Pathophysiology and classification of pseudomyxoma peritonei

Introduction

The term pseudomyxoma peritonei (PMP) was first described by Werth in 1884 [1] when describing a mucinous tumour of the ovary and subsequently in 1901 by Frankel [2] who described a cyst of the appendix. There has been much debate since then over the term PMP but a recent consensus of world experts agreed that PMP should be thought of as a clinical entity characterised by the presence of mucinous ascites, omental cake, peritoneal implants and possibly ovarian involvement [3]. It generally originates from mucinous appendiceal tumours but has been associated with other mucinous neoplasms of the ovary, colon, and pancreas [4, 5].

Incidence of appendiceal tumours and PMP

Appendiceal mucinous neoplasms are unusual and their incidence is difficult to define as estimates are based on retrospective analysis following appendicectomy for assumed appendicitis or following synchronous removal at the time of colonic resection for adenocarcinoma [6–8]. The incidence of appendiceal tumours in appendicetomy specimens is estimated at 1 % in appendicetomies [6, 7] and this incidence increases to 4 % in the more selective population of patients undergoing colonic resection for adenocarcinoma [8]. There is an estimated incidence of mucinous neoplasia (including mucinous adenocarcionoma) of 35 % [7, 8] amongst the abnormal appendices. PMP is uncommon. 20 % of epithelial neoplasms of the appendix analysed in population-based epidemiological studies progressed to PMP and this suggested an incidence of 2–3 cases per million of population [7, 9] but this may be an underestimate with data from specialist centres suggesting rates up to 3–4 cases per million [10].

Clinical presentation and origin of PMP

The clinical presentation of appendiceal tumours is either incidental when an appendicetomy is performed as part of another procedure or when presenting as appendicitis. PMP has a much more varied presentation. Even when the disease burden is high the associated symptoms can often be vague. Esquivel and Sugarbaker in 2000 [11] reported that appendicitis (27 %), increasing abdominal distension (23 %) and a new onset hernia (14 %) were the most common presenting features but in 2014 Glaysher et al [12] reported that 50 % of patients were diagnosed on CT. Only 21 % of patients were diagnosed at laparotomy or laparoscopy for either appendicitis or for abnormal histology and 5 % were confirmed after the onset of a new abdominal wall hernia where the presence of mucin within the sac warrants further investigation with a CT scan [13]. This disparity between the two series highlights both the impact that CT has had in the investigation of non-specific abdominal symptoms but also the indolent nature that this disease can take [14] (Figure 1).

Figure 1:

CT scan showing widespread mucinous ascites.

The primary tumour was the appendix. Mucinous ascites can be seen in the abdominal cavity but also in both subdiaphragmatic spaces (arrows).

Origin and natural history of PMP

It is now widely accepted that the appendix is the primary site in the vast majority of PMP in both men and women. Historically it was felt that the ovary was a common primary site in women but recent work has shown that the disease in the ovary is often metastatic and that the appendix is usually the primary site. Smeenk [7] identified the appendix as the primary site in 82 % of patients with PMP in whom a primary site was identified (68 % of PMP patients) (Figure 2). Immunohistochemistry samples have looked at the expression of mucins in PMP, in particular MUC2 and MUC5AC as these mucins are gel-forming, similar to that seen in PMP [15]. Ovarian mucinous tumours typically over express MUC5AC and to a lesser degree MUC 2 [16] compared with appendiceal tumours which over express MUC 2 more than MUC5AC [16–20]. The expression of CDX-2 and CK-20 and the lack of expression of CK-7 also support the hypothesis of an appendiceal primary in the majority of PMP cases. In a study by Chu that assessed cytokeratin expression in epithelial neoplasms CK-20 was expressed in 100 % of colonic cancers but only 4 % of ovarian cancers whilst CK-7 which was expressed in 100 % of ovarian cancers and only 5 % of colon cancers [21].CDX-2 is a gene that encodes an intestinal-specific transcription factor which is expressed in the nuclei of epithelial cells throughout the intestine. CK-7 (cytokeratin-7) and CK-20 (cytokeratin-20) are proteins produced by epithelial cells. In most cases of PMP, CK-20, CDX-2, and MUC-2 are diffusely positive supporting the hypothesis that PMP is usually secondary to an appendiceal neoplasm [16–22]. This pattern of gene expression demonstrates the similarity of appendiceal PMP to colorectal adenocarcinoma however PMP disseminates widely throughout the abdomen yet seldom metastasizes outside the peritoneal cavity. It has been proposed that the increased expression of N-cadherin and reduction in E-cadherin expression reduces cell adhesion and subsequent metastatic disease [23]. This over expression of mucin and lack of adherence may explain the natural history of PMP.

