Exploring the peritoneal surface malignancy phenotype—a pilot immunohistochemical study of human pseudomyxoma peritonei and derived animal models

Human Pathology (2010) 41, 1109–1119

www.elsevier.com/locate/humpath

Original contribution

Exploring the peritoneal surface malignancy phenotype— a pilot immunohistochemical study of human pseudomyxoma peritonei and derived animal models☆ Kjersti Flatmark MD, PhD a,b,⁎, Ben Davidson MD, PhD c,d , Alexandr Kristian MSc a , Helene Tuft Stavnes MSc c , Mette Førsund MSc c , Wenche Reed MD, PhD e a

Department of Tumor Biology, Institute for Cancer Research, The Norwegian Radium Hospital, Oslo University Hospital, N-0310 Oslo, Norway b Department of Surgical Oncology, The Norwegian Radium Hospital, Oslo University Hospital, Radiumhospitalet, Oslo, Norway c Division of Pathology, Norwegian Radium Hospital, Oslo University Hospital, N-0310 Oslo, Norway d The Medical Faculty, University of Oslo, N-0310 Oslo, Norway e Department of Research Services, Oslo University Hospital, Rikshospitalet, N-0027 Oslo, Norway Received 21 October 2009; revised 17 December 2009; accepted 29 December 2009

Keywords: Peritoneal surface malignancy; Pseudomyxoma peritonei; Animal model; Immunohistochemistry

Summary Peritoneal surface malignancies are characterized by the propensity for tumor growth on peritoneal surfaces without development of extraperitoneal metastases, but the molecular basis for this phenomenon is incompletely understood. Five human tumors and corresponding orthotopic animal models of human pseudomyxoma peritonei and peritoneal mucinous carcinomatosis from colorectal carcinoma were extensively characterized by immunohistochemical analysis of molecular markers of tissue differentiation (carcinoembryonal antigen, CK20, CK7, and vimentin), proliferation and metastasis (Ki-67, vascular endothelial growth factor, and S100A4), mucins (MUC1, MUC2, MUC4, MUC5AC), and adhesion molecules (E-cadherin, N-cadherin, P-cadherin, claudin 1, claudin 3, and claudin 4). Macro- and microscopic growth patterns of implanted human tissues were preserved through passages in the animals, as were with few exception immunohistochemical staining profiles, supporting the relevance of the models as tools for studying the human disease. Tissue differentiation marker expression was in accordance with previously published results and high Ki-67 score confirmed high proliferative capacity, whereas absence of metastatic capacity was supported by low expression levels of the studied metastasis markers. These mucinous tumors expressed high levels of MUC2 and MUC4, whereas MUC1 was not expressed and MUC5AC expression was variable. Similarly, specific adhesion molecules from the cadherin and claudin families were shown to be of relevance in the investigated samples. The results indicate that mucinous peritoneal surface malignancies of intestinal origin are characterized by the presence of specific molecular markers and represent a step toward understanding the complexity of this intriguing phenotypic entity. © 2010 Elsevier Inc. All rights reserved.

☆ The work was supported by a postdoctoral grant from the Norwegian Foundation for Health and Rehabilitation (grant no. HR 2007/0305). ⁎ Corresponding author. Department of Tumor Biology, Institute for Cancer Research, The Norwegian Radium Hospital, 0310 Oslo, Norway. E-mail address: [email protected] (K. Flatmark).

0046-8177/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.humpath.2009.12.013

1110

1. Introduction The term peritoneal surface malignancy is used to describe isolated intraperitoneal (IP) tumor growth without development of extraperitoneal metastases and rests on the conception of the peritoneum as a barrier against disease spread [1]. The peritoneal cavity is then viewed as an isolated compartment, and localized disease may be amenable to potentially curative local treatment strategies, such as complete surgical removal of tumor tissue combined with local (IP) chemotherapy [2,3]. However, in, for instance, peritoneal carcinomatosis (PC) from colorectal cancer (CRC), only a small group of patients actually exhibit isolated peritoneal tumor growth and are candidates for such treatment [4,5]. The predilection for extensive growth on the peritoneal surfaces without metastasizing suggests the existence of a distinct phenotype that might be present in some tumors and absent in others and which might be explored by studying tumor cells growing in the peritoneal cavity. Pseudomyxoma peritonei (PMP) is a rare, clinically uniform but histologically heterogeneous, progressive malignant disease characterized by the accumulation of mucinous tumor tissue in the peritoneal cavity without metastasis development and, as such, a peritoneal surface malignancy in its purest form [6]. Abdominal distension because of large amounts of extracellular mucin is the main cause of morbidity, and untreated PMP will slowly lead to compression of intraabdominal organs and ultimately, the patients' demise. The primary lesion is in most cases a mucinous tumor of the appendix from which tumor cells are shed to the peritoneal cavity, but in some cases, clinical PMP may originate from mucinous tumors of the colorectum [7,8]. Current standard treatment is cytoreductive surgery combined with hyperthermic IP chemotherapy using the cytostatic agent mitomycin C, although the efficacy of adding IP chemotherapy has not been verified in randomized controlled trials. This treatment strategy has produced excellent long-term results in the most benign histologic variant, disseminated peritoneal adenomucinosis, whereas peritoneal mucinous carcinomatosis (PMCA) responds less well [9]. As clinical studies are challenging to perform in a disease as rare as PMP, the possibility of studying disease characteristics and treatment response in representative model systems is appealing. In the present study, we show that 5 models of PMP, established in nude mice from human tumors, have retained macro- and microscopic growth and similar immunohistochemical expression patterns compared to their human counterparts. Furthermore, a pilot study was performed using immunohistochemistry, examining the reactivity toward a panel of antibodies against markers of differentiation, proliferation and metastasis, mucins, and adhesion molecules, all of potential relevance for the peritoneal surface malignancy phenotype.

