Pancreatic cancer (PC) is the seventh leading cause of cancer-related death globally (1). Pancreatic ductal adenocarcinoma (PDAC) accounts for >90% of PC cases and is the most typical type of PC. Despite advances in treatment, PDAC has a low survival rate, as the 5-year overall survival rate is <10% (2). Sarcomatoid (spindle cell) carcinoma (SC) is an aggressive form of carcinoma composed of malignant spindle cells, with or without a coexisting epithelial cell component. Though it can occur in all organs of the body, it mainly affects the respiratory tract, lungs, breasts and kidneys, and in extremely rare cases, the pancreas (3). Undifferentiated SC of the pancreas (SCP) is an aggressive malignant neoplasm originating in the pancreas with a poor prognosis. Following the World Health Organization (WHO) classification (Fifth edition, 2019), SCP represents a histologically undifferentiated PDAC subtype, accounting for 2–3% of all PDACs and its variants (4–6); however, the histogenesis of this carcinoma type remains debatable. Currently, the clinicopathological features, molecular landscape and therapeutic strategies for SCP are poorly understood due to its low incidence. The present review aimed to describe the histogenesis, diagnosis, genetic characteristics, prognosis and treatment of SCP, specifically focusing on the molecular alterations to elucidate potential targets for precision therapy. The eligible cases of SCP from the Affiliated Lihuili Hospital of Ningbo University (Ningbo, China) were retrospectively collated to summarize the single-center experience and a literature review was performed by searching PubMed (https://pubmed.ncbi.nlm.nih.gov) and the China Science Periodical Database (wanfangdata.com.cn) from January 1, 1990 up to August 31, 2023 with a combination of the following keywords: ‘sarcomatoid carcinoma’; ‘undifferentiated carcinoma’; and ‘pancreatic cancer’. Studies that reported an explicit histopathological diagnosis of SCP with follow-up data were considered eligible for inclusion in the present review. The present review excluded certain articles where terminologies such as ‘carcinosarcoma’, ‘sarcoma-like’ or ‘carcinosarcomatous histology’ were used, and collected and analyzed data from 38 patients with SCP (Table SI) (7–28).

