- Email: [email protected]
Intent-to-treat analysis of stage Ib and IIa cervical cancer in the United States: radiotherapy or surgery 1988–1995
Intent-to-Treat Analysis of Stage Ib and IIa Cervical Cancer in the United States: Radiotherapy or Surgery 1988 –1995 WENDY R. BREWSTER, MD, PhD, BRADLEY J. MONK, MD, ARGYRIOS ZIOGAS, PhD, HODA ANTON-CULVER, PhD, S. DIANE YAMADA, MD, AND MICHAEL L. BERMAN, MD Objective: To estimate the patterns of care and outcome of women with early cervical cancer in the United States based on surgical or radiation intent-to-treat principles. Methods: The Surveillance, Epidemiology, and End Results 1995 public-use file was the data source. Subjects between the ages of 15 and 80 years at diagnosis who were treated for stage Ib or IIa cervical cancer were identified. The 1039 women who comprised the study group were stratified according to age at diagnosis (40 years or less, older than 40 years), primary treatment intent (surgery, radiotherapy), tumor size (4 cm or less, over 4 cm), registry site, and ethnicity. Survival analyses included 784 women who had at least 2 years of follow-up. Results: There were 276 cancers (26.5%) over 4 cm, and 586 (56%) women were older than 40 years at diagnosis. There were 741 (71%) subjects in the surgical intent-to-treat group, and the remainder (298) were in the radiation intent-to-treat group. Kaplan-Meier analysis indicated a 5-year survival advantage for women with tumors 4 cm or less who were in the surgical intent-to-treat group compared with the radiation intent-to-treat group (86% and 71%, P < .001). Treatment group was not prognostic for cervical cancers over 4 cm (surgical intent-to-treat compared with radiation intent-totreat; 72% and 68% survival, respectively). Multivariable analysis confirmed a survival advantage for women with surgical intent-to-treat and tumors of 4 cm or less. Conclusion: In the United States there is a survival advantage for surgical intent-to-treat compared with radiation intent-to-treat for women with tumors 4 cm or less, independent of ethnicity, adjuvant therapy, or age. (Obstet Gynecol 2001;97:248 –54. © 2001 by The American College of Obstetricians and Gynecologists.) From the Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, and Division of Epidemiology, Department of Medicine, University of California Irvine, Irvine, California; and Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, University of Chicago, Chicago, Illinois. Presented at the Society of Gynecologic Oncologists 31st Annual Meeting, February 2000, San Diego, California.
248 0029-7844/01/$20.00 PII S0029-7844(00)01117-0
The effect of widespread application of Papanicolaou smears in cervical cancer screening has resulted in a dramatic reduction of cervical cancer rates since the 1950s. Nevertheless, approximately 12,800 new cases of invasive cervical carcinoma will be diagnosed this year in the United States, and an estimated 4800 women will die of this disease during the same period.1 The conventional treatment of stage Ib and IIa cervical carcinoma, which constitutes at least half of all newly diagnosed cases,2 consists of radical hysterectomy and bilateral pelvic lymph node dissection, or radiation therapy combining whole pelvic teletherapy with local brachytherapy. Several studies comparing those treatment modalities, typically performed in academic settings, found both options to be equally efficacious with respect to local control and survival.3–5 Definitive surgical treatment may be selected more often than radiotherapy in the treatment of early-stage cervical cancer in younger women because there is cessation of ovarian function in younger women after radiation but not after surgery. In addition, the late complications of radiation are avoided when patients are treated with surgery alone. During the past 10 years the delivery of health care has become increasingly fragmented. Cancer care often has been directed away from tertiary-care institutions to smaller community-based centers that might be less likely to use radiotherapeutic techniques, dosing schedules, and state-of-the-art equipment that is used in larger academic institutions. In contrast, the rigorous requirements in a limited number of approved training programs for standardized training of gynecologic oncologists who perform most operations in women with cervical cancers might help preserve the excellent survival results for early-stage cervical cancer patients who
Obstetrics & Gynecology
have primary operative treatment in the general population. It is possible, therefore, that the change in these patterns of care might affect outcome adversely for women treated with radiotherapy compared with those who have surgery in the community setting. In order to characterize the population-based treatment patterns and survival of women with early cervical cancers treated in the United States with radiation or surgery, we reviewed the 1973–1995 data from the Surveillance, Epidemiology, and End Results program of the National Cancer Institute. This program includes nine regional tumor registries that abstract information from medical records on all cancers diagnosed within specific geographic areas. Most patients in the program are treated in the community setting; however, definitive numbers of patients treated in the community compared with those cared for in academic settings cannot be determined because the site of care is not mentioned in the database. The regions include Connecticut; Iowa; Arizona; Hawaii; Detroit, Michigan; San Francisco and Oakland, California; Salt Lake City, Utah; Atlanta, Georgia; and Seattle and Puget Sound, Washington, which together comprise 13.9% of the United States population. These regional registries are selected for their ability to accurately abstract data from their unique geographic areas. These nine areas are believed to provide a representative cross-section of the United States population. Therefore, this database permits population-based analyses of treatment practices, the results of which can be generalized throughout the United States.
