| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Departments of Obstetrics and Gynecology (AA, GG), Medical Oncology (FF, DA), and Pathology (LS) and the Romagna Cancer Registry (LB), Luigi Pierantoni Hospital, Forlì, Italy.
Correspondence: Address correspondence and reprint requests to: Lauro Bucchi, MD, Romagna Cancer Registry, Medical Oncology Department, Luigi Pierantoni Hospital, 47100 Forlì, Italy; Fax: 039-0543-731-736; E-mail: i.o.r{at}fo.nettuno.it
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: In this study of operable endometrial carcinoma cases registered by the population-based Romagna Cancer Registry (northern Italy) between 1987 and 1994, the association of demographic (age, time period, place of birth, place of residence, place of treatment, and marital status) and pathological factors (histological type, tumor grade, myoinvasion, and extension of disease to cervix, serosa, adnexa, and vagina) with the probability of lymphadenectomy was evaluated by multiple logistic regression analysis.
Results: Of the 300 potentially eligible cases, sufficient information was obtained for 276 (92%; median age, 63 years; range, 33–87 years). No case of para-aortic lymphadenectomy was observed. Pelvic lymphadenectomy was performed in 86 (31%) cases. The probability of pelvic lymphadenectomy was related to tumor grade (positive association), place of treatment, and marital status. All other variables, including myoinvasion and extension of disease to the cervix and beyond the uterus, had no effect whatsoever.
Conclusions: The most likely interpretations of results include poor acceptance of current surgical pathological staging criteria and insufficient use of standard diagnostic techniques for preoperative and intraoperative assessment of myoinvasion and extrauterine spread.
Key Words: Endometrial carcinoma • Staging • Lymphadenectomy • Tumor grade • Myoinvasion
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
It is conceivable that such controversies may adversely affect the translation of the new FIGO criteria to clinical practice. Many reports have shown that lymphadenectomy is a viable and customary procedure at highly specialized centers. In contrast, little is known of the management of EC patients on a large scale. Available data have been obtained with questionnaire surveys.11–14 These have shown that tumor grade and myometrial invasion influence the frequency of lymphadenectomy. Such investigations, however, have generally targeted selected categories of professionals or leading centers for gynecological oncology. To the best of our knowledge, the question of lymphadenectomy has never been the subject of a population-based study.
At the cancer registry that covers the Romagna region of northern Italy (Romagna Cancer Registry, RTRO), we retrieved surgical and pathological information on a series of EC cases. Our aim was to evaluate the patterns and correlates of staging and the prognostic value of stage and other known or suspected predictors of survival. This article focuses on the frequency and determinants of lymphadenectomy.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In 1999, a computer file was obtained from the RTRO database that included all cases (n = 377) that met the following requirements: (1) topography code 182 (corpus uteri), behavior code 3 (malignant, primary site), and morphology codes 8010 to 8572 (adenocarcinoma and other specified carcinoma types) according to the International Classification of Diseases for Oncology (ICD-O)16; (2) certainty of diagnosis according to standard criteria of cancer registration17; (3) availability of information other than death certificate only; and (4) residence in the districts of Forlì, Ravenna, and Faenza (total female population 219,567 on December 31, 1994). The date of diagnosis varied from 1987 (the first year of cancer registration in Romagna) to 1994 (the most recent year of complete registration available when the study was undertaken).
Items of Information
The computer file comprised all routine registration variables.17 Among these, the following were used for this analysis: date of birth, place of birth, municipality of residence, marital status, date of diagnosis, modalities of diagnosis (histology, surgery, peritoneal cytology, and others), place (hospital and department) of primary surgical treatment, tumor grade, and morphology code according to ICD-O.
Information regarding surgical treatment and stage—routinely collected but not computerized—was abstracted by one of us (A.A.) from the original clinical and histological reports stored in the paper archives of the registry. Patients undergoing hysterectomy with or without salpingo-oophorectomy were considered surgically treated. Lesions reported to invade the middle third of the myometrium were recorded as such. All other stage data were classified according to tumor, node, metastasis/FIGO criteria. The number of pelvic and para-aortic lymph nodes dissected was recorded. Lymphadenectomy was defined as the removal of at least one lymph node. For 48 of 377 (13%) cases, the available data were considered inconclusive or incomplete, thus prompting further inquiries at the referral hospitals.
Elegibility
On the basis of data collected, 77 patients were excluded from the study. These were accounted for by patients not treated surgically (n = 65), patients with surgical FIGO stage IV (n = 7), and patients with T0 disease according to tumor, node, metastasis staging (no evidence of primary tumor on hysterectomy after the diagnosis of EC on biopsy; n = 5). There were thus 300 potentially eligible cases. Adequate information regarding stage parameters and lymphadenectomy was available for 276 (92%) patients. These were included in data analysis.