Histologically, SCP predominantly comprises mesenchymatoid spindle-shaped tumor cells originating from pancreatic ducts and acinus, but it does not exhibit glandular differentiation. The tumor displays a distinct biphasic component of carcinoma and sarcoma. Carcinosarcoma of the pancreas (CSP) also originates in the pancreas and has similar biphasic features (29). Previously published studies often use the terms ‘carcinosarcoma’ and ‘sarcomatoid carcinoma’ interchangeably and the definitions of these terms vary in the literature (30,31). According to the fifth edition of the WHO classification of exocrine pancreatic tumors from 2019, SCP and CSP are classified as undifferentiated carcinoma of the pancreas (UCP) (32). UCP, a subtype of PDAC, represents a group of rare tumors that account for ~5% of PC (30). The primary difference between UCP and PDAC is that UCP is a hypercellular tumor with minimal stroma and scant desmoplastic reaction, whereas conventional PDAC has a considerable amount of desmoplastic stroma with few neoplastic cells/glands. Based on the aforementioned WHO classification of exocrine pancreatic tumors, SCP consists of spindle-shaped cells that may contain allogenic components, such as bone and cartilage. The microscopic description is a critical indicator for differentiating SCP from CSP. SCP is defined as a neoplasm composed of >80% atypical spindle cells, with or without heterogenic differentiation. Pathologically, CSP is defined as a UCP subtype composed of a combination of round epithelioid cells and spindle sarcoma cells, with each component constituting ~30% of the tumor (Table I) (32). In addition, there is a transitional zone between the epithelioid and sarcomatoid cells in SCs, whilst in carcinosarcomas, these two portions are separated without such a transitional zone (Table I) (16,31). The sarcomatous tissues of the SCs exhibit biphasic expression of mesenchymal markers and epithelial markers, and ultrastructures of epithelial cells (11,16). Sarcomatoid components in carcinosarcomas do not have this feature; they express only mesenchymal markers and are negative for epithelial-derived markers (Table I) (13,32). However, the pathogenesis of SC remains unclear. Researchers have presented the following hypotheses to explain the phenomenon that a particular tumor exhibits both epithelial and sarcomatous traits biphasically: i) Conversion: The sarcomatous components are transformed from cancerous components by metaplasia; ii) collision: The sarcoma and carcinoma grow independently adjacent to each other; and iii) combination: Bidirectional differentiation of primitive totipotent stem cells into epithelium and sarcoma tissue (33–36). The genetic alterations in sarcoma and epithelial components have been reported to be nearly identical in pancreatic mucinous cystic neoplasms and sarcomatous stroma, which is consistent with the ‘combination theory’ that the two components of the neoplasms have the same clonal origin and subsequently differentiate into the epithelial and sarcomatous components of the carcinosarcomas (37). Since the emergence of the epithelial-mesenchymal transition (EMT) theory in cancerous tissues in the 1990s, EMT has been reported by certain researchers to explain the histogenesis mechanism of carcinosarcoma. They have pointed out that the complete transformation of epithelial cells into mesenchymal cells is a continuous process driven by the EMT program, generating cells that exhibit a series of intermediate phenotypic states. This process is regulated by contextual signals and the intracellular gene circuitry of the cells. Therefore, cells in intermediate states may exhibit the characteristics of mesenchymal cells but can retain certain epithelial markers (38). In certain instances, transforming growth factor-β (TGF-β) can act as an oncogene to promote the proliferation of normal epithelial cells. Therefore, TGF-β is considered to induce EMT in pancreatic cells and to promote the formation of SCP (Fig. 1) (39). In other tumor types, TGF-β is also a powerful tumor suppressor inhibiting the multiplication of pre-malignant cells by triggering apoptosis. This dual effect of TGF-β is mainly mediated by the Smad pathway (40). Furthermore, the pathological type, cellular context and specific environment determine the tumor responsiveness to TGF-β (41–43). Ren et al (10) reported that the plasma interleukin (IL)-11 and TGF-β levels were notably higher in patients with SCP compared with those in healthy controls and patients with PDAC. IL-11 is a TGF-β target gene and TGF-β induces IL-11 production in several cell types. The Smad tumor suppressor pathway mediates the expression of IL-11 and connective tissue growth factor via TGF-β (44). Furthermore, TGF-β may be a critical driver of sarcomatoid transdifferentiation in renal clear cell carcinoma (45). Kimura et al (26,27) assessed the expression of fibronectin, Snail and phosphorylated (p)Smad2/3 in the sarcomatoid tissues of 3 patients with SCP. Fibronectin is an extracellular marker of spindle metaplasia during EMT (46). Snail is a zinc-finger transcription factor that represses the transcription of E-cadherin, which is involved in the regulation of EMT during embryonic development. pSmad2 and pSmad3 are regarded as critical intracellular transduction molecules that transmit TGF-β signals from the cell surface into the nucleus (47,48). The aforementioned studies reported that sarcomatoid components may be converted from cancerous components via EMT mediated by TGF-β. However, further studies are required to elucidate the molecular mechanisms underlying the processes of cellular differentiation leading to SCP.

According to the aforementioned WHO guidelines, the diagnosis of SCP is highly dependent on pathology and immunohistochemistry (IHC); however, it is challenging to distinguish SCP from other pancreatic tumors preoperatively. Potential preoperative diagnosis includes PDAC, mucinous cystic neoplasm, pseudocyst and solid pseudopapillary tumor (18). Several studies have also reported certain specific imaging features of SCP, such as large irregular cystic-solid masses, which were prone to invading adjacent organs (9,12,14,17,26). Furthermore, biopsy is crucial for the preoperative diagnosis of SCP and fine needle aspiration guided by endoscopic ultrasound has relatively high sensitivity and specificity for the diagnosis of solid pancreatic masses. The final diagnosis requires histological examination and immunohistochemical analysis. Due to the poor prognosis and aggressive clinical behavior, the early diagnosis of SCP is critical (15,20,29).

