ANTICANCER RESEARCH 25: 2705-2712 (2005)

Myofibroblasts Correlate with Lymphatic Microvessel Density and Lymph Node Metastasis in Early-stage Invasive Colorectal Carcinoma

PIN LIANG1, JIAN-WEI HONG2, HIDEYUKI UBUKATA1, GE LIU1, MOTONOBU KATANO1, GYO MOTOHASHI1, TERUHIKO KASUGA1, YOSHINORI WATANABE1, ICHIRO NAKADA1 and TAKAFUMI TABUCHI1

1Fourth Department of Surgery and 2Department of Pathology, Tokyo Medical University Kasumigaura Hospital, Ibaraki, Japan

Abstract. Background: Recent studies have shown that the Myofibroblasts are the main component cells in tumor interactions between tumor cells and stromal cells are stroma, and alpha-smooth muscle actin (·-SMA)-positive important in tumor development. A possible correlation myofibroblasts were found to participate in the synthesis of between tumor-activated myofibroblasts, the main component extracellular matrix components of tumor stroma, and to cells of tumor stroma, and lymphatic microvessel density produce lytic enzymes able to degrade the basement (LMVD) or other clinical parameters in carcinoma was membrane surrounding tumor glands. Although present in investigated. Materials and Methods: Immunohistochemical the progressive tumor nodules, they disappear during tumor examination of alpha-smooth muscle actin and podoplanin regression (5, 6). The correlation between microvessel were performed in 83 cases of early-stage invasive colorectal density and myofibroblasts was shown by Zidar et al. (7), but carcinoma. Results: There was a good correlation between the influence of myofibroblasts in lymphagiogenesis remains proliferation of myofibroblasts (PMpt) and LMVD (LMVDpt) unclear. Only recently, podoplanin, a 43-kd glomerular in the peri-tumoral area (p=0.0034). Increased PMpt was also podocyte membrane mucoprotein and a specific lymphatic associated with lymphatic invasion (p=0.0051) and with vessel marker, has enabled the investigation of the lymph node metastasis (p=0.011). However, proliferation of lymphatic microvessel density (LMVD) in paraffin- myofibroblasts in intra-tumoral (PMit) areas was not embedded sections (8, 9). associated with these clinical parameters. Conclusion: In this study, the correlation between the proliferation of Proliferation of myofibroblasts in peri-tumoral areas seem to ·-SMA-positive myofibroblasts (PM) and LMVD, as well as play an important role in lymphangiogenesis, and is also other clinical parameters in early-stage invasive colorectal associated with lymph node metastasis. carcinomas, were investigated.

The proliferation, invasion and metastasis of tumor cells Materials and Methods require complex interactions between the tumor cells and the surrounding matrix. Recent studies have shown that the Patients and tissues. Eighty-three patients with early-stage invasive interaction between tumor cells and stromal cells in colorectal carcinoma, defined as carcinoma that invades only the different organs is important in the development, submucosal layer of the bowel wall (10, 11), who underwent resection from 1985 to 2004 in our department, were studied. The differentiation and proliferation of tumor cells (1-4). group consisted of 57 men and 26 women, with a mean age 67.5 years (range 39~88). Of these, 64 tumors were in the colon and 19 in the rectum. The macroscopic and histological findings including gross type, depth of tumor invasion, histology, Correspondence to: Pin Liang, 3-20-1 Chuo, Ami, Inashiki, Ibaraki, lymphatic invasion and lymph node metastasis were evaluated 300-0395 Japan. Tel: +81-29-887-1161, Fax: +81-29-840-2089, e- based on the criteria of the Japanese Research Society for mail: [email protected] colorectal cancer (12). The depth of submucosal invasion was divided into three grades: (i) Sm1, carcinoma invading into the Key Words: Myofibroblast, microvessel density, early-stage upper one-third, thus showing minimum submucosal invasion; (ii) colorectal carcinoma, podoplanin, lymphangiogenesis, lymph node Sm2, carcinoma invading into the middle part of the submucosa; metastasis. (iii) Sm3, carcinoma invading into the lower layer of the