K. Flatmark et al.

2. Materials and methods 2.1. Patients Tumor tissue was acquired from patients undergoing surgical treatment at The Norwegian Radium Hospital (Oslo, Norway) after obtaining written informed consent. The study was approved by the Regional Committee for Medical Research Ethics of Southern Norway. The clinical presentation in all the patients was that of PMP with abdominal distension as one of the main symptoms, but the primary lesions and histopathologic characteristics varied as detailed below. 2.1.1. Patients 1 (PMP-1) and 2 (PMP-2) Both had appendiceal primary lesions and relatively benign histopathologic characteristics of PMCA of intermediate type (PMCA-I type), presented in detail [10]. 2.1.2. Patient 3 (PMCA-1) The surgical specimen used to establish the PMCA-1 model was obtained from a 68-year-old woman who received palliative cytoreduction with removal of large amounts of mucinous tumor tissue. Her primary lesion was a rectal carcinoma treated by abdominoperineal resection 4 years previously, followed by a local recurrence that was removed after preoperative radiotherapy with free surgical margins. One year subsequent to surgery for recurrent rectal carcinoma, she presented with monstrously distended abdomen caused by mucinous ascites. Curative surgery was deemed impossible, and repeated palliative debulking procedures were performed. 2.1.3. Patient 4 (PMCA-2) Tissue for the PMCA-2 model was collected from disseminated abdominal mucinous carcinomatosis in a 74-year-old man. The primary lesion was a mucinous adenocarcinoma of the descending colon that was removed by left-sided colectomy 3 years earlier. One year after primary surgery, peritoneal metastases were diagnosed. An attempt at curative surgery was performed followed by 5-fluorouracil and oxaliplatin-based postoperative systemic chemotherapy. Two years after this procedure, a bulky mucinous recurrence was diagnosed and explorative laparotomy revealed extensive disease not amenable to curative cytoreduction, and the current sample was retrieved at this procedure. 2.1.4. Patient 5 (PMCA-3) Patient 5 was a 62-year-old man who, 3 years before the main surgical procedure, had his appendix removed for suspected appendicitis, but the resected specimen was not evaluated by routine histologic examination. In the course of investigations for abdominal pain, computer tomographic scans revealed bulky mucinous tumor in the abdomen,

Exploring the peritoneal surface malignancy phenotype

1111

but no primary tumor was detected during subsequent preoperative workup. At surgery, disseminated peritoneal tumor was observed, in part consisting of loosely attached, mucinous tissue that was used for implantation in the animal model, whereas large, bulky masses contained solid tumor components. Complete cytoreduction was performed, but because one regional lymph node metastasis was detected, systemic rather than IP chemotherapy was administered postoperatively.

omy, and 6 tumor fragments were placed in the peritoneal cavity in upper and lower abdominal quadrants as well as in both flanks. After implantation, the abdominal wall was closed in 2 layers with Dexon 5/0. After a few IP passages, mucinous ascites dominated the growth pattern, and transfer of tissue was performed by IP injection of 250 μL mucinous ascites, and no laparotomy was required. The well-being of the mice was carefully monitored, and animals were killed if and when signs of disease were detected. The main indicator of IP tumor growth was increased abdominal volume, and animals were killed when a distinct increase of abdominal size was detectable, typically 1 to 3 months after implantation. Autopsy was performed and macroscopic assessment was made for the presence of tumor tissue or metastatic lesions and sampled for histologic examination.

2.2. Animals Locally bred female BALB/c nude (nu/nu) mice, 6 to 8 weeks of age at implantation, were used. The animals were maintained under specific pathogen-free conditions, and food and water were supplied ad libitum. Housing and all procedures involving animals were performed according to protocols approved by the animal care and use committee, in compliance with the National Committee for Animal Experiments guidelines on animal welfare. During surgical procedures, mice were anaesthetized using subcutaneous injections of 0.075 mL/10 g of a mixture of tiletamine, 2.4 mg/ mL, and zolazepam, 2.4 mg/mL (Zoletil vet, Virbac Laboratories, Carros, France); xylazine, 3.8 mg/mL (Narcoxyl vet, Roche, Basel, Switzerland); and butorphanol, 0.1 mg/mL (Torbugesic, Fort Dodge Laboratories, Fort Dodge, IA).

2.4. Histologic examination Formalin-fixed tissue from primary lesions (if available), main surgical specimens (from the later peritoneal involvement), and samples from tumor-bearing mice were hematoxylin-eosin stained using standard protocols. Histologic classification was performed according to Ronnett et al [9].

