Correlation between Expression of EGFR and the Prognosis of Patients with Cervical Carcinoma

Gynecologic Oncology 87, 84 – 89 (2002) doi:10.1006/gyno.2002.6803

Correlation between Expression of EGFR and the Prognosis of Patients with Cervical Carcinoma Young Tae Kim,* ,1 Sang Won Park,† and Jae Wook Kim† *Department of Obstetrics and Gynecology, BK21 Project for Medical Sciences, Yonsei University College of Medicine, Seoul, Korea; and †Department of Obstetrics and Gynecology, College of Medicine Pochon CHA University, Seoul, Korea Received February 7, 2002; published online September 6, 2002

the quantitation of their expressed products will undoubtedly find increased clinical utility [1–3]. Epidermal growth factor (EGF) is one of the well-characterized growth factors which interacts with specific cell-surface receptors, transducing intracellular signals to stimulate DNA synthesis and cell division [4, 5]. Epidermal growth factor receptor (EGFR) is a 170-kDa membrane-bound glycoprotein encoded by the c-erbB-1 oncogene. It was first purified from the A431 cell line which was derived from an epidermoid carcinoma of the vulva [6 – 8]. EGFR is a member of the growth factor receptor family of protein tyrosine kinases, a class of cell cycle regulatory molecules. The gene for the human EGFR is located on chromosome 7 [9]. The receptor is capable of binding several ligands such as EGF, transforming growth factor (TGF) ␣ [3], amphiregulin [10], and heparinbinding EGF [11]. It is activated when its ligand binds to the extracellular domain, resulting in autophosphorylation of the receptor’s intracellular tyrosine kinase domain [6, 12]. EGFR has been investigated by several methods, such as radioligand-binding assays, immunohistochemistry, enzyme immunoassay, and multiparameter flow cytometry [13–17]. We previously confirmed the high level of EGFR expression in carcinoma of the uterine cervix by employing an enzyme immunoassay [18, 19]. Overexpression of EGFR has been demonstrated on many human tumors, and increase in receptor expression levels has been linked with a poor clinical outcome [20 –22]. The prognosis of cervical cancer patients seems to be significantly affected by clinicopathological parameters, which are frequently intercorrelated, but a number of biological factors have been shown to correlate with outcome in patients with cervical carcinoma. However, controversy exists in the literature with respect to relationship between expression of EGFR and prognosis of patients with cervical cancer. It has been reported that overexpression of EGFR in cervical carcinoma correlates with poor prognosis [23–26], although this association is controversial [27–29]. The current study aimed to evaluate the clinical value of EGFR with respect to predicting prognosis in patients with

Objective. Overexpression of epidermal growth factor receptor (EGFR) gene has been detected in a large number of human tumors, in most of which the association between overexpression of EGFR and poor prognosis of the disease has been reported. However, the prognostic role of EGFR oncoprotein in cervical carcinoma remains controversial. The current study aimed to determine the prognostic value of EGFR in patients with cervical cancer. Methods. We measured EGFR oncoprotein with an enzymelinked immunosorbent assay (ELISA). Results were correlated to clinical data. Results. The presence of measurable levels of EGFR in the tumor was found in all of the explored tumors. The levels varied widely from 31 to 2874 fmol/mg protein with a median at 582 fmol/mg protein, and Q1, Q2, and Q3 quartiles were 156, 562, and 1047, respectively. Overexpression of EGFR was associated with an impaired prognosis with respect to disease-free interval (P ⴝ 0.03) and overall survival (P ⴝ 0.04). Multivariate analysis revealed that tumor EGFR provided prognostic information with respect to disease-free interval (P ⴝ 0.002) and overall survival (P ⴝ 0.005) independently of the two established prognosticators, FIGO stage and lesion size. Conclusion. Our results are consistent with the concept that EGFR confers prognostic information in addition to that provided by the established clinicopathologic parameters of FIGO stage and lesion size. © 2002 Elsevier Science (USA) Key Words: EGFR; cervical cancer; enzyme immunoassay; prognosis.