0250-7005/2005 $2.00+.40 2705 ANTICANCER RESEARCH 25: 2705-2712 (2005) submucosa (13). Of these 83 patients, 7 (8.4%) patients had Results lymph node metastasis. Myofibroblasts appeared as ·-SMA-positive spindle-shaped Immunohistochemistry. Immunostaining for ·-SMA and podoplanin stromal cells with long cytoplasmic extensions. Their was performed using the Envision+/HRP method with heat- induced antigen retrieval. One or two representative blocks of each distribution in normal mucosa was scarce (Figure 1), with tumor were selected for immunostaining analysis. Serial 3-Ìm periglandular myofibroblasts forming a layer around the paraffin sections containing the tumor margin were dewaxed in glandular epithelium, as described by Hewitt et al. (17). In xylene, rehydrated in alcohol, and for staining podoplanin the carcinomas, myofibroblasts were very much more abundant sections were heated to 95ÆC in an oven (650W) for 45 min to within the tumor stroma, in moderate to high abundance, reactivate the antigen. Endogenous peroxidase activity was and were stained more intensively in the intra-tumoral zone suppressed by a solution of 3% hydrogen peroxide in methanol for than in the peri-tumoral zone (Figure 2). The PMit was 20 min. After being rinsed three times in phosphate-buffered saline (PBS), sections were incubated for 90 min at room temperature identified in all 83 patients, while the PMpt was identified in with monoclonal antibody against ·-SMA (1A4, DAKO, Glostrup, 68 cases (82%). The Kruskal-Wallis test revealed a Denmark), or monoclonal antibody against podoplanin (11-003, significant difference in LMVDpt in cases with different AngioBio, Dal Mar, CA, USA). After washing in PBS, the sections degrees of PMpt (Figures 3, 4 and Table I). Subsequent were treated with goat anti-mouse immunoglobulins conjugated to Mann-Whitney U-tests revealed a significant difference in peroxidase labelled-dextran polymer (Dako, Carpinteria, CA, LMVDpt between cases with score+ and score+++ USA) for 1 h at room temperature. Then the sections were washed (p<0.001) and between cases with overall score++ and in PBS and developed in 0.05 M Tris-HCl buffer (PH=7.5) containing 0.6 mg/ml 3,3’-diaminobenzidine tetrahydrochloride score+++ (p<0.001). No significant difference in LMVDpt (DAB) for 4 min for myofibroblast staining and 10 min for between cases with overall score+ and score++ was podoplanin staining at room temperature. After washing in water, observed (p=0.203). The PMpt was also significantly the nuclei were counterstained with Mayer’s hematoxylin. Negative associated with lymphatic invasion and lymph node control sections were stained by omitting the primary antibody. metastasis (Figure 5 and Table I). In addition, multivariate Tumor size was measured by minor axis x major axis (mm2) on analysis of different effects, presumably associated with the hematoxylin and eosin specimen. Determination of PM was lymph node metastasis, is shown in Table II. As a result, the performed according to the Quickscore method (14). Briefly, the staining intensity was scored on a scale of 3 (1=weak, 2=moderate, degree of submucosal invasion lymphatic invasion and of 3=strong), and the proportion of the stroma in or adjacent to the PMpt emerged as independent predictors of lymph node tumor staining positively was scored out of 4 (1<25%, 2=25-50%, metastasis. The degree of PMit was positively related to that 3=51-75%, 4=76-100%).The score for intensity was added to the of PMpt (p<0.001 Spearman test), but no significant score for proportion to give a score in the range of 0-7 and grouped correlation was found between PMit and different clinical as –(score=0), +(score=1-3), ++(score=4-5), +++(score=6-7). parameters (Table I). Degrees of both intra-tumoral and peri-tumoral PM were scored As shown in Figures 4 (B) and 5, lymphatic capillaries on each section. Determination of LMVD was performed as suggested by were identified as weakly stained, inconspicuous vessels Weidner et al. and Straume et al. (15, 16). Briefly, the sections when stained with anti-podoplanin antibody. LMVDit was were scanned at low magnifications (x40 and x100) to identify the rare, whereas the majority of intra-tumor hot spots were tumor areas with the highest amount of lymphatic vessels ("hot most frequently found in close proximity to the tumor spots"). Within these areas, five fields at x200 magnification (high border, as reported by Peter et al. (18). Evaluation of the power field (HPF), 0.708 mm2/field) were examined, and the mean LMVDit gave a median count of 2.1 per millimeter (range value of these fields were calculated. Vessels more than one-half 1.0 to 9), while the median count of LMVDpt was 7.3 HPF (x200) away from the invasive front, were not counted. Any endothelial cell or cell cluster, highlighted by podoplanin reactivity (range 2.0 to 22). The difference was significant between and clearly separate from adjacent vessels, tumor cells and LMVDpt and LMVDit (p<0.01 Mann-Whitney test). connective tissue elements, was regarded as a distinct countable vessel (14). Special LMVD counts were established for LMVD in Discussion intra-tumoral (LMVDit) and LMVD in peri-tumoral (LMVDpt) areas for each case, and the counts are given as vessels per We set out to determine whether ·-SMA-positive PM is millimeter squared. The predominant appearance of peri-tumoral associated with LMVD as well as with other clinical lymphatic vessels was also recorded (compressed and/or angulated parameters in early-stage invasive colorectal carcinomas. versus rounded). In all 83 colorectal carcinomas, sections stained for ·-SMA and Myofibroblasts have heterogenous cytoskeletal phenotypes podoplanin were scored independently by Pin Liang and J. W Hong. with regard to their content of intermediate filaments (vementin and desmin), ·-SMA, ‚ actin, Á actin and myosin Statistics. The Kruskal-Wallis test , Mann-Whitney test, Spearman (19). Our study focused only on ·-SMA-positive stromal cells. test and logistic regression were used as appropriate. For all tests, a In colorectal carcinoma, myofibroblasts have been reported to two-sided value of 0.05 or less was considered statistically significant. proliferate at the invasive front, thereby altering the adhesive

