- Email: [email protected]
The association between nm23-H1 expression and survival in patients with esophageal squamous cell carcinoma
Cancer Letters 138 (1999) 139±144
The association between nm23-H1 expression and survival in patients with esophageal squamous cell carcinoma Norio Iizuka a, b,*, Akira Tangoku b, Hiroto Hayashi b, Shigehumi Yosino b, Toshihiro Abe b, Takayuki Morioka b, Masaaki Oka b a
Department of Bioregulatory Function, Yamaguchi University School of Medicine, Ube, Yamaguchi 755-8505, Japan b Department of Surgery II, Yamaguchi University School of Medicine, Ube, Yamaguchi 755-8505, Japan Received 23 October 1998; received in revised form 7 December 1998; accepted 8 December 1998
Abstract The nm23 gene is a potential metastasis suppressor gene originally identi®ed using a murine melanoma cell line. The expression of nm23-H1 protein was examined immunohistochemically in 50 eligible patients with esophageal squamous cell carcinoma (ESCC). The expression was not correlated with other prognostic factors including lymph node metastases; however, overall survival rates of nm23-H1-negative patients were signi®cantly shorter than those of nm23-H1-positive patients (P , 0:05). Furthermore, reduced expression of nm23-H1 was associated with shorter overall survival in patients with involved lymph nodes (P , 0:01), but not in patients without involved lymph nodes. These data support the conclusion that reduced expression of nm23-H1 may be associated with poor prognosis of ESCC patients, suggesting the value of nm23-H1 expression as a prognostic marker for ESCC patients, especially ESCC patients with involved lymph nodes. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: NM23; Prognosis; Esophageal cancer; Metastasis
1. Introduction Although signi®cant progress has been made in the surgical treatment of esophageal squamous cell carcinoma (ESCC) as well as in the detection of early stage ESCC by diagnostic techniques, the prognosis of patients with ESCC remain poor. Furthermore, even if they have locoregional disease at diagnosis, about half of them die of metastatic relapse within the ®rst 2 years after tumor resection [1]. The high metastatic potential of ESCC may be attributable to the anatomical structure of the esophagus without serosa or to properties of this cancer cell type, indicating that it * Corresponding author. Tel.: 1 81-836-222194; fax: 1 81836-222195.
may be necessary to predict the metastatic potential of ESCC in order to determine a more effective therapy and to improve the poor prognosis of this disease. The nm23 gene was originally identi®ed by means of differential screening of a murine K1735 melanoma cell line cDNA library with RNA from cell lines of differing metastatic potentials [2]. Subsequently, several studies have demonstrated the favorable clinical outcome of overexpression of nm23-H1 in some tumors [3±5] but not other tumors [6±8]. On the other hand, there have been only a few studies on the role of nm23-H1 in esophageal squamous cell carcinoma in the past number of years [9]. In addition, con¯icting data were reported regarding squamous cell carcinomas derived from other organs [10,11]. Therefore,
0304-3835/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0304-383 5(98)00387-5
140
N. Iizuka et al. / Cancer Letters 138 (1999) 139±144
Fig. 1. Immunohistochemical staining for nm23-H1 protein in esophageal squamous cell carcinoma. A and B demonstrate staining of nm23-H1 protein in a representative nm23-H1-positive case and a representative nm23-H1-negative case, respectively. (original magni®cation for A, £200; B, £200).
in order to evaluate the clinical signi®cance of nm23H1 protein expression in ESCC, the present study was undertaken using the resected specimens of 50 ESCC patients.
and magnetic resonance imaging every 3 months. These imaging techniques revealed metastatic relapse or tumor progression of ESCC in 29 (58%) of the 50 patients and 26 of the 50 patients died of recurrence within the median follow-up period of 63 months (range: 21±105 months).