Materials and Methods Abstracted cervical cancer information was extracted from the Surveillance, Epidemiology, and End Results 1995 public-use CD-ROM. Because cervical cancer staging as defined by the International Federation of Gynecology and Obstetrics (FIGO) was not coded before 1988, the study period extended from 1988 to 1995, thereby excluding all cancers diagnosed between 1973 and 1987 in that database. The database did not identify any subjects treated with a simple hysterectomy followed by a radical parametrectomy with or without lymphadenectomy. Only women between age 15 and 80 years with stage Ib or IIa cervical cancer as the first or only invasive malignancy were selected. Subjects who had standard hysterectomy before radiotherapy were excluded based on the assumptions that the diagnosis of cervical cancer was not made before surgery and that stage of disease for those women was uncertain. Also excluded were women who did not receive treatment for their cervical cancer. Furthermore, there was no way to determine which patients were treated by planned
VOL. 97, NO. 2, FEBRUARY 2001
combined multimodality therapy of radiotherapy followed by surgery. Abstracted data that provided information on the type of surgery, type of radiation, and surgeryradiation sequence were analyzed simultaneously to stratify subjects on the basis of the intent to treat, either with surgery or radiotherapy. Subjects were excluded if they received both radiation and surgery and the treatment sequence was unknown. Subjects were stratified into the surgical intent-to-treat arm if any abdominal or pelvic exploration was first in the treatment sequence. Subjects who had minor cervical procedures (eg, examination under anesthesia or biopsies) before radiotherapy, received only radiation therapy, or had radiotherapy followed by surgery were assigned to the radiation intent-to-treat arm. The subset of patients included in the survival analyses were those with cervical cancer diagnosed before 1994 for whom follow-up data were available, thereby permitting a minimum of 2 years follow-up for all study subjects. In the database, only one International Classification of Disease (ICD) code was used for cause of death. The coded cause of death for each deceased woman within this cohort was reviewed. Eight subjects were excluded from survival analyses because the cause of death was not recorded. One subject was reported to have died of chronic renal failure. There were no documented causes of death from hemorrhage, obstructive renal failure, or sepsis. All survival analyses were adjusted for cause of death from cervical cancer. Given the large size of this study group, it is unlikely that there was unequal misclassification of cause of death from cervical cancer between the two treatment groups. Statistical analyses were done using SAS (SAS Institute, Cary, NC). 2 tests were used for categorical group comparisons. Survival analyses were performed using the Kaplan-Meier method, for which the Wilcoxon log rank test was used to test survival hypotheses. Cox regression analysis, adjusted for cause of death from cervical cancer, was used for multivariable analyses. Only P ⬍ .05 by two-tailed analysis was considered statistically significant.
Results By the methods of subject selection described above, 1228 subjects whose cancer was diagnosed between 1988 and 1995 were identified from the 1973–1995 Surveillance, Epidemiology, and End Results publicuse file. One hundred eighty-nine subjects were excluded, including 34 with tumors invading less than 5 mm, 88 who received no further treatment after simple hysterectomy, 55 who received radiotherapy after a simple hysterectomy, and 12 who did not receive
Brewster et al
Treatment of Cervical Cancer
249
Table 1. Demographic Characteristics of the Study Population of 1039 Women
Table 2. Tumor Size and Treatment Intent by Age Category Variable n (%)
Age at diagnosis (y) ⱕ40 ⱖ41 Cervix cancer tumor size (cm) ⱕ4 ⬎4 Treatment intent Surgery Non-Hispanic white (n ⫽ 509) Black (n ⫽ 74) Asian (n ⫽ 66) Hispanic (n ⫽ 55) Unknown (n ⫽ 37) Radiotherapy Non-Hispanic white (n ⫽ 187) Black (n ⫽ 40) Asian (n ⫽ 30) Hispanic (n ⫽ 10) Unknown (n ⫽ 31)
453 (43.6) 586 (56.4) 763 (73.4) 276 (26.6) 741 (71.3)
298 (29.8)
treatment. Of the remaining 1039 subjects, 869 had stage Ib (84%) and 170 had stage IIa (16%) disease. The race or ethnicity was known for 992 (95%) of the women in this study; 696 (67%) were non-Hispanic white, 114 (11%) were black, 86 (8%) were Hispanic, and 96 (9%) were Asian. Other demographic characteristics are presented in Table 1. Pelvic and/or periaortic lymph node removal was done in 96% of the subjects who had primary treatment by surgical intent with a mean of 23 lymph nodes examined. There was no significant difference in the frequency of lymph node evaluation based on tumor size or age at diagnosis. When the entire study cohort was stratified by age and treatment intent, a greater percentage of younger women (40 years or less) had a surgical intent-to-treat preference than radiation intent-to-treat (82% and 18%, respectively). There was also a surgical treatment preference over radiation for women older than 40 years of age (63% and 37%, respectively); however, the younger women were significantly more likely than the older women to be treated surgically than with radiotherapy (82% compared with 63%, P ⫽ .001). When subjects were stratified by tumor size, the younger cohort with tumors 4 cm or less was more likely to be treated surgically than by radiotherapy (93% and 7%, respectively), as was the older cohort (primary surgery compared with radiotherapy, 71% and 29%). Younger women with smaller cancers were significantly more likely than their older counterparts to have a surgical intent-to-treat preference (93% compared with 71%, P ⫽ .001). For subjects with cervical cancers greater than
250 Brewster et al
Treatment of Cervical Cancer
Age (y) ⱕ40 ⱖ41 Tumor size (cm) ⱕ4 ⱕ40 y ⱖ41 y ⬎4 ⱕ40 y ⱖ41 y
Surgery n (%)
Radiation n (%)
371 (82) 370 (63)
82 (18) 216 (37)
312 (93) 303 (71)
23 (7) 125 (29)
59 (50) 67 (42)
59 (50) 91 (58)
4 cm, there was no significant difference in treatment choices of primary surgery compared with radiotherapy for either the younger (50% and 50%) or older (42% and 58%) cohorts (Table 2). For black, Hispanic, and non-Hispanic white women, there was a primary surgical preference compared with radiotherapy for those with tumors of 4 cm or less (P ⬍ .01); however, the differences were not significant among Asian subjects (P ⫽ .5). Of 1039 subjects, 784 (75%) met the criteria for inclusion in the survival analyses; 577 had tumors smaller than 4 cm (74%) and 207 had larger tumors. Five hundred forty-eight of the 784 patients were in the surgical intent-to-treat group, 27% of whom (150) received postoperative radiotherapy. Twenty-two percent (102 of 460) of the women with a surgical intent-to-treat who had small cervical cancers and 55% (48 of 88) of the women with large tumors in the surgical intent-to-treat group received adjuvant radiotherapy. Two hundred thirty-six patients were treated with primary radiotherapy, 21% of whom (50) had elective surgery after radiotherapy. The 5-year actuarial survival for the entire cohort, surgical intent-to-treat compared with radiation intentto-treat, was 84% and 69%, respectively (P ⬍ .001). Among women with cervical cancers of 4 cm or less, there was a statistically significant survival advantage for the women in the surgical intent-to-treat category. The 5-year survival adjusted for cause of death from cervical cancer for the surgery intent-to-treat group was 86% compared with 71% for radiation intent-to-treat (P ⬍ .001) (Figure 1). However the 5-year survival for women with cervical cancers larger than 4 cm was not significantly different for the intent-to-treat categories (surgery 72%, radiotherapy 68%) (P ⫽ .28) (Figure 2). Age did not influence 5-year survival in the surgical intent-to-treat group when analyzed for each tumor size category (Figures 3 and 4). The survival advantage for women with smaller cancers treated primarily with surgery persisted in multivariable survival analyses controlling for age as a