INTRODUCTION It has been established over the past 20 years that the molecular mechanisms implicated in the development and uncontrolled proliferation of cancers involve abnormalities of oncogenes and growth factor-receptor systems. As our knowledge of the role of regulatory genes in oncogenesis increases, 1

To whom correspondence should be addressed at Department of Obstetrics and Gynecology, Yonsei University College of Medicine, C.P.O. Box 8044, Seoul, Korea 120-752. Fax: 82-2-313-8357. E-mail: ytkchoi@yumc. yonsei.ac.kr. 0090-8258/02 $35.00 © 2002 Elsevier Science (USA) All rights reserved.

84

EGFR AND PROGNOSIS IN CERVICAL CANCER

cervical cancer on a large prospective cohort of a patient population with a long follow-up period, and belonging to a single institution. PATIENTS AND METHODS Patients This study involved 73 consecutive patients treated at Yonsei University College of Medicine from January 1993 to June 1995. Staging of the primary tumor was performed according to the criteria of FIGO (International Federation of Gynecology and Obstetrics) staging system for cervical carcinoma. Tumor lesion size was estimated by pelvic examination, colposcopy, and abdomino-pelvic MRI (magnetic resonance imaging). Patient age ranged from 34 to 78 years with a mean of 49.8 years. The study group consists of 7 patients with stage Ia disease, 19 with stage Ib, 8 with stage IIa, 7 with stage IIb, 29 with stage III, and 3 with stage IV. The 7 patients with stage Ia were treated with either cervical conization or a modified radical hysterectomy. Twenty-seven patients with stage Ib or IIa were primarily treated with radical abdominal hysterectomy and systematic pelvic lymphadenectomy plus selective paraaortic lymph node sampling. Radical surgery was not fulfilled in 1 patient with stage IIa disease because of positive pelvic lymph nodes found at laparotomy. The remaining patients with higher FIGO stages primarily received cis-platinum-based neoadjuvant chemotherapy and radiotherapy to the pelvis. The patients were treated with the following chemotherapeutic schedule: cisplatin 50 mg/m 2 intravenously (iv) on Day 1 with previous hydration, and vincristine 1 mg/m 2 iv on Day 2. Treatment courses were repeated every 10 days for a total of three cycles. After chemotherapy, a standard combined approach existing of external beam radiotherapy (46 Gy in 4 – 6 weeks) followed by brachytherapy was given. After treatment, all of the patients were followed at our department or at local hospitals until May 30, 2000. All of the patients who relapsed have been submitted to our department for therapy. No patient has been lost to follow-up. During follow-up, 17 patients had recurrent cervical carcinoma, 13 patients died of cancer-related causes, and 1 patient died of intercurrent disease whereas 59 are still alive without evidence of relapse. Median time of follow-up was 59 months (range 9 to 78 months). EGFR Measurement In this study, no patients had been treated with radiotherapy or chemotherapy prior to tissue sampling. Shortly after removal of cervical tissue colposcopically directed biopsy, tissue specimens were stored at ⫺70°c until they were processed and a portion of each tissue to be assayed for EGFR status was examined to confirm the histological diagnosis of cervical lesion. For the tissue preparation, after stripping of blood and necrotic tissue, the obtained tissue samples were thawed on ice,

