Multivariate analysis of the prognostic factors and outcomes in early cervical cancer patients undergoing radical hysterectomy

https://doi.org/10.1016/j.ygyno.2004.01.026Get rights and content

Abstract

Objective. This study was performed to identify pathologic and clinical risk factors that best predicted 5-year recurrence-free survival (RFS) among patients with early-stage cervical carcinoma, treated by radical hysterectomy and pelvic lymphadenectomy.

Methods. The records of 197 patients with early-stage invasive cervical carcinoma who underwent radical hysterectomy and pelvic lymphadenectomy from 1990 to 1999 were retrospectively reviewed. Clinical and pathologic variables including age, tumor size (TS), clinical stage, depth of invasion (DI), lymphovascular space involvement (LVSI), cell type, tumor grade, lymph node metastases (LNM), parametrial invasion, surgical margin involvement, and pattern of adjuvant therapy were analyzed using univariate and multivariate methods to define those variables that best predicted RFS.

Results. Outer 1/3 invasion, LVSI, and LNM were identified as independent poor prognostic factors, which were used to define three prognostic groups: patients (n = 104) with good prognoses (LVSI (−) and LNM (−)), patients (n = 46) with intermediate prognoses (either LVSI (+) without outer 1/3 invasion or LNM (+) without LVSI), and patients (n = 47) with poor prognoses (LVSI (+) patients with outer 1/3 invasion). The estimated 3-year RFS for patients with LVSI and deeply invasive tumors regardless of nodal status and/or nodal metastases receiving adjuvant CT + RT was significantly greater than that for patients who received only adjuvant radiotherapy (80% vs. 49%, P = 0.048 in the group of patients with LVSI and deeply invasive tumors with positive nodes and without positive nodes; 87% vs. 36%, P = 0.013 in the group of patients with LVSI and deeply invasive tumors with positive nodes only).

Conclusions. The multivariate analysis and prognostic grouping system maximally separated patients with early-stage invasive cervical carcinoma into groups with good, intermediate, or poor prognoses, with 3-year RFSs of 90%, 82%, 67%; and 5-year RFSs of 89%, 69%, 43%, respectively. CT + RT played a role in improving RFS among patients with LVSI and deeply invasive tumors and poor prognoses.

Introduction

Early-stage cervical cancer has assumed increased importance for gynecological oncologists because of the rising proportion of early-stage invasive cervical cancer among patients. Although the majority of patients with early-stage cancer are cured with radical hysterectomy and pelvic lymphadenectomy or with radiotherapy, the 20% mortality rate among these patients has not improved during the last two decades [1], [2]. The clinical staging system for cervical cancer promoted by the International Federation of Gynecology and Obstetrics (FIGO) is effective in predicting the outcomes of this disease. However, disparate prognoses have been observed in patients with tumors of the same stage. To clarify this discrepancy, many investigators have tried to identify prognostic factors obtained by surgery. Pelvic lymph node metastases (LNM), tumor size (TS), cervical stromal invasion, the presence of lymphovascular space involvement (LVSI), histological subtype, and direct extension of the tumor to the uterine corpus, vagina, parametrium, or to the surgical margins of resection are reported to have prognostic value [3], [4], [5], [6], [7], [8], [9].

It is clear that many of these risk factors are interrelated. Studies have been designed to statistically identify the best combination of these risk factors using multivariate analysis techniques [5], [6], [7]. Using Cox's proportional hazards regression model, these studies have produced different sets of combined risk factors for surgically treated patients. However, the reported variables of combined risk factors are inconclusive, and the definitions of prognostic groups are complicated in the reports by Delgado et al.[5]], Sevin et al. [6], and Kamura et al. [7]. We used univariate and multivariate analysis with Cox's regression model to statistically identify the best combination of these risk factors to predict prognoses. Finally, we used a prognostic grouping system stratified by different survival curves and grouped similar recurrence-free survival (RFS) rates together to maximally separate patients with early-stage invasive cervical cancer into three subgroups. Although adjuvant radiotherapy and/or chemotherapy have been prescribed for different high-risk groups, the role of postoperative adjuvant therapy is only defined in high-risk patients with early-stage disease. For patients who were assigned to poor prognostic factor groups in our study, we attempted to evaluate the impact of postoperative adjuvant therapy and to see if adjuvant CT + RT could improve RFS compared to radiotherapy alone.

Section snippets

Materials and methods

From 1990 to 1999, there were 213 patients with stages IB and II cervical carcinoma who underwent type III radical hysterectomy and bilateral pelvic lymphadenectomy as primary treatment at Cathay General Hospital in Taipei, Taiwan. Paraaortic lymph node sampling was performed concurrently if there was any suspicion of metastasis. Pelvic lymphadenectomy was performed as described by Ho et al. [10].

All clinical and pathological variables were retrospectively reviewed, using a predefined data

Results

The 5-year overall survival (OS) and RFS of all 197 patients were 82% and 75%, respectively. Univariate analysis showed that deep stromal invasion (outer one-third vs. inner or middle one third of the tumor, P < 0.001), larger TS (P = 0.005), LVSI (P < 0.001), LNM (P < 0.001), higher clinical stage (II vs. IB, P = 0.022), surgical margin involvement (P < 0.003), and adjuvant RT (P < 0.001) were significant predicts of poor RFS. Patients' characteristics and univariate analysis of factors

Discussion

The prognostic grouping system reflects maximal differences in the power of risk factors based on Cox's regression analyses. For example, the presence of LNM is accepted as high risk for treatment failure. In the model, LNM was a poor prognostic factor only in patients with deeply invasive tumors and LVSI. The incidence of LNM for all groups was 23.4%. The incidence of patients with LNM and without deeply invasive tumors was 5.6% in contrast to 17.8% of those with LNM and deeply invasive

Conclusion

Using Cox's multivariate analysis and a prognostic grouping system, we maximally separated patients with early-stage invasive cervical carcinoma into good, intermediate, and poor prognosis subgroups, with corresponding 5-year RFSs of 89%, 69%, and 43%, respectively. The prognostic grouping system was easy to apply, and only required measurement of three independent risk factors: LVSI (+, −), LNM (+, −), and outer 1/3 invasion (+, −) to assign an individual patient to one of three risk groups.

References (30)

  • C.P Morrow

    Is pelvic radiation beneficial in the postoperative management of stage IB squamous cell carcinoma of the cervix with pelvic node metastasis treated by radical hysterectomy and pelvic lymphadenectomy?

    Gynecol. Oncol.

    (1980)
  • W.K Kinney et al.

    Value of adjuvant whole-pelvis irradiation after Wertheim hysterectomy for early-stage squamous carcinoma of the cervix with pelvic nodal metastasis: a matched-control study

    Gynecol. Oncol.

    (1989)
  • D.H Kenneth et al.

    Cervical and vaginal cancer

  • J.G Boyce et al.

    Vascular invasion in stage I carcinoma of the cervix

    Cancer

    (1984)
  • B.U Sevin et al.

    Surgically defined prognostic parameters in patients with early cervical carcinoma

    Cancer

    (1996)
  • Cited by (0)

    View full text