Obstetrics & Gynecology
Figure 3. Five-year survival for women with tumors smaller than 4 cm. Thin line indicates age 40 years or less and thick line indicates age over 40 years. Figure 1. Five-year survival adjusted for death caused by cervical cancer. Thin line indicates surgery intent-to-treat and thick line indicates radiation intent-to-treat.
cancers was not significantly different for either treatment category (RR 1.17, P ⫽ .6) (Table 3).
Discussion
continuous variable and tumor size in millimeters. Women who received primary radiotherapy were more likely to die of cancer than those treated with surgery (relative risk [RR] 2.75, 95% confidence interval [CI] 1.53, 4.94). The RR of death for women with larger
Morley et al6 retrospectively analyzed survival statistics for 446 women with stage Ib cervical cancer treated primarily with radiotherapy or surgery over a period of 31 years at the University of Michigan Medical Center.
Figure 2. Five-year survival for women with tumors larger than 4 cm. Thin line indicates surgery intent-to-treat and thick line indicates radiation intent-to-treat.
Figure 4. Five-year survival for women with tumors larger than 4 cm. Thin line indicates age 40 years or less and thick line indicates age over 40 years.
VOL. 97, NO. 2, FEBRUARY 2001
Brewster et al
Treatment of Cervical Cancer
251
Table 3. Risk of Death Variable Cervical tumor size ⱕ4 cm (n ⫽ 577) Age at cervical cancer diagnosis (y) Treatment Surgery (n ⫽ 460) Radiotherapy (n ⫽ 117) Tumor size (mm) Cervical tumor size ⬎4 cm (n ⫽ 207) Age at cervical cancer diagnosis (y) Treatment Surgery (n ⫽ 88) Radiotherapy (n ⫽ 119) Tumor size (mm)
Risk ratio
95% Confidence interval
0.99
0.97, 1.01
1.00 2.42 1.04
– 1.41, 4.12 1.02, 1.07
1.00
0.98, 1.03
1.00 1.26 1.00
0.72, 2.22 0.998, 1.04
Cox proportional analyses. Age and tumor size in this model are continuous variables. The risk ratio indicates the unit increase in risk associated with a unit increase in the continuous variable controlling for the other variables within the statistical model. Registry site and race or ethnicity were included in the model but were not prognostic variables.
Although they did not separate patients into groups based on prognostic indicators such as tumor size or volume, there was no difference in the 5-year survival for women treated by radical hysterectomy and pelvic lymphadenectomy compared with those treated with traditional radiotherapy. A prospective randomized trial at the University of Milan also compared radiation therapy with radical hysterectomy and pelvic lymph node dissection in patients with stage Ib or IIa cervical cancer. Multivariable analyses controlling for tumor size (4 cm or smaller compared with larger than 4 cm) found equivalent overall survival rates for the two treatment groups.7 Despite the consistency between these two large trials and many other similar studies, the treatment results from the Surveillance, Epidemiology, and End Results database did not confirm that surgery and radiotherapy are equally effective treatment modalities for women with stage Ib or IIa cervical cancer. In this study, after controlling for age and tumor size within the cohort of women with smaller tumors, there was a significant 5-year survival advantage for those treated surgically. Among the women with larger cancers there was no significant difference in 5-year survival between the groups. Treatment results of early-stage disease (Ib1) typically reflect the ability to control disease locally, whereas for bulkier disease, failures often occur regionally or at distant sites. The different observations might reflect better local control of early lesions but equivalent regional control of surgery and radiotherapy. The treatment of bulky cervical cancers is controversial. Indeed this is shown by the high percentage of surgery intentto-treat subjects who received adjuvant radiotherapy. It
252 Brewster et al
Treatment of Cervical Cancer
is possible that future trials will show that chemoradiation is superior to surgery. The benefit of surgery in the treatment of smaller cervical cancers shown in this study might represent the difference in standardization of treatments between surgery and radiation in the general population. Radical hysterectomy with pelvic lymph node dissection is the surgical procedure used to treat cervical cancer, and a standardized technique has been taught through fellowship training at only 24 –26 institutions throughout the United States.8 This procedure rarely is performed in the community setting by physicians who have not had formal training in a gynecologic oncology fellowship. This operation involves resection of the uterus, parametria, upper vagina, cardinal and uterosacral ligaments, and pelvic and periaortic lymph nodes in a consistent fashion. Although the technique can be modified on the basis of prognostic factors, such as tumor volume, to remove less parametrial tissue and less vaginal tissue in women with very early cancers, these differences have not been shown to adversely affect prognosis.9 Conversely, delivery of appropriate radiotherapy can be associated with subtle differences in treatment planning technique, portal design, radiation target volume, vaginal brachytherapy, total dose administered, duration of treatment, and technique, all of which can influence treatment results markedly. This assertion is supported by a recent report by Eifel et al10 on the method of radiotherapy in use for patients with cervical cancer at academic and nonacademic facilities. A review of 61 facilities that treat cervical cancer found that 57% treated fewer than three cases per year. Eightythree percent of nonacademic and 5% of academic centers treated fewer than three newly diagnosed cases per year. Patients at academic centers consistently received higher doses of radiation than those in the community setting. Thus, the differences between radiotherapy survival statistics from the Surveillance, Epidemiology, and End Results database and those cited in numerous publications reflecting the experience of investigators in academic settings might have resulted from different treatment approaches used in the general population compared with those used in academic settings. The reduction in survival seen with radiotherapy might not have been found in patients who were treated surgically in the community setting because surgical treatment approaches do not differ appreciably between academic and community settings. The analysis of this trial was based on an intent-totreat design, which often classifies subjects into treatment groups that do not reflect the actual treatment received by subjects in a given treatment category.