85

placed in 10 vol of ice-cold receptor buffer, 10 mM Tris-HCl (pH 7.4), 1.5 mM EDTA, 10% glycerol, and 0.1% sodium azide, and homogenized. The resulting mixture was centrifugated at 1000 –2000g for 10 min at room temperature and the supernatant was recovered. The total protein in the prepared supernatant was measured by the method of Lowry et al. [30]. The EGFR oncoprotein enzyme immunoassay EGFR EIA kit (Oncogene Science, Inc., NY) used is a sandwich-type immunoassay involving a mouse monoclonal capture antibody [31] with a rabbit antiserum as detector. The microtiter assay plate coated with anti-EGFR monoclonal antibody was incubated with specimen (tissue supernatant) and the standard. During this incubation, the receptor protein present in the specimen or the standard was bound to the solid phase and the unbound materials present in the specimen were removed by aspiration of fluid and washing of the plate. Anti-EGFR conjugated with horseradish peroxide was incubated with the plate and if the receptor proteins were present in the specimen, the EGFR conjugates were bound to the receptors in the plate. Unbound conjugate was removed by aspiration, and the plate was washed. The plate was next incubated with enzyme substrate solution (hydrogen peroxide and O-phenylenediamine) to develop color, which reflected the amount of bound EGFR conjugate. The enzyme reaction was stopped by the addition of 2.5 N sulfuric acid and the intensity of the color developed was read using a spectrophotometer set at 490 nm. The intensity of the color formed was proportional to the concentration of receptors in the sample. A standard curve was obtained by plotting the EGFR concentration of the standards according to the absorbance and the values of specimens were determined from the curve. The concentration of the EGFR was calculated in femtomoles per milligram cytosol protein by dividing the concentration of receptors by the concentration of cytosol protein. We used a cutoff point of 250 fmol/mg protein to indicate the overexpression of EGFR as previously described [19]. Statistical Analysis Nonparametric tests including the Mann-Whitney U test and Kruskal-Wallis test were used to evaluate the relationship between EGFR and clinico-pathological parameters of the patients. Differences were considered significant when the probability of error was below 5% (P ⬍ 0.05). The following variables were studied: age, menopausal status, FIGO stage, tumor size, cell type, histological subtype, and lymph node status. The prognosis of patients was determined by diseasefree interval (DFI) and overall survival (OS) after treatment. Patient overall survival was calculated from the time of pathological diagnosis to the time of death of last event. DFI was estimated from the date of initial treatment until the establishment of local recurrence or distant metastasis. Recurrent disease was confirmed by pathology or imaging study including computed tomography scan and MRI. Estimates of survival

86

KIM, PARK, AND KIM

TABLE 1 Epidermal Growth Factor Receptor Status According to Clinicopathological Parameters

tistical analyses were carried out using SPSS software (SPSS version 9.0, NC).

EGFR (fmol/mg protein) Parameters Total number of patients Age (years) ⱖ50 ⬍50 Menopausal status Premenopause Postmenopause FIGO stage I–II III–IV Lesion size (cm) ⱖ4 ⬍4 Histology Squamous Others a Subtype LCNK b LCK c Lymph node involvement Yes No

RESULTS

No. of cases

Median

Range

73

582

31–2874

37 36

582 693

31–2004 69–1982

0.74

32 41

578 467

47–1419 52–1223

0.19

41 32

762 942

47–2004 68–2874

0.06

47 26

792 238

31–2874 68–1223

0.002

64 9

794 582

31–1982 42–1223

0.20

35 29

782 698

31–2004 47–2874

0.34

7 20

624 598

31–1982 47–2004

0.11

P value

a

5 cases with adenocarcinoma, 4 cases with adenosquamono carcinoma. Large cell nonkeratinizing. c Large cell keratinizing. b

probability (DFI and OS) were obtained using the KaplanMeier nonparametric method [32]. Cox proportional hazards regression model was used to detect the impact of variables on DFI and OS [33]. Differences in survival duration between subgroups were analyzed using a two-sided log-rank test. Sta-

