An Update on Clinicopathological and Molecular Features of Plexiform Fibromyxoma

Plexiform fibromyxoma is a rare and newly described gastric mesenchymal tumor with only 121 reported cases in the literature. Our understanding of plexiform fibromyxoma requires updating since the first case has been reported by Takahashi et al. 12 years ago. The present review summarized reported cases in the literature, and both clinical and pathological aspects of plexiform fibromyxoma were comprehensively discussed. Plexiform fibromyxoma usually causes nonspecific or bleeding signs or symptoms, and therefore clinical recognition of the disease is challenging. Plexiform fibromyxoma is of benign nature without any metastasis or recurrence reported, and more conservative surgical treatment should be considered.


Introduction
Plexiform fibromyxoma (PF), also known as plexiform angiomyxoid myofibroblastic tumor (PAMT), was first described in 2007 by Takahashi et al. This rare tumor of mesenchymal origin is typically seen in the stomach and shows clinically benign behavior. As implied by the name, it features a plexiform cellular architecture of a myofibroblastic nature, encircled by a myxoid intercellular matrix with rich vasculature [1].
To date, only 121 cases have been reported in the literature with a designation of PF or PAMT; the clinical features of reported cases are summarized in Table 1. Specious case studies were published before the Takahashi's first report in 2007, terming the condition "gastric fibromyxoma," "gastric myxoma," or "fibromyxoangioma" [2][3][4][5][6][7][8], but these are not included in the present review because the diagnoses could not be ascertained by immunohistochemical staining despite the similarity of the clinical features. Many cases have been reported since Takahashi first defined the entity 12 years ago; however, as new cases emerge, many facts about PF require updating, including the terminology, epidemiological data, various clinical presentations, diagnosis, diverse strategy on treatment, and prognosis. Correct updated information will help in the clinical recognition of the disease and improve the outcomes of treatments. The aim of this review is to provide a comprehensive updating of PF including the published cases to summarize what we know about PF, to identify what we still need to investigate, and to achieve consensus on all aspects of this disease. Hence, we present a review of the published cases in the literature and discuss clinically significant issues about PF. Takahashi et al., in 2007, reported 2 cases of gastric plexiform fibromyxoma using the term "plexiform angiomyxoid myofibroblastic tumor (PAMT)" because the condition was morphologically distinct from other gastrointestinal mesenchymal tumors due to its bland spindle cells in a plexiform pattern, myxoid intercellular matrix with hypervascularity, and myofibroblastic nature [1]. In 2008, Yoshida et al. reported 2 additional cases and modified the designation to "plexiform angiomyxoid tumor," deleting the 2 Canadian Journal of Gastroenterology and Hepatology  Canadian Journal of Gastroenterology and Hepatology Canadian Journal of Gastroenterology and Hepatology term "myofibroblastic" due to the evident differentiation into smooth muscle cells that had been absent from the previous report and emphasizing the differentiating potential of myofibroblasts into smooth muscle cells. Yoshida et al. also further characterized the spectrum of fibrous, fibromyxoid, and myxoid stromal patterns of the disease [9]. Takahashi et al. continued to use the designation "myofibroblastic," because this presentation was seen in the majority of cases [10].