2. Materials and methods 2.1. Patients
2.2. Immunohistochemical analysis for NM23-H1 protein
The expression of nm23-H1 protein was examined immunohistochemically in 50 patients who underwent surgical treatment for ESCC in the Department of Surgery II, Yamaguchi University School of Medicine between April 1989 and July 1996. All patients had pathologically proven esophageal squamous cell carcinoma and were classi®ed according to the pathologic tumor node metastasis (TNM) classi®cation system [12]. None of them had preoperative radiotherapy or chemotherapy. Of the 50 patients examined, 30 (60%) had lymph node metastasis and 20 (40%) had no lymph node metastasis at surgery. Of the 30 patients with involved lymph nodes, 28 underwent cisplatin-based chemotherapy following resection and 2 did not undergo the adjuvant therapy at all. Of the 20 patients without involved lymph nodes, 8 underwent cisplatin-based chemotherapy and 4 underwent a different chemotherapy regimen or radiation therapy alone following resection. Eight did not undergo the adjuvant therapy at all. All patients were subjected to computed tomography
Following resection, tumor samples were ®xed in 10% formaldehyde solution and embedded in paraf®n. Sections (4 mm thick) were deparaf®nized in xylene and progressively rehydrated in decreasing concentrations of alcohol. Immunohistochemical staining was performed by the avidin±biotin af®nity method with the use of OmniTags (Lipshaw, Pittsburgh, PA). Brie¯y, the sections were immersed in protein blocking agent for 5 min to reduce non-speci®c staining and incubated at room temperature for 120 min with 1 mg/ml antihuman nm23-H1 monoclonal antibody (H1-229, Seikagaku Corp., Tokyo, Japan) [5,6]. These sections were washed in chilled PBS three times and were reacted with biotinylated polyvalent antibody at room temperature for 30 min. After being washed in PBS, the sections were reacted with streptavidin±alkaline phosphatase reagent at room temperature for 30 min. Finally, the sections were treated with fast red chromogen for 10 min. Normal mouse IgG was used as a negative control instead of the primary antibodies. The sections were counter-
N. Iizuka et al. / Cancer Letters 138 (1999) 139±144
141
Table 1 Relationship between nm23-H1 expression and clinicopathologic characteristics in eligible 50 ESCC patients Factors a Years , 60 ^ 60 Sex Male Female Location Upper thoracic Mid-thoracic Lower thoracic PT pT1 pT2 pT3 pT4 PN pN0 pN1 PM pM0 pM1 pTNM Stage I II III IV Differentiation Well Moderate Poor Resectability R0 R1 R2 a b
nm23-H1(1) (n 23)
nm23-H1(2) (n 27)
9 14
13 14
20 3
23 4
3 15 5
5 15 7
4 5 9 5
6 4 13 4
9 14
11 16
19 4
21 6
4 6 8 5
6 8 6 7
4 14 5
5 13 9
12 6 5
15 7 5
P value NS b NS NS
NS
NS NS NS
NS
NS
All patients were classi®ed according to the pathologic tumor node metastasis (TNM) classi®cation system. NS, not signi®cant.
stained lightly with hematoxylin. The expression of nm23-H1 protein was classi®ed into the following two grades: negative (weaker or equal intensity compared with stromal cells) and positive (staining was more intense than stromal cells) (Fig. 1).
Meier method using Cox±Mantel comparisons for statistical signi®cance [13,14].
2.3. Statistical analysis
Of 50 patients included in this study, 23 (46%) were positive for nm23-H1 staining and 27 (54%) were negative. As shown in Table 1. there was no signi®cant difference in the other clinicopathologic characteristics between nm23-H1-positive and nm23-H1negative groups. The overall survival curves for
The x 2 test with Yates' correction or Fisher's exact test were used to elucidate the correlation between nm23-H1 expression and clinicopathologic characteristics. All survival data were analyzed by the Kaplan±
3. Results
142
N. Iizuka et al. / Cancer Letters 138 (1999) 139±144
lymph nodes, overall 1-, 3- and 5-year survival rates of nm23-H1-negative patients (n 16) were 43.8, 12.5 and 0%, respectively. By contrast, those of nm23-H1-positive patients (n 14) were 85.7, 56.2 and 46.9%, respectively. Thus, reduced expression of nm23-H1 was associated with poor prognosis of ESCC patients with involved lymph nodes (Fig. 2B, P , 0:01). On the other hand, of the 20 ESCC patients without involved lymph nodes. the overall survival rate of nm23-H1-positive patients (n 9) was similar to that of nm23-H1-negative patients (n 11) (Fig. 2C). 4. Discussion
Fig. 2. Overall survival curves of ESCC patients according to nm23-H1 status. A, Overall survival (OS) curves of 50 ESCC patients according to nm23-H1 status. B and C demonstrate OS curves of 30 ESCC patients with involved lymph nodes and 20 ESCC patients without involved lymph nodes according to nm23H1 status respectively.
