA cases need to be analyzed to characterize their TP53 variant profile.

The terminology for SG IPMN was proposed because the lesions were well-demarcated, mimicking intraductal proliferations. However, in fact, most of the reported SG IPMN cases have lacked surrounding myoepithelial/basal cells suggesting that they are, in fact, not true intraductal proliferations [3, 4]. In the current study, except for partial rimming by myoepithelial/basal cells in 1 case (Fig. 3D), there was no evidence of surrounding myoepithelial/basal cells present in any cases. These results may support the idea that SG IPMN has a malignant potential. Furthermore, invasion into the surrounding parenchyma and lymphovascular invasion in SG IPMN have been reported in several studies, including here in three of our cases (Case 1, 5, 11, Table 2) [19, 22]; and, in our cohort, lymph node metastasis was demonstrated in 2 cases (17%), both of which were larger in size with unique genetics (Cases 1, 3; Table 1). Overall, histologic features suggestive of aggressive biological potential, including severe cellular atypia, brisk mitotic activity, high Ki-67 proliferating index, lymphatic invasion, and lymph node metastasis were detected in 6 SG IPMN cases (50%), despite no evidence of residual disease in our cohort. Except for the absence of unequivocal invasion, other histologic features including cell composition and architectural growth pattern overlapped with those of MA [2]. Although long-term follow-up data are not yet available in all cases, it is reasonable to consider that SG IPMN represent a low-grade counterpart of MA [5, 6].

Previous reported cases of MA and SG IPMN have been exclusively from the intraoral salivary glands including the sublingual gland. In 2018, a morphologically similar mucinous tumor was reported in the trachea and was followed by lung metastases detected 2 years after the primary surgery [23]. The authors considered it as the first case of primary mucinous adenocarcinoma of the trachea arising from the bronchial seromucinous glands. Interestingly, this tumor was also diffusely positive for NKX3.1, but molecular analysis was not performed for the case. For our cohort, the distinctive intraoral localization of SG IPMN and NKX3.1 expression might suggest the possibility that SG IPMN may arise from the mucinous acinar cells of the minor salivary gland.

The current study is associated with certain limitations. First, comprehensive molecular analysis using next generation sequencing was performed for only 2 of our cohort cases (17%), while variant analysis was limited to AKT1, HRAS and a subset of other genes for the other 10 SG IPMN cases in our cohort. To better understand the true relationship between SG IPMN and MA, the genetic landscape of these 2 entities will need to be studied using comprehensive molecular analysis. Second, the current cohort did not include colloid carcinoma or signet-ring cell carcinoma cases, namely non-IPMN MA. Despite their interest as MA subtypes, they are morphologically distinct, and typically not a pitfall in the diagnosis of SG IPMN. However, the most common mucous-cell-containing salivary carcinomas in this location, mucoepidermoid carcinomas, were included. To better elucidate the relationship between SG IPMN and MA, and their histogenesis, NKX3.1 expression of additional non-IPMN MA cases will need to be investigated in the future.

In conclusion, SG IPMN showed consistently diffuse and strong immunohistochemical expression of NKX3.1 that is not shared by its main histologic mimics, indicating that it may be useful as a diagnostic immunohistochemical marker. Additionally, the histologic features and gene alterations in the PI3K/AKT pathway, including the AKT1 p.E17K variant, indicate that SG IPMN has significant biological potential and may represent a low-grade subtype of MA.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

MN, PMS, and WCF contributed to the study conception and design. Material preparation, data collection and analysis were performed by MN, PMS, RH, KK, TT, MU, TN, and WCF. The first draft of the manuscript was written by MN, PMS, and WCF. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This study was funded by NIH/NHS 1P01CA240239-01 (WCF, PMS).

Data Availability

The data that support the findings of this study are available from the corresponding author, MN, upon reasonable request.

Code Availability

Not applicable.

Declarations

Conflict of interest

The other authors have no relevant financial or non-financial interests to disclose.

Ethical Approval

The present study was approved by the Institutional Review Board (IRB) of each collaborating institution (Mass General Brigham IRB 2015P001749, Nagoya University IRB 2018–019014195).

Consent to Participate

The need to obtain informed consent was waived due to the retrospective nature of the analysis.

Consent for Publication

The need to obtain informed consent was waived due to the retrospective nature of the analysis.