History and Nomenclature
In 2009, Miettinen et al. described 12 cases of the disease and designated the tumors as "plexiform fibromyxoma" simply due to their cellular architecture and fibromyxoid nature [11]. They also identified previously reported diseases from 1959 to 1986 that shared similar characteristics with PF by the names "gastric fibromyxoma" or "gastric myxoma" [2,5,6,8,12]. In 2010, Takahashi et al. argued that the designation "plexiform fibromyxoma" could lead to confusion, since "gastric fibromyxoma" or "gastric myxoma" describes a relatively narrow entity of purely fibroblastic tumors that present a different immunoprofile from that of PAMT; consequently, those cases reported as "gastric fibromyxoma" or "gastric myxoma" might actually differ from PAMT and would require further pathological or immunological evidence [10]. However, the WHO classification of tumors of the digestive system later adopted the term "plexiform fibromyxoma" in 2010 to designate the entity [13]. Since then, both "plexiform fibromyxoma" and "plexiform angiomyxoid myofibroblastic tumor" have both been used as the nomenclature of the disease, with controversy in the literature. Despite the nomenclature set by the WHO classification, many authors still preferred the term PAMT as a better description of both the histogenesis and histology of the tumor [14][15][16]. In the present review, 85 cases were designated as PF, while 36 were designated as PAMT.
Sing et al. proposed that PF and PAMT are two related, but different, entities at two respective fibroblastic and myofibroblastic ends of a spectrum [20]. Yet, due to the generally close similarity of the two, Duckworth et al. considered the two to be a single entity with an acceptable range of histologic, immunohistochemical, and ultrastructural presentations [34]. By contrast, Sing et al. argued that PF occupies the "fibroblastic" end while PAMT occupies the fully differentiated "myofibroblastic" end of the spectrum, and that the size, female exclusiveness of PF, vascular invasion, and extragastric extension are distinguishable between the two, despite the similar location of occurrence and age of onset [20]. Immunohistochemically, Sing et al. suggested that desmin and caldesmon could be used to distinguish PAMT as it shows focally positive results, whereas PF shows negative results. However, on the grounds of the limited case numbers for PF and PAMT and that desmin and caldesmon were not specific for the myofibroblastic form only, we suggest that discrimination of the two designations is unnecessary. In the literature, the cases designated as PF and those as PAMT showed no significant difference in age of onset or mean tumor size and both had female predominance and extragastric involvement. Although vascular invasion had only been reported in PF cases, the number of cases with vascular invasion is very small.
Since the WHO classification had designated the nomenclature PF for this entity, to avoid confusion in the literature, we suggest the use of PF in consideration of epidemiological surveillance and scientific communication purposes. The term PAMT literally describes the features of the disease in more detail, but PF could be viewed sensu lato as a broader nomenclature that covers the disease as well as other variants. We consider this to be acceptable, given that the disease is a spectrum with variations. The final designation or subsets under PF could be further categorized in the future after consideration of sufficient case numbers and advanced investigations.

Epidemiology
According to the 110 cases reported in the literature from 2007 to 2018, the frequency of PF is more than 9.17 cases per year worldwide. However, we feel this is an underestimate, because clinical recognition of the disease has only been increasing since 2007, and some cases are assumed to have been misidentified as gastrointestinal stromal tumors (GISTs) or other entities. Miettinen et al. estimated that the frequency of PF is 150 times lower than gastric GIST [11]; however, about 3,000 cases of gastric GIST are diagnosed yearly in the United States [79]. Therefore, we consider a proportion of 1:150 to be a rather conservative estimation and should be far smaller.
The patient ages showed a broad range, from 5 to 81 years (mean age 43.17±18.00 years; median age 46 years). Most patients were middle-aged, with a peak around 30-60 years old. This age distribution of PF was in approximate accordance with previous reviews [10,51]; however, the adult-tochild ratio was 8:1 by our estimation, unlike the 5:1 ratio stated by Morris's [51] but more similar to the 7:1 ratio proposed by Fukuzawa's in a recent systematic review [75]. The previous studies all reported a 1:1 male-female ratio for PF [10,51,59], although we found a slight female predominance, with male patients accounting for 43% and female patients for 57%.
The races of patients were not provided in most of the studies; consequently, the countries of the studies were substituted for the ethnicity of the patients unless the ethnic information of the patients was given. In terms of regions, most cases were reported from East Asia (N=58; 47.9%), followed by North America (N=29; 24.0%), Europe (N=26; 21.5%), South-East Asia (N=4; 3.3%), South Asia (N=3; 2.5%), and Africa (N=1, 0.8%). In terms of countries, most cases were reported from China (N=41; 33.9%), followed by the United States (N=29; 24.0%), Belgium (N=17; 14.0%), Japan (N=10; 8.3%), and Korea (N=6; 5.0%). This distribution may not reflect the genuine epidemiological status of PF, since a higher frequency may result from a higher quality of healthcare and from a larger population. Nevertheless, it still suggests that PF is a worldwide disease with a predominance in East Asia.