Obstetrics & Gynecology
Consequently, therapeutic benefits might be underestimated or overestimated and clinical reality distorted. One possible conclusion of the results of this analysis is that the two treatment modalities might be equally efficacious rather than the surgery intent-to-treat strategy being superior to the radiation intent-to-treat, but the differences in outcome might be attributable to other factors that could not be evaluated. In this analysis several important factors such as age, ethnicity, and tumor size were controlled, but no adjustment could be made for factors such as geometric distribution of the cancer within the cervix. Large exophytic tumors might respond differently to radiotherapy, for example, than barrel-type lesions within the cervix. Furthermore, subjects were not stratified by histologic type, and no adjustment could be made for covariables such as smoking history, vascular disease, obesity, and socioeconomic differences. Nevertheless, all those factors are generally viewed as of lesser prognostic importance than tumor size, which is the most important prognostic variable for a given stage of disease.11 This analysis confirmed that tumor size in general and a 4-cm separation of tumor size in particular are important independent prognostic variables, consistent with the 1994 FIGO staging system. A physician’s choice of primary treatment modality for stage Ib and IIa cervical cancer is based on multiple factors, including individual interpretation of published trials, regional resources for delivery of care, patient bias, physician tradition or bias, and concurrent comorbid patient factors, including obesity and age. Furthermore, treatment differences lead to various types of morbidities that can also influence the choice. For example, radical surgery permits the preservation of gonadal function. In addition, fistulae are more easily corrected when they complicate surgery compared with radiotherapy. Surgery also allows better assessment of the lymph nodes, which is of great prognostic importance. Only recently have data challenged the assumption that the adverse effects on sexual function can be minimized in patients who have radical hysterectomy compared with those treated with radiotherapy.12 Conversely, radiotherapy is an excellent treatment choice for patients who are not good surgical candidates by virtue of comorbidities of age, weight, and concurrent diseases. In this population-based analysis, women with cervical tumors of 4 cm or less were more likely to be treated surgically than by radiotherapy irrespective of age. Although surgery was preferred to radiotherapy among older women with smaller cervical cancers, the difference was even more dramatic among younger women. For larger cervical cancers, primary radiotherapy was used significantly more frequently than surgery in those
VOL. 97, NO. 2, FEBRUARY 2001
older than 40 years, and as often as surgery in women 40 years or less. This distribution of treatment choices might reflect the higher incidence of comorbidities in the older population, thereby increasing the risk of surgical complications. This is further complicated by the uncertainty of the relative efficacy of both approaches with larger cancers. Presumably the decision to treat women with larger tumors with radiotherapy reflects the belief that radiotherapy is at least equally effective as surgery for these lesions. In fact, the Surveillance, Epidemiology, and End Results data suggest that the two approaches might be approximately equal in the general population of women with this condition despite data to the contrary from many academic centers. It is now recognized that the addition of chemotherapy to the radiotherapeutic regimen for cervical cancer improves survival in women with large Ib or IIa cancers or pelvic lymph node metastases compared with radiotherapy alone.13 It is possible that the indications for chemotherapy will extend its use to women with smaller tumors also receiving primary radiotherapy. However, the study interval of our report did not include the period during which chemotherapy was considered standard of care in conjunction with radiotherapy. Thus, it is possible that radiotherapy combined with chemotherapy might provide survival results for the general population that equal or exceed those of surgery. The current study suggests that surgery as the primary treatment modality for women with smaller stage Ib (ie, stage Ib1) and IIa cervical cancer as delivered in the general United States population can provide a significantly better 5-year survival than radiotherapy.