2.5. Immunohistochemistry Formalin-fixed paraffin-embedded sections selected from the main surgical specimen, and 3 murine passages (early, middle, and late) were chosen for immunohistochemical analysis. Antibodies were selected to examine 4 groups of

2.3. Animal experiments Fresh tumor tissue was cut into 3 × 3 × 3-mm pieces and implanted intraperitoneally through a small midline laparot-

Table 1

Antibodies and conditions used for immunohistochemistry

Target

M/P Clone

Source a

Dilution Pretreatment

Conditions Positive control

CK20 CK7 CEA Vimentin Ki-67 VEGF S100A4 E-Cadherin N-Cadherin P-Cadherin

M M M M M M P M M M

IT-Ks 20,8 OV-TL 12/30 II-7 Vim3B4 Ki-55 VG1 – HECD-1 3B9 56

1/100 1/100 1/100 1/200 1/25 1/50 1/500 1/100 1/100 1/100

10′ Pronase MWO, TrisEDTA, pH = 9.0 MWO, low pH, pH = 6.0 MWO, TrisEDTA, pH = 9.0 MWO, TrisEDTA, pH = 9.0 MWO, TrisEDTA, pH = 9.0 No pretreatment MWO, EDTA, pH = 8.0 MWO, EDTA, pH = 8.0 MWO, EDTA, pH = 8,0

30′, RT 30′, RT 30′, RT 30′, RT 30′, RT 30′, RT 30′, RT 30′, RT 30′, RT 30′, RT

Colon carcinoma Pancreas Colon carcinoma Appendix Tonsil Placenta Mesothelioma Normal skin Mesothelioma Ovarian carcinoma TMA

MUC1 MUC2 MUC4 MUC5AC Claudin 1 Claudin 3 Claudin 4

M M M M P P P

SPM492 M53 – 45MI – – –

Progen Dako Dako Dako Dako Dako Dako Zymed Zymed Transduction Laboratories Thermo Scientific Thermo Scientific Abcam Thermo Scientific Zymed Zymed Zymed

1/100 1/100 1/100 1/100 1/200 1/100 1/100

MWO, citrate, pH = 6.0 MWO, citrate, pH = 6.0 MWO, citrate, pH = 6.0 MWO, citrate, pH = 6.0 MWO, TrisEDTA, pH = 9.0 MWO, citrate, pH = 6.0 MWO, EDTA, pH = 8.0

30′, RT 30′, RT 30′, RT 30′, RT 30′, RT 30′, RT 30′, RT

Breast carcinoma Colon carcinoma Ovarian carcinoma TMA Colon carcinoma Ovarian carcinoma Breast carcinoma Ovarian carcinoma TMA

Abbreviations: M indicates monoclonal; P, polyclonal (rabbit); MWO, microwave oven; RT, room temperature; TMA, tissue microarray. a Progen Biotechnik GMBH, Heidelberg, Germany; Zymed Laboratories, San Francisco, CA; Transduction Laboratories, Lexington, KY; Abcam, Cambridge, United Kingdom; Thermo Scientific, Freemont, CA.

1112

K. Flatmark et al.

Table 2 Staining patterns considered relevant for examined immunohistochemical targets

3.2. Microscopic assessment

Immunohistochemical target Relevant staining pattern

We have previously described the microscopic evaluation of the PMP-1 and PMP-2 models [10]. Briefly, the primary tumors of PMP-1 and PMP-2 were appendiceal, one mucinous cystadenoma with low-grade atypia (PMP-1), and one mucinous adenocarcinoma (PMP-2), whereas peritoneal lesions from the main surgical specimens were in both cases classified as PMCA-intermediate features exhibiting moderately atypical glandular epithelium. The patient from which the PMCA-1 model was derived had a primary mucinous, moderately differentiated adenocarcinoma of the rectum, with infiltration of tumor cells into perirectal tissues (T3), which exhibited glandular formation with moderate atypia partially lining large pools of mucin (Fig. 1). The PMCA-2 patient had a primary mucinous colon adenocarcinoma infiltrating pericolonic fatty tissues (T3) with glandular and cribriform mucinous elements. Histologic examination of the main surgical specimens from PMCA-1 and PMCA-2 revealed extensive peritoneal lesions with invasive cribriform epithelial tumor foci and abundant pools of extracellular mucin, partially lined with atypical strips of glandular epithelium. Atypical glands were also found floating in the mucin, and in the case of PMCA-1, occasional signet ring cells were present. In both cases, the peritoneal lesions were classified as PMCA. All examined xenografts from animal passages from PMCA-1 and PMCA-2 exhibited the same histopathologic growth pattern as the peritoneal lesions, with moderately atypical glands and pools of mucin, locally lined with atypical glandular epithelium. In contrast to the PMCA-1 and 2 models, the primary lesion of the PMCA-3 model was most likely appendiceal, although the appendectomy specimen was unfortunately not examined histologically. The massive peritoneal lesions were characterized by signet ring cells, either as single cells or as small clusters of tumor cells surrounded by mucin (Fig. 1F and G). One periintestinal regional lymph node metastasis was found in the main surgical specimen. The animal passages from PMCA-3 displayed the same microscopic growth pattern with clusters of tumor cells differentiated as signet ring cells surrounded by mucin.