Clinical Presentation
PF has a benign nature but is associated with hypervascularity; therefore, the clinical presentation can range from  [11,53,74] or hemodynamic instability [58] was also reported. Reflux symptoms might be present, with or without other complaints, and are likely to mask PF if presenting alone or with nonspecific complaints [22,26,68,71,80]. Among the cases that presented with reflux symptoms, only one tumor was located at the gastric fundus and might have been the cause of the symptoms [22]; other tumors were located at the gastric pylorus or antrum [26,53,68,71,80]. The likelihood that PF could directly or indirectly provoke reflux symptoms is doubtful, and the reflux symptoms were more likely to be concurrent with PF because of the high GERD prevalence worldwide. Ten cases were diagnosed incidentally, even with ulcerative PF lesions [25,33,59]; the size of incidentally diagnosed tumor ranged from 0.8 to 4.5 cm, and 4 tumors were located at gastric body, 1 was located at gastric fundus, and 5 were located at gastric antrum or pylorus [1,16,25,27,33,37,56,59,61]. Two cases with hemorrhagic perforation were reported by Takahashi et al. and Tan et al., with maximal tumor diameters of 4 and 3.5 cm, respectively [1,21]. Other signs or symptoms at presentation included amenorrhea with cushingoid appearance in a 35-year-old female, but her symptoms were actually caused by polycystic ovary syndrome [20]. Chest pain with shortness of breath and finger numbness in a 16-year-old female was probably a result of a mass effect of the 3.2 cm PF in the mediastinum [34], while cholelithiasis was reported in a 55-year-old female with a PF in the gallbladder [42]. Fever was reported in a 42-yearold female with a fistulating abscess formation connecting the tumor and gastric lumen, suggesting possible infection sequelae of PF if left untreated [36]. The pathogenic association between these signs or symptoms and PF could not be proven; however, physicians are reminded of the nonspecific presentation of the disease. If a gastric neoplasm is clinically suspected, further endoscopic diagnostic intervention is indicated.

Endoscopic Findings/Macroscopic Pathological Findings
The size of the tumors, available in 98 cases, ranged from 0.8 to 17 cm in the maximal diameter, with an average size of 4.81±3.30 cm and a median size of 4.0 cm. Endoscopic visualization reveals that PFs are typically pink or reddish and glistening tumors, elastic in texture, and covered with ulcerative, erosive, or smooth mucosa. Endoscopic ultrasonography indicates that PFs are hypoechoic with mild heterogenicity. Macroscopic examination shows a classical PF appearance as a lobulated tan-white or grayish-whitish mass, gelatinous on the cut surface, cystic, with mucinous fluids, a multinodular or polypoidal growth pattern, unencapsulated, and with well-defined (but sometimes ill-defined) margins.
Hemorrhage is commonly encountered. PFs mostly originate from the submucosa and muscularis propria, with extension ranging from the mucosa to the serosa, causing ulcer and/or perforation. The condition of the tumor surface was reported in 76 cases: 50 (65.8%) were ulcerated, 22 (28.9%) were nonulcerated, and 4 (5.3%) were covered with eroded mucosa. Ulceration or erosion of the tumor was significantly associated with hemorrhage-related signs or symptoms, as determined by Pearson's chi-squared test (p<0.0001). The difference in tumor size between ulcerative or erosive lesions and nonulcerative lesions was not statistically significant, as calculated by an independent Student's t-test (p=0.184).