Figure 2:

(a) CT showing a large pelvic mass with a calcified rim which extends to the caecal pole and represents an appendiceal mucocele seen in (b).

Mucin producing appendiceal neoplasms typically progress and eventually occlude the lumen of the appendix which is followed by distension of the appendix with the development of a sterile mucocele. This will ultimately perforate spilling the mucin and mucinous tumour cells into the peritoneal cavity. This is often an asymptomatic process but if the occlusion is rapid it may present as acute appendicitis. Once within the peritoneal cavity the mucinous tumour cells will continue to proliferate and produce the characteristic “jelly belly”. Mucin and tumour cells are distributed according to the “redistribution phenomenon” described by Sugarbaker (1994) [24] which dictates that large volumes of disease will be found at certain predetermined anatomic sites within the peritoneal cavity but will be absent or greatly reduced in others (Figures 3–5). For PMP there will be high-volume disease within the greater omentum (“omental cake”), the undersurface of the right hemidiaphragm, Morison’s Pouch (Hepatorenal space), the left paracolic gutter, the ligament of Treitz and the recto-vesical pouch whilst there will be relative sparing of the peritoneal surfaces of stomach and bowel due to continued peristaltic activity in these areas reducing mucin and tumour adherence. The mucin and tumour cells circulate because of this lack of adhesion due to two physiological mechanisms. The first is gravity resulting in accumulation in gravity-dependent areas within the abdominal cavity such as Morrison’s Pouch (Hepatorenal space), the left paracolic gutter, the ligament of Treitz and the recto-vesical pouch. The second mechanism involves the physiological process of peritoneal fluid reabsorption by the greater omentum and right hemidiaphragm and at which sites mucin and tumour cells accumulate. The concentrated tumour masses that occur as a result of the filtration of peritoneal fluid result in “scalloping” of the liver and the development of an “omental cake”. It is this distribution within the peritoneal cavity as a result of the mechanism described above, and the sparing of small bowel in particular, which allows complete cytoreductive surgery (CRS), combined with heated intraperitoneal chemotherapy (HIPEC) to be effective in the management of this disease. The nature of this disease is that it does not metastasise; however, it can still result in significant symptoms with progressive abdominal distension and small bowel involvement making complete cytoreductive surgery impossible and if left untreated, gastrointestinal tract compression that leads to starvation and death.

Figure 3:

Mucinous disease involving the right hemidiaphragm (a), spleen (b) and omental cake with small bowel sparing (c).

Figure 4:

Omental cake – surgical specimen, formalin-fixed tissue.

Figure 5:

(a) Low grade mucinous carcinoma peritonei (DPAM) of peritoneum. Abundant mucus is present with a small amount of low-grade columnar epithelium. ×4 magnification. Haematoxylin and Eosin stain. (b) High grade mucinous carcinoma peritonei (PMCA) of peritoneum. There is a cribriform pattern of growth. ×20 magnification. Haematoxylin and Eosin stain.

