Research advances and treatment perspectives of pancreatic adenosquamous carcinoma

Abstract

Background

As a malignant tumor, pancreatic cancer has an extremely low overall 5-year survival rate. Pancreatic adenosquamous carcinoma (PASC), a rare pancreatic malignancy, owns clinical presentation similar to pancreatic ductal adenocarcinoma (PDAC), which is the most prevalent pancreatic cancer subtype. PASC is generally defined as a pancreatic tumor consisting mainly of adenocarcinoma tissue and squamous carcinoma tissue. Compared with PDAC, PASC has a higher metastatic potential and worse prognosis, and lacks of effective treatment options to date. However, the pathogenesis and treatment of PASC are not yet clear and are accompanied with difficulties.

Conclusion

The present paper systematically summarizes the possible pathogenesis, diagnosis methods, and further suggests potential new treatment directions through reviewing research results of PASC, including the clinical manifestations, pathological manifestation, the original hypothesis of squamous carcinoma and the potential regulatory mechanism. In short, the present paper provides a systematic review of the research progress and new ideas for the development mechanism and treatment of PASC.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13402-022-00732-2.

Introduction of PASC

In 2017, 441,000 individuals were diagnosed with pancreatic cancer worldwide, compared to 196,000 in 1990 [1]. The universal burden of pancreatic cancer has more than doubled over the past 20 years. This digestive tract tumor has a poor prognosis with a 5-year survival rate of less than 11% [2]. Pancreatic cancer consists of a variety of histological types with differentiated statuses, including pancreatic ductal adenocarcinoma (PDAC), pancreatic adenosquamous carcinoma (PASC), colloid carcinoma, hepatoid carcinoma, medullary carcinoma, signet ring cell carcinoma, undifferentiated carcinoma with osteoclast-like giant cells, and undifferentiated carcinoma [3]. Approximately 90% of pancreatic cancer is PDAC, and 1-4% is PASC, an enigmatic and aggressive tumor that has a worse prognosis and higher metastatic potential than its adenocarcinoma counterpart [4].

PASC has been referred to as adenoacanthoma, mixed squamous and adenocarcinoma, and mucoepidermoid carcinoma. In 1907, Gotthold Herxheimer first reported this histological subtype, referring to it as “Cancroide” [5]. Histologically, PASC of the pancreas is strictly defined as a neoplasm with 30% or more malignant squamous cell carcinoma mixed with ductal adenocarcinoma [6]. However, it is unknown whether the degree of squamous cell differentiation in PASC (e.g., < 30% versus ≥ 30%) is clinically relevant. Although pancreatic cancer lesions with squamous differentiation lead to a worse prognosis for patients, the proportion of squamous differentiation is not associated with median overall survival (< 30% versus ≥ 30%, P = 0.82) [7–9].

Some clinical features of PASC have been reported and compared with that of PDAC [4, 7, 10–12]. The mean age at diagnosis for both PASC and PDAC patients is around 67 years, but the incidence is slightly higher in men than in women [4, 11]. The most common primary tumor location is the head of the pancreas in both tumors. The clinical presentation is similar across each disease with abdominal pain, obstructive jaundice, weight loss, and anorexia [4, 6, 10–12]. However, the larger tumor size, higher proportion of grade III and IV tumors, poorer differentiation and higher angiolymphatic and perineural invasion rate result in that PASC is more malignant and has a worse prognosis than PDAC [4, 7, 10–13] (Table 1). In addition, several case reports have revealed the presence of malignancy-associated hypercalcemia in PASC, which is rarely observed in PDAC [9, 14, 15]. Interestingly, hyperglycemia has been reported in up to 80% of patients with PDAC, but the incidence of hyperglycemia in PASC is rather low [9, 16].

Table 1.

Summarization of PASC clinicopathologic features from the existing large clinical studies