CK20 CK7 CEA Vimentin Ki67 VEGF S100A4 Mucins Cadherins Claudins

Cytoplasm and/or cell membrane Cytoplasm and/or cell membrane Cytoplasm and/or cell membrane Cytoplasm Nucleus Cytoplasm Cytoplasmic and nuclear staining Cytoplasm and/or cell membrane Cell membrane Cell membrane

markers: (i) tissue differentiation: CK20, CK7, and carcinoembryonal antigen (CEA); vimentin (ii) proliferation and metastasis: Ki-67, vascular endothelial growth factor (VEGF), and S100A4; (iii) mucins: MUC1, MUC2, MUC4, and MUC5AC; and (iv) cell adhesion: E-cadherin, N-cadherin, P-cadherin, claudin 1, claudin 3, and claudin 4 (for details see Table 1). Visualization was achieved using the EnVision + peroxidase system (Dako, Glostrup, Denmark). Negative control consisted of sections that underwent similar staining procedures with nonrelevant rabbit immunoglobulins or a monoclonal antibody of the same isotype as the primary antibody used. For all antibodies, positive controls were included with satisfactory results. The number of immunoreactive tumor cells was semiquantitatively evaluated: 0 (no positive cells); 1 = 1% to 5%; 2 = 6% to 25%; 3 = 26% to 75%; and 4 = 75% to 100%. Staining was considered positive only when unequivocally localized according to the staining patterns described in Table 2. Evaluation of immunoreactivity was performed independently by 2 surgical pathologists (BD and WR) with only minor disagreements with respect to scoring (interobserver agreement N 80%, disagreements solved by consensus).

3. Results 3.1. Tumor growth in animal models The initial growth pattern in all models was characterized by a combination of mucinous ascites and a variable number of “solid” mucinous tumor lesions of varying size. The solid tumor components were attached to the peritoneum and the serosa of IP organs, such as urinary bladder, omentum majus, ovaries, liver, and splenic hilum, but invasive growth was never observed. No metastatic lesions were observed in any animal throughout the entire series. After several passages, the growth pattern was increasingly dominated by mucinous ascites with less frequent appearance of lesions adhering to abdominal organs, and passage could be reliably performed using an injection strategy with take rates close to 100%.

3.3. Immunohistochemistry Nearly all sections were easily evaluated and the immunohistochemical reaction was generally remarkably conserved through xenograft passages compared to the main surgical specimens (Table 3 and Figs. 2 and 3). The main exception was passage 12 of the PMCA-3 model in which the differential staining between tumor cells and stroma cells was difficult to assess with some of the antibodies. To study tumor differentiation in patient tumors and xenograft passages, immunoreactivity toward antibodies against classic differentiation markers was assessed. The 2 intestinal tumor markers, CEA and CK20, were highly

Exploring the peritoneal surface malignancy phenotype

1113

Fig. 1 Microscopic growth assessed by hematoxylin-eosin staining. Top panels (A, B, and C) are from the PMCA-1 model and represent the primary rectal tumor, main surgical specimen, and xenograft, respectively. Arrows in panel 1B indicate small vessels in stromal tissue between mucinous areas. Panels D, E, and F show the primary colon tumor, main surgical specimen, and xenograft from PMCA-2, and panels G and F are from the main surgical specimen and xenograft of PMCA-3.

expressed in 4 of the 5 models (PMP-1, PMP-2, PMCA-2, and PMCA-3) in the main surgical specimens and all examined sections from corresponding animal passages, with membranous and cytoplasmic staining observed in more than 75% of tumor cells (Table 3 and Fig. 2B and C). The PMCA-1 model, which originated from a primary rectal adenocarcinoma, was negative for CK20, whereas CEA was highly expressed. CK7, a cytokeratin commonly associated with gynecologic malignancies, was essentially not detectable in the PMP-1, PMCA-2, and PMCA-3 models, whereas the PMP-2 and PMCA-1 models exhibited membranous CK7 expression in all examined sections. The mesenchymal

marker vimentin was not expressed in tumor cells of any of the models. Nuclear presence of the proliferation marker Ki-67 was detected in more than 25% of tumor cells in the PMP-1, PMCA-1, and PMCA-2 main surgical specimens, and this feature was conserved throughout the animal passages. In the PMP-2 and PMCA-3 models, pKi-67 expression was less uniform, with low and intermediate Ki-67-positive cell fractions in some examined sections. Staining for VEGF was negative in all sections from PMP-1, PMCA-2, and PMCA3, whereas sections from the late passages of the PMP-2 and PMCA-1 exhibited slight positivity for this marker (b25%

1114

Table 3

Scoring of immunoreactivity in examined sections Tissue differentiation

Proliferation and metastasis

Mucins

Cell adhesion

CK20 CK7 CEA Vimentin K i - VEGF S100A4 S100A4 MUC1 MUC2 MUC4 MUC5AC EPNClaudin Claudin Claudin 67 N C Cadherin Cadherin Cadherin 1 3 4 PMP-1

MSS E (p0) a M (p2) L (p15) PMP-2 MSS E (p0) M (p6) L (p14) PMCA-1 MSS E (p0) M (p7) L (p18) PMCA-2 MSS E (p0) M (p5) L (p17) PMCA-3 MSS E (p0) M (p3) L (p12)