patients with positive or negative nm23-H1 staining are shown in Fig. 2A. Overall 1-, 3- and 5-year survival rates of nm23-H1-negative patients were 63.0, 36.7 and 32.6%, respectively. By contrast, those of nm23-H1-positive patients were 91.3, 67.9 and 62.2%, respectively. Thus, nm23-H1-negative patients had signi®cantly shorter overall survival than did nm23-H1-positive patients (Fig. 2A, P , 0:05). Of the 30 ESCC patients with involved
The present study showed that 23 (46%) of 50 ESCC patients were positive for nm23-H1 staining. However, Patel et al. have shown that 13 (29%) of 45 ESCC patients with T 3±4N 1M 0 were positive for nm23 staining [9]. The percentage of nm23-H1-positive cases in our series was rather similar to that found in patients with breast [4] and uterine cervical carcinoma [15], in which the antibody was the same as that used here. Therefore, this discrepancy may be due to difference in the speci®city of the antibody used. Our study showed that expression of nm-23H1 protein was associated inversely with overall survival rates of ESCC patients following resection. Steeg et al. reported that nm23 RNA levels in metastatic sites were lower than those in primary sites in NMUinduced rat mammary carcinomas [2]. This means that the reduction of nm23 expression plays an important role in the metastatic process. This concept was also supported by the results obtained from some immunohistochemical studies [16±18]. However, our preliminary study demonstrated that nm23-H1 protein levels in metastatic lymph nodes were almost consistent with those in primary tumors of patients with ESCC (data not shown). Moreover, in this study, it was con®rmed that nm23-H1 expression in ESCC was not related to lymph node metastasis, a representative marker for predicting the metastatic potential of ESCC [19] (Table 1). Our data also showed that reduced expression of nm23-H1 was associated with shorter overall survival in patients with involved lymph nodes (P , 0:01), but not with patients without involved lymph nodes. Thus, nm23-
N. Iizuka et al. / Cancer Letters 138 (1999) 139±144
H1 expression was associated with poor prognosis of advanced ESCCs with involved lymph nodes. Likewise, Patel et al. have reported that nm23 protein expression was associated with poor prognosis of ESCCs with T 3±4N 1M 0 [9]. Taken together, one possibility is that nm23-H1 in ESCC may play a role in the metastatic process following lymphatic spread. More interestingly, nm23-H1 has been shown to relate to sensitivity to cisplatin by transfection assay [20,21]. Scambia et al. have con®rmed a direct correlation between nm23-H1 expression and complete/partial responses of ovarian carcinoma to cisplatin-based chemotherapy by means of second-look laparotomy [22]. In this study, 28 (93.3%) of the 30 ESCC patients with involved lymph nodes underwent cisplatin-based chemotherapy following resection, suggesting that nm23-H1 status may be closely associated with drug sensitivity rather than metastatic potential in ESCC. In addition, a lack of correlation between nm23-H1 expression and prognosis in the 20 ESCC patients without involved lymph nodes may be due to the fact that only 8 (40%) of them underwent cisplatinbased chemotherapy following resection. Since our cohort was too small to clarify the correlation between nm23-H1 expression and clinical prognosis or responsiveness to cisplatin of patients with ESCC, further large studies should be required to clarify the role of nm23-H1 in both metastasis and drug sensitivity of ESCCs. Our data support the conclusion that reduced expression of nm23-H1 may be associated with poor prognosis of ESCC patients with involved lymph nodes; furthermore, analysis of nm23-H1 status in patients with ESCC will enable us to determine a more effective and suitable therapy.
Acknowledgements This work was supported in part by a Grant-in-Aid from the Ministry of Education, Science and Culture of Japan.
[2]
[3]
[4]
[5]
[6] [7]
[8] [9]
[10]
[11]
[12] [13] [14] [15]
References [1] J.D. Roder, R. Busch, H.J. Stein, U. Fink, J.R. Siewert, Ratio of invaded to removed lymph nodes as a predictor of survival
[16]
143