	2022 Braun R [10]		2022 Lv [11]		2021 Lee [7]		2018 Hester [12]		2012 Boyd [4]
	PASC	PDAC	PASC	PDAC	PASC	PDAC	PASC	PDAC	PASC	PDAC
n (%)	278	37,938	784	1568	56	100	1745	205,328	415	45,693
Sex
Female	123/44.2	17,712/46.7	374/47.7	733/46.7	21/37.5	/	813/46.6	101,779/49.6	45.50%	49.40%
Male	155/55.8	20,226/53.3	410/52.3	835/53.3	35/62.5	/	932/53.4	103,549/50.4	54.50%	50.60%
Age	70	70	67.9	67.5	63	/	/	/	66.6	66.9
Sizes	/	/	/	/	4.75 cm	4.17 cm	> 4cm	2-4 cm	5.7 cm	4.3 cm
n (%)	142	12,768	784	1568	56	100	1745	205,328	241	24,796
Grade
G1	0/0	510/4.0	3/0.38	7/0.45	47/83.9	86/86	277/15.8	42,813/20.8	28.60%	55%
G2	40/28.2	5895/46.2	100/12.8	177/11.3
G3	95/66.9	5201/40.7	299/38.1	605/38.6	9/16.1	14/14	709/40.6	35,474/17.3	71.40%	45%
G4	0/0	62/0.5	14/1.79	28/1.79
GX	7/4.9	1100/8.6	368/46.9	751/47.9	0	0	759/43.5	127,041/61.9	0	0
	2022 Braun R [10]		2022 Lv [11]		2021 Lee [7]		2018 Hester [12]		2012 Boyd [4]
	PASC	PDAC	PASC	PDAC	PASC	PDAC	PASC	PDAC	PASC	PDAC
n (%)	142	12,768	784	1568	56	100	1745	205,328	267	27,545
Lymph node metastasis
L0	44/31	6868/30.3	282/36	608/38.8	7/12.5	33/33	280/16	18,858/9.2	47.20%	52.90%
L1	97/68.3	8558/67	311/39.7	599/38.2	49/87.5	67/67	383/21.9	30,441/14.8	52.80%	47.10%
Lx	1/0.7	342/2.7	191/24.4	361/23	0	0	1082/62	156,029/76	0	0
Vascular metastasis
V0	72/50.7	6086/47.7	/	/	20/35.7	71/71	150/8.6	12,643/6.2	/	/
V1	40/28.2	2037/16.0	/	/	36/64.3	29/29	140/8	7386/3.6	/	/
Vx	30/21.1	4645/36.4	/	/	/	/	1455/83.4	185,299/90.2	/	/
Nerve metastasis
N0	/	/	/	/	7/12.5	11/11.0	/	/	/	/
N1	/	/	/	/	49/87.5	89/89	/	/	/	/
Nx	/	/	/	/	/	/	/	/	/	/
Distant metastases
M0	117/82.4	10.101/79.1	365/46.6	702/44.8	46/82.1	94/94	/	/	/	/
M1	21/14.8	1966/15.4	297/37.9	633/40.4	10/17.9	6/6.0	/	/	/	/
Mx	4/2.8	701/5.5	122/15.6	233/14.9	/	/	/	/	/	/

Note: “/” represent no data available

In recent years, a large number of innovative works on PDAC have been conducted; however, there have been few systematic attempts to evaluate PASC. This article reviews and systematically summarizes the literature on the clinical manifestation, diagnosis, and pathological features of PASC, and it provides a comprehensive summary and new ideas for further study of its pathogenesis and new therapy.

Pathological characteristics

As the gold standard for diagnosing various diseases, pathological manifestation also plays an important role in the diagnosis of PASC. Adenosquamous carcinoma consists mainly of adenocarcinoma tissue and squamous carcinoma tissue with an occasional sarcomatous component. Despite the irregular arrangement of the components, there are significant differences in cytomorphology and molecular markers, and pathologists can make an accurate diagnosis of adenosquamous carcinoma by these differences [17] (Fig. S1).

Morphologically speaking, adenocarcinoma forms glands or glandular structures, often containing intracellular mucosa or lumen [18]. Squamous differentiation is characterized by dense and eosinophilic cytoplasm, clear cell boundaries, aggregation or stratification of polygonal cells, and varying degrees of keratinization [6]. Compared to the nuclei of benign squamous cells, the nuclei of malignant squamous cells are hyperchromic and pleomorphic. The cells of sarcomatous tissue lack cell adhesion and exhibit morphologies from round to spindle-shaped [6, 19, 20]. In addition to morphological differences, molecular markers for each component of adenosquamous carcinoma are also distinctive. Glandular tissue expresses the typical PDAC markers, namely, CA19-9, CEA, CK7, MUC1, and MUC5AC [17, 18]. Squamous cell carcinoma has been shown to express TP63, TP40, and CK5/6, which is identical to the squamous cell carcinoma immunophenotype of other organs [17, 18]. It is worth noting that PD-L1 expression is limited to the squamous component but not to the adenocarcinoma component [7, 21, 22]. For sarcoma tissue, although P40 staining for the sarcomatous component is positive along with squamous carcinoma, E-cadherin expression is absent, while Vimentin is expressed, which is related to the ability of tumor invasion, suggesting that sarcomatoid tissue may be a signal of increasing malignancy of adenosquamous carcinoma [19] (Table 2).

Table 2.

Comparisons of PDAC and PASC

	PDAC	PASC	Ref.
Percentage of pancreatic cancer	> 90%	1%-4%	[4]
Clinical Presentation	Abdominal pain, obstructive jaundice, weight loss, and anorexia	Similar but more metastatic, larger lesions and worse prognosis	[4, 7, 8, 12, 23–25]
Components	Adenocarcinoma tissue	Adenocarcinoma tissue: squamous carcinoma tissue with the occasional sarcomatous component	[6, 18]
Cytomorphology	Forming glands or glandular structures, often containing intracellular mucosa or lumen	Adenocarcinoma tissue: forming glands or glandular structures, often containing intracellular mucosa or lumen. Squamous differentiation: dense and eosinophilic cytoplasm, clear cell boundaries, aggregation or stratification of polygonal cells, and varying degrees of keratinization Cells of sarcomatous tissue: lacking cell adhesion and exhibiting morphologies from round to spindle-shaped	[6, 18–20]
Molecular Markers	CA19-9, CEA, CK7, MUC1, and MUC5AC	Adenocarcinoma tissue: CA19-9, CEA, CK7, MUC1, and MUC5AC Squamous carcinoma tissue: TP63, TP40, CK5/6, and PD-L1 Sarcomatous component: P40	[7, 17–19, 21, 22]

Mysterious origin hypotheses of squamous carcinoma

Because squamous epithelial cells do not exist in normal pancreatic tissue, it is important to understand where the squamous component of adenosquamous carcinoma comes from. There are three main popular hypotheses as follows: the collision theory, pancreatic cancer stem cell differentiation theory, and squamous metaplasia of adenocarcinoma theory (Fig. 1).