2706 Liang et al: Myofibroblasts and Lymph Node Metastasis in Invasive Colorectal Carcinoma

Figure 1. Myofibroblasts detected by ·-smooth muscle actin (·-SMA). There is generally little evidence of myofibroblasts in the connective tissues of normal mucosa. Original magnification x100.

Figure 2. Myofibroblasts detected by ·-smooth muscle actin(·-SMA), are abundant and stained more intensively in the intra-tumor zone than in the peri- tumoral zone. Original magnification x100.

2707 ANTICANCER RESEARCH 25: 2705-2712 (2005)

Figure 3. Distribution of low proliferation peri-tumoral myofibroblasts accompanied with low lymphatic microvessel density (LMVD). ·-Smooth muscle actin (A) and podoplanin (B) detection on serial sections of early-stage of colon carcinoma. Original magnification x200.

2708 Liang et al: Myofibroblasts and Lymph Node Metastasis in Invasive Colorectal Carcinoma

Figure 4. Distribution of highly proliferating myofibroblasts (PMpt) accompanied with high lymphatic microvessel density (LMVDit). ·-Smooth muscle actin (A) and podoplanin (B) detected in serial sections of early-stage of colon carcinoma. Original magnification x200.

2709 ANTICANCER RESEARCH 25: 2705-2712 (2005)

Figure 5. The same sample of colon carcinoma shown in Figure 4. Lymphatic vessel invaded by tumor cells as assessed by immunostaining for podoplanin. Original magnification x400.