in squamous cell carcinoma of the oesphagus, Br. J. Surg. 81 (1994) 410±413. P.S. Steeg, G. Bevilacqua, L. Kopper, U.P. Thorgeirsson, J.B. Talmadge, L.A. Liotta, M. Sobel, Evidence for a novel gene associated with low tumor metastatic potential, J. Natl. Cancer Inst. 80 (1988) 200±204. V.A. Florenes, S. Aamdal, O. Myklebost, G. Maelandsmo, O.S. Bruland, O. Fodstad, Levels of nm23 messenger RNA in metastatic malignant melanomas: inverse correlation to disease progression, Cancer Res. 52 (1992) 6088±6091. Y. Tokunaga, T. Urano, K. Furukawa, H. Kondo, T. Kanematsu, H. Shiku, Reduced expression of nm23-H1, but not of nm23-H2 is concordant with the frequency of lymph node metastasis of human breast cancer, Int. J. Cancer 55 (1993) 66±71. N. Iizuka, M. Oka, T. Noma, A. Nakazawa, K. Hirose, T. Suzuki, NM23-H1 and NM23 H2 messenger RNA abundance in human hepatocellular carcinoma, Cancer Res. 55 (1995) 652±657. G. Lindmark, NM-23 H1 immunohistochemistry is not useful as predictor of metastatic potential of colorectal cancer, Br. J. Cancer 74 (1996) 1413±1418. M. Higashiyama, O. Doi, H. Yokouchi, K. Kodama, S. Nakamori, R. Tateishi, N. Kimura, Immunohistochemical analysis of nm23 gene product/NDP kinase expression in pulmonary adenocarcinoma: Lack of prognostic value, Br. J. Cancer 66 (1992) 533±536. C.S. Wang, K.H. Lin, Y.C. Hsu, S. Hsueh, Distant metastasis of gastric cancer is associated with elevated expression of the antimetastatic nm23 gene, Cancer Lett. 128 (1998) 23±29. D.D. Patel, J.M. Bhatavdekar, P.R. Chikhlikar, Y.V. Patel, N.G. Shah, N. Ghosh, T.P. Suthar, D.B. Balbar, Clinical signi®cance of pS3, nm23, and bc1-2 in T3-4N IMO oesophageal carcinoma: an immunohistochemical approach, J. Surg. Oncol. 65 (1997) 111±116. R. Holm, G. Tanum, Evaluation of the prognostic signi®cance of nm23/NDP kinase and cathepsin D in anal carcinomas. An immunohistochemical study, Virchows Arch. 428 (1996) 85± 89. S. Gazzeri, E. Brambilla, A. Negoescu, D. Thoraval, M. Veron, D. Moro, C. Brambilla, Overexpression of nucleoside diphosphate/kinase A/nm23-H1 protein in human lung tumors: association with tumor progression in squamous carcinoma, Lab. Invest. 74 (1996) 158±167. L.H. Sobin, C. Wittekind, UICC International Union Against Cancer. TNM Classi®cation of Malignant Tumors, 5th edn., Wiley-Liss, 1997 pp. 54-58. D.R. Cox, Regression models and life table, J. R. Stat. Soc. 34 (1972) 187±220. B.L. Kaplan, P. Meier, Nonparametric estimation from incomplete observations, J. Am. Stat. Assoc. 53 (1958) 457±481. M. Mandai, I. Konishi, M. Koshiyama, T. Komatsu, S. Yamamoto, K. Nanbu, T. Mori, M. Fukumoto, Altered expression of nm23-H1 and e-erbB-2 proteins have prognostic signi®cance in adenocarcinoma but not in squamous cell carcinoma of the uterine cervix, Cancer 75 (1995) 2523±2529. H. Nakayama, W. Yasui, H. Yokozaki, E. Tahara, Reduced
144
[17]
[18]
[19] [20]
N. Iizuka et al. / Cancer Letters 138 (1999) 139±144 expression of nm23 is associated with metastasis of human gastric carcinomas, Jpn. J. Cancer Res. 84 (1993) 184±190. N. Konishi, S. Nakaoka, T. Tsuzuki, K. Matsumoto, Y. Kitahori, Y. Hiasa, T. Urano, H. Shiku, Expression of nm23-H1 and nm23-H2 proteins in prostate carcinoma, Jpn. J. Cancer Res. 84 (1993) 1050±1054. T. Nakayama, A. Ohtsuru, K. Nakao, M. Shima, K. Nakata, K. Watanabe, N. Ishli, N. Kimura, S. Nagataki, Expression in human hepatocellular carcinoma of nucleoside diphosphate kinase, a homologue of the nm23 gene product, J. Natl. Cancer Inst. 84 (1992) 1349±1354. H. Aklyama, M. Tsurumaru, H. Udagawa, Y. Kajiyama, Radical lymph node dissection for cancer of the thoracic esophagus, Ann. Surg. 220 (1994) 364±372. A.W. Ferguson, U. Flatow, N.J. MacDonald, F. Larminat,
V.A. Bohr, P.S. Steeg, Increased sensitivity to cisplatin by nm23-transfected tumor cell lines, Cancer Res. 56 (1996) 2931±2935. [21] J.M. Freije, J.A. Lawrence, M.G. Hollingshead, A. De la Rosa, Y. Narayanan, M. Grever, E.A. Sausville, K. Paull, P.S. Steeg, Identi®cation of compounds with preferential inhibitory activity against low-Nm23-expressing human breast carcinoma and melanoma cell lines, Nature Med. 4 (1997) 395± 401. [22] G. Scambia, G. Ferrandina, M. Marone, P.P. Benedetti, C. Giannitelli, M. Piantelli, A. Leone, S. Mancuso, Nm23 in ovarian cancer: correlation with clinical outcome and other clinicopathologic and biochemical prognostic parameters J. Clin. Oncol. 14 (1996) 334±342.