4 4 4 4 4 4 4 4 0 0 0 0 4 4 4 4 4 4 4 4

0 0 0 0 4 4 4 4 4 3 4 4 0 0 0 0 1 2 0 NA

4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

3 3 3 3 2 3 2 3 3 3 3 3 3 3 4 4 3 1 0 3

0 0 0 0 0 0 0 2 0 0 0 1 0 0 0 0 0 0 0 0

2 0 0 0 0 0 2 1 0 0 0 0 0 0 0 0 0 0 0 2

2 3 1 1 0 0 2 1 0 0 1 2 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 NA

4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4

4 4 4 4 4 4 4 3 4 4 2 4 4 4 3 2 4 3 3 NA

1 4 4 4 4 4 4 NA 4 4 4 4 0 3 3 4 0 3 3 4

4 4 4 3 2 4 4 4 4 3 4 4 4 3 3 2 2 1 0 1

2 3 3 3 2 3 3 3 1 3 3 3 2 1 0 1 2 3 1 NA

0 0 0 0 0 0 NA 1 0 0 0 0 0 0 0 0 0 0 0 NA

4 4 4 4 4 4 4 4 4 4 4 4 4 4 3 4 3 2 2 4

0 1 2 2 2 2 3 4 1 4 3 3 2 3 3 3 0 0 3 0

4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 3 2 2 3 4

Abbreviation: The designation of E (early), M (middle), and L (late) refer to xenograft passages in mice of respective models with actual passage numbers in parentheses. NA indicates not assessable; MSS, main surgical specimen. a The terms early, middle, and late refer to xenograft passages in mice of respective models with actual passage numbers in parentheses.

K. Flatmark et al.

Exploring the peritoneal surface malignancy phenotype

1115

Fig. 2 Representative immunohistochemical reactivity in selected specimens. A represents CK7; B, CK20; C, CEA; D, vimentin; E, Ki-67; F, VEGF; G, E-cadherin; H, N-cadherin; and I, P-cadherin.

and b5% of the tumor cells were stained, respectively). To conclude the assessment of markers associated with proliferation and metastasis development, nuclear and cytoplasmic staining of S100A4 was registered. Most cases were completely negative for S100A4, with the exception of the main surgical specimen of PMP-1 and the middle and late passages of PMP-2 that exhibited both nuclear and cytoplasmic staining in 1% to 25% of the cells, some positive nuclei in the late passage of PMCA-3 (2+), and slight cytoplasmic staining of tumor cells from the middle and late passage of PMCA-1 (1+ and 2+). As mucin production is one of the predominant features of PMP and mucinous PC from CRC, reactivity toward a panel of mucins was investigated. MUC1 was not detected in any

of the models, with the exception of a few positive cells in the late passages of PMP-2 (1%-5%). In contrast, MUC2 and MUC4 were consistently and highly expressed in all surgical specimens and xenograft passages, with positivity rates of more than 75% and more than 25%, respectively. MUC5AC was detected in high amounts in the xenografts, whereas the main surgical specimens of PMP-1, PMCA-2, and PMCA-3 were essentially negative for MUC5AC reactivity. Of the cell adhesion proteins investigated, E-cadherin was generally highly expressed, mostly at scoring level 3 and 4, with the exception of the PMCA-3 model that exhibited low or no expression of this molecule. N-cadherin was essentially not expressed in any of the sections, although a few tumor cells in the late passage of PMP-2 were positive, whereas P-

1116

K. Flatmark et al.

Fig. 3 Representative immunohistochemical reactivity in selected specimens. A represents claudin 1; B, claudin 3; C, claudin 4; D, S100A4; E, MUC1; F, MUC2; G, MUC4; and H, MUC5AC.

cadherin was present in variable proportions of tumor cells in all the models. The expression of 3 members of the claudin family was also assessed, and claudins 1 and 4 were generally expressed at high levels in all models, whereas claudin 3 expression was more variable and generally less abundant. Notably, the PMCA-3 model had low levels of all the examined adhesion molecules compared to the other models.

4. Discussion The orthotopic animal models studied in the present work were all established by implanting tumor tissue from patients

with the clinical presentation of PMP, although the primary tumors had variable origin and microscopic characteristics. Interestingly, the propensity to grow intraperitoneally without invasive growth and metastasis development was preserved in the animals throughout the passages, even in the PMCA-1, 2, and 3 models that in fact by histologic assessment represent models of mucinous PC. As we previously published for the PMP-1 and PMP-2 models [10], the microscopic growth pattern of implanted tumor tissue was preserved in the xenografts, as was, with few exceptions, the immunohistochemical staining profile. All tumors could be passed to new generations of mice with almost 100% take rate by simple IP injections of mucinous ascites with reasonable lag times. The models have the

Exploring the peritoneal surface malignancy phenotype

1117

potential of being powerful tools for studying the peritoneal surface malignancy phenotype and exploring the potential value of treatment alternatives, and the present work focuses on the molecular characterization of these 5 human tumors and the derived xenografts. Cytokeratins CK7 and CK20, CEA, and vimentin are classic tissue markers used in routine histopathologic examination for classification of the origin of malignant lesions. The unequivocal CEA positivity is consistent with the intestinal origin of the tumors, whereas 2 models did not express CK20, which is another commonly used intestinal tumor marker. Interestingly, 2 of the models expressed CK7 in 50% to 100% of tumor cells in all sections, one concomitantly with CK20 (PMP-2), whereas one was completely negative for CK20 (PMCA-1). The results are in agreement with previously published findings that demonstrate that although CK20 is expressed in most mucinous intraabdominal malignancies of intestinal origin, it is not obligatory, and coexpression or isolated expression of CK7 is found in a subset of tumors [7,11,12]. The mesenchymal marker vimentin was not expressed in any of the sections, which is to be expected because these tumors are of epithelial origin. One remarkable feature of PMP is the propensity for extensive growth on the peritoneal surfaces without extraperitoneal metastases, thus, displaying high proliferative and low metastatic ability. The analysis of pKi-67 expression is a well-established method for determining the fraction of actively proliferating cells in a tissue, and the fraction of Ki-67 positive cells has been suggested as a marker to predict disease prognosis and therapy response [13]. In 16 of 20 examined sections, nuclear pKi-67 was detected in more than 25% of tumor cells, which clearly demonstrates the consistently high proliferative capacity of the tumors. The development of vasculature is essential for tumor growth beyond 1 to 2 mm and molecules involved in angiogenesis are being explored as potential targets for new anticancer drugs [14]. VEGF is implicated in the metastatic process, mainly through its functions as a regulator of angiogenesis, and VEGF inhibitors are currently being integrated into standard treatment of metastatic CRC [15]. In the main surgical specimens as well as in xenograft sections, stromal elements containing fibroblasts and small vessels were observed throughout the mucinous tissues, suggesting that angiogenesis must have occurred to establish the nutritional microenvironment necessary for tumor growth (Fig. 1B, arrows). However, VEGF expression was detected in only 2 examined sections, a finding that would be consistent with the nonmetastatic nature of these tumors, but which suggests that the observed vessel formation could have been promoted by other proangiogenic molecules or that VEGF, if involved, might be derived from other cells in the peritoneal cavity. Interestingly, and suggesting some involvement of VEGF, one case report describes long-term (24 months) stable disease in a