At present, there are numerous reports of the relationship between PDAC carcinogenesis and several gene alterations, including mutations in genes such as KRAS, TP53, SMAD4 and CDKN2A (29,53). Due to the rarity of SCP, only a few individual cases have been reported (20,54,55), without any large prospective studies, and its genetic characterization has not been fully described. As a subtype of PDCA, SCP has a genetical landscape mimicking PDAC (29). Recently, Ding et al (54) assessed the clinicopathological and genetic characterization of 71 patients with USC, which included five USCs of the pancreas. This study reported the presence of highly frequent TP53, RB1, TER0054 and KRAS alterations. In particular, mutations in TP53 and KRAS were identified in all cases and the KRAS mutation was reported to be associated with a poor prognosis (54). Furthermore, a next-generation sequencing (NGS) analysis of 10 SCP samples revealed that SCP was genetically similar to PDAC. It was also reported that 100% of these samples exhibited KRAS mutations, 90% exhibited TP53 mutations and 60% exhibited CDKN2A mutations. However, SCP also exhibited several critical genetic characteristics that were distinct from PDAC. SMAD4, a tumor suppressor gene that is altered in 50–60% of conventional PDAC cases (56,57), was mutated in only 1 SCP case (10%). This type of mutation may indicate early metastasis of the tumor (56). Furthermore, tumor cells in PDAC rarely revealed recurrent KRAS amplification, whilst it was present in 3/10 cases. Similarly, rhabdoid carcinoma of the pancreas (RCP) is a subtype of UCP. A cohort study on RCP revealed this amplification in 5/13 (38%) patients (58). Moreover, in at least a subset of patients with PDAC, KRAS amplification may be a genetic driver for the acquisition of undifferentiated morphology. Finally, two potential molecular therapeutic targets such as the POLQ mutation and MCL1 amplification that did not belong to the typical PDAC molecular landscape were detected in 2 cases (55).

PDAC is occasionally associated with germline BRCA mutations. The BRCA genes, including BRCA1 and BRCA2, encode proteins involved in repairing the broken DNA double strands via the homologous recombination pathway (20,59). A previous large prospective study reported that patients with pathogenic BRCA1/2 variants may benefit from poly-ADP ribose polymerase (PARP) inhibitor treatment, an emerging therapy targeting the genes involved in DNA maintenance (59). A study reported mutations in TP53 and KRAS, as well as BRCA2 (20). The molecular profiles of the SCP cases we collected are summarized in Table II. Overall, these genomic profiling results indicate encouraging outcomes for precise targeted therapy in SCP.

At present, validated predictive biomarkers for immunotherapy include programmed death ligand 1 (PD-L1), as well as microsatellite instability and tumor mutation load. As presented in Table II, all previously mentioned cases exhibited low levels of tumor mutation burden with microsatellite-stable states, indicating poor outcomes of immunotherapy. Notably, Lehrke et al (60) reported that patients with UCP with osteoclast-like giant cells (OCGCs) expressed PD-L1 significantly more frequently than patients with PDAC (63 vs. 15%; P<0.01). Another study retrospectively evaluated PD-L1 and Notch expression in 6 cases of SCP (61). The combined positive score (CPS) is an index that can be used to evaluate PD-L1 expression in tumor's and is obtained by dividing the number of PD-L1-stained cells, namely tumor cells, macrophages and lymphocytes, by the total number of viable tumor cells multiplied by 100 (61). A CPS ≥1 was typical in 5 cases of SCP (83%) and 3 of the subjects (50%) had a CPS ≥50. This finding indicates an improved effectiveness in SCP compared with conventional PDAC. However, high expression levels of Notch1 and Notch3 were also reported in all cases. Further immunofluorescence analysis revealed that, when the expression levels of PD-L1 and Notch3 were upregulated within the cytoplasmic or membranous compartments of the sarcomatoid cells, both proteins were co-localized in the same cells, providing a rationale for future research in anticipation of evaluating the potential crosstalk between the PD-L1/programmed cell death protein 1 axis and the Notch3 pathway (61). Therefore, further studies on the significance of immunotherapy are required.