A summary of EGFR expression according to the clinicopathological parameters is shown in Table 1. The presence of measurable levels of EGFR in the tumor was found in all of the explored tumors. The levels varied widely from 31 to 2874 fmol/mg protein with a median at 582 fmol/mg protein, and Q1, Q2, and Q3 quartiles were 156, 562, and 1047, respectively. Using cutoff values of 250 fmol/mg protein, 71.2% of tumors were considered as being EGFR⫹. An analysis to search for the relationship of EGFR levels with several parameters was performed. Patients with lesions of 4 cm and larger had significantly higher receptor levels than those with lesions under 4 cm (median levels were 792 fmol/mg protein for patients with lesions of 4 cm and larger and 238 fmol/mg protein for those with lesions under 4 cm, P ⫽ 0.002). No significant relationship was found between age, menopausal status, FIGO stage, histology, subtype, and tumor EGFR expression. EGFR expression did not correlate with lymph node involvement in patients who received radical hysterectomy. Table 2 gives the results of the univariate and the multivariate analysis searching for prognostic parameters related to disease-free interval and overall survival. In the univariate analysis, the significant parameters linked to DFI were: age (P ⫽ 0.04), FIGO stage (P ⫽ 0.02), lesion size (P ⫽ 0.002), and EGFR status (P ⫽ 0.03). By applying a multivariate analysis, FIGO stage (P ⫽ 0.003), lesion size (P ⫽ 0.004), and EGFR status (P ⫽ 0.002) remained as independent prognostic factors. Menopausal status (P ⫽ 0.005), FIGO stage (P ⫽ 0.04), lesion size (P ⫽ 0.02), and EGFR status (P ⫽ 0.04) were the significant parameters for OS in the univariate analysis. In the multivariate analysis, FIGO stage (P ⫽ 0.01), lesion size (P ⫽ 0.001), and EGFR status (P ⫽

TABLE 2 Univariate Log-Rank Test and Multivariate Cox Regression Model of Prognostic Parameters Predicting Disease-Free Interval (DFI) and Overall Survival (OS) DFI

OS

Parameters

Univariate P value

Multivariate P value

RR 95% CI

Univariate P value

Multivariate P value

RR 95% CI

Age Menopausal status FIGO stage Lesion size Histology Subtype EGFR status

0.04 0.6 0.02 0.002 0.6 0.2 0.03

0.3 — 0.003 0.004 — — 0.002

— — 2.3 (1.3–3.9) 2.4 (1.4–4.1) — — 3.1 (1.5–6.9)

0.07 0.05 0.04 0.02 0.9 0.3 0.04

— — 0.01 0.001 — — 0.005

— — 2.3 (1.2–4.2) 3.2 (1.8–5.6) — — 3.5 (1.5–8.3)

Note. RR, relative risk; CI, confidence interval.

EGFR AND PROGNOSIS IN CERVICAL CANCER

FIG. 1.

Disease-free interval related to EGFR positivity.

0.005) were independent significant prognostic factors for the length of OS. Disease-free interval related to EGFR positivity for patients with cervical carcinoma is shown in Fig. 1. A stratification of patients was performed on the basis of EGFR cutoff values at the different quartiles. Patients with tumor EGFR levels ⬍156 fmol/mg protein, between 156 and 1047 fmol/mg protein, and ⬎1047 fmol/mg protein had a 5-year DFI rate of 95, 51, and 28%, respectively (log-rank test: P ⫽ 0.003). The three subgroups had a 5-year OS rate of 89, 61, and 49%, respectively (log-rank test: P ⫽ 0.013) (Fig. 2). DISCUSSION Worldwide, cervical carcinoma is second only to breast cancer as the most common malignancy in women with respect to both incidence and mortality [34]. Close to half a million patients with cervical cancer are diagnosed annually. In both underdeveloped and industrialized nations, the majorities of women who develop the disease are most often economically disadvantaged and have poor access to medical care and screening programs. Despite significant survival improvements in patients with advanced cervical carcinoma due to the development of chemoradiotherapy regimens, their survival remains rather poor. The 5-year survival rate varied from 63 to 75% in stage IB2 to 18 to 34% in stage IVA [35]. Identification of reliable tumor parameters that reflect tumor aggressiveness could provide useful tools for choosing more or less aggressive treatment modalities. Cervical carcinoma reveals the plethora of potential clinical and biological prognostic factors with parameters associated with oncogenes and growth factor-receptor systems [36, 37]. However, the most frequently explored biological factors for growth factor-receptor systems in cervical carcinoma remain EGFR and c-erbB2. Furthermore, there is controversy in the literature with respect