patient with advanced PMP treated with the VEGF inhibitor bevacizumab in combination with capecitabine [16]. Another metastasis-related protein that was recently shown to play a role in peritoneal dissemination of ovarian carcinomas is the small, acidic, calcium binding protein S100A4, belonging to the S100 protein family [17]. In CRC, nuclear but not cytoplasmic presence of S100A4 was associated with advanced tumor stage and adverse prognosis [18]. Only occasional tumor samples in our panel exhibited low-grade nuclear or cytoplasmic S100A4 staining, which is consistent with the conception of these lesions being localized IP malignancies with low or no metastatic potential. Mucinous ascites is responsible for one of the most apparent clinical features of PMP, namely the gross abdominal enlargement, rendering the composition of intracellular and secreted mucins an area of interest. The mucin protein family of carbohydrate-rich glycoproteins constitutes an integral part of mucosal barriers in among others the digestive tract, with MUC2 and MUC5AC belonging to the gel-forming and MUC1 and MUC4 the membrane-associated subgroups [19,20]. MUC2 is reported to be ubiquitously expressed in PMP [21], which is in accordance with the findings of the present study, as a consistently high expression of MUC2 was detected. In contrast, MUC1 was not expressed in implanted tumor tissue or xenografts, which is reasonable, as high MUC1 expression was previously detected in nonmucinous CRC but shown to be down-regulated in mucinous carcinomas [20]. MUC4 was expressed in all examined specimens, consistent with previous findings of up-regulation in peritoneal effusions in ovarian carcinoma compared to corresponding primary tumors and solid metastases [22,23]. In our study, MUC5AC was highly expressed in the xenografts of all models and in the main surgical specimens of PMP-2 and PMCA-1 but not in the surgical specimens of PMP-1, PMCA2, and PMCA-3. Lack of expression in 3 of 5 human samples is in contrast to the one previous report describing MUC5AC presence in 95 of 100 PMP samples [21], whereas in another study, expression was more variable with less than 50% of tumors exhibiting high levels of MUC5AC [24]. This finding represents the most conspicuous molecular discrepancy observed between implanted human tissues and xenografts, the only other involving claudin 3 that was also expressed to lesser extent in the original human tissue. One might speculate that the microenvironment of the murine peritoneal cavity may have influenced the expression of these molecules, and the findings underline the importance of animal model validation. In fact, the observed high degree of similarity between the xenografts and the human tissue counterparts is somewhat surprising but in agreement with the impression rendered from the “clinical presentation” and micro- and macroscopic growth pattern in the animals. Deregulation of cellular adhesion is a common feature of cancer, and as maintaining cellular adhesion must be essential to support extensive local growth without metastasis development,

1118 one would expect a number of adhesion molecules to be involved in peritoneal surface malignancies. Cadherins are Ca2+-dependent membrane proteins involved in cell-cell adhesion, with a central role in differentiation and tissue organization in the embryo and in maintaining adult tissue structure [25]. In epithelial-mesenchymal transition, which is thought to be a key process in metastasis development, Ecadherin expression is typically lost and may be followed by expression of N-cadherin, which is normally present only in neural and mesenchymal cells [26]. In our study, Ecadherin was consistently and strongly expressed in 4 of the 5 models, which is in line with E-cadherin's role as a tumor suppressor and that loss of E-cadherin correlates with dedifferentiation, invasive tumor growth, and metastasis development [27]. P-cadherin expression and function is less well described in cancer and its expression has never previously been studied in PMP or mucinous PC, but the molecule has been shown to be expressed in multiple adult tissues and overexpressed in tumors, often colocalizing with E-cadherin [28]. P-cadherin was present in low to moderate levels in most of the examined sections, suggesting that this molecule may also be relevant for adhesive properties in the studied tumors. As would be expected for these epithelialderived tumors, N-cadherin was generally not expressed or exhibited very slight staining. Our findings are somewhat in contrast to results from a previous study of 26 PMP samples, in which E-cadherin was expressed at lower and N-cadherin at higher levels than in simultaneously analyzed CRC samples and normal colon [11]. To determine the relevance of cadherins in PMP and mucinous PC, the expression of these molecules should be studied in larger clinical panels. Another family of adhesion proteins is the multigene claudin family, which plays an integral role in the formation and function of tight junctions, displaying organ- and tumor-specific expression [29-31]. Unlike many other tight junction molecules, claudins are highly expressed in many cancers, exemplified by claudins 1, 3, and 4 in CRC [32,33] and claudin 3 and 4 in ovarian cancer [34]. Interestingly, claudins 1, 3, and 7 were up-regulated in ovarian carcinoma effusions relative to their solid counterparts, with claudin 3 and 7 overexpression as independent predictors of adverse prognosis [35]. In our panel, claudin 1 and 4 were abundantly present in most cases, whereas claudin 3 was more variably expressed between and within series. Decreased expression of claudins would support the leading hypothesis that disruption of tight junctions and loss of cellcell adhesion would promote cancer progression, and increased expression of some claudins in cancer might be considered a paradox [32]. In CRC, one study reported that claudin 4 expression was decreased at the invasive front while well preserved at the luminal tumor surface and in metastatic lesions [36], whereas another group noted cytoplasmic and nuclear redistribution of claudin 1 associated with increased expression in primary tumors and metastatic lesions [37]. In the specimens studied in the