References 1. Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics. CA Cancer J Clin 1999;49:12–3. 2. National Cancer Institute. Surveillance, Epidemiology, and End Results Cancer statistics review 1973–1995. Bethesda, Maryland: National Cancer Institute, 1995. 3. Delgado G, Bundy BN, Zaino R, Sevin BU, Creasman WT, Major F, et al. A prospective surgical pathological study of stage I squamous carcinoma of the cervix: A Gynecologic Oncology Group study. Gynecol Oncol 1990;38:352–7. 4. Piver MS, Marchetti DL, Patton T, Halpern J, Blumenson L, Driscoll DL. Radical hysterectomy and pelvic lymphadenectomy versus radiation therapy for small (⬍3 cm) stage IB cervical carcinoma. Am J Clin Oncol 1988;11:21– 4. 5. Hopkins MP, Morley GW. Radical hysterectomy versus radiation therapy for stage IB squamous cell cancer of the cervix. Cancer 1991;98:272–7. 6. Morley GW, Seski JC. Radical pelvic surgery versus radiation therapy for stage I carcinoma of the cervix (exclusive of microinvasion). Am J Obstet Gynecol 1976;126:785–98. 7. Landoni F, Maneo A, Colombo A, Placa F, Milani R, Perego P, et al.
Brewster et al
Treatment of Cervical Cancer
253
8.
9.
10.
11.
12.
13.
Randomised study of radical surgery versus radiotherapy for stage Ib-IIa cervical cancer. Lancet 1997;350:535– 40. Piver MS, Rutledge F, Smith JP. Five classes of extended hysterectomy for women with cervical cancer. Obstet Gynecol 1974;44:265– 72. Photopulos GJ, Zwaag RV. Class II radical hysterectomy shows less morbidity and good treatment efficacy compared to class III. Gynecol Oncol 1991;40:21– 4. Eifel PJ, Moughan J, Owen J, Katz A, Mahon I, Hanks GE. Patterns of radiotherapy practice for patients with squamous carcinoma of the uterine cervix: Patterns of care study. Int J Radiat Oncol Biol Phys 1999;43:352– 8. Perez CA, Grigsby PW, Nene SM, Camel HM, Galakatos A, Kao MS, et al. Effect of tumor size on the prognosis of carcinoma of the uterine cervix treated with irradiation alone. Cancer 1992;69:2796 – 806. Bergmark K, Avall-Lundquvist E, Dickman PW, Henningson L, Steineck G. Vaginal changes and sexuality with a history of cervical cancer. N Engl J Med 1999;340:1383–9. Morris M, Eifel P, Lu J, Grigsby PW, Levenback C, Stevens RE, et al. Pelvic radiation with concurrent chemotherapy compared with pelvic and para-aortic radiation for high risk cancer. N Engl J Med 1991;325:1371–2.
NEW,
Address reprint requests to:
Wendy R. Brewster, MD, PhD University of California, Irvine Medical Center The Chao Family Comprehensive Cancer Center Department of Obstetrics and Gynecology Division of Gynecologic Oncology 101 The City Drive, Building 23, Room 107 Orange, CA 92868 E-mail: [email protected]
Received March 23, 2000. Received in revised form July 21, 2000. Accepted August 10, 2000.
Copyright © 2001 by The American College of Obstetricians and Gynecologists. Published by Elsevier Science Inc.
THIRD EDITION OF WRITING GUIDE AVAILABLE
To help prospective authors, especially those who are beginning in medical journal writing, to produce better papers, this journal has developed “A Guide to Writing for Obstetrics & Gynecology.” Copies of the third edition of this booklet are available, free of charge, from: Editorial Office, Obstetrics & Gynecology, PO Box 70410, Washington, DC 20024-0410. Requests may also be submitted by FAX: (202) 479-0830, or E-mail: [email protected].
254 Brewster et al
Treatment of Cervical Cancer
Obstetrics & Gynecology
248 0029-7844/01/$20.00 PII S0029-7844(00)01117-0
The effect of widespread application of Papanicolaou smears in cervical cancer screening has resulted in a dramatic reduction of cervical cancer rates since the 1950s. Nevertheless, approximately 12,800 new cases of invasive cervical carcinoma will be diagnosed this year in the United States, and an estimated 4800 women will die of this disease during the same period.1 The conventional treatment of stage Ib and IIa cervical carcinoma, which constitutes at least half of all newly diagnosed cases,2 consists of radical hysterectomy and bilateral pelvic lymph node dissection, or radiation therapy combining whole pelvic teletherapy with local brachytherapy. Several studies comparing those treatment modalities, typically performed in academic settings, found both options to be equally efficacious with respect to local control and survival.3–5 Definitive surgical treatment may be selected more often than radiotherapy in the treatment of early-stage cervical cancer in younger women because there is cessation of ovarian function in younger women after radiation but not after surgery. In addition, the late complications of radiation are avoided when patients are treated with surgery alone. During the past 10 years the delivery of health care has become increasingly fragmented. Cancer care often has been directed away from tertiary-care institutions to smaller community-based centers that might be less likely to use radiotherapeutic techniques, dosing schedules, and state-of-the-art equipment that is used in larger academic institutions. In contrast, the rigorous requirements in a limited number of approved training programs for standardized training of gynecologic oncologists who perform most operations in women with cervical cancers might help preserve the excellent survival results for early-stage cervical cancer patients who
Obstetrics & Gynecology
have primary operative treatment in the general population. It is possible, therefore, that the change in these patterns of care might affect outcome adversely for women treated with radiotherapy compared with those who have surgery in the community setting. In order to characterize the population-based treatment patterns and survival of women with early cervical cancers treated in the United States with radiation or surgery, we reviewed the 1973–1995 data from the Surveillance, Epidemiology, and End Results program of the National Cancer Institute. This program includes nine regional tumor registries that abstract information from medical records on all cancers diagnosed within specific geographic areas. Most patients in the program are treated in the community setting; however, definitive numbers of patients treated in the community compared with those cared for in academic settings cannot be determined because the site of care is not mentioned in the database. The regions include Connecticut; Iowa; Arizona; Hawaii; Detroit, Michigan; San Francisco and Oakland, California; Salt Lake City, Utah; Atlanta, Georgia; and Seattle and Puget Sound, Washington, which together comprise 13.9% of the United States population. These regional registries are selected for their ability to accurately abstract data from their unique geographic areas. These nine areas are believed to provide a representative cross-section of the United States population. Therefore, this database permits population-based analyses of treatment practices, the results of which can be generalized throughout the United States.