Fig. 1.

Three main popular hypotheses for the origin of squamous epithelial cells in adenosquamous carcinoma. The collision theory (upper): two histologically distinct tumors arise independently in the pancreas and are joined together. Pancreatic cancer stem cell differentiation theory (middle): Adenocarcinoma and squamous carcinoma derived from the differentiation of pancreatic stem cells. Squamous metaplasia of adenocarcinoma theory (lower): squamous carcinoma is derived from adenocarcinoma

Collision theory

The collision theory suggests that two histologically distinct tumors arise independently in the pancreas and are joined together, leading to PASC. However, as the research has progressed, the evidence supporting the hypothesis has become less and less, it is no longer accepted by scientists.

Differentiation of pancreatic stem cells

Adenocarcinoma and squamous carcinoma, both derived from the differentiation of pancreatic stem cells, are the main components of the second theory, which is supported by the presence of pancreatic cancer stem cells in PASC lesions [9, 26]. Zhao et al. performed single-cell RNA sequencing (scRNA-seq) to profile sample tissues from a healthy donor pancreas, an intraductal papillary mucinous neoplasm, and a patient with PASC, and they certificated a subset of adenosquamous carcinoma cells expressing characteristic proteins of stem cells, such as UBE2C, ASPM, CENPF and TOP2A, on their surface as stem cell-like carcinoma cells [27]. Lenkiewicz et al. performed ATAC-seq sequencing of preclinical model samples of PASC, and they reported active expression of the pancreatic cancer stem cell regulator, RORC, supporting the conclusion that PASC is enriched in cell populations with cancer stem cell features [28]. More importantly, to analyzing single-nucleotide variants (SNVs) and indels as well as identified the copy number aberrations (CNAs) of each component, Fang et al. separate adenocarcinoma tissue and squamous carcinoma components from three PASC by using laser capture microdissection (LCM). Fang et al. had reported that the two components showed similar patterns of recurrent CNAs and comparable mutation frequencies for several detected SNVs/indels [29, 30]. These results suggest that the two cellular components of PASC contain similar genomic alterations and may develop from the same progenitor tumor cells.

Squamous metaplasia of adenocarcinoma

The theory of squamous metaplasia of adenocarcinoma suggests that squamous carcinoma is derived from adenocarcinoma as researchers have identified the transition zone region and the important role of TP63 in the transformation of adenocarcinoma to squamous carcinoma [9, 31]. Boecker et al. identified a transition zone region located midway between the K8/18 + adenocarcinoma component and the TP63⁺/TP40⁺/K5/K14⁺ squamous components, and they reported that this region expressed both K8/18 and TP63 proteins [17]. Thus, the K8/18 + glandular cells of PASC transform to TP63⁺/TP40⁺/K5/K14⁺ cells, which constitutes squamous metaplastic epithelial proliferation in these tumors. These findings indicate that TP63 is a key player in this metaplastic process.

Ji et al. reported that the transdifferentiation of adenocarcinoma to squamous carcinoma is promoted by the upregulation of TP63 in lung cancer [32, 33]. TP63 also plays a role in pancreatic cancer. TP63 is considered a marker of pancreatic squamous differentiation and can be used to distinguish squamous metaplasia/migratory chemosis from pancreatic intraepithelial neoplasia (PanIN) [34–38]. At the molecular level, high TP63 expression promotes squamous differentiation of the pancreas by promoting enhancer reprogramming and synergizing with a series of downstream factors [34–36].

To date, the origin of squamous carcinoma in adenosquamous carcinoma remains unclear, and more studies are needed to elucidate the origin and mechanisms of adenosquamous carcinoma in the pancreas.

Pathogenesis of PASC

Due to the limited incidence of adenosquamous carcinoma and the difficulty of collecting sufficient cases, the genomic features of this disease have rarely been explored. In recent years, researchers have conducted additional studies on PASC at the molecular level and have used multiple sequencing approaches to primarily analyze PASC. By comparing adenosquamous carcinoma with ductal adenocarcinoma at the molecular level, researchers have attempted to identify specific pathogenesis of PASC, laying the foundation for the discovery of additional potential therapeutic targets. In this article, we combine the reported data on the possible pathogenesis of adenosquamous carcinoma from three different aspects as follows: genetic mutations, chromosomal fragment deletion, and the tumor microenvironment.

Genetic mutations

Several studies have detected the status of the following genetic variants in adenosquamous carcinoma, which are similar to those in ductal adenocarcinoma: CDKN2A deletions, SMAD4 deletions, KRAS mutations, TP53 mutations, and MYC amplification [28, 29]. KRAS, TP53, and SMAD4 are the most commonly mutated genes in both cancers. Notably, TP53 is more frequently mutated in adenosquamous carcinoma than in ductal adenocarcinoma and is often prevalent together with KRAS mutations (88% of adenosquamous carcinoma patients have both TP53 and KRAS mutations), suggesting that TP53 may play an important role in the development of adenosquamous carcinoma [29].