The research into molecular classifications and genetic signatures has long spurred the development of novel therapeutic strategies, enabling medical practitioners to make accurate and personalized decisions. Table III summarizes gene alterations in SCP compared with in PDAC. Nevertheless, due to the low incidence of SCP, there are limited genomic profiling data available. Thus, further studies based on larger cohorts of patients with SCP are warranted to explore the genetic features of SCP.

Due to its rarity, the surgical protocols, postoperative adjuvant treatments and overall prognosis of SCP are insufficiently described in the literature. Notably, no direct comparisons between SCP and PDAC have been made; however, patients with SCP tend to have worse survival rates than those with PDAC (11,28). Generally, most patients present with an advanced, unresectable state of the disease, with only ~12% of the patients surviving >5 years (62). For patients presenting with resectable disease (10–15%), surgery followed by adjuvant chemotherapy is the standard therapeutic approach, with an anticipated median overall survival of 54.4 months (53).

The poor prognosis of SCP makes identifying effective treatments a top priority. Notably, total R0 surgical extirpation is the sole opportunity for a radical cure (28,63) and it has been reported that patients who did not undergo complete R0 tumor extirpation had an early recurrence, leading to mortality in <3 months. Surgical procedures mainly include pancreaticoduodenectomy and distal pancreatectomy. Occasionally it is necessary to remove the surrounding invaded organs to ensure the complete removal of the tumor. In addition to surgery, patients can benefit from postoperative adjuvant treatments, especially chemotherapy (29). A study analyzing the prognoses of 261 patients with UCP indicated that surgery was the first choice for resectable UCP and that adjuvant therapies needed to be introduced immediately (64). Generally, patients with UCP were administered the same regimens as those with more common PDACs. Albumin paclitaxel and gemcitabine, and fluorouracil, irinotecan, leucovorin and oxaliplatin, the first-line chemotherapy regimens for PDAC, are also the preferred choices of adjuvant therapy for SCP (53,65). Furthermore, gemcitabine has been reported to be effective in patients with tumor recurrence or portal vein thrombosis (65). A multicenter cohort study retrospectively analyzed the outcomes of 50 patients with unresectable UCP and assessed the efficacy of several chemotherapies. It was reported that the median overall survival of these patients was 4.08 months and a paclitaxel-containing regimen was associated with a relatively longer survival (65). Gkountakos et al (29) also reported that complete surgical resection followed by PDAC-standardized adjuvant chemotherapy was the only tangible possibility for long-term survival in patients with SCP. Another retrospective study reported 8 patients with SCP in a single center. 2/8 cases underwent R0 resection and received adjuvant therapy with the tumors located in the body/tail of the pancreas, surviving >5 years. Furthermore, one of the aforementioned cases had a survival of ~16 years in spite of lymph node metastasis, representing the longest survival time of patients with SCP in the literature, to the best of our knowledge (28). Additionally, immune checkpoint inhibitors are increasingly being administered in several types of cancer; however, PDAC has shown a limited response to immunotherapy compared with other tumor types. It has been reported that PD-L1 expression is more frequent in SCP and UCP. Therefore, immunotherapy has become a promising treatment option (29,59); however, its effectiveness in SCP needs to be confirmed in large prospective studies.