Objectives and Rationale
Our primary objective was to determine whether factors listed in Table 1 influenced the probability of pelvic lymphadenectomy, para-aortic lymphadenectomy, or both. Major factors included tumor grade and two indicators of disease spread, namely, the depth of myometrial invasion (IA–IB, IC) irrespective of invasion of the cervix or extension beyond the uterus (hereby referred to as myoinvasion) and extension of disease (I, II–IIIA–IIIB) irrespective of lymph node status (hereby referred to as extension). The rationale for such definitions was to obtain two distinct indicators of tumor spread as potentially resulting from preoperative and intraoperative assessment. On the basis of reported data on the relationship between depth of muscle invasion and risk of node metastasis,6 lesions reported to invade the middle third of the myometrium were coded as IB. The secondary aim of the study was to determine the number of lymph nodes dissected and the frequency of nodal involvement.
|
Data Analysis
Univariate associations were evaluated with the 
2 test. Multivariate associations were evaluated with the multiple regression analysis by using a backward stepwise approach. Two separate multiple logistic models were built. Model 1 included myoinvasion and grade, as well as a term for the myoinvasion x grade interaction. Model 2 included extension and grade, as well as a term for the extension x grade interaction. All other factors listed in Table 1 (i.e., histological type, patient age, time period, place of birth, place of residence, place of treatment, and marital status) were included in both models. Age was treated as a continuous variable. Covariates were removed from each model at a level of significance of P > .05. Odds ratios not including the unity were considered statistically significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Table 3 shows the results of the two logistic models. In model 1, myoinvasion did not influence the probability of pelvic lymphadenectomy. This was strongly and linearly associated with tumor grade. Also, major differences were related to the place of treatment. At one of three hospitals of the study area, as well as in the private sector, lymphadenectomy was performed in a negligible proportion of cases. It is noteworthy that a two-thirds reduction in the probability of lymphadenectomy was observed among the 32 patients treated in surgical centers situated elsewhere in northern Italy (only one at a leading institution for gynecological oncology). Finally, a significant association with marital status was observed. Histological type, patient age, time period, place of birth, and place of residence exerted no independent effect. No myoinvasion x grade interaction was demonstrated.
|
In model 2, myoinvasion was replaced by extension, and the extension x grade interaction was evaluated. The results were exactly the same as those of model 1. Extension had no effect on the probability of lymphadenectomy, nor did it interact with grade.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
As far as the preoperative and intraoperative determinants of lymphadenectomy are concerned, we found a relationship only with tumor grade. Myoinvasion and extension had no influence whatsoever. Although the statistical power of the study was limited, the myoinvasion x grade and extension x grade interactions were evaluated to test the null hypothesis that the effect of both indicators was the same across the tumor grades. The results suggested that this was the case. One further limitation of the study was the arbitrary classification of tumors with invasion of the middle third of the myometrium as IB cases. Thus, model 1 was rebuilt after placing such cases among those coded as IC. No change in results was observed (data not shown). In fact, lymphadenectomy was equally distributed even in the subset of cases reported by using a three-tier classification of myoinvasion (data not shown). In our opinion, such findings were compatible with four essential interpretations not mutually exclusive.
First, one may hypothesize that tumor grade was really considered as the only relevant parameter. If so, such a view is only partially in line with the most widespread concepts regarding lymphadenectomy. Although the value of grade is generally accepted6,12 and particularly emphasized by some authors,7 most European gynecological oncologists stated that they consider myoinvasion as a factor of importance.11 In the United States, myoinvasion has been reported to be a major determinant of lymphadenectomy, especially among grade 1 and 2 tumors.6,12 As has been pointed out, however, continuing controversy about those criteria is expected to adversely affect their acceptance on a large scale, with an inevitable lack of uniformity in surgical management of EC patients.12
Second, the results are compatible with the hypothesis that preoperative assessment of myoinvasion and extrauterine spread was less than optimal. Before the implementation of the new FIGO system, clinical staging was based on the identification of gross invasion of the cervix and positive endocervical curettage.6,20 Over the years, presurgical staging has been improved by accurate pelvic examination, blood tests, pelvic washing, and imaging evaluations of the pelvis and abdomen.3 Among these, transvaginal ultrasonography has been demonstrated to predict the depth of invasion as well as the tumor volume.21 At least in part, our data are likely to reflect insufficient use of such techniques. In particular, transvaginal ultrasonography was still incompletely available during the study period, and the computed tomography services were often saturated.
Third, it cannot be excluded that our results are at least partially related to problems with intraoperative assessment of myometrial invasion. During surgery, this can be determined either grossly, by bivalving the uterus, or on frozen sections.6,22 In the United States, bivalving and gross examination of the uterus have been reported to be an almost customary practice, whereas frozen section analysis is performed in relation to tumor grade.12 In Romagna, no routine information is available. According to personal communications from the pathology departments in the area, both procedures are seldom practiced. It has been suggested that a low frequency of lymphadenectomy may reflect a tendency to individualize surgical staging on the basis of the intraoperative evaluation of the uterine specimen.12 In this case series, lymphadenectomy was infrequent, but no relation with intraoperative parameters was observed. Following this line of reasoning, it is suggested that the strong effect of tumor grade depended only on the availability of that information as reported on biopsy.
Fourth, our observations are likely to suggest that surgical stage information had only a limited effect on treatment planning. Although the frequency of adjuvant therapy was not an end point of the study, we collected data (not shown) for patients living in the district of Forlì (37% of the study population). Among these, only 52% stage IC cases and 44% cases with extrauterine (stage II–III) disease underwent radiotherapy. The frequency of chemotherapy was negligible (3%).
In conclusion, we found that the average frequency of lymphadenectomy was poorer than that expected on the basis of the figures reported from United States and that tumor grade was the only factor taken into consideration. The second part of this study will be aimed at survival analysis. Stage misclassifications possibly related to the observed pattern of surgical management of the disease will be investigated by evaluating the consistency of the stage-specific 5-year survival rates.
|
Acknowledgments |
|---|
Received for publication February 15, 2001. Accepted for publication June 20, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