Microscopic Findings
The signature of PF, as disclosed in the name PAMT, is the presence of bland ovoid to spindle cells arranged in irregular plexiform or multinodular pattern and separated by abundant myxoid and a variably collagenized extracellular background, interwoven with rich, arborizing, capillarysized vasculature. The myxoid matrix is consistently Alcian blue positive. The tumor cells demonstrate monomorphous oval nuclei containing indistinct nucleoli and fine chromatin, surrounded by mildly eosinophilic cytoplasm with indistinct borders. Delicate and indistinct nucleoli and fine chromatin may be present. Cellular atypia and mitosis are both rare and are not seen in the majority of the cases. Microscopically, the tumor margin is infiltrative and unencapsulated; in 33 cases reporting the condition tumor margin, 20 were illdefined and the rest were well-circumscribed. Necrosis was reported in the literature in only 2 cases: a 42-year-old female with fistulating abscess showed central necrosis as well as gas-fluids level [36], and a 31-year-old female with an ulcerative lesion exhibited only surface necrotic tissue coverage without central necrosis [74]. Some lymph nodes display enlargement with reactive changes [19,67]. Vascular or lymphatic involvement was observed in 5 cases [11,60].
Significantly or diffusely reactive immunostainings included vimentin (100%), SMA (89.1%), and MSA (90%); however, these are nonspecific markers for mesenchymal and myofibroblastic lineages and were therefore sensitive but specific for diagnosing PF. Equivocal staining results were demonstrated for desmin, caldesmon, calponin, CD10, estrogen receptor (ER), and progesterone receptor (PR). Desmin and caldesmon are more specific markers for muscular lineage toward terminal muscle cell differentiation and exhibited limited and focal reactive results in PF, consistent with the proposed myofibroblastic spectrum of PF cell development.
Calponin is a nonspecific muscular marker for differentiated smooth muscle cells, while CD10 indicates cells with fibroblastic traits. The calponin and CD10 results confirmed the variably myofibroblastic nature of PF in the muscular and fibroblastic axes, respectively. The ER showed all negative staining, but the PR were diffusely or focally reactive in 8 of 10 cases [20,31,59], one of which revealed a prominent PR positivity in 80% of the tumor cells [20]. The PR positivity implied that PF may be sensitive to hormonal therapy [20,59], but this might not be clinically practical. PF is reminiscent of extrauterine or metastatic endometrial stromal sarcoma, which is rare but most commonly occurs in the GI tract [20]. However, it typically presents ER positivity [81], while PF consistently shows negative ER immunostaining result.
The rest of the markers mainly had negative results in PF. DOG-1 and CD117 were always negative in the reported cases, which allowed PF to be well-distinguished from GIST [82]. CD34 was mostly negative (95.1%), but focally positive results have been shown in several cases. CD34 could be labeled in cells with fibroblastic nature [83], but CD34 shows strong positive staining in vascular endothelial tissues; therefore, we suggest the possibility that focally positive or equivocal results originate from the rich vascularity of PF or as a result of technical or interpretation errors. The PF cases showed almost entirely negative results according to S-100 protein staining, indicating that PF is not derived from the neural crest; the 2 cases reporting positive results we highly suspect to be biased by technical error. Except for 1 case with a positive EMA staining result and 1 case with focally positive cytokeratin AE1/AE3 staining result, negative outcomes were reported for EMA, neurofilament, cytokeratin, -catenin, ALK, cytokeratin AE1/AE3, and synaptophysin, thereby excluding epithelial, perineural, neuronal, and neuroendocrine cell lineage and ruling out some of the important

22
Canadian Journal of Gastroenterology and Hepatology differential diagnoses, such as fibromatosis and inflammatory myofibroblastic tumor.
Genetic mutation has also been examined in some studies. The C-KIT and PDGFRA gene mutations are important and characteristic in GIST [82], but they were both negative in all PF cases reported, thereby further enhancing the differentiation between PF and GIST. Genetic mutations involving glioma-associated oncogene homologue 1 (GLI1) and metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) have been identified in a subset of PF cases [54,59], with GLI1 gene translocation reported in 6 cases (24%) and GLI1 polysomy reported in 2 cases (8%), out of overall 25 cases with GLI1 genetic analysis [54,59,78]. The gene translocation t(11;12)(q11;q13) producing functional MALAT1-GLI1 chimeric proteins and the polysomy of GLI1/12q13 both will lead to overexpression of GLI1 protein. The overproduction of GLI1 protein has been recognized in a wide range of neoplasms [84,85] and occurs via activation of the Hedgehog signaling pathway [54], which plays important roles in gastrointestinal developments, diseases, and neoplasms [86]. Apart from the canonical pathway of Hedgehog signaling, a noncanonical, Patched-dependent, and Smoothenedindependent pathway has been recently described, which may be vital for the maintenance of gastrointestinal neoplasms [87]. The molecular pathway of PF development implicates that such neoplasms with GLI1 oncogenesis may be sensitive to Hedgehog pathway inhibitors, targeting either Patched, Smoothened, GLI1, or other Hedgehog pathway components [88]. PF remains an emerging disease entity, and we are far from understanding its genetic profile with such a limited case number. Nevertheless, the present results of genetic analysis that aimed at benefiting differential diagnosis confirm a different genetic profile from GIST, and further research on the GLI1 mutation and Hedgehog signaling may bring about improved understanding of its pathogenesis and even novel therapeutic strategies.