and migratory properties of colorectal carcinoma cells (5, 20). lymphatic vessels to retain their pattern in the center of Myofibroblasts were more abundant within the tumor stroma tumors. According to Curti et al. (23), interstitial pressure is than in normal tissues; furthermore they did not show a significantly higher within the tumor than outside. The uniform distribution. This is the first study where tumor- average size (cross-sectional area) of the lymphatic vessels activated PMpt was found to be significantly associated with in the center of tumors was found to be less than that in LMVDpt as well as with lymphatic invasion and lymph node peri-tumoral areas. LMVDpt is greater than LMVDit, and metastasis in early-stage invasive colorectal carcinoma. the fact that intra-tumor hot spots were most frequently Some earlier studies have proven that cancer cells and present in close proximity to the tumor border also support stromal cells could promote the secretion of pro-angiogenic our speculation. Padera et al. have also found that, although and anti-angiogenic molecules in human carcinoma. These there were fewer lymphatic vessels in intra-tumoral areas, pro-angiogenic molecules such as VEGF and VEGF-c they were not functional because of the higher pressure (24). secreted by tumor cells are believed to be the main factors The difference between PMpt and PMit with regard to inducing blood and lymph vessel neoangiogenesis (21, 22). This maturity and origin could also be responsible for their study has revealed that myofibroblasts in peri-tumoral areas different functions. The function of tumor-activated also play an important role in lymphangiogenesis, at least in myofibroblasts around the tumor cells differs with the early stage invasive colorectal carcinoma. passage of time. Recently, Bourreyron et al. have shown that Our analysis showed no association between PMit and myofibroblasts in culture before the fifth passage induce LMVDit (p=0.812 Kruskal-Wallis test). Furthermore, the LoVoC5 colon cancer cell proliferation, spread and score of PMit was higher than that of PMpt, whereas that of adhesion. After the 15th passage, myofibroblasts have been LMVDit was less than that of LMVDpt. These data could found to lose these functions (25). Studies have also shown be explained by the fact that the resistance of lymphatic that the origin of myofibroblasts at the invasive edge of a vessels, to pressure is lower than that of blood vessels which tumor and within tumors is different in early stage possess smooth muscle cells, thus rendering it difficult for colorectal carcinomas, and that myofibroblast maturity is

2710 Liang et al: Myofibroblasts and Lymph Node Metastasis in Invasive Colorectal Carcinoma

Table I. Correlation between proliferation of myofibroblast (PM) and clinical parameters.

No. of patients

Intra-tumoral PM Peri-tumoral PM

+ ++ +++ P-value - + ++ +++ P-value

Macroscopic typea Elevated 7 31 18 6 37 12 1 Depressed 8 12 5 0.081 8 13 4 0 0.079 Flat 1 1 0 1 1 0 0 Histological typeb Well-differentiated 11 21 12 0.409 11 25 8 0 0.144 Moderately-differentiated 5 23 11 4 26 8 1 Wall invasiona sm 1 5 18 7 7 17 6 0 sm2 4 11 6 0.786 4 12 5 0 0.608 sm3 7 15 10 4 22 5 1 Tumor sizeb 21mm or greater 7 35 17 0.093 5 42 11 0 0.097 less than 21 mm 9 9 6 10 9 5 1 Lymphatic invasionb positive 2 16 9 0.115 2 15 9 1 0.005 negative 14 28 14 13 36 7 0 Lymph node metastasisb positive 1 4 2 0.761 0 3 3 1 0.011 negative 15 40 21 15 48 13 0 LMVD(median±SD/mm2)a 6.4±4.3 6.8±3.3 7.7±4.8 0.725 5.7±3.3 6.4±3.4 10.1±4.7 14 0.003 aKruskal-Wallis test bMann-Whitney test

also different in different types of gastric carcinomas (26, Table II. Logistic regression analysis in relation to lymph node metastasis. 27). Our results revealed that the staining intensity of PMit Coefficient of ¯2 P-value was significantly higher than that of PMpt. Differences in regression maturity and origin may also lead to different PMpt and PMit functions in lymphangiogenesis. Further studies of the Histological type mechanism of function in lymphangiogenesis are required. well-differentiated vs. Our findings also indicate that the degree of PMpt, but moderately-differentiated –0.62 0.264 0.608 not of PMit, was closely linked to aggressive biological Wall invasion behavior, including lymphatic invasion and lymph node sm1 or sm2 vs. sm3 1.65 4.254 0.039 metastasis. Moreover, multivariate analysis demonstrated the degree of PMpt to be an independent predictor of lymph Lymphatic invasion node metastasis. It has been emphasized that myofibroblasts positive vs. negative 16.41 9.421 <0.001 have a role in the up-regulation of production of the LMVD –0.09 0.332 0.565 MMP-2 and U-PA enzymes, which have been proven to be connected with tumor invasion and metastasis (5, 28, 29). PMpt 2.21 4.381 0.036 Furthermore, the myofibroblast cells both in the intra- tumoral and peri-tumoral areas may prevent physical contact between tumor cells and immune cells, which is an essential phenomenon for the effective destruction of tumor cells (30). These results point to tumor cells defined by a higher peri-tumoral lymphangiogenesis, and that the degree of degree of PMpt and LMVDpt in peri-tumoral areas, and proliferation of myofibroblasts in peri-tumoral areas is a new more likely to invade adjacent normal and lymphatic tissue. and strong predictor of lymph node metastasis. Further In conclusion, the results of our study indicate that studies dealing with peri-tumoral myofibroblasts may suggest myofibroblasts in peri-tumoral areas seem to play a role in new therapeutic strategies to limit tumor spread.