In addition to similar genetic variants of two pancreatic cancers, adenosquamous carcinoma exhibits unique genomic landscape mutations in chromatin regulators and activation of oncogenic pathways. Through whole-exome sequencing of PASC, Lenkiewicz et al. reported additional heterogeneous lesions, including SMARCA2, ARID2, and ASXL2 homozygous deletion and mutations of key regulators of the epigenome, such as the TET1 DNA demethylase, the MSL2 histone ubiquitin E3 ligase, and the KANSL1 chromatin regulator. In addition, Lenkiewicz et al. detected amplification of TLK2, ASH2L, MPL, and FRS2 as well as FGFR2 (A322G)-activating mutations and FGFR1-ERLIN2 fusion mutations. Each of the latter three lesions has been shown to activate the FGFR/FRS2 signaling pathway in cancer, which plays an important role in tumorigenesis and is an important target of targeted therapy [28]. Recently, Ma et al. identified genes with a high mutation frequency through whole-exome sequencing of 12 adenosquamous carcinoma lesions, including the germline mutated genes MAP3K1 (9 cases), PDE4DIP (7 cases), BCR (7 cases), ALK (6 cases), USP6 (5 cases), AR (4 cases), HLA-A (4 cases), SPEN (4 cases), KMT2D (3 cases), NUTM2B (3 cases), ZFHX3 (3 cases), and MN1 (3 cases) as well as the somatically mutated genes TP53 (5 cases), KRAS (3 cases), HRNR (3 cases), and OBSCN (3 cases) [39]. Among these mutated genes, somatic mutations in KRAS, OBSCN, and HRNR as well as germline mutations in USP6 are correlated with the invasion and metastasis of PASC (Table 3).

Table 3.

Genetic mutations of PASC

Authors	Sequencing method	Sequencing outcomes	Genetic Mutations of PASC	Ref.
Lenkiewicz et al. Fang et al.	Whole-exome sequencing	Mutations similar to PDAC	CDKN2A deletions, SMAD4 deletions, KRAS mutations, TP53 mutations, and MYC amplification	[28]
	Whole-genome sequencing			[29]
Lenkiewicz et al.	Whole-exome sequencing	Mutations in chromatin regulators	SMARCA2, ARID2, and ASXL2 homozygous deletion TET1, MSL2, KANSL1 mutation.	[28]
		Mutations activating oncogenic pathways	TLK2, ASH2L, MPL, FRS2 amplification FGFR2 (A322G) activating mutations FGFR1-ERLIN2 fusion mutations
Ma et al.	Whole-exome sequencing	Germline mutations genes	MAP3K1, PDE4DIP, BCR, ALK, USP6, AR, HLA-A, SPEN, KMT2D, NUTM2B, ZFHX3, and MN1	[39]
		Somatic mutation genes	TP53, KRAS, HRNR, and OBSCN

Loss of chromosomal fragments

Through whole-genome sequencing analysis, Fang et al. found that compared to PDAC, loss of the chr 3 3p region, including 3p26.3-26.1, 3p25.1-21.31, and 3p21.2-11.1, is altered in a greater proportion of cases and that 3p21.2-11.1 has significantly more copy number losses than in PDAC [29]. Analysis of the CNA data of 34 types of cancer downloaded from The Cancer Genome Atlas (TCGA) has revealed that squamous cell carcinomas (i.e., lung, 81.8%; head and neck, 72.4%) harbor an increased frequency of chr 3 3p loss. These findings suggest that frequent 3p loss is a distinct genomic feature of squamous cell carcinoma and contains genes important for the progression of squamous carcinogenesis.

Changes in the composition of the tumor microenvironment

By studying the single-cell RNA sequencing profiles of PASC, Zhao et al. revealed the heterogeneous variety of ductal and stromal cells, defined cancer-associated signaling pathways, and deciphered intercellular interactions following PASC progression [27].

The cell subpopulations within adenosquamous carcinoma contain mainly cancer cells, myeloid cells, fibroblasts, and T cells. Compared to normal pancreatic ductal cells, LGALS1, NPM1, RACK1, and PERP are upregulated from ductal cells to cancer cells. Furthermore, the copy number variations in ductal and cancer cells are greater than those in the control cells. The level of EREG, FCGR2A, CCL4L2, and CTSC increases in myeloid cells from the normal pancreas to PASC. The gene sets expressed by cancer-associated fibroblasts are enriched in immunosuppressive pathways. All of these newly identified genes may be related to the progression and remodeling of the immunosuppressive microenvironment surrounding PASC. Consequently, targeting the tumor microenvironment is expected to enhance the therapeutic effectiveness of this cancer [27].

Furthermore, Zhao noticed that cancer cells participate in cellular communication more often with myeloid cells and fibroblasts than other cells as evidenced by the high activation of cancer cell surface receptor‒ligand pairs, such as EGFR/TGFB1, ANXA1/FPR1, ICAM-1/AREG, C5AR1/RPS19, and HLA-C/FAM3C, with EGFR, in particular, being predominant. EGFR and its ligands, including TGFB1, MIF, HBEGF, GRN, COPA, and AREG, are upregulated in both cancer cells and fibroblasts. EGFR is overexpressed in cancer cells, thus activating fibroblasts and myeloid cells through molecules, such as AREG. Activated fibroblasts and myeloid cells react with cancer cells, in turn forming an EGFR-associated feedback loop that promotes the malignant transformation of normal ductal cells to cancer cells and leads to enhanced malignancy in PASC [27] (Fig. 2).