The association between nm23-H1 expression and survival in patients with esophageal squamous cell carcinoma Norio Iizuka a, b,*, Akira Tangoku b, Hiroto Hayashi b, Shigehumi Yosino b, Toshihiro Abe b, Takayuki Morioka b, Masaaki Oka b a
Department of Bioregulatory Function, Yamaguchi University School of Medicine, Ube, Yamaguchi 755-8505, Japan b Department of Surgery II, Yamaguchi University School of Medicine, Ube, Yamaguchi 755-8505, Japan Received 23 October 1998; received in revised form 7 December 1998; accepted 8 December 1998
Abstract The nm23 gene is a potential metastasis suppressor gene originally identi®ed using a murine melanoma cell line. The expression of nm23-H1 protein was examined immunohistochemically in 50 eligible patients with esophageal squamous cell carcinoma (ESCC). The expression was not correlated with other prognostic factors including lymph node metastases; however, overall survival rates of nm23-H1-negative patients were signi®cantly shorter than those of nm23-H1-positive patients (P , 0:05). Furthermore, reduced expression of nm23-H1 was associated with shorter overall survival in patients with involved lymph nodes (P , 0:01), but not in patients without involved lymph nodes. These data support the conclusion that reduced expression of nm23-H1 may be associated with poor prognosis of ESCC patients, suggesting the value of nm23-H1 expression as a prognostic marker for ESCC patients, especially ESCC patients with involved lymph nodes. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: NM23; Prognosis; Esophageal cancer; Metastasis
1. Introduction Although signi®cant progress has been made in the surgical treatment of esophageal squamous cell carcinoma (ESCC) as well as in the detection of early stage ESCC by diagnostic techniques, the prognosis of patients with ESCC remain poor. Furthermore, even if they have locoregional disease at diagnosis, about half of them die of metastatic relapse within the ®rst 2 years after tumor resection [1]. The high metastatic potential of ESCC may be attributable to the anatomical structure of the esophagus without serosa or to properties of this cancer cell type, indicating that it * Corresponding author. Tel.: 1 81-836-222194; fax: 1 81836-222195.
may be necessary to predict the metastatic potential of ESCC in order to determine a more effective therapy and to improve the poor prognosis of this disease. The nm23 gene was originally identi®ed by means of differential screening of a murine K1735 melanoma cell line cDNA library with RNA from cell lines of differing metastatic potentials [2]. Subsequently, several studies have demonstrated the favorable clinical outcome of overexpression of nm23-H1 in some tumors [3±5] but not other tumors [6±8]. On the other hand, there have been only a few studies on the role of nm23-H1 in esophageal squamous cell carcinoma in the past number of years [9]. In addition, con¯icting data were reported regarding squamous cell carcinomas derived from other organs [10,11]. Therefore,
0304-3835/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0304-383 5(98)00387-5
140
N. Iizuka et al. / Cancer Letters 138 (1999) 139±144
Fig. 1. Immunohistochemical staining for nm23-H1 protein in esophageal squamous cell carcinoma. A and B demonstrate staining of nm23-H1 protein in a representative nm23-H1-positive case and a representative nm23-H1-negative case, respectively. (original magni®cation for A, £200; B, £200).
in order to evaluate the clinical signi®cance of nm23H1 protein expression in ESCC, the present study was undertaken using the resected specimens of 50 ESCC patients.
and magnetic resonance imaging every 3 months. These imaging techniques revealed metastatic relapse or tumor progression of ESCC in 29 (58%) of the 50 patients and 26 of the 50 patients died of recurrence within the median follow-up period of 63 months (range: 21±105 months).