2711 ANTICANCER RESEARCH 25: 2705-2712 (2005)

Acknowledgements 15 Weidner N, Semple JP, Welch WR and Folkman J: Tumor angiogenesis and metastasis-correlation in invasive breast The authors thank Kr. Tsutomu Kohda for technical assistance, carcinoma. N Engl J Med 324: 1-8, 1991. and thank Prof. J. Patrick Barron of the International Medical 16 Straume O, Jackson DG and Akslen LA: Independent Communications Center of Tokyo Medical University, Japan, for prognostic impact of lymphatic vessel density and presence of his review of this manuscript. low-grade lymphangiogenesis in cutaneous melanoma. Clin Cancer Res 9: 250-256, 2003. References 17 Hewitt RE, Powe DG, Carter GI and Turner DR: Desmoplasia and its relevance to colorectal tumor invasion. Int J Cancer 53: 62-69, 1993. 1 Hom YK, Young P, Wiesen JF et al: Uterine and vaginal organ 18 Peter B, Monika S, Andreas O, Gerhard B, Heinrich K, Georg growth requires epidermal growth factor receptor signaling O et al: Lymphatic microvessel density as a novel prognostic from stroma. Endocrinology 139: 913-921, 1998. factor in early-stage invasive cervical cancer. Int J Cancer 95: 2 Ogawa H, Iwaya K, Izumi M, Kuroda M, Serizawa H, Koyanagi 29-33, 2001. Y et al: Expression of CD10 by stromal cells during colorectal 19 Schurich W, Seemayer TA and Gbbiani G: The myofibroblast: tumor development. Hum Pathol 33: 806-811, 2002. a quarter century after its discovery. Am J Surg Pathol 22: 141- 3 Li C, Gardy R, Seor BK et al: Both high intratumoral 147, 1998. microvessel density determined using CD105 antibody and 20 Balazs M and Kovacs A: Electron microscopic study of elevated plasma levels of CD105 in colorectal cancer patients carcinoma of the colon. Exp Path 20: 203-214, 1981. correlate with poor prognosis. Br J Cancer 88: 1424-1431, 21 Fukumura D, Xavier R, Sugiura T, Chen Y, Park EC, Lu N, 2002. Selig M, Nielsen G, Taksir T, Jain RK and Seed R: Tumor 4 Nakada I, Tasaki T, Ubukata H, Goto Y, Wantanabe Y, Sato induction of VEGF promoter activity in stromal cells. Cell 94: S, Tabuchi T et al: Desmoplastic response in biopsy specimens 715-725, 1998. of early colorectal carcinoma is predictive of deep submucosal 22 Skobe M, Hawighorst T, Jackson DG, Prevo R, Janes L, invasion. Dis Colon Rectum 41: 896-900, 1998. Velasco P, Riccardi L, Alitalo K, Claffey K and Drtmar M: 5 Martin M, Pujuguet P and Martin F: Role of stromal Induction of tumor lymphangiogenesis by VEGF-C promoters myofibroblasts infiltrating colon cancer in tumor invasion. Path breast cancer metastasis. Nat Med 7: 192-198, 2001. Res Pract 192: 712-717, 1996. 23 Curti BD, Urba WJ, Alvord WG, Janik JE, Smith JW II, 6 Lieubeau B, Garrigue L, Barbieux I, Meflash K and Gregoire Madara K and Longo DL: Interstitial pressure of subcutaneous M: The role of transforming growth factor beta 1 in the nodules in melanoma and lymphoma patients: changes during fibroblastic reaction associated with rat colorectal tumor treatment. Cancer Res 53: 2204-2207, 1993. development. Cancer Res 54(24): 6526-6532, 1994. 