87

to correlation between expression of EGFR and prognosis of patients with cervical cancer [21, 24 –29]. This fact prompted us to evaluate the respective prognostic power of EGFR in this study. Eight different groups have reported on the relationship between EGFR and survival (Table 3). The size of the series of patients investigated ranged between 52 and 216 with EGFR positivity in 6 –71.2%. The median (18 –96 months) follow-up showed great differences. Using univariate analysis, three of five study groups found a significant relationship between EGFR status and DFI or OS at a certain time point of followup. This finding was confirmed in three studies using multivariate analysis. In agreement with our previous study where a higher EGFR expression in cervical carcinoma than in normal cervix and many positive cases among the advanced stage tumor and large lesion size tumor were found [19], we found that FIGO stage, lesion size, and EGFR tumor levels had a prognostic impact, suggesting that the expression of EGFR oncogene product may correlate with aggressive and proliferative course of cervical carcinoma. Pfeiffer et al. investigated 52 patients with squamous cell carcinoma of all clinical stages and reported that overexpression of EGFR correlates with a worse prognosis. They found that the level of EGFR may be indicative of the biological aggressiveness of cervical carcinomas [23]. In a previous study of 62 cases, Hale et al. reported that a strong correlation was found between EGFR overexpression and mortality, especially in patients without lymph node metastasis [25]. Conversely, Hayashi et al. [27] reported EGFR overexpression in only 3 of 52 cases with squamous cell carcinoma, the number of patients being too small for a confirmative evaluation of the prognostic significance. Recently, Kersemaekers et al. indicated that by multivariate analysis, overexpression of EGFR is an independent predictor for prognosis in

FIG. 2. levels.

Overall survival according to the three subgroups of EGFR tumor

88

KIM, PARK, AND KIM

TABLE 3 Relationship between EGFR and Prognosis in Patients with Cervical Cancer

First author

Year

No. of cases

Method

Cut off (fmol/mg)

EGFR⫹ (%)

Follow-up (months) median

Significance of EGFR

Pfeiffer et al. [23] Hayashi et al. [27] Hale et al. [25] Kristensen et al. [24] Oka et al. [29] Scambia et al. [28] Kersemaekers et al. [26] Kim et al. [current study]

1989 1991 1993 1996 1997 1997 1999 2001

52 52 62 132 216 90 136 73

RBA IH IH IH IH RBA IH EA

100 ⱖ1⫹ ⱖ2⫹ ⱖ1⫹ ⱖ1⫹ 6 ⱖ1⫹ 250

14 6 34 25.8 33 51 54 71.2

18 NM 96 73.9 NM 41 48 58

Indicator of biological aggressiveness NS Correlation with mortality P ⫽ 0.014 (RFS: multivariate) NS NS P ⫽ 0.04 (OS: multivariate) P ⫽ 0.005 (OS: multivariate)

Note. NS, not significant; NM, not mentioned; RFS, relapse-free survival; OS, overall survival; RBA, radioligand-binding assay; IH, immunohistochemistry; EA, enzyme immunoassay.

earlier stages of cervical cancer [26]. The differences in results and conclusions might partly be due to variability in patient number, cutoff levels of EGFR, application of definitive therapy, and duration of patient follow-up— generally concerns affecting the evaluation of other prognostic indicators. Furthermore, the lack of a standardized method for determining EGFR status of cervical cancer is particularly a major aspect to be solved. Tumor EGFR status has been investigated in several ways, including radioligand-binding assays, immunocytohistochemistry, gene amplification, enzyme immunoassay, and multiparameter flow cytometry. Of the range of techniques that can be used to detect aberrant EGFR protein expression, there are several advantages to using the enzyme immunoassay with monoclonal antibodies in contrast to the radioligand-binding method. With this enzyme immunoassay, it is possible to detect the receptor protein whether or not it is occupied by ligand: it requires only a small volume of tumor tissue, and it can be performed in a short period [18, 19]. In addition, the enzyme immunoassay is less affected by degradation of the receptor protein during a period of devascularization than the binding assay. However, it would be essential that future studies consider the standardization of a universal method to evaluate tumor EGFR status. In summary, the current data confirm previously published results of a prognostic value of tumor EGFR level in a univariate analysis. A multivariate analysis of the survival data showed that tumor expression of EGFR provides independent information in addition to the established clinicopatholgic parameters of FIGO stage and lesion size. A clinical use of EGFR as an additional stratification indicator in multiinstitutional prospective studies involving patients with cervical carcinoma remains to be determined.