Genetic mutations in PMP

Genetic mutations associated with PMP may be the key to understanding the natural history of the disease and open other therapeutic avenues (Table 1). When looking for genetic mutations in PMP, researchers have principally looked for those that are seen in colorectal cancer as a starting point. KRAS, part of the RAS family of genes, has been studied extensively. Mutation of the KRAS gene leads to unregulated downstream growth, and the mutation is associated with 40 % of colorectal cancers but also 50 % of adenomas and is thought of as an early event in the pathogenesis of colorectal cancer [25, 26]. Its involvement with PMP is much higher with a mutation detected in 58–94 % of cases [27–33]. This mutation appears mainly on codon 12 and is seen in both low and high grade tumours. The presence of the mutation does not have any impact on survival [30] unlike colorectal cancer where the presence of the mutation has been associated with a poorer prognosis [34–36]. The OPUS and CRYSTAL trials demonstrated that colorectal cancers with the KRAS mutation gain little benefit from anti-epidermal growth factor receptor (EGFR) agents such as cetuximab and that their addition to first-line regimes may even be detrimental [37–39]. It is therefore likely that similar agents are unlikely to have any benefit in PMP and it seems logical that agents which target the RAS-MAPK pathway may have a therapeutic benefit in the treatment of PMP and may delay recurrence.

GNAS regulates the expression of MUC2 and MUC5AC in colon cancer cells when a mutant form of the gene is introduced into mice with a colorectal cell line. It has been identified in varying degrees in low grade PMP with an incidence ranging from 12 to 100 % [27, 30–33] and a significantly lower distribution in high grade PMP (19/23 low grade vs. 1/6 high grade, p=0.005) [32] although this is not supported in all analyses [33]. GNAS mutations have also been identified in intraductal papillary mucinous neoplasms of the pancreas [40] and it has therefore been postulated that activation of the GNAS gene may induce mucin production seen in PMP and IPMN [27]. However MUC2 is seen in all patients, regardless of GNAS status and not all mucinous tumours derived from other organs display the GNAS mutation [40]. Therefore whilst the dysregulation of the GNAS gene may explain some of the phenotype seen in PMP it is certainly not the only the explanation for a disease that has a heterogeneity in clinical presentation.

Whilst KRAS mutation is distributed among both low and high grade PMP and GNAS is more commonly seen in low grade disease, there are some genetic mutations that are specific to high grade disease. Mutations in the TP53 gene and genes related to the PI3-AKT pathway are the most common of these with a prevalence of over 40 % in colorectal cancers [41]. It is of little surprise therefore that mutations are also seen in mucinous neoplasms of the appendix and PMP. Whilst they have been demonstrated in low grade disease they are significantly more common in high grade disease. Noguchi [27] found TP53 mutations exclusively in peritoneal mucinous carcinomatosis (PMCA) whilst Shetty [28] showed that distribution was significantly greater in high grade PMP (35.5 % vs. 54.4 %, p=0.009). This was supported by Nummela [33] in their analysis of 74 cases of PMP with an incidence of 31.3 % in high grade PMP compared with 7.1 % in low grade disease (p=0.012). This may explain the aggressive nature of some forms of PMP and this is supported by the finding that TP53 over expression is associated with female sex, higher-grade disease, and a poorer survival [28, 33]. The PIK3-AKT pathway is an intracellular signaling pathway that is important in regulating the cell cycle and therefore has been directly linked to cancer formation and proliferation. Within PMP there are isolated cases of PIK3-AKT mutation that are commonly seen in high grade disease with an incidence of 5–10 % [27–31]. The significance of this is unclear.

The findings in PMP are in contrast to colorectal cancer. KRAS and GNAS mutations are common in PMP but seen far less often in colorectal cancers [42, 43]. PIK3-AKT and TP53 mutations are seen in 20 % and 50 % of colorectal cancers respectively, but this figure is significantly lower in PMP and is associated with high grade disease. The literature in this area is difficult to interpret with studies of small numbers and significant heterogeneity in the histological classification of the specimens. This makes direct comparison difficult in terms of relative numbers but does not detract from the major findings that there are genetic mutations peculiar to high grade disease that may explain their more aggressive nature and reduced survival. Furthermore management with anti-EGFR agents will not be effective in the presence of KRAS mutation, an almost universal finding in both high and low grade disease. Further work using the recently described classification system will allow reproducible work to be undertaken and assist in a better understanding of the genetics of PMP:

Classification of appendiceal mucinous neoplasms, PMP and impact on prognosis

The majority of PMP is derived from tumours of the appendix and the biology of these tumours allows disseminated spread within the peritoneal cavity to both the peritoneum and other abdominal organs. The classification of PMP and its primary appendiceal neoplasm has been a source of confusion for a number of years with several different classification systems proposed to describe both the primary tumour and peritoneal disease. Carr in 1995 [44], and later through the World Health Organisation in 2010 [45], classified tumours without invasion as adenoma, whilst those with infiltrative invasion as mucinous adenocarcinoma. Tumours with pushing invasion were initially termed mucinous tumour of uncertain malignant potential until Misdrajil [46] in 2003 proposed a new term, low-grade appendiceal mucinous neoplasm (LAMN), which was further divided into LAMN with a low risk of recurrence and LAMN with a high risk of recurrence by Pai [47] in 2009. Whilst the term LAMN was adopted into the WHO classification system in 2010 there was no comment on risk of recurrence. More recently an attempt by McDonald [48] to determine the need for cytoreductive surgery for early LAMNs identified two subtypes, LAMN I and LAMN II with subtype II having pathological features that increased the risk of dissemination and for which cytoreductive surgery was recommended whereas type I lesions were deemed suitable for surveillance.

Controversy also exists over the classification of peritoneal disease. PMP was first divided into disseminated peritoneal adenomucinosis (DPAM) and peritoneal mucinous carcinomatosis (PMCA) by Ronnett [49]. This was a retrospective analysis of patients who had undergone cytoreductive surgery with HIPEC at a single institution. DPAM was characterised by peritoneal lesions with an abundance of extracellular mucin with little cytologic atypia or mitotic activity whereas PMCA described peritoneal lesions with a mucinous epithelium and architectural and cytologic features of carcinoma. A further group was classified as PMCA-I with intermediate or discordant features (based on the finding of at least focal areas of carcinoma in peritoneal lesions). Five year survival rates were 84 % for DPAM, 37.6 % for PMCA-I and 6.7 % for PMCA (p > 0.0001) and 10 years survival rates of 68 %, 21 %, and 3 % respectively (p=0.0001). Ronnett [49] proposed that PMP should be a clinical descriptor, DPAM should be used as a pathological term for patients with bland peritoneal mucinous tumours associated with ruptured low grade mucinous lesions they termed “adenomas” and that PMCA should be considered as mucinous carcinoma. Subsequently, the Ronnett group classified PMCA-I with PMCA based on similar outcomes in a later series [50]. Bradley [51] were unable to reproduce these results demonstrating no statistical difference between the survival outcomes of DPAM and PMCA-I at 1-year, 3-year, and 5-year following cytoreductive surgery with HIPEC at their institution. They proposed the terms mucinous carcinoma peritonei-low grade for low-grade histology (DPAM and PMCA-I) and mucinous carcinoma peritonei-high grade for moderately to poorly differentiated cases (PMCA). The presence of signet rings in their classification automatically classified the case as high grade. Shetty [52] suggested a grading system with three tiers, PMP 1, PMP2 and PMP3. PMP1 subjects consisted of cases with low grade histological features and were analogous to DPAM whilst PMP3 consisted of all patients with signet cells. PMP2 was any case that did not meet the criteria for PMP1 or PMP3. Survival rates of 86 %, 63 % and 32 % for PMP1, PMP2, and PMP3 respectively at 5 years were described. Carr reviewed 257 appendix tumours treated in Basingstoke by cytoreductive surgery with HIPEC and classified them using The WHO classification [25] into low and high grade [53]. In the majority of cases there was concordance between the primary lesion and peritoneal disease. The overall 5-year survival following complete cytoreduction with HIPEC was 84 % for low grade PMP and 48 % in the high grade group (p < 0.001).