K. Flatmark et al. present work, claudin staining was generally quite uniform and strictly membranous, suggesting that these molecules might contribute to the noninvasive phenotype in accordance with the general hypothesis, although the relevance of claudin expression and function in peritoneal surface malignancies remains incompletely understood. Interestingly, expression of most of the adhesion molecules (claudins as well as the cadherins) was clearly lower in the PMCA-3 human tumor sample and xenografts, which might be related to the signet-ring cell differentiation that dominated all examined samples from this model. The results are consistent with previous reports describing low expression levels of E-cadherin in colorectal and gastric carcinomas with signet-ring cell differentiation [38,39]. The results from the present study first of all confirm the impression that these orthotopic models are indeed representative of the original human tumors and that they hold promise as potential tools for further studies of PMP, in particular, for assessment of in vivo drug sensitivity. Although the peritoneal surface malignancy phenotype remains elusive, findings from this pilot study including a limited number of samples suggest some directions for future studies in larger clinical panels. Glandular tumor elements lining mucin pools tended to retain adhesive properties, whereas the tumor model with predominant signet-ring cell differentiation exhibited loss of expression of several adhesion molecules. These interesting observations suggest that expression of adhesion molecules should be undertaken in larger patient cohorts to assess the potential value as prognostic markers as well as their contribution to the nonmetastatic phenotype of PMP and possibly other peritoneal surface malignancies. Another finding that deserves further investigations is the absence of VEGF expression in these highly vascularized tumors, as antiangiogenic therapy could potentially supplement traditional treatment of PMP.

References [1] Sugarbaker PH. Peritoneum as the first-line of defense in carcinomatosis. J Surg Oncol 2007;95:93-6. [2] Miner TJ, Shia J, Jaques DP, Klimstra DS, Brennan MF, Coit DG. Long-term survival following treatment of pseudomyxoma peritonei: an analysis of surgical therapy. Ann Surg 2005;241:300-8. [3] Smeenk RM, Verwaal VJ, Zoetmulder FA. Toxicity and mortality of cytoreduction and intraoperative hyperthermic intraperitoneal chemotherapy in pseudomyxoma peritonei–a report of 103 procedures. Eur J Surg Oncol 2006;32:186-90. [4] Esquivel J, Elias D, Baratti D, Kusamura S, Deraco M. Consensus statement on the loco regional treatment of colorectal cancer with peritoneal dissemination. J Surg Oncol 2008;98:263-7. [5] Verwaal VJ, Bruin S, Boot H, van Slooten G, van Tinteren H. 8-year follow-up of randomized trial: cytoreduction and hyperthermic intraperitoneal chemotherapy versus systemic chemotherapy in patients with peritoneal carcinomatosis of colorectal cancer. Ann Surg Oncol 2008;15:2426-32. [6] Smeenk RM, Bruin SC, van Velthuysen ML, Verwaal VJ. Pseudomyxoma peritonei. Curr Probl Surg 2008;45:527-75.