ACKNOWLEDGMENT This study was supported by BK21 Project for Medical Sciences Yonei University in 2002.

REFERENCES 1. Hynes NE, Stern DF. The biology of erbB-2/neu/HER-2 and its role in cancer. Biochim Biophys Acta 1994;1198:165– 84. 2. Cowley GP, Smith JA, Gusterson BA. Increased EGF receptors on human squamous carcinoma cell lines. Br J Cancer 1986;53:223–29. 3. Fisher DA, Lakshmanan J. Metabolism and effects of epidermal growth factor and related growth factors in mammals. Endocr Rev 1990;11:418 – 42. 4. Ozanne B, Richards CS, Hendler F, Burns D, Gusterson B. Over-expression of the EGF receptor is a hallmark of squamous cell carcinomas. J Pathol 1986;149:9 –14. 5. Stoscheck CM, King LE Jr. Role of epidermal growth factor in carcinogenesis. Cancer Res 1986;46:1030 –37. 6. Carpenter G. Receptors for epidermal growth factor and other polypeptide mitogens. Annu Rev Biochem 1987;56:881–14. 7. Cohen S. The epidermal growth factor (EGF). Cancer 1983;51:1787–91. 8. Yarden Y, Ullrich A. Growth factor receptor tyrosine kinases. Annu Rev Biochem 1988;57:443–78. 9. Fukushige S, Matsubara K, Yoshida M. Localization of a novel v-erbBrelated gene, c-erbB-2, on human chromosome 17 and its amplification in a gastric cancer cell line. Mol Cell Biol 1986;6:955–58. 10. Shoyab M, Plowman GD, McDonald VL, Bradley JG, Todaro GJ. Structure and function of human amphiregulin: a member of the epidermal growth factor family. Science 1989;243:1074 –76. 11. Higashiyama S, Abraham JA, Miller J, Fiddes JC, Klagsbrun M. A hepar in-binding growth factor secreted by macrophage-like cells that is related to EGF. Science 1991;251:936 –39. 12. Schreiber AB, Libermann TA, Lax I, Yarden Y, Schlessinger J. Biological role of epidermal growth factor-receptor clustering. Investigation with monoclonal anti-receptor antibodies. J Biol Chem 1983;258:846 –53. 13. Battaglia F, Scambia G, Benedetti Panici P. Epidermal growth factor receptor expression in gynecological malignancies. Gynecol Obstet Invest 1989;27:42– 44. 14. Kohler M, Janz I, Wintzer HO, Wagner E, Bauknecht T. The expression of EGF receptors, EGF-like factors and c-myc in ovarian and cervical carcinomas and their potential clinical significance. Anticancer Res 1989; 9:1537– 47. 15. Berchuck A, Rodriguez G, Kamel A, Soper JT, Clarke-Pearson DL, Bast RC Jr. Expression of epidermal growth factor receptor and HER-2/neu in normal and neoplastic cervix, vulva, and vagina. Obstet Gynecol 1990; 76:381– 87.