2. Materials and methods 2.1. Patients
2.2. Immunohistochemical analysis for NM23-H1 protein
The expression of nm23-H1 protein was examined immunohistochemically in 50 patients who underwent surgical treatment for ESCC in the Department of Surgery II, Yamaguchi University School of Medicine between April 1989 and July 1996. All patients had pathologically proven esophageal squamous cell carcinoma and were classi®ed according to the pathologic tumor node metastasis (TNM) classi®cation system [12]. None of them had preoperative radiotherapy or chemotherapy. Of the 50 patients examined, 30 (60%) had lymph node metastasis and 20 (40%) had no lymph node metastasis at surgery. Of the 30 patients with involved lymph nodes, 28 underwent cisplatin-based chemotherapy following resection and 2 did not undergo the adjuvant therapy at all. Of the 20 patients without involved lymph nodes, 8 underwent cisplatin-based chemotherapy and 4 underwent a different chemotherapy regimen or radiation therapy alone following resection. Eight did not undergo the adjuvant therapy at all. All patients were subjected to computed tomography
Following resection, tumor samples were ®xed in 10% formaldehyde solution and embedded in paraf®n. Sections (4 mm thick) were deparaf®nized in xylene and progressively rehydrated in decreasing concentrations of alcohol. Immunohistochemical staining was performed by the avidin±biotin af®nity method with the use of OmniTags (Lipshaw, Pittsburgh, PA). Brie¯y, the sections were immersed in protein blocking agent for 5 min to reduce non-speci®c staining and incubated at room temperature for 120 min with 1 mg/ml antihuman nm23-H1 monoclonal antibody (H1-229, Seikagaku Corp., Tokyo, Japan) [5,6]. These sections were washed in chilled PBS three times and were reacted with biotinylated polyvalent antibody at room temperature for 30 min. After being washed in PBS, the sections were reacted with streptavidin±alkaline phosphatase reagent at room temperature for 30 min. Finally, the sections were treated with fast red chromogen for 10 min. Normal mouse IgG was used as a negative control instead of the primary antibodies. The sections were counter-
N. Iizuka et al. / Cancer Letters 138 (1999) 139±144
141
Table 1 Relationship between nm23-H1 expression and clinicopathologic characteristics in eligible 50 ESCC patients Factors a Years , 60 ^ 60 Sex Male Female Location Upper thoracic Mid-thoracic Lower thoracic PT pT1 pT2 pT3 pT4 PN pN0 pN1 PM pM0 pM1 pTNM Stage I II III IV Differentiation Well Moderate Poor Resectability R0 R1 R2 a b
nm23-H1(1) (n 23)
nm23-H1(2) (n 27)
9 14
13 14
20 3
23 4
3 15 5
5 15 7
4 5 9 5
6 4 13 4
9 14
11 16
19 4
21 6
4 6 8 5
6 8 6 7
4 14 5
5 13 9
12 6 5
15 7 5
P value NS b NS NS
NS
NS NS NS
NS
NS
All patients were classi®ed according to the pathologic tumor node metastasis (TNM) classi®cation system. NS, not signi®cant.
stained lightly with hematoxylin. The expression of nm23-H1 protein was classi®ed into the following two grades: negative (weaker or equal intensity compared with stromal cells) and positive (staining was more intense than stromal cells) (Fig. 1).
Meier method using Cox±Mantel comparisons for statistical signi®cance [13,14].
2.3. Statistical analysis
Of 50 patients included in this study, 23 (46%) were positive for nm23-H1 staining and 27 (54%) were negative. As shown in Table 1. there was no signi®cant difference in the other clinicopathologic characteristics between nm23-H1-positive and nm23-H1negative groups. The overall survival curves for
The x 2 test with Yates' correction or Fisher's exact test were used to elucidate the correlation between nm23-H1 expression and clinicopathologic characteristics. All survival data were analyzed by the Kaplan±
3. Results
142
N. Iizuka et al. / Cancer Letters 138 (1999) 139±144
lymph nodes, overall 1-, 3- and 5-year survival rates of nm23-H1-negative patients (n 16) were 43.8, 12.5 and 0%, respectively. By contrast, those of nm23-H1-positive patients (n 14) were 85.7, 56.2 and 46.9%, respectively. Thus, reduced expression of nm23-H1 was associated with poor prognosis of ESCC patients with involved lymph nodes (Fig. 2B, P , 0:01). On the other hand, of the 20 ESCC patients without involved lymph nodes. the overall survival rate of nm23-H1-positive patients (n 9) was similar to that of nm23-H1-negative patients (n 11) (Fig. 2C). 4. Discussion
Fig. 2. Overall survival curves of ESCC patients according to nm23-H1 status. A, Overall survival (OS) curves of 50 ESCC patients according to nm23-H1 status. B and C demonstrate OS curves of 30 ESCC patients with involved lymph nodes and 20 ESCC patients without involved lymph nodes according to nm23H1 status respectively.