Materials and Methods Abstracted cervical cancer information was extracted from the Surveillance, Epidemiology, and End Results 1995 public-use CD-ROM. Because cervical cancer staging as defined by the International Federation of Gynecology and Obstetrics (FIGO) was not coded before 1988, the study period extended from 1988 to 1995, thereby excluding all cancers diagnosed between 1973 and 1987 in that database. The database did not identify any subjects treated with a simple hysterectomy followed by a radical parametrectomy with or without lymphadenectomy. Only women between age 15 and 80 years with stage Ib or IIa cervical cancer as the first or only invasive malignancy were selected. Subjects who had standard hysterectomy before radiotherapy were excluded based on the assumptions that the diagnosis of cervical cancer was not made before surgery and that stage of disease for those women was uncertain. Also excluded were women who did not receive treatment for their cervical cancer. Furthermore, there was no way to determine which patients were treated by planned
VOL. 97, NO. 2, FEBRUARY 2001
combined multimodality therapy of radiotherapy followed by surgery. Abstracted data that provided information on the type of surgery, type of radiation, and surgeryradiation sequence were analyzed simultaneously to stratify subjects on the basis of the intent to treat, either with surgery or radiotherapy. Subjects were excluded if they received both radiation and surgery and the treatment sequence was unknown. Subjects were stratified into the surgical intent-to-treat arm if any abdominal or pelvic exploration was first in the treatment sequence. Subjects who had minor cervical procedures (eg, examination under anesthesia or biopsies) before radiotherapy, received only radiation therapy, or had radiotherapy followed by surgery were assigned to the radiation intent-to-treat arm. The subset of patients included in the survival analyses were those with cervical cancer diagnosed before 1994 for whom follow-up data were available, thereby permitting a minimum of 2 years follow-up for all study subjects. In the database, only one International Classification of Disease (ICD) code was used for cause of death. The coded cause of death for each deceased woman within this cohort was reviewed. Eight subjects were excluded from survival analyses because the cause of death was not recorded. One subject was reported to have died of chronic renal failure. There were no documented causes of death from hemorrhage, obstructive renal failure, or sepsis. All survival analyses were adjusted for cause of death from cervical cancer. Given the large size of this study group, it is unlikely that there was unequal misclassification of cause of death from cervical cancer between the two treatment groups. Statistical analyses were done using SAS (SAS Institute, Cary, NC). 2 tests were used for categorical group comparisons. Survival analyses were performed using the Kaplan-Meier method, for which the Wilcoxon log rank test was used to test survival hypotheses. Cox regression analysis, adjusted for cause of death from cervical cancer, was used for multivariable analyses. Only P ⬍ .05 by two-tailed analysis was considered statistically significant.
Results By the methods of subject selection described above, 1228 subjects whose cancer was diagnosed between 1988 and 1995 were identified from the 1973–1995 Surveillance, Epidemiology, and End Results publicuse file. One hundred eighty-nine subjects were excluded, including 34 with tumors invading less than 5 mm, 88 who received no further treatment after simple hysterectomy, 55 who received radiotherapy after a simple hysterectomy, and 12 who did not receive
Brewster et al
Treatment of Cervical Cancer
249
Table 1. Demographic Characteristics of the Study Population of 1039 Women
Table 2. Tumor Size and Treatment Intent by Age Category Variable n (%)
Age at diagnosis (y) ⱕ40 ⱖ41 Cervix cancer tumor size (cm) ⱕ4 ⬎4 Treatment intent Surgery Non-Hispanic white (n ⫽ 509) Black (n ⫽ 74) Asian (n ⫽ 66) Hispanic (n ⫽ 55) Unknown (n ⫽ 37) Radiotherapy Non-Hispanic white (n ⫽ 187) Black (n ⫽ 40) Asian (n ⫽ 30) Hispanic (n ⫽ 10) Unknown (n ⫽ 31)
453 (43.6) 586 (56.4) 763 (73.4) 276 (26.6) 741 (71.3)
298 (29.8)
treatment. Of the remaining 1039 subjects, 869 had stage Ib (84%) and 170 had stage IIa (16%) disease. The race or ethnicity was known for 992 (95%) of the women in this study; 696 (67%) were non-Hispanic white, 114 (11%) were black, 86 (8%) were Hispanic, and 96 (9%) were Asian. Other demographic characteristics are presented in Table 1. Pelvic and/or periaortic lymph node removal was done in 96% of the subjects who had primary treatment by surgical intent with a mean of 23 lymph nodes examined. There was no significant difference in the frequency of lymph node evaluation based on tumor size or age at diagnosis. When the entire study cohort was stratified by age and treatment intent, a greater percentage of younger women (40 years or less) had a surgical intent-to-treat preference than radiation intent-to-treat (82% and 18%, respectively). There was also a surgical treatment preference over radiation for women older than 40 years of age (63% and 37%, respectively); however, the younger women were significantly more likely than the older women to be treated surgically than with radiotherapy (82% compared with 63%, P ⫽ .001). When subjects were stratified by tumor size, the younger cohort with tumors 4 cm or less was more likely to be treated surgically than by radiotherapy (93% and 7%, respectively), as was the older cohort (primary surgery compared with radiotherapy, 71% and 29%). Younger women with smaller cancers were significantly more likely than their older counterparts to have a surgical intent-to-treat preference (93% compared with 71%, P ⫽ .001). For subjects with cervical cancers greater than