24 Padera TP, Kadambi A, Tomaso ED, Carreira CM, Brown EB, 7 Zidar N, Gale N, Kambic V and Fischinger J: Proliferation of Boucher Y, Choi NC, Mathisen D and Wain J: Lymphatic myofibroblasts in the stroma of epithelial hyperplastic lesions metastasis in the absence of functional intratumor lymphatics. and squamous carcinoma of the larynx. Oncol 62: 381-385, Science 296: 1883-1886, 2002. 2002. 25 Pourreyron C, Dumortier J, Ratineau C, Nejjari M, Beatrix O, 8 Weninger W, Partanen TA, Breiteneder-Geleff S, Mayer C, Jacquier MF, Remy L, Chayvialle JA and Scoazec JY: Age- Kowalski H, Mildner M et al: Expression of vascular endothelial dependent variations of human and rat colon myofibroblasts in growth factor receptor-3 and podoplanin suggests a lymphatic culture: influence on their functional interactions with colon endothelial cell origin of Kaposi’s sarcoma tumor cells. Lab cancer cells. Int J Cancer 104(1): 28-35, 2003. Invest 79: 243-251, 1999. 26 Ban S, Kamada K, Mitsuki N, Goto Y, Shimizu Y, Tskahama M 9 Valencak J, Heere-Ress E, Kopp T, Schoppmann SF, Kittler H, and Shibata T: Phenotypic change of muscosae in early invasive Pehamberger H et al: Selective immunohistochemical staining colorectal adenocarcinoma. J Clin Pathol 53: 878-881, 2000. shows significant prognostic influence of lymphatic and blood 27 Nakayama H, Enzan H, Miyazaki E and Toi M: ·-Smooth vessels in patients with malignant melanoma. Eur J Cancer 40: muscle actin positive stromal cells in gastric carcinoma. J Clin 358-364, 2004. Pathol 55: 741-744, 2002. 10 Kitamura K, Taniguchi H, Yamaguchi T, Sawai K, Takahashi T 28 Koretz K, Moller P and Schwartz-Albiez R: Plasminogen et al: Clinical outcome of surgical treatment for invasive early activators and plasminogen inhibitors in human colorectal colorectal cancer in Japan. Hepatogast 44: 108-115, 1997. carcinoma tissues are not expressed by the tumor cells. Eur J 11 Hermanek P and Gall FP: Early (microinvasive) colorectal Cancer 29A: 1184-1189, 1993. carcinoma, pathology, diagnosis, surgical treatment. Int J 29 Pyke C, Ralfkiaer E, Tryggvason K and Dano K: Messenger Colorect Dis 1: 79-84, 1986. RNA for two type IV collagenases is located in stromal cells in 12 Japanese Society for Cancer of the Colon and Rectum. human colon adenocarcinomas. Am J Pathol 142: 359-365, 1993. Japanese Classification of Colorectal Carcinoma. First English 30 Brodsky JT, Richard GK, Cohen AM and Minisky BD: Edition. Kanehara & Co., Lid., Tokyo, 1997. Variables correlated with the risk of lymph node metastasis in 13 Kudo S: Endoscopic mucosal resection of flat and depressed early rectal cancer. Cancer 69(2): 322-326, 1992. type of early colorectal cancer. Endoscopy 25: 455-461, 1993. 14 Lee H, Douglas-Jones AG, Morgan JM and Jasani B: The effect of fixation and processing on the sensitivity of oestrogen receptor assay by immunohistochemistry in breast carcinoma. J Received December 21, 2004 Clin Pathol 55: 236-238, 2002. Accepted April 13, 2005

2712