Whilst it is widely accepted that histological grade of both the primary appendiceal mucinous neoplasm and PMP affect prognosis there has been no widely accepted classification system until The Peritoneal Surface Oncology Group International (PSOGI) aimed to establish one in 2015 [3]. Using a modified Delphi process, a consensus method that is applicable when there is a lack of concordance of opinion because scientific evidence is lacking or contradictory (Kusamura, Jones) [54, 55], a panel of 71 international experts were questioned. The aim of the consensus was to develop a uniform classification system with prognostic significance in both clinical evaluation and therapeutic decision making.

Clinically agressive pseudomyxoma peritonei – the limitations of histological classification and the role of tumour markers

A pathological classification of a tumour does not always accurately predict clinical behaviour with presumed good prognosis tumours sometimes metastasing rapidly and unexpectedly. Bad tumour biology is often used to explain these cases and to describe changes at a genetic level that make the tumour more aggressive and at odds with the histological staging. PMP is no exception to this phenomenon. The controversies in histological classification discussed earlier are testament to the difficulties that pathologists and surgeons have had in relating clinical outcome to microscopic findings. The PMCA-I group (low grade with any high grade features) proposed by Ronnett [49], which was combined with DPAM by Bradley [51] to form low-grade mucinous carcinoma peritonei, probably represents this low grade group that act aggressively. Maximal tumour debulking (MTD) is performed when complete cytoreduction cannot be achieved usually due to extensive small bowel serosal involvement or widespread upper abdominal disease encasing the stomach and porta. One might expect these tumours to be high grade histologically but a recent study reported that 70 % of patients who underwent MTD were low grade [10]. Mohamed [57] identified 11 patients with clinically aggressive low grade disease from a series of 501 and case matched them to look for specific differences. No differences in clinical factors or in mucin expression were seen between the two groups to explain the clinical outcomes. The genetic mutations associated with PMP have been outlined above and one key point to note is the heterogeneity in the genetic mutations between individuals. This is clearly seen with TP53. A mutation in this gene has been shown in PMP to confer a worse prognosis but has been seen in low and high grade disease. What is not known is the impact this or the other mutations have in clinically aggressive disease.

The ability to predict disease behaviour is important as it allows clinicians to discuss realistic outcomes with patients and may influence adjuvant treatments. Tumour markers, namely carcinoembryonic antigen (CEA), carbohydrate antigens 125 (CA 125) and 19–9 (CA 19–9) are often raised in PMP and normalise rapidly in most patients within a week of CRS and may even be an indicator of effective cytoreduction [10, 58, 59]. At least one tumour marker is raised in 70–76 % [59, 60] of patients and elevation of all three tumour markers (CEA, CA125 and CA 19–9) increases the risk of recurrent disease and shortens survival after complete cytoreduction with HIPEC for appendiceal tumours [59–61].

It is important to note that although raised tumour markers are correlated with the peritoneal carcinomatosis index (PCI) score they do not reflect the ability to achieve a complete cytoreduction and should not be considered a contraindication to surgery [60]. The pathway of CEA, CA125 and CA19-9 production including half life are yet to be established clearly and may relate to volume of peritoneal disease. If the new PSOGI classification system for appendiceal tumours is widely accepted the next step may be to study these tumour markers in greater depth as they seem to predict tumour behaviour after complete cytoreduction sometimes more accurately than histology alone.

Conclusion

PMP is a disease that is widely understood in terms of clinical presentation but the lack of a consensus in classification has hindered comparison of treatment efficacy and molecular understanding. We now have a classification that has been agreed upon by the leading experts in PMP and that should lead towards homogenous comparisons between study groups. Research into PMP is still in its infancy and is often based on retrospective case series with the inherent limitations of this kind of work. Attention now needs to be focused on understanding the factors associated with poor prognosis through well designed multi-centred prospective studies. This will allow us to identify patients with bad tumour biology so that targeted treatment based on likely prognosis may then become a reality.

Author contributions: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.

Research funding: None declared.

Employment or leadership: None declared.

Honorarium: None declared.

Competing interests: The funding organization(s) played no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the report for publication.