250 Brewster et al
Treatment of Cervical Cancer
Age (y) ⱕ40 ⱖ41 Tumor size (cm) ⱕ4 ⱕ40 y ⱖ41 y ⬎4 ⱕ40 y ⱖ41 y
Surgery n (%)
Radiation n (%)
371 (82) 370 (63)
82 (18) 216 (37)
312 (93) 303 (71)
23 (7) 125 (29)
59 (50) 67 (42)
59 (50) 91 (58)
4 cm, there was no significant difference in treatment choices of primary surgery compared with radiotherapy for either the younger (50% and 50%) or older (42% and 58%) cohorts (Table 2). For black, Hispanic, and non-Hispanic white women, there was a primary surgical preference compared with radiotherapy for those with tumors of 4 cm or less (P ⬍ .01); however, the differences were not significant among Asian subjects (P ⫽ .5). Of 1039 subjects, 784 (75%) met the criteria for inclusion in the survival analyses; 577 had tumors smaller than 4 cm (74%) and 207 had larger tumors. Five hundred forty-eight of the 784 patients were in the surgical intent-to-treat group, 27% of whom (150) received postoperative radiotherapy. Twenty-two percent (102 of 460) of the women with a surgical intent-to-treat who had small cervical cancers and 55% (48 of 88) of the women with large tumors in the surgical intent-to-treat group received adjuvant radiotherapy. Two hundred thirty-six patients were treated with primary radiotherapy, 21% of whom (50) had elective surgery after radiotherapy. The 5-year actuarial survival for the entire cohort, surgical intent-to-treat compared with radiation intentto-treat, was 84% and 69%, respectively (P ⬍ .001). Among women with cervical cancers of 4 cm or less, there was a statistically significant survival advantage for the women in the surgical intent-to-treat category. The 5-year survival adjusted for cause of death from cervical cancer for the surgery intent-to-treat group was 86% compared with 71% for radiation intent-to-treat (P ⬍ .001) (Figure 1). However the 5-year survival for women with cervical cancers larger than 4 cm was not significantly different for the intent-to-treat categories (surgery 72%, radiotherapy 68%) (P ⫽ .28) (Figure 2). Age did not influence 5-year survival in the surgical intent-to-treat group when analyzed for each tumor size category (Figures 3 and 4). The survival advantage for women with smaller cancers treated primarily with surgery persisted in multivariable survival analyses controlling for age as a
Obstetrics & Gynecology
Figure 3. Five-year survival for women with tumors smaller than 4 cm. Thin line indicates age 40 years or less and thick line indicates age over 40 years. Figure 1. Five-year survival adjusted for death caused by cervical cancer. Thin line indicates surgery intent-to-treat and thick line indicates radiation intent-to-treat.
cancers was not significantly different for either treatment category (RR 1.17, P ⫽ .6) (Table 3).
Discussion
continuous variable and tumor size in millimeters. Women who received primary radiotherapy were more likely to die of cancer than those treated with surgery (relative risk [RR] 2.75, 95% confidence interval [CI] 1.53, 4.94). The RR of death for women with larger
Morley et al6 retrospectively analyzed survival statistics for 446 women with stage Ib cervical cancer treated primarily with radiotherapy or surgery over a period of 31 years at the University of Michigan Medical Center.
Figure 2. Five-year survival for women with tumors larger than 4 cm. Thin line indicates surgery intent-to-treat and thick line indicates radiation intent-to-treat.
Figure 4. Five-year survival for women with tumors larger than 4 cm. Thin line indicates age 40 years or less and thick line indicates age over 40 years.
VOL. 97, NO. 2, FEBRUARY 2001
Brewster et al
Treatment of Cervical Cancer
251
Table 3. Risk of Death Variable Cervical tumor size ⱕ4 cm (n ⫽ 577) Age at cervical cancer diagnosis (y) Treatment Surgery (n ⫽ 460) Radiotherapy (n ⫽ 117) Tumor size (mm) Cervical tumor size ⬎4 cm (n ⫽ 207) Age at cervical cancer diagnosis (y) Treatment Surgery (n ⫽ 88) Radiotherapy (n ⫽ 119) Tumor size (mm)
Risk ratio
95% Confidence interval
0.99
0.97, 1.01
1.00 2.42 1.04
– 1.41, 4.12 1.02, 1.07
1.00
0.98, 1.03
1.00 1.26 1.00
0.72, 2.22 0.998, 1.04
Cox proportional analyses. Age and tumor size in this model are continuous variables. The risk ratio indicates the unit increase in risk associated with a unit increase in the continuous variable controlling for the other variables within the statistical model. Registry site and race or ethnicity were included in the model but were not prognostic variables.