Fig. 2.

Cellular communication in normal pancreatic tissue (left) or in adenosquamous tumor tissue (right). When the normal pancreatic tissue transforms to PASC, compared to normal pancreatic ductal cells, LGALS1, NPM1, RACK1, and PERP are upregulated from ductal cells to cancer cells, the expression of EREG, FCGR2A, CCL4L2, and CTSC increases in myeloid cells. Besides, cancer cells participate in cellular communication more often with myeloid cells and fibroblasts than other cells

Diagnosis of PASC

PASC is often incorrectly diagnosed as ductal adenocarcinoma preoperatively due to similar clinical features. Currently, the diagnosis of PASC is primarily based on clinical observations, additional laboratory tests (such as serological tests), and radiography as well as histological examinations of lesions.

Potential diagnostic markers

Altered levels of certain proteins in the blood may indicate a higher risk of tumors, and serum alpha-fetoprotein levels are significantly higher in patients with early-stage hepatocellular carcinoma than in patients with cirrhosis. Regrettably, similar specific proteins have not been identified in pancreatic adenosquamous carcinoma. Patients with PASC have elevated CA19-9 levels in the blood, which has also been observed in the blood of ductal adenocarcinoma patients [40]. Thus, CA19-9 levels cannot separate patients with PASC from PDAC, thereby failing to become an adenosquamous carcinoma biomarker.

Researchers have been persistent in their work of identifying deliberate markers of PASC, and find some possible candidates. Upframeshift1/upstream shifting protein 1 (UPF1) somatic mutations have been previously considered the first molecular marker for adenosquamous carcinoma [41]. UPF1 is a core component of the RNA degradation pathway, nonsense-mediated RNA decay (NMD) [42, 43]. In 2014, Liu et al. reported that exon 10–11 and exon 21–23 regions of UPF1 are mutated at high frequencies in somatic cells of patients with adenosquamous carcinoma. These mutations affect the splicing of UPF1 precursor mRNA, resulting in decreased UPF1 protein expression and reduced efficiency of NMD [41]. Mutations in NMD lead to a range of genetic diseases, such as the development of peripheral demyelinating degeneration and the formation of many cancers [42, 43]. However, some recent studies do not support the findings of UPF1 mutations [28, 44]. First, no mutations in the UPF1 gene have been reported in other relevant studies of PASC [28, 44]. Second, UPF1 mutations are not unique to adenosquamous carcinomas and have been detected in pancreatic ductal adenocarcinomas [44] (Liu et al. reported 18/23 detected in adenosquamous carcinomas and 29/29 not detected in nonadenosquamous pancreatic adenocarcinomas). Most importantly, UPF1 mutation models constructed in human and mouse cells, deleting expression of the UPF1 exon 10–11 region, do not inhibit tumor growth, acquisition of squamous features, UPF1 RNA splicing, UPF1 protein expression, and NMD catalysis [44]. These most recent results suggest that the role of UPF1 in adenosquamous carcinoma needs to be explored in future studies.

With advanced research on pancreatic adenosquamous carcinoma, SMYD2, RORC, phTERT, and S100A2 are also considered possible markers to distinguish adenosquamous carcinoma from ductal adenocarcinoma. SET and MYND domain-containing protein 2 (SMYD2) is a lysine methyltransferase, and retinoid-acid receptor-related orphan receptor (RORC) is an essential nuclear hormone receptor that regulates pancreatic stem cells. Lenkiewicz et al. reported that the levels of H3K4me1, denoting the degree of chromatin activation in a murine PDX model, are higher in these loci in PASC than in PDAC [28]. Phosphorylated human telomerase reverse transcriptase (TERT), a telomerase reverse transcriptase, is expressed in a wide range of cancers, tends to increase with pathological grade, and is related to poor patient prognosis [45]. Matsuda et al. reported that the level of TERT expression in the squamous carcinoma fraction is higher than that in the adenocarcinoma fraction and that it is positively correlated with the expression of the squamous carcinoma cell markers, DNP63 and CK5/6 [23, 45]. In mice, DNP63 overexpression induces TERT promoter activation and reduces intracellular telomerase reverse transcriptase activity by inducing RNA splicing of TERT, suggesting that high expression of phTERT may promote squamous cell differentiation in pancreatic adenosquamous carcinoma [46]. Through single-cell transcriptome sequencing analysis of PASC, Zhao et al. recently reported that S100 calcium-binding protein A2 (S100A2) is highly expressed in cancer cells and may also be a characteristic cellular marker [27]. The high expression of S100A2 in pancreatic cancer tissues is associated with poor prognosis, and Yuan et al. reported that the expression of S100A2 in pancreatic cancer is positively correlated with the expression of PD-L1 [47]. Because PD-L1 is selectively expressed in squamous carcinoma components, these results suggest that S100A2 may be a potential diagnostic marker for PASC.