patients with positive or negative nm23-H1 staining are shown in Fig. 2A. Overall 1-, 3- and 5-year survival rates of nm23-H1-negative patients were 63.0, 36.7 and 32.6%, respectively. By contrast, those of nm23-H1-positive patients were 91.3, 67.9 and 62.2%, respectively. Thus, nm23-H1-negative patients had signi®cantly shorter overall survival than did nm23-H1-positive patients (Fig. 2A, P , 0:05). Of the 30 ESCC patients with involved
The present study showed that 23 (46%) of 50 ESCC patients were positive for nm23-H1 staining. However, Patel et al. have shown that 13 (29%) of 45 ESCC patients with T 3±4N 1M 0 were positive for nm23 staining [9]. The percentage of nm23-H1-positive cases in our series was rather similar to that found in patients with breast [4] and uterine cervical carcinoma [15], in which the antibody was the same as that used here. Therefore, this discrepancy may be due to difference in the speci®city of the antibody used. Our study showed that expression of nm-23H1 protein was associated inversely with overall survival rates of ESCC patients following resection. Steeg et al. reported that nm23 RNA levels in metastatic sites were lower than those in primary sites in NMUinduced rat mammary carcinomas [2]. This means that the reduction of nm23 expression plays an important role in the metastatic process. This concept was also supported by the results obtained from some immunohistochemical studies [16±18]. However, our preliminary study demonstrated that nm23-H1 protein levels in metastatic lymph nodes were almost consistent with those in primary tumors of patients with ESCC (data not shown). Moreover, in this study, it was con®rmed that nm23-H1 expression in ESCC was not related to lymph node metastasis, a representative marker for predicting the metastatic potential of ESCC [19] (Table 1). Our data also showed that reduced expression of nm23-H1 was associated with shorter overall survival in patients with involved lymph nodes (P , 0:01), but not with patients without involved lymph nodes. Thus, nm23-
N. Iizuka et al. / Cancer Letters 138 (1999) 139±144
H1 expression was associated with poor prognosis of advanced ESCCs with involved lymph nodes. Likewise, Patel et al. have reported that nm23 protein expression was associated with poor prognosis of ESCCs with T 3±4N 1M 0 [9]. Taken together, one possibility is that nm23-H1 in ESCC may play a role in the metastatic process following lymphatic spread. More interestingly, nm23-H1 has been shown to relate to sensitivity to cisplatin by transfection assay [20,21]. Scambia et al. have con®rmed a direct correlation between nm23-H1 expression and complete/partial responses of ovarian carcinoma to cisplatin-based chemotherapy by means of second-look laparotomy [22]. In this study, 28 (93.3%) of the 30 ESCC patients with involved lymph nodes underwent cisplatin-based chemotherapy following resection, suggesting that nm23-H1 status may be closely associated with drug sensitivity rather than metastatic potential in ESCC. In addition, a lack of correlation between nm23-H1 expression and prognosis in the 20 ESCC patients without involved lymph nodes may be due to the fact that only 8 (40%) of them underwent cisplatinbased chemotherapy following resection. Since our cohort was too small to clarify the correlation between nm23-H1 expression and clinical prognosis or responsiveness to cisplatin of patients with ESCC, further large studies should be required to clarify the role of nm23-H1 in both metastasis and drug sensitivity of ESCCs. Our data support the conclusion that reduced expression of nm23-H1 may be associated with poor prognosis of ESCC patients with involved lymph nodes; furthermore, analysis of nm23-H1 status in patients with ESCC will enable us to determine a more effective and suitable therapy.
Acknowledgements This work was supported in part by a Grant-in-Aid from the Ministry of Education, Science and Culture of Japan.
[2]
[3]
[4]
[5]
[6] [7]
[8] [9]
[10]
[11]
[12] [13] [14] [15]
References [1] J.D. Roder, R. Busch, H.J. Stein, U. Fink, J.R. Siewert, Ratio of invaded to removed lymph nodes as a predictor of survival
[16]
143