Molecular alterations may serve as targets for precise therapy. These abnormal genetic events can be detected by NGS and can be used to find approaches to selectively kill tumor cells (66). SCP is genetically similar to PDAC (29,55). In general, the main stages of tumorigenesis include oncogene activation and tumor suppressor inactivation. Notably, numerous researchers are working to develop strategies to target oncogenes such as KRAS; however, no KRAS inhibitor has reached the clinical application stage at present (67). Advances have been made in clinical and preclinical trials of treatments targeting TP53, CDKN2A and SMAD4, the three major tumor suppressors of PDAC (53), and further studies are warranted to assess whether the reactivators clinically improve the prognosis of patients. Moreover, the genes involved in chromatin stabilization and remodeling, such as BRCA and KDM6A, have been reported to be deficient in patients with PDAC and SCP (66). It is encouraging that administering PARP inhibitors to block base-excision repair leaves both double- and single-stranded DNA breaks unrepaired, leading to death of the cells with BRCA dysfunction (68). Furthermore, a phase III trial reported that, among patients with germline BRCA mutations and metastatic PC, the progression-free survival was longer in patients with maintenance olaparib administration than in those with placebo administration (69). Therefore, precise therapy based on molecular alterations is a promising approach.

In PDAC, several histopathological factors have been reported as prognostic factors, including tumor grade, R0 resection margin, lymph node status and adjuvant therapy (28,70). In SCP, a comparison analysis of these factors is not adequate. Notably, evidence has suggested a possible association between cellular senescence induced by TGF-β and long-term survival could be interpreted as a promising finding. The study reported positive staining for fibronectin, Snail and pSmad2/3 at the IHC level in the tumor cells of 3 patients with SCP. γ-H2AX, p53 and p21, typically used as markers of cellular senescence, were observed in the sarcomatoid component of a case with long-term survival but not in the others. Consistent with this finding, the Ki-67 labelling index of the long-term survivor was the lowest compared with that of the other 2 patients (26,27). The Ki-67 labelling index is a strong prognostic factor in pancreatic neuroendocrine tumor (71); however, its clinical significance in PDAC has not been thoroughly evaluated. Therefore, TGF-β-mediated senescence and a low Ki-67 labelling index may be critical in reducing the proliferation and metastasis of sarcomatous cells. Furthermore, OCGCs, the multinucleated giant cells with abundant cytoplasm resembling giant cell tumors of the bone, have previously been reported to protect against anaplastic carcinoma, with long-term survival reported in ~50% of patients in a previous study (30,72).

We included cases with histological diagnosis of PDAC and excluded cases with a previous history of malignant tumors. Between August 2015 and August 2023, 603 cases of PDAC, including 7 cases of SCP, were pathologically confirmed at the Affiliated Lihuili Hospital of Ningbo University (Ningbo, China) and the prevalence of SCP in all PDACs was 1.16%, which is lower than that previously reported in the literature (1,2). Of the 7 cases (Table SI), 3 exhibited abdominal pain, 3 exhibited jaundice and 1 was asymptomatic. Serum bilirubin, mainly direct bilirubin, alanine aminotransferase and aspartate aminotransferase increased in all 3 cases with jaundice. CA 19-9 was increased in 5 cases, and CA 12-5 was increased in 3 cases (Table IV). All the patients underwent contrast-enhanced CT. CT revealed that the pancreatic mass was cystic-solid or cystic, with inhomogeneous or mild enhancement, and the boundary was mostly unclear (data not shown). Fig. 2 presents the imaging features of a typical case.