Diagnosis and Differential Diagnoses
Clinical findings are indecisive for the diagnosis of PF, as a solid, elastic tumor with or without ulceration is most likely to be encountered first during endoscopy, with typically nonspecific or hemorrhagic gastrointestinal signs or symptoms. Diagnosis of PF depends on pathological and immunochemical examinations, which are not achieved by clinical examination alone [73], and a minimum of three microscopic signatures should be observed, including spindle or oval cells in a plexiform growth pattern, rich and small-sized vasculature, and abundant myxoid matrix. The benignity of PF could be speculated by its limited atypia and low mitotic rate [89]. The plexiform pattern is highly characteristic of PF, so the designation of a tumor without plexiform growth pattern as PF is highly unlikely. Immunohistochemically, PF is almost always positive for vimentin and SMA and variably positive for desmin, caldesmon, and CD10. Negativities for DOG-1, CD117, CD34, S-100 protein, EMA, and ALK should be examined to exclude similar mimicking conditions. The most important differential diagnosis of PF is GIST, especially myxoid variants of GIST with spindle-cell histology [77,89]. Most PF cases were initially treated based on a suspicion of GIST, which is the most common gastrointestinal mesenchymal tumor and potentially fatal [77,82]. Microscopically, although not totally impossible [89], a plexiform pattern is highly unusual for GIST [9]. In addition, the characteristic immunoprofiles of CD117, CD34, and DOG-1 immunoreactivities and C-KIT or PDGFRA mutations of GIST are adopted widely in routine pathological practice [82].
Other pathological differential diagnoses, based on spindle cell morphology and myxoid stromal background, include sarcomatoid carcinoma, peripheral nerve sheath tumor, inflammatory fibroid polyp, and inflammatory myofibroblastic tumor. Primary or secondary sarcomatoid carcinoma with spindle tumor cells and myxoid stroma could be sometimes mistaken as gastrointestinal mesenchymal neoplasm. A clinical cancer history, a more malignant cytological feature, such as brisk mitotic or apoptotic activities, positive epithelial immunoreactivities, and markers for possible metastatic origins, such as TTF-1 for pulmonary origin, are significantly conducive to the correct diagnosis [90].
Inflammatory fibroid polyps consist of bland-looking spindle cells arranged in perivascular whorled pattern in a myxoedematous inflammatory stroma with mostly eosinophils. Positive CD34 immunoreactivity is the key characteristic of this rare tumor, which should be negative in PF except for vascular endothelial cells [93].
Inflammatory myofibroblastic tumor is characterized by monotonous spindle cells arranged in fascicles or vague whorls in an inflammatory and edematous stroma full of neutrophils, eosinophils, or lymphoplasma cells. Although the myofibroblastic nature of an inflammatory myofibroblastic tumor is similar to that of PF by morphology and immunohistochemical studies, the lack of a plexiform growth pattern, a predominantly inflammatory cellular microenvironment, and positive ALK immunoreactivity help to distinguish inflammatory myofibroblastic tumor from PF [94].