Although they did not separate patients into groups based on prognostic indicators such as tumor size or volume, there was no difference in the 5-year survival for women treated by radical hysterectomy and pelvic lymphadenectomy compared with those treated with traditional radiotherapy. A prospective randomized trial at the University of Milan also compared radiation therapy with radical hysterectomy and pelvic lymph node dissection in patients with stage Ib or IIa cervical cancer. Multivariable analyses controlling for tumor size (4 cm or smaller compared with larger than 4 cm) found equivalent overall survival rates for the two treatment groups.7 Despite the consistency between these two large trials and many other similar studies, the treatment results from the Surveillance, Epidemiology, and End Results database did not confirm that surgery and radiotherapy are equally effective treatment modalities for women with stage Ib or IIa cervical cancer. In this study, after controlling for age and tumor size within the cohort of women with smaller tumors, there was a significant 5-year survival advantage for those treated surgically. Among the women with larger cancers there was no significant difference in 5-year survival between the groups. Treatment results of early-stage disease (Ib1) typically reflect the ability to control disease locally, whereas for bulkier disease, failures often occur regionally or at distant sites. The different observations might reflect better local control of early lesions but equivalent regional control of surgery and radiotherapy. The treatment of bulky cervical cancers is controversial. Indeed this is shown by the high percentage of surgery intentto-treat subjects who received adjuvant radiotherapy. It
252 Brewster et al
Treatment of Cervical Cancer
is possible that future trials will show that chemoradiation is superior to surgery. The benefit of surgery in the treatment of smaller cervical cancers shown in this study might represent the difference in standardization of treatments between surgery and radiation in the general population. Radical hysterectomy with pelvic lymph node dissection is the surgical procedure used to treat cervical cancer, and a standardized technique has been taught through fellowship training at only 24 –26 institutions throughout the United States.8 This procedure rarely is performed in the community setting by physicians who have not had formal training in a gynecologic oncology fellowship. This operation involves resection of the uterus, parametria, upper vagina, cardinal and uterosacral ligaments, and pelvic and periaortic lymph nodes in a consistent fashion. Although the technique can be modified on the basis of prognostic factors, such as tumor volume, to remove less parametrial tissue and less vaginal tissue in women with very early cancers, these differences have not been shown to adversely affect prognosis.9 Conversely, delivery of appropriate radiotherapy can be associated with subtle differences in treatment planning technique, portal design, radiation target volume, vaginal brachytherapy, total dose administered, duration of treatment, and technique, all of which can influence treatment results markedly. This assertion is supported by a recent report by Eifel et al10 on the method of radiotherapy in use for patients with cervical cancer at academic and nonacademic facilities. A review of 61 facilities that treat cervical cancer found that 57% treated fewer than three cases per year. Eightythree percent of nonacademic and 5% of academic centers treated fewer than three newly diagnosed cases per year. Patients at academic centers consistently received higher doses of radiation than those in the community setting. Thus, the differences between radiotherapy survival statistics from the Surveillance, Epidemiology, and End Results database and those cited in numerous publications reflecting the experience of investigators in academic settings might have resulted from different treatment approaches used in the general population compared with those used in academic settings. The reduction in survival seen with radiotherapy might not have been found in patients who were treated surgically in the community setting because surgical treatment approaches do not differ appreciably between academic and community settings. The analysis of this trial was based on an intent-totreat design, which often classifies subjects into treatment groups that do not reflect the actual treatment received by subjects in a given treatment category.
Obstetrics & Gynecology
Consequently, therapeutic benefits might be underestimated or overestimated and clinical reality distorted. One possible conclusion of the results of this analysis is that the two treatment modalities might be equally efficacious rather than the surgery intent-to-treat strategy being superior to the radiation intent-to-treat, but the differences in outcome might be attributable to other factors that could not be evaluated. In this analysis several important factors such as age, ethnicity, and tumor size were controlled, but no adjustment could be made for factors such as geometric distribution of the cancer within the cervix. Large exophytic tumors might respond differently to radiotherapy, for example, than barrel-type lesions within the cervix. Furthermore, subjects were not stratified by histologic type, and no adjustment could be made for covariables such as smoking history, vascular disease, obesity, and socioeconomic differences. Nevertheless, all those factors are generally viewed as of lesser prognostic importance than tumor size, which is the most important prognostic variable for a given stage of disease.11 This analysis confirmed that tumor size in general and a 4-cm separation of tumor size in particular are important independent prognostic variables, consistent with the 1994 FIGO staging system. A physician’s choice of primary treatment modality for stage Ib and IIa cervical cancer is based on multiple factors, including individual interpretation of published trials, regional resources for delivery of care, patient bias, physician tradition or bias, and concurrent comorbid patient factors, including obesity and age. Furthermore, treatment differences lead to various types of morbidities that can also influence the choice. For example, radical surgery permits the preservation of gonadal function. In addition, fistulae are more easily corrected when they complicate surgery compared with radiotherapy. Surgery also allows better assessment of the lymph nodes, which is of great prognostic importance. Only recently have data challenged the assumption that the adverse effects on sexual function can be minimized in patients who have radical hysterectomy compared with those treated with radiotherapy.12 Conversely, radiotherapy is an excellent treatment choice for patients who are not good surgical candidates by virtue of comorbidities of age, weight, and concurrent diseases. In this population-based analysis, women with cervical tumors of 4 cm or less were more likely to be treated surgically than by radiotherapy irrespective of age. Although surgery was preferred to radiotherapy among older women with smaller cervical cancers, the difference was even more dramatic among younger women. For larger cervical cancers, primary radiotherapy was used significantly more frequently than surgery in those
VOL. 97, NO. 2, FEBRUARY 2001
older than 40 years, and as often as surgery in women 40 years or less. This distribution of treatment choices might reflect the higher incidence of comorbidities in the older population, thereby increasing the risk of surgical complications. This is further complicated by the uncertainty of the relative efficacy of both approaches with larger cancers. Presumably the decision to treat women with larger tumors with radiotherapy reflects the belief that radiotherapy is at least equally effective as surgery for these lesions. In fact, the Surveillance, Epidemiology, and End Results data suggest that the two approaches might be approximately equal in the general population of women with this condition despite data to the contrary from many academic centers. It is now recognized that the addition of chemotherapy to the radiotherapeutic regimen for cervical cancer improves survival in women with large Ib or IIa cancers or pelvic lymph node metastases compared with radiotherapy alone.13 It is possible that the indications for chemotherapy will extend its use to women with smaller tumors also receiving primary radiotherapy. However, the study interval of our report did not include the period during which chemotherapy was considered standard of care in conjunction with radiotherapy. Thus, it is possible