Advance on imaging diagnosis

Significant differences in imaging parameters between the two cancers have been observed through various imaging studies, including CT, CT Texture Analysis (CTTA), 18 F-fluorodeoxyglucose positron emission tomography/computed tomography (18 F-FDG PET/CT), and ultrasound-guided fine-needle aspiration (US-FNA), and have contributed to identifying histopathological subtypes of pancreatic cancer. The CT results of PASC patients show more abundant tumor necrosis in the center of the tumor than those of PDAC patients [48–50]. Imaoka et al. reported that PASC shows a higher smooth contour, annular enhancement pattern, and cystic changes compared to PDAC, and the most significant feature of PASC is the annular enhancement pattern (the sensitivity was 65.2%, and the sex specificity was 89.6%) [50]. Compared to CT, MRI depicts ring-enhancement in PASC with greater reader confidence. The concurrent presence of these two imaging features should raise high suspicion for PASC [51] (Fig. S2-S4). CTTA is a new area of ​​research that can accurately describe tumors by assessing the grayscale of tissue images and quantifying tumor heterogeneity [52, 53]. CTTA has been conducted to identify various cancers with promising preliminary results. Using the random forest (RF) method, Ren et al. built and validated 10 radiomics features at the late arterial phase (named A_) and portal venous phase (named V_) suitable for differential diagnostics of PASCs and PDACs from radiomics features derived from dual-phase CT images with 94.5% accuracy, 98.3% sensitivity, and 90.1% specificity [54].

Currently, 18 F-FDG PET/CT has been used in the prediction of the effect of chemotherapy [55, 56]. Su et al. found that the technique could also be applied in the diagnosis of PASC, and they reported that adenosquamous carcinoma absorbed more FDG (deoxyglucose) with a positive mean retention index (RI) (18/99 cases of ductal adenocarcinoma had a negative RI). In addition, the ki-67 tumor cell proliferation index was highly correlated with the SUVmax (maximum standardized value) imaging index, and the RI value of patients in the organ metastasis group was higher than that of patients in the metastasis-free group [57]. Finally, US-FNA for sampling is a sensible and safe method for diagnosing solid pancreatic tumors [58]. Although this method provides a complete tissue block, it is deficient in the diagnosis for the reason that biopsy samples may not be of the rare tumor because the full spectrum of the entire tumor phenotype is not reflected due to sampling errors caused by the internal heterogeneity of PASC [54].

Treatment of PASC

Current status of PASC treatment

Due to the poor incidence of PASC, clinical studies on this disease have been focused on single-case reports, small sample case analyses, or meta-clinical analyses for years, and large sample studies are lacking to help guide clinical treatment decisions (Table 4).

Table 4.

The therapeutic effect of current clinical treatments of PASC in the existing studies

	Surgery alone	Chemotherapy alone	Radiotherapy alone	Adjuvant therapy	Adjuvant chemotherapy	Adjuvant radio therapy	Adjuvant chemo radiology therapy	Neoadjuvant chemotherapy	Ref.
Numbers	86			48					[10]
Median OS(m)	6.03			22.37
Numbers									[11]
Median OS(m)		5		12			31
Numbers	112	139	7	283				63	[59]
Median OS(m)	7.2	9.2	2.3	19.4				19.6
Numbers	25				30				[40]
Median OS(m)	6				14
Numbers	68				72	5	44		[60]
Median OS(m)	8				12	13	23

At present, there is no exceptional clinical treatment for PASC. Local control and aggressive resection are the first-choice treatments for patients with PASC [61]. Surgical resections for patients with PASC include partial pancreaticoduodenectomy, distal pancreatectomy, and total pancreatectomy etc., among which partial pancreaticoduodenectomy is the most widely used at present [10]. There have been multiple studies showing that surgical resection offers significant survival benefits for PASC. Surgical resection is verified to be the best option for PASC patients who can achieve R0 resection [10, 24]. Moreover, a study from Mayo Clinic showed that PASC patients undergone R1 resection have better survival rates than those patients without surgery [62]. Although both PDAC and PASC can benefit from surgical excision, PASC patients have worse postsurgical median survival times than the PDAC patients [4, 10–13, 63, 64].

While surgery is the most routine treatment, chemotherapy and radiotherapy as well as combination therapy are gradually being applied in pancreatic cancer. Recently study showed that median survival is similar for patients treated with chemotherapy alone (gemcitabine) and surgery alone (9.2 months vs. 7.2 months, P = 0.504), but the overall survival rate is higher for patients treated with surgery compared to chemotherapy alone (two-year survival rate-surgery alone: 23% vs. chemotherapy alone: 10%) [59]. The FOLFIRINOX regimen and the gemcitabine plus albumin-bound paclitaxel regimen have significantly improved the prognosis of metastatic pancreatic cancer patients [11, 65, 66]. Similar with PDAC, adjuvant therapy significantly prolonged the median overall survival of patients with PASC [11, 40, 60]. However, FOLFIRINOX regimen is not popular in Asian countries due to the high incidence of myelosuppression and hepatic impairment in most Asian patients treated with FOLFIRINOX regimen [40]. Actually, Shanghai Ruijin Hospital in China recommended Abraxane + Gemcitabine (AG) or Gemcitabine + S-1 (GS) regimens for the treatment of patients with PASC [40]. What is more, Platinum-based drugs have been shown to improve the outcome of squamous cell carcinomas such as esophageal and ovarian cancers [67]. A recent study showed that the median of survival time（MOS） of PASC patients received platinum-based adjuvant therapy was significantly higher than that of patients received standard adjuvant therapy (19.1 months vs. 10.7 months) [67].