in squamous cell carcinoma of the oesphagus, Br. J. Surg. 81 (1994) 410±413. P.S. Steeg, G. Bevilacqua, L. Kopper, U.P. Thorgeirsson, J.B. Talmadge, L.A. Liotta, M. Sobel, Evidence for a novel gene associated with low tumor metastatic potential, J. Natl. Cancer Inst. 80 (1988) 200±204. V.A. Florenes, S. Aamdal, O. Myklebost, G. Maelandsmo, O.S. Bruland, O. Fodstad, Levels of nm23 messenger RNA in metastatic malignant melanomas: inverse correlation to disease progression, Cancer Res. 52 (1992) 6088±6091. Y. Tokunaga, T. Urano, K. Furukawa, H. Kondo, T. Kanematsu, H. Shiku, Reduced expression of nm23-H1, but not of nm23-H2 is concordant with the frequency of lymph node metastasis of human breast cancer, Int. J. Cancer 55 (1993) 66±71. N. Iizuka, M. Oka, T. Noma, A. Nakazawa, K. Hirose, T. Suzuki, NM23-H1 and NM23 H2 messenger RNA abundance in human hepatocellular carcinoma, Cancer Res. 55 (1995) 652±657. G. Lindmark, NM-23 H1 immunohistochemistry is not useful as predictor of metastatic potential of colorectal cancer, Br. J. Cancer 74 (1996) 1413±1418. M. Higashiyama, O. Doi, H. Yokouchi, K. Kodama, S. Nakamori, R. Tateishi, N. Kimura, Immunohistochemical analysis of nm23 gene product/NDP kinase expression in pulmonary adenocarcinoma: Lack of prognostic value, Br. J. Cancer 66 (1992) 533±536. C.S. Wang, K.H. Lin, Y.C. Hsu, S. Hsueh, Distant metastasis of gastric cancer is associated with elevated expression of the antimetastatic nm23 gene, Cancer Lett. 128 (1998) 23±29. D.D. Patel, J.M. Bhatavdekar, P.R. Chikhlikar, Y.V. Patel, N.G. Shah, N. Ghosh, T.P. Suthar, D.B. Balbar, Clinical signi®cance of pS3, nm23, and bc1-2 in T3-4N IMO oesophageal carcinoma: an immunohistochemical approach, J. Surg. Oncol. 65 (1997) 111±116. R. Holm, G. Tanum, Evaluation of the prognostic signi®cance of nm23/NDP kinase and cathepsin D in anal carcinomas. An immunohistochemical study, Virchows Arch. 428 (1996) 85± 89. S. Gazzeri, E. Brambilla, A. Negoescu, D. Thoraval, M. Veron, D. Moro, C. Brambilla, Overexpression of nucleoside diphosphate/kinase A/nm23-H1 protein in human lung tumors: association with tumor progression in squamous carcinoma, Lab. Invest. 74 (1996) 158±167. L.H. Sobin, C. Wittekind, UICC International Union Against Cancer. TNM Classi®cation of Malignant Tumors, 5th edn., Wiley-Liss, 1997 pp. 54-58. D.R. Cox, Regression models and life table, J. R. Stat. Soc. 34 (1972) 187±220. B.L. Kaplan, P. Meier, Nonparametric estimation from incomplete observations, J. Am. Stat. Assoc. 53 (1958) 457±481. M. Mandai, I. Konishi, M. Koshiyama, T. Komatsu, S. Yamamoto, K. Nanbu, T. Mori, M. Fukumoto, Altered expression of nm23-H1 and e-erbB-2 proteins have prognostic signi®cance in adenocarcinoma but not in squamous cell carcinoma of the uterine cervix, Cancer 75 (1995) 2523±2529. H. Nakayama, W. Yasui, H. Yokozaki, E. Tahara, Reduced
144
[17]
[18]
[19] [20]
N. Iizuka et al. / Cancer Letters 138 (1999) 139±144 expression of nm23 is associated with metastasis of human gastric carcinomas, Jpn. J. Cancer Res. 84 (1993) 184±190. N. Konishi, S. Nakaoka, T. Tsuzuki, K. Matsumoto, Y. Kitahori, Y. Hiasa, T. Urano, H. Shiku, Expression of nm23-H1 and nm23-H2 proteins in prostate carcinoma, Jpn. J. Cancer Res. 84 (1993) 1050±1054. T. Nakayama, A. Ohtsuru, K. Nakao, M. Shima, K. Nakata, K. Watanabe, N. Ishli, N. Kimura, S. Nagataki, Expression in human hepatocellular carcinoma of nucleoside diphosphate kinase, a homologue of the nm23 gene product, J. Natl. Cancer Inst. 84 (1992) 1349±1354. H. Aklyama, M. Tsurumaru, H. Udagawa, Y. Kajiyama, Radical lymph node dissection for cancer of the thoracic esophagus, Ann. Surg. 220 (1994) 364±372. A.W. Ferguson, U. Flatow, N.J. MacDonald, F. Larminat,
V.A. Bohr, P.S. Steeg, Increased sensitivity to cisplatin by nm23-transfected tumor cell lines, Cancer Res. 56 (1996) 2931±2935. [21] J.M. Freije, J.A. Lawrence, M.G. Hollingshead, A. De la Rosa, Y. Narayanan, M. Grever, E.A. Sausville, K. Paull, P.S. Steeg, Identi®cation of compounds with preferential inhibitory activity against low-Nm23-expressing human breast carcinoma and melanoma cell lines, Nature Med. 4 (1997) 395± 401. [22] G. Scambia, G. Ferrandina, M. Marone, P.P. Benedetti, C. Giannitelli, M. Piantelli, A. Leone, S. Mancuso, Nm23 in ovarian cancer: correlation with clinical outcome and other clinicopathologic and biochemical prognostic parameters J. Clin. Oncol. 14 (1996) 334±342.