Context.—Pneumatosis cystoides intestinalis (PCI) is a condition with multiple gas-filled cysts within the bowel wall, associated with diverse background diseases. Its pathogenesis is still a mystery. Some previous observations scattered in the literature have suggested an association of the cystic spaces in PCI with the lymphatics.
Objective.—To further investigate whether PCI results from the ballooning of gas-filled lymphatic channels.
Design.—We did immunostaining of podoplanin, a mucoprotein preferentially expressed in lymphatic endothelial cells, in 13 cases (8 men, 5 women; age range, 18–80 years) of PCI. Ten cases were diagnosed in resected segments of bowel and 3 in biopsies. Pneumatosis was seen in the right side of the colon (9 cases), transverse colon (1 case), sigmoid colon (1 case), and small bowel (2 cases). In addition, immunostaining for CD31, calretinin, WT1, CD68, smooth muscle actin, desmin, vimentin, and cytokeratins was also performed for comparison and correlation.
Results.—A strong immunopositivity of podoplanin was seen in a condensed linear structure in the pericystic interstitium in 100% of the cases, but was not seen in the overlying giant and flat cells that were all CD68-positive histiocytes. Meanwhile, the podoplanin-expressing structure was negative for calretinin and WT1, which ruled out the possible mesothelial origin. There were coexistent variable immunopositivity of smooth muscle actin, which suggests an admixture of myofibroblasts. These findings indicated that the PCI cases were gas-distended lymphatics with the lymphatic epithelium ruptured and embedded in the reactive histiocytes and giant cells.
Conclusion.—Our findings support the lymphatic theory about the pathogenesis of PCI.
Pneumatosis cystoides intestinalis (PCI), also termed pneumatosis intestinalis, pneumatosis coli (when occurring in colon only), and several other uncommon names in the literature, is a rare condition characterized by multiple gas-filled cystic spaces within the bowel wall. It was first described by Du Vernoi1 in 1783 and later documented in more detail by Bang2 in 1876. Pneumatosis cystoides intestinalis is associated with a large variety of diseases and clinical conditions in which the loss of bowel mucosal integrity and/or luminal high pressure usually occur. Pneumatosis cystoides intestinalis per se is mostly asymptomatic. A spectrum of possible manifestations, usually related to the underlying pathologic conditions, includes vomiting, abdominal distention, abdominal pain, diarrhea, bloody stool, constipation, and tenesmus. These days, PCI is often identified incidentally on radiologic imaging or on pathologic examination.
The etiology and pathogenesis of PCI have been a mystery. Several theories have been proposed based on different observations, including (1) luminal gas penetrating the bowel mucosa into the bowel wall following mucosal injury,3–5 (2) pulmonary gas released from the ruptured alveoli dissecting along vascular channels in mediastinum and tracking through the retroperitoneum into the mesenteric root,6,7 (3) gas production of some bacteria that invades through the mucosal breaches into the intramural compartments and/or grows within the microscopic cysts,8 and (4) gas production related to defects in food digestion causing excessive fermentation.9
In addition, a suggestion that the cystic spaces in PCI are dilated lymphatic vessels has been lingering in literature. In the early days, this condition was also termed lymphopneumatosis cystoides intestinalis or cystic lymphopneumatosis based purely on assumption. In an experiment conducted half a century ago, Staudacher and Bencini10 injected a dyed fluid into the lymphatics of hogs and demonstrated that the distended lymphatics formed cystic spaces. Some investigators also noticed a spatial relationship of pneumatosis to lymphoid tissue,11–13 which is indeed a frequent histologic finding. By light and electron microscopic study in one case of PCI, Haboubi et al14 claimed to identify some evidence of the cystic spaces being dilated lymphatic vessels in nature. All of these have supported the lymphatic theory in the pathogenesis of PCI.13–17 However, to our knowledge, no conclusion has been made.
Following an incidental finding in an index case, we have noticed a strong immunohistochemical positivity of podoplanin (D2-40 as antibody), a transmembrane mucoprotein commonly expressed in lymphatic endothelial cells, in a prominent linear structure surrounding the cystic spaces of all PCI cases that we have collected. In order to further understand the nature of the podoplanin-expressing structures and to clarify if they are the ruptured lymphatic epithelium, we have conducted an immunohistochemical study in a small series.