Given that patients receiving radiotherapy alone exhibited an extremely short survival time of 2.3 months, adjuvant radiotherapy is an option in the treatment of patients with PASC [59]. As a study reports that there is no difference in the efficacy of adjuvant chemotherapy alone and adjuvant radiotherapy alone (12 months vs. 13 months) [60]. Actually, the application of adjuvant chemoradiotherapy was more common than that of adjuvant radiotherapy or adjuvant chemotherapy alone in clinic practice [60]. Indeed, adjuvant chemoradiotherapy significantly improved the overall survival time of patients with PASC compared to adjuvant chemotherapy or adjuvant radiotherapy (23 months vs. 12 months vs. 13months) [60]. Sun reported that the postoperative MOS of regional PASC patients undergone adjuvant chemoradiotherapy reached 31 months, superior to 12 months with adjuvant chemotherapy [11]. Surprisingly, patients with PDAC did not benefit from this combined treatment (18 months versus 17 months), which could be explained as squamous carcinomas in PASC are more sensitive to radiotherapy [11] .

In recent years, neoadjuvant chemotherapy has been more and more used in patients with nonmetastatic pancreatic cancer [59, 68–70]. A recent study reported the significantly improved outcome of PASC patients received neoadjuvant therapy compared with chemotherapy or pancreatic resection alone, and it was similar with the outcome of patients received adjuvant chemotherapy [59]. Many PASC patients failed to receive adjuvant therapy after pancreatectomy because of the postoperative complications, the neoadjuvant is therefore very important for these patients to achieve better prognosis. Noteworthy, the patients received neoadjuvant therapy had higher proportion of pathological T0 and T1 tumors, lower proportion of pathological T4 tumors and lower rate of positive margins, compared to those patients undergone pancreatectomy alone. Compared with the patients received neoadjuvant chemotherapy, patients undergone surgery alone had higher rates of unexpected readmissions (6.6% VS 15.7%), 30-day mortality (1.6% VS 9.8%) and 90-day mortality (4.8% VS 23.2%) [59]. A prospective clinical trial of patients with pancreatic cancer receiving neoadjuvant therapy is ongoing. This study includes the patients with PASC and the results may provide more information on the effectiveness of neoadjuvant therapy in PASC [59, 71].

Future directions in PASC treatment

To date, clinical treatments for this rare cancer lack specificity and provide less benefit to patients. In this article, we integrate the molecular mechanisms currently identified and the routine treatment directions of pancreatic cancer, and we propose possible new directions for the development of PASC treatment in terms of interfering with downstream signaling pathways within tumor cells, targeting the immune response, and targeting components of the microenvironment surrounding the tumor (Fig. 3).

Fig. 3.

Microenvironmental characteristics and treatment of PASC. When the normal pancreatic tissue transforms to PASC, the compositions of tumor and microenvironment have some changes and unique molecular features. According to the molecular mechanisms currently identified and the routine treatment directions of pancreatic cancer, we propose possible new directions for the development of PASC treatment in terms of interfering with downstream signaling pathways within tumor cells, targeting the immune response, and targeting components of the microenvironment surrounding the tumor

Targeted Therapies

As research into the pathogenesis of PASC continues, novel potential therapeutic targets, such as EGFR and FGFR, have been identified that may provide an alternative choice for treatment [27]. EGFR is a transmembrane receptor for a tyrosine kinase that is activated upon binding to the EGF and transforming growth factor-α (TGF-α) ligands [72, 73]. EGFR is overexpressed in 90% of pancreatic tumors and exerts an influential effect on the metastasis and recurrence of pancreatic cancer. When combined with gemcitabine in the treatment of pancreatic cancer, numerous targeted agents fail to provide a survival benefit. The only targeted agent to exhibit a statistically significant yet clinically marginal effect on patient survival is the EGFR inhibitor, erlotinib [74]. In a randomized clinical trial of advanced pancreatic cancer, Moore et al. reported that the mean survival of patients treated with the combination of gemcitabine and erlotinib relative to gemcitabine alone improved the mean survival by only approximately 2 weeks [75]. This may be due to the presence of KRAS mutations in 90% of patients with pancreatic cancer, where inhibition of EGFR, an upstream molecule of KRAS, has only a weak therapeutic effect [76]. In PASC, EGFR is persistently activated at the molecular level and EGFR protein is highly expressed at the tissue level [28]. The high expression of EGFR may render EGFR inhibitors useful in PASC, but no relevant clinical reports or basic studies have been reported.

Similarly, FGFR contributes to the development of pancreatic cancer. The binding of FGFR to the FGF ligand induces phosphorylation of the FRS2 substrate protein and subsequent recruitment and activation of downstream pathways, such as Ras/MAPK and PI3K/Akt, by phosphorylated FRS [77]. In pancreatic cancer, the elevated expression of FGFR1 and FGFR2 is linked to an advanced grade of tumors and short-term survival of patients [73]. Sensitivity to treatment with the FGFR inhibitor, infigratinib, has been observed in preclinical models of adenosquamous carcinoma bearing FGFR-ERLIN2 fusion mutations [28].