EGFR AND PROGNOSIS IN CERVICAL CANCER 16. Gullick WJ. Prevalence of aberrant expression of the epidermal growth factor receptor in human cancers. Br Med Bull 1991;47:87–98. 17. van Dam PA, Lowe DG, Watson JV. Multiparameter flow-cytometric quantitation of epidermal growth factor receptor and c-erbB-2 oncoprotein in normal and neoplastic tissues of the female genital tract. Gynecol Oncol 1991;42:256 – 64. 18. Kim JW, Kim YT, Kim DK. Correlation between EGFR and c-erbB-2 oncoprotein status and response to neoadjuvant chemotherapy in cervical carcinoma. Yonsei Med J 1999;40:207–14. 19. Kim JW, Kim YT, Kim DK, Song CH, Lee JW. Expression of epidermal growth factor receptor in carcinoma of the cervix. Gynecol Oncol 1996; 60:283– 87. 20. Nicholson RI, McClelland RA, Gee JM. Epidermal growth factor receptor expression in breast cancer: association with response to endocrine therapy. Breast Cancer Res Treat 1994;29:117–25. 21. Nicholson RI, Gee JM, Harper ME. EGFR and cancer prognosis. Eur J Cancer 2001;37 Suppl 4:S9 –15. 22. Scambia G, Benedetti Panici P, Battaglia F. Significance of epidermal growth factor receptor in advanced ovarian cancer. J Clin Oncol 1992;10: 529 –35. 23. Pfeiffer D, Stellwag B, Pfeiffer A, Borlinghaus P, Meier W, Scheidel P. Clinical implications of the epidermal growth factor receptor in the squamous cell carcinoma of the uterine cervix. Gynecol Oncol 1989;33:146 – 50. 24. Kristensen GB, Holm R, Abeler VM, Trope CG. Evaluation of the prognostic significance of cathepsin D, epidermal growth factor receptor, and c-erbB-2 in early cervical squamous cell carcinoma. An immunohistochemical study. Cancer 1996;78:433– 40. 25. Hale RJ, Buckley CH, Gullick WJ, Fox H, Williams J, Wilcox FL. Prognostic value of epidermal growth factor receptor expression in cervical carcinoma. J Clin Pathol 1993;46:149 –53. 26. Kersemaekers AM, Fleuren GJ, Kenter GG. Oncogene alterations in

89

carcinomas of the uterine cervix: overexpression of the epidermal growth factor receptor is associated with poor prognosis. Clin Cancer Res 1999; 5:577– 86. 27. Hayashi Y, Hachisuga T, Iwasaka T. Expression of ras oncogene product and EGF receptor in cervical squamous cell carcinomas and its relationship to lymph node involvement. Gynecol Oncol 1991;40:147–51. 28. Scambia G, Ferrandina G, Distefano M, D’Agostino G, Benedetti-Panici P, Mancuso S. Epidermal growth factor receptor (EGFR) is not related to the prognosis of cervical cancer. Cancer Lett 1998;123:135–39. 29. Oka K, Nakano T, Arai T. Expression of cathepsin D and epidermal growth factor receptor in stage III cervical carcinoma. Intl J Gynecol Cancer 1997;7:122–26. 30. Lowry OH, Rosebrough NJJ, Farr AL. Protein measurement with the folin phenol reagent. J Biol Chem 1951;193:265– 69. 31. Waterfield MD, Mayes EL, Stroobant P. A monoclonal antibody to the human epidermal growth factor receptor. J Cell Biochem 1982;20:149 – 61. 32. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457– 81. 33. Cox DR. Regression models and life-tables. J R Stat Soc B 1972;34:187– 20. 34. Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics, 1999. CA Cancer J Clin 1999;49:8 –31, 1. 35. Bosch FX. Trends in cervical cancer mortality. J Epidemiol Community Health 1999;53:392. 36. Dell G, Gaston K. Human papillomaviruses and their role in cervical cancer. Cell Mol Life Sci 2001;58:1923– 42. 37. Beger M, Butz K, Denk C, Williams T, Hurst HC, Hoppe-Seyler F. Expression pattern of AP-2 transcription factors in cervical cancer cells and analysis of their influence on human papillomavirus oncogene transcription. J Mol Med 2001;79:314 –20.