Exploring the peritoneal surface malignancy phenotype [7] Ronnett BM, Shmookler BM, Diener-West M, Sugarbaker PH, Kurman RJ. Immunohistochemical evidence supporting the appendiceal origin of pseudomyxoma peritonei in women. Int J Gynecol Pathol 1997;16:1-9. [8] Bradley RF, Stewart JHT, Russell GB, Levine EA, Geisinger KR. Pseudomyxoma peritonei of appendiceal origin: a clinicopathologic analysis of 101 patients uniformly treated at a single institution, with literature review. Am J Surg Pathol 2006;30:551-9. [9] Ronnett BM, Yan H, Kurman RJ, Shmookler BM, Wu L, Sugarbaker PH. Patients with pseudomyxoma peritonei associated with disseminated peritoneal adenomucinosis have a significantly more favorable prognosis than patients with peritoneal mucinous carcinomatosis. Cancer 2001;92:85-91. [10] Flatmark K, Reed W, Halvorsen T, et al. Pseudomyxoma peritonei— two novel orthotopic mouse models portray the PMCA-I histopathologic subtype. BMC Cancer 2007;7:116. [11] Bibi R, Pranesh N, Saunders MP, et al. A specific cadherin phenotype may characterise the disseminating yet non-metastatic behaviour of pseudomyxoma peritonei. Br J Cancer 2006;95:1258-64. [12] Guerrieri C, Franlund B, Fristedt S, Gillooley JF, Boeryd B. Mucinous tumors of the vermiform appendix and ovary, and pseudomyxoma peritonei: histogenetic implications of cytokeratin 7 expression. HUM PATHOL 1997;28:1039-45. [13] Brown DC, Gatter KC. Ki67 protein: the immaculate deception? Histopathol 2002;40:2-11. [14] Cao Y. Molecular mechanisms and therapeutic development of angiogenesis inhibitors. Adv Cancer Res 2008;100:113-31. [15] Hurwitz H. Integrating the anti-VEGF-A humanized monoclonal antibody bevacizumab with chemotherapy in advanced colorectal cancer. Clin Colorectal Cancer 2004;4(Suppl 2):S62-68. [16] Sun WL, Hutarew G, Gradl J, Gratzl M, Denz H, Fiegl M. Successful antiangiogenic combination therapy for pseudomyxoma peritonei with bevacizumab and capecitabine. Cancer Biol Ther 2009;8. [17] Maelandsmo GM, Florenes VA, Nguyen MT, Flatmark K, Davidson B. Different expression and clinical role of S100A4 in serous ovarian carcinoma at different anatomic sites. Tumour Biol 2009;30: 15-25. [18] Flatmark K, Pedersen KB, Nesland JM, et al. Nuclear localization of the metastasis-related protein S100A4 correlates with tumour stage in colorectal cancer. J Pathol 2003;200:589-95. [19] Nonaka D, Kusamura S, Baratti D, Casali P, Younan R, Deraco M. CDX-2 expression in pseudomyxoma peritonei: a clinicopathological study of 42 cases. Histopathol 2006;49:381-7. [20] Byrd JC, Bresalier RS. Mucins and mucin binding proteins in colorectal cancer. Cancer Metastasis Rev 2004;23:77-99. [21] O'Connell JT, Hacker CM, Barsky SH. MUC2 is a molecular marker for pseudomyxoma peritonei. Mod Pathol 2002;15:958-72. [22] Singh AP, Senapati S, Ponnusamy MP, et al. Clinical potential of mucins in diagnosis, prognosis, and therapy of ovarian cancer. Lancet Oncol 2008;9:1076-85.

1119 [23] Davidson B, Baekelandt M, Shih Ie M. MUC4 is upregulated in ovarian carcinoma effusions and differentiates carcinoma cells from mesothelial cells. Diagn Cytopathol 2007;35:756-60. [24] Baratti D, Kusamura S, Nonaka D, Cabras AD, Laterza B, Deraco M. Pseudomyxoma peritonei: biological features are the dominant prognostic determinants after complete cytoreduction and hyperthermic intraperitoneal chemotherapy. Ann Surg 2009;249:243-9. [25] Shirayoshi Y, Hatta K, Hosoda M, Tsunasawa S, Sakiyama F, Takeichi M. Cadherin cell adhesion molecules with distinct binding specificities share a common structure. Embo J 1986;5:2485-8. [26] Christofori G. New signals from the invasive front. Nature 2006;441: 444-50. [27] Christofori G, Semb H. The role of the cell-adhesion molecule Ecadherin as a tumour-suppressor gene. Trends Biochem Sci 1999;24: 73-6. [28] Paredes J, Correia A, Ribeiro A, Albergaria A, Milanezi F, Schmitt F. P-cadherin expression in breast cancer: a review. Breast Cancer Res 2007;9:214. [29] Hewitt KJ, Agarwal R, Morin PJ. The claudin gene family: expression in normal and neoplastic tissues. BMC Cancer 2006;6:186. [30] Swisshelm K, Macek R, Kubbies M. Role of claudins in tumorigenesis. Adv Drug Deliv Rev 2005;57:919-28. [31] Morin PJ. Claudin proteins in human cancer: promising new targets for diagnosis and therapy. Cancer Res 2005;65:9603-6. [32] de Oliveira SS, de Oliveira IM, De Souza W, Morgado-Diaz JA. Claudins upregulation in human colorectal cancer. FEBS Lett 2005; 579:6179-85. [33] Miwa N, Furuse M, Tsukita S, Niikawa N, Nakamura Y, Furukawa Y. Involvement of claudin-1 in the beta-catenin/Tcf signaling pathway and its frequent upregulation in human colorectal cancers. Oncol Res 2001; 12:469-76. [34] Morin PJ. Claudin proteins in ovarian cancer. Dis Markers 2007;23: 453-7. [35] Kleinberg L, Holth A, Trope CG, Reich R, Davidson B. Claudin upregulation in ovarian carcinoma effusions is associated with poor survival. HUM PATHOL 2008;39:747-57. [36] Ueda J, Semba S, Chiba H, et al. Heterogeneous expression of claudin4 in human colorectal cancer: decreased claudin-4 expression at the invasive front correlates cancer invasion and metastasis. Pathobiol 2007;74:32-41. [37] Dhawan P, Singh AB, Deane NG, et al. Claudin-1 regulates cellular transformation and metastatic behavior in colon cancer. J Clin Invest 2005;115:1765-76. [38] Börger ME, Gosens M, Jeuken JWM, et al. Signet ring cell differentiation in mucinous colorectal carcinoma. J Pathol 2007;212: 278-86. [39] Humar B, Blair V, Charlton A, More H, Martin I, Guilford P. Ecadherin deficiency initiates gastric signet-ring cell carcinoma in mice and man. Cancer Res 2009;69:2050-6.