Targeted immunotherapy

In cancer, local and systemic immune responses constitute major determining elements of tumor resistance and progression [78]. Immune escape and successful progression of cancer cells suppressing immune responses with themselves or other cells are among the 14 hallmarks of cancer [79]. To date, immunotherapy has transformed the outlook of advanced patients with solid tumors, including melanoma and pulmonary cancer, but has yielded little progress in pancreatic cancer. In pancreatic cancer, the formation of a characteristic immunosuppressive microenvironment driven by oncogenic KRAS mutations at the early stages of cancer, the disruption of innate and adaptive antitumor immunity, and the restricted initiation of immune responses by T cells all give pancreatic cancer a vigorous immune resistance, unlike that of other cancers [80]. Therapy aimed at CTLA4 and the PD-1/PD-L1 immune checkpoint as well as cancer vaccines and other immunotherapies have failed to achieve success [74]. In adenosquamous carcinoma, it is noteworthy that the PD-L1 immune checkpoint is expressed only in squamous carcinoma, and immune checkpoint therapy may be effective in patients with this feature of PASC [7, 21, 22]. However, Lee et al. reported no significant difference in survival between the PD-L1-positive and PD-L1-negative groups of patients with PASC, which may be related to the insensitivity of pancreatic cancer to immunotherapy alone [7].

Targeting the peritumor microenvironment

The microenvironment surrounding pancreatic tumors is enriched with the fibrotic matrix (which includes multiple cell types, such as inflammatory cells, blood vessels, and cancer-associated fibroblasts) and extracellular interstitial components (such as collagen, fibronectin, and hyaluronic acid) [81, 82]. As characterizations of the cancer microenvironment, dense and solid fibrous tissue proliferation and extensive immunosuppression play an invaluable role in tumor growth, invasion, and chemotherapy resistance. Hence, there has been an increasing focus on the role played by the tumor microenvironment in therapy.

Currently, therapies targeting angiogenesis in pancreatic cancer have achieved modest efficacy. The long list of targeted compounds tested in trials and found to be ineffective includes vascular endothelial growth factor (VEGF) inhibitors, such as bevacizumab and aflibercept, and multikinase inhibitors with antiangiogenic activity, such as sunitinib, sorafenib, and axitinib [74, 76, 83–87]. It can be speculated that the ineffectiveness of all antiangiogenetic approaches tested to date is attributed to the largely hypovascular nature of the stroma surrounding cancer cells.

Activated fibroblasts produce collagen, fibronectin, laminin, and hyaluronic acid, resulting in the deposition of collagen fibers around tumor cells, making the tumor stiff and anoxic [81, 82]. Pancreatic stellate cells (PSCs), a type of tumor-associated fibroblast, are continuously activated and express high levels of vitamin D receptors during the development of pancreatic cancer. Inhibition of calcitriol, the vitamin D ligand, induces a return of activated pancreatic stellate cells to a quiescent state in vitro, resulting in reduced fibrous deposition and increased uptake of the chemotherapeutic agent, gemcitabine [88]. Another barrier separating the tumor cells from the drug is the deposition of the extracellular matrix. The barrier alters the physical properties of pancreatic tumors, resulting in increased stiffness and hydrostatic pressure [81]. This physical barrier in conjunction with insufficient vascularity around the tumor hinders the absorption of drugs. Consequently, breaking the stromal barrier and allowing higher internal concentrations of drugs is a promising therapeutic direction [74, 89]. In addition to the positive effect of inhibiting pericyte mesenchymal formation, such as collagen deposition, and increasing vascularity, inhibition of pericyte mesenchymal formation also triggers accelerated tumor growth and an elevated rate of metastasis [90, 91]. Future studies on the peritumor microenvironment balance and the bilateral effects of inhibiting mesenchymal production should be performed.

Conclusion

In recent years, studies concerning PASC have gradually expanded from the perspectives of macroscopic pathology and microscopic molecular sequencing. To better provide evidence for the preoperative diagnosis of PASC, scientists have attempted to distinguish PASC from PDAC in terms of clinical presentation, imaging features, and pathological features as well as to identify the origin of squamous tissue, unique molecular markers, and genomic features in PASC. The present article provides a comprehensive review of the literature on PASC and discusses potential therapeutic perspectives in the context of the characteristics of adenosquamous carcinoma itself with the goal of aiding intensive research on this rare cancer to improve patient survival.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

JD and XJ L had the idea for the article, WZ performed the literature search and finished the manuscript and figures; WZ finished the tables; JD, XJ L and JZ made critical revisions and proofread the manuscript. All authors read and approved the final manuscript.

Funding

This study was supported by Program of the Shanghai Key Laboratory of Cell Engineering (14DZ2272300), the Sailing project (22YF1458800) from the Science and Technology Commission of Shanghai Municipality. The study was also supported by National Natural Science Foundation of China (82202912) and Shanghai Natural Science Foundation of China (22ZR1477600).

Data availability

No data was used for the research described in the article.

Declarations

Ethics approval and consent to participate

No ethical approval and consent to participate was required for this study.

Consent for publication

There is no conflict of interest (either financial or personal). All authors and acknowledged contributors have read and approved the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.