Treatment
The mainstream treatment of PF is surgical removal, while medical treatment serves an assistant role for symptomatic management. Among the 121 reported cases, treatments were available in 99 cases and not provided in 22 cases. Partial gastrectomy (N=30) and distal gastrectomy (N=26) are the most frequently performed surgical treatments for PF, followed by local resection (N=12), subtotal gastrectomy (N=10), wedge resection (N=7), submucosal dissection (N=6), antrectomy (N=3), partial duodenectomy (N=2), nonsurgical follow-up with only endoscopic biopsy (N=2), total gastrectomy (N=1), and laparoscopic cholecystectomy (N=1). A laparoscopic operative technique was performed in 10 cases, endoscopic resection in 9 cases (including 1 thoracoscopic resection), and laparoscopic endoscopic cooperative surgery (LECS) in 4 cases. None of the surgically treated cases presented with severe postoperative complications, except for a 63year-old male patient who had experienced recurrent GI bleeding for 17 years with a duodenal PF and had undergone pancreas-preserving duodenectomy of the first 2 portions, complicated by a low output pancreatic-skin fistula which had totally drained after 4 months. Secondary procedures included Billroth's operation I (N=3), Billroth's operation II (N=2), Roux-en-Y anastomosis (N=2), and omentectomy (N=1).
Conservative management included endoscopic resection and nonsurgical intervention. Since PF is currently considered a benign disease (though not yet verified by convincing evidence), conservative management may be suitable for PF patients, and particularly for the elderly or selected patients with surgically contraindicated comorbidities [72]; at least, more conservative surgical management than partial gastrectomy could be feasible [23]. However, vascular and lymphatic invasion were reported by Miettinen et al. and Kawara et al. [11,60], and PF usually develops in a nodular and plexiform pattern with unclear tumor margins [1,30,64], suggesting that the possibility of malignancy or local recurrence cannot be fully excluded [9,73]. Therefore, regardless of the operative technique, we suggest that complete resection still be the first consideration when treating PF rather than consideration of the conservativeness of the operative technique [30,64,71]. PF cases with severe clinical presentations, such as perforation [1,21], infection [36], or considerable hemorrhage [53,58,63,74], and with malignant suspicion, such as significant body weight loss [11,72] or rapid tumor growth [16], should be addressed with aggressive radical surgical treatment rather than conservative management, despite the benign pathological results.

Prognosis
PF shows a benign biological behavior [18,73,74], with a low proliferation rate and low mitotic rate, and no local recurrence or distant metastasis has been reported so far [74]. However, no consensus has been reached whether PF is actually a benign tumor, and no cases have confirmed that malignant change does not occur, so confirmation of the benign nature of PF still requires longitudinal observational studies with sufficient case numbers. Nevertheless, for the time being, PF should be considered a benign tumor, in response to the strategy of conservative management.
Of all the 121 reported cases, follow-up was not reported in 37 cases, and the follow-up period was not reported in 4 cases. No malignant change, local recurrence, or PF-related mortality was reported. Most of the cases were uneventful after treatments, and among the cases that had follow-up periods, the uneventful or alive duration ranged from 0.75 to 396 months, with an average of 44.29±72.5 months and a median of 15 months. An 81-year-old female who underwent nonsurgical management also had her symptoms resolved 3 months after [72], and a 45-year-old male in Lai's study who underwent only endoscopic biopsy without further resection also led an uneventful life at 0.2-year follow-up [77]. Three cases were reported to have died for unknown causes at 2 months, 14.5 years, and 25.5 years after diagnosis [11]. Vascular or lymphatic invasion was noted in 5 cases, none of which had adverse significance on prognosis [11,60].

Conclusion
PF is a rare mesenchymal tumor with increasing clinical attention and occurs mostly but not exclusively in the GI tract. Typical clinical presentations are nonspecific GI signs or symptoms, or upper GI bleeding. Endoscopic biopsy is recommended for visualizing the microscopic features of PF with benign cytological traits, and immunohistochemical staining is required for diagnosis as well as for exclusion of GIST. Surgical local excision is the main treatment; however, more conservative management is suggested within an individually reasonable range because of its very good prognosis.