As indicated in Table SI, 6 patients underwent radical surgery and histological examination (Fig. 3), and all had lymphovascular and perineural invasion. The tumor invaded adjacent organs (duodenum, n=2; stomach, n=1; and colon, n=1) in 4 patients (57.1%). A total of 2 patients (28.6%) had no lymph node metastases, whereas the remaining patients had ≥1 positive lymph node metastasis. All samples examined by immunohistochemical staining were positive for vimentin and CK (Fig. 4). The protocol was as follows: Samples of all cases were fixed with 3.7% neutral formaldehyde solution, dehydrated, embedded in paraffin and slices were subjected to H&E staining according to routine procedures to prepare slides observed under a light microscope. All paraffin blocks containing tumors were stained with 34BE12+P540s double labelling. Immunohistochemical staining for certain markers in the cases was performed at the time of diagnosis. Immunohistochemical analysis was performed at the clinical laboratory of our institution using the Roche BenchMark automated system (Roche Diagostics) with appropriate controls. The following primary antibodies were applied: CK7 [cat. no. ZM-0071; 1:200 dilution; Zhongshang Goldenbridge (ZSGB)-Bio], CK19 (cat. no. l1006; pre-diluted antibody; Biolynx), CK20 (cat. no. ZA-0574; 1:100 dilution; ZSGB-Bio), CK8/18 (cat. no. ZM-0315; pre-diluted antibody; ZSGB-Bio), MUC1 (cat. no. CMM-0251; pre-diluted antibody; Celnovte), smooth muscle actin (cat. no. CAM-0191; pre-diluted antibody; Celnovte), human melanoma black 45 (cat. no. ZM-0187; 1:1 dilution; ZSGB-Bio), vimentin (cat. no. ZM-0260; 1:200 dilution; ZSGB-Bio) and Ki-67 (cat. no. ZM-0166; 1:800 dilution; ZSGB-Bio). All antibodies were incubated for 30 min at room temperature. Next, a conjugated secondary antibody was added (cat. no. DS0003; pre-diluted antibody; ZSGB-Bio) and incubated at room temperature for 30 min. Subsequently, visualization was performed by applying 0.1% 3,3′-di-aminobenzidine tetrahydrochloride solution for 5 min at room temperature. The sections were finally counterstained with Mayer's hematoxylin for 1 min at room temperature, dehydrated and mounted with coverslips after being embedded in mounting medium. The slides were stored at room temperature. The sections were viewed under a light microscope by two independent pathologists blinded to the patients' clinical data. Immunoreactivity was evaluated in a semiquantitative manner to assess the percentage of immunopositive tumor cells: Negative (−), 0%; focal, <25%; moderate, 25–75%; and diffuse, >75%.

In all 3 surviving patients, the tumor was located in the distal pancreas rather than in the pancreatic head, without distant metastases. Among them, 1 patient underwent a gene test. The NGS revealed TP53 and KRAS mutations, and a pathogenic variant of the germline BRCA1 gene. Therefore, the patient received a PARP inhibitor because of their poor tolerance to chemotherapy; to the best of our knowledge, this is the first report of this drug administration to a patient with SCP in the literature. It was encouraging that a good result was obtained after administering olaparib to the patient with a germline BRCA mutation and this prompted continuation of the genetic testing of patients with rare tumors.

Accordingly, as presented in Table SI, the 3-month and 1-year mortality rates of the patients with SCP exceeded 23.3 and 46.7%, respectively, despite aggressive surgical management, with many succumbing to early metastasis. Subsequently, Kaplan-Meier curves of the survival outcomes of all patients with SCP were plotted (Fig. 5), and the median overall survival time was 10 months. Despite the small sample size and incomplete follow-up data, the data indicates that SCP is associated with a worse prognosis.

SCP is a rare subtype of PDAC and is generally considered to be an aggressive neoplasm with a poor prognosis. Nevertheless, the low incidence and the incomplete understanding of its clinical course hinder the possibility of performing large-scale studies on patients with SCP. At present, the treatment strategy for SCP is empirical therapy based on medical research of PDAC. Similarly, surgical resection followed by PDAC-standardized adjuvant chemotherapy is the most likely treatment option for achieving long-term survival. A considerable portion of patients with SCP may benefit from emerging immunotherapy-based strategies in the near future. Notably, patients with SCP frequently exhibit TP53 and KRAS mutations, highlighting the hereditary homogeneities with PDAC; however, there are also certain crucial distinctions. Particularly, certain molecular alterations in SCP, including BRCA mutation, MCL1 amplification and POLQ mutation, uncover more genetic features and provide novel therapeutic targets. For example, PARP inhibitors aim to selectively kill carcinoma cells with BRCA mutation. Lately, several clinical trials have confirmed the partial efficacy of olaparib, prompting further investigation to achieve synthetic lethality in PC. For this reason, there is an urgent need for genomic and transcriptomic studies based on larger cohorts of patients with SCP to explore its molecular profile in greater depth and to identify its histogenesis.