MATERIALS AND METHODS
Thirteen cases of PCI (8 men, 5 women; age range, 18–80 years) were retrospectively retrieved from the surgical pathology files of Calgary Laboratory Services, including one outside consult case. As shown in Table 1, 10 cases were diagnosed on surgical resection specimens and 3 on colonic endoscopic biopsies. The pneumatosis cystoides were seen in various regions of the small and large intestines, mostly in the right side of the colon (9 cases), including pan–right colon (5 cases), ascending colon (3 cases), cecum (1 case), transverse colon (1 case), sigmoid colon (1 case), and small bowel (2 cases). The associated diseases/clinical conditions included ischemic bowel (3 cases), diverticulosis (2 cases), sepsis (1 case), Clostridium difficile colitis (1 case), Crohn disease after ileocolic resection (1 case), volvulus (1 case), endoscopically resected polyps (1 case), lymphoma (1 case), bowel perforation (1 case), and rectal bleeding (1 case). In 1 case, the associated disease was not clearly identified.
One representative section/tissue block from each case was selected for histologic analysis and immunohistochemical study, in which the pneumatosis cystoides were confirmed microscopically. Immunohistochemistry of the following panel was performed: podoplanin (D2-40 as antibody, commonly known as a marker of lymphatic endothelial cells), CD31 (known as a marker of pan-endothelial cells), CD68 (a marker of macrophages/histiocytes), smooth muscle actin (SMA; a marker of smooth muscle differentiation), and WT1 and calretinin (common markers of mesothelial cells). Additionally, immunostains for pan-cytokeratins and vimentin were also performed. The antibodies used for the immunohistochemistry are listed in Table 2. The intensity of the immunohistochemical positivity of each of the markers was semiquantitated as negative (−), weak (+), moderate (++), and strong (+++).
The study was approved by the Health Research Ethics Board in the authors' institution.
On gross or endoscopic examination, all cases showed typical pathologic features of PCI characterized by polypoid appearance of mucosal surface (in 8 cases) and/or cystlike empty spaces within bowel wall (submucosa +/− mucosa and subserosa) on cut surfaces (Figure 1). Microscopically, a variable number of variously sized cystic spaces were present in the mucosal, submucosal, and/or subserosal layer(s) of the bowel wall. Almost all of the cystic spaces in all of the cases were lined by a layer of variably prominent foreign body–type multinucleated giant cells and/or epithelioid macrophages (Figure 2, A). Some areas of the inner surface of the cystic spaces were lined by mononucleated flat cells (Figure 3, A), and some areas were denuded without lining cells (Figure 3, C). Underlying the giant cell or flat cell lining, there were condensed spindling cells or layering linear structures inseparable from the surrounding stroma.
On immunohistochemistry, a strong immunopositivity for podoplanin (D2-40) was seen in the condensed linear structures within the pericystic interstitium (ie, the inner part of the stroma surrounding the cystic spaces) of every cystic space (Table 3; Figures 2, B, and 4, B), superficially resembling that seen in the adjacent lymphatic spaces (Figure 4, B). Moreover, a close anatomic relationship between the lymphatic spaces and cystic spaces was frequently noted in many areas, as demonstrated in Figure 4, A.
On higher magnification, the podoplanin-expressing structures were identified to be linear fibers, spindle cells, and small cuboidal cells (Figure 3, B through D).
A weak immunopositivity of podoplanin was also seen in most of the denuded areas of the inner surfaces of the cystic spaces, that is, the compressed stroma with linear structures and/or spindle stromal cells immediately under the detached lining cells or exposed directly to the cystic spaces (Figure 3, B and C).
However, for the overlying cells, either giant or flat lining cells, essentially no expression of podoplanin was seen (Tables 4 and 5; Figures 2, B, and 3, A and B), except for rare cells in rare cystic spaces (seen in 1 of the 13 cases and in 2 subserosal cysts) in which some podoplanin-positive cells were noted among the giant cells.
On the expanded panel of immunohistochemistry (Tables 4 and 5), the lining cells, both giant and flat cells, were all strongly positive for CD68 (Figures 2, D, and 5, A and B) and vimentin as well as weakly positive (faint stain in most cases) for CD31 (Figures 2, C, and 5, C and D). However, these cells were all negative for cytokeratins, WT1, calretinin, SMA, and desmin (Tables 4 and 5).
The pericystic stromal cells/fibers, on the other hand, were essentially negative for CD68, CD31, and cytokeratins (Table 3). In 2 of the 13 cases, a few scattered stromal cells were also weakly positive for WT1 and/or calretinin (Table 3). However, the stromal cells were all positive for vimentin and some were positive for SMA (Table 3).
The present study started with the first author's incidental finding (in case 1) that a condensed podoplanin-immunopositive linear structure was seen around the cystic spaces of pneumatosis coli and superficially resembled the adjacent lymphatic spaces. The author's initial interpretation of the finding was that the cystic spaces of pneumatosis intestinalis were the gas (air)-distended lymphatic channels secondary to the entry of bowel luminal gas (air) into the bowel wall in various conditions in which the mucosal integrity is lost.18 The additional finding of a dual immunopositivity of podoplanin and CD31 that was observed in 1 of the cases (case 5; micrographs not shown) also seemed to support the idea. Moreover, this notion arose with reference to some observations reported in the literature.10–17 By electron microscopic study in one case of PCI, Haboubi et al14 identified some ultrastructural evidence of the cystic spaces being dilated lymphatics in nature. The evidence was 2-fold. First, some of the lining cells were endothelial cells with characteristic ultrastructural features (high pinocytic activity, overlapping cytoplasmic edges, and lack of basement membrane) consistent with the lymphatic endothelial cells. Secondly, abundant fatty globules, presumably from the lymphatic leakage, were found in the cystic lumina and within the lining histiocytes/multinucleated giant cells. From their single case study, the authors concluded that the pneumatosis cystoides were derived from lymphatics that ruptured because of the increased intraluminal pressure caused by air entry, and the leakage of the fatty lymph contents initiated the histiocytic reaction with the formation of foreign body giant cells. In 2 other supportive studies, Staudacher and Bencini10 recovered in the cystic spaces the dyed liquid that had been injected into the lymphatics of hogs. Sibbons et al17 demonstrated the development of PCI in piglets with ligation of mesenteric lymphatic vessels.
Our further observations in more cases with expansion of the immunohistochemistry panel, however, have generated complex data and brought the initial finding under a wider scope. First of all, the podoplanin expression was essentially not seen in the cells that line the cystic spaces, except in a few, but was basically in the underlying interstitium, including some interstitial spindle cells, small cuboidal cells, and interstitial fibers. Second, the cyst lining cells, either flat or giant cells, were all histiocytic in nature, evident by their strong CD68 immunopositivity. Parenthetically, they were also weakly positive for CD31. Third, the podoplanin-immunopositive elements also showed variable immunopositivity of SMA. Considering all of the findings together, the interpretation of podoplanin immunopositivity becomes less straightforward than we initially thought, and it deserves a careful analysis.
If pneumatosis cystoides were truly distended and ruptured lymphatic vessels caused by the entry of bowel luminal gas (air), it would be reasonable to expect to see some residual lymphatic endothelial cells remaining on the lining of the cystic spaces, and thus there should be at least some podoplanin-expressing lining cells. Additionally, the reactive histiocytes/macrophages may have taken up the debris of the lymphatic endothelial cells, and thus they would be immunopositive for podoplanin. However, these conditions seem to be noted in only 2 cysts in 1 of the cases, and not identified in the rest of the 13 cases that have been thoroughly reviewed. In a study of 25 PCI cases reported by Gagliardi et al,13 they stated that “in no case was an endothelial component identified immunohistochemically” (by detecting factor VIII, CD34, and Ulex europeus type 1 lectin), and that “in two cases, mildly dilated vascular channels consistent with lymphatics were present adjacent to the cysts, but no definite continuity was identified with the cysts.” With their findings in mind, we tend to be cautious in interpreting the finding in the single positive case in our observations.
If those flat lining cells are the residual lymphatic endothelial cells, in consideration of their weak CD31 positivity, they should be podoplanin positive and CD68 negative. However, what we saw in this study is clearly opposite. All of the flat cells showed an immunohistochemical pattern identical to that of the multinucleated giant cells, which is clearly supportive of their being macrophagic/histiocytic in nature.
The weak immunopositivity of CD31 seen in the flat and giant lining cells is of interest. CD31, also known as platelet endothelial cell adhesion molecule 1, is known to be a pan-endothelial marker, although it is nonspecific to and less expressed in lymphatic endothelial cells.19 On immunohistochemistry, lymphatics are expected to be positive for both podoplanin and CD31. Our study clearly showed a disassociated pattern in the pneumatosis cystic spaces. The adjacent normal lymphatic vessel was clearly immunopositive for both podoplanin and CD31. However, the cystic space showed a different pattern, in which the lining cells were weakly positive for CD31 but negative for podoplanin, whereas the underlying interstitium showed an opposite pattern of stain. Such an opposite CD31/podoplanin immunostain pattern between the pneumatosis cystic space and the adjacent lymphatic channel seems to negate the possibility that the lining cells of the cystic spaces are the lymphatic endothelial cells. On the other hand, the weak immunopositivity of CD31 seen in the lining cells is actually not against their apparent histiocytic nature, because CD31 expression is known to exist in monocytes, macrophages, and histiocytes as well.20–22
Thus the question is exactly what the underlying podoplanin-positive structure is. First of all, could it be that the ruptured lymphatic epithelium remains embedded by the reactive histiocytes? In an early histologic study, Koss8 proposed a 5-stage histogenesis of PCI: (1) formation of cysts with simple endothelial cells, (2) formation of giant cells, (3) inflammatory reaction in the surrounding tissue, (4) sloughing of giant cells with concomitant fibrosis of the surrounding stroma, and (5) diminution of the size of the cysts to the point of complete fibrosis with ultimate disappearance as scarring progresses. Although we did not confidently identify stages 1 and 5 in our cases, we did see the mixed morphology of the other changes. It is most reasonable to assume that the histiocytes/multinucleated giant cells that line the cystic spaces came in response to some type of chemotactic factor(s) occurring in the formation of the gas-distended spaces and the process of interstitium dissection, and the histiocytes in this setting function as a sealant to cover and repair the injured interstitium. In line with this, if the pneumatosis cystoides are the result of gas-distended and ruptured lymphatics, the destroyed lymphatic endothelial lining would be covered by the reactive histiocytes and embedded within the other inflammatory cells in the surrounding stroma. As a result, an immunopositivity for the lymphatic endothelial cell components, including some specific molecules, such as podoplanin, deposited in the pericystic interstitium would be expected. Would some of those cells and/or the condensed linear structure immediately under the overlying histiocytes, and those cells on the seemingly denuded surface, in fact be the residual lymphatic endothelial cells or their remains? When we look carefully at some of the higher-magnification micrographs, at least some areas look quite similar to the lymphatic epithelium of normal lymphatics. In our opinion, this explanation sounds very plausible.
The other question is how specific the immunoreactive podoplanin is to the lymphatic endothelial cells, in terms of being a marker. Podoplanin is an Mr 40 000 O-linked transmembrane sialoglycoprotein originally detected in association with germ cell neoplasia and fetal testicular gonocytes. With the development of a monoclonal antibody, clone D2-40, podoplanin was later found to be expressed in the endothelium of lymphatic capillaries but not in the blood vasculature, and therefore it became an immunohistochemical marker of lymphatic endothelial cells.23–26 Further studies, however, have found that podoplanin is also detected in a variety of other cells, including mesothelial cells,23 reticular cells, follicular dendritic cells, myoepithelial cells,27,28 and myofibroblasts.29,30 Interestingly, we found in our study that the areas with podoplanin-expressing elements in the pericystic stroma also showed more or less immunopositivity of SMA but not WT1, calretinin, or desmin. This additional immunoprofile seems to be suggestive of the presence of myofibroblasts. Additionally, we noted that in some areas of the uninvolved intestinal mucosa there are scattered podoplanin-expressing spindle cells in the lamina propria, and these cells are located and arranged around the intestinal glands/crypts in a pattern resembling pericryptal myofibroblasts (micrographs not shown). However, in the majority of our cases this type of podoplanin-expressing spindle cell was not seen in the normal intestinal mucosa. Our hypothesis for the latter puzzling findings is that a subpopulation of activated pericryptal myofibroblasts may gain podoplanin expression. These additional findings unfortunately further complicate our data and make it more difficult to draw a conclusion. In our further analysis, we have noted that the positivities of podoplanin and SMA around the cystic spaces are not exactly identical. In some cases the pericystic stroma develops a florid inflammatory reaction or fibrosis, where the myofibroblasts are presumably prominent and the immunopositivity of SMA appears stronger than that of podoplanin. When the surrounding inflammation and/or fibrosis is mild or nonexistent, the immunopositivity of SMA usually appears much weaker and sparse, as compared with that of podoplanin. Could this discrepancy of podoplanin and SMA immunostains also suggest the presence of lymphatic epithelial elements in the background? We favor this hypothesis, although we feel that it is very difficult to completely clarify without a definitively specific lymphatic marker.
In summary, although our immunohistochemical study still cannot definitively confirm or reject the proposal that pneumatosis cystoides are distended and ruptured lymphatic channels associated with bowel gas entry, we would like to argue that the lymphatic theory is most likely true based on our findings and analyses. The podoplanin-expressing condensed linear structure and cells under the overlying reactive histiocytes are most likely the remains of the damaged lymphatic epithelium. In line with this, we also postulate that the lymphatic channels become the common gas-transporting pipeline in various conditions in which bowel mucosal integrity is impaired and bowel luminal gas leaks into and permeates the bowel wall. The clinical presentation most likely depends on the location where the gas enters and the extent as well as degree of the ballooning of the bowel wall lymphatics. Further study by using a definitively specific and sensitive marker of lymphatic epithelium, when it is available in the future, will ultimately clarify this issue.
Ms Michelle Darago, BS, provided assistance with the study materials. This study was supported by Calgary Laboratory Services (CLS internally supported project RS11-510).
The authors have no relevant financial interest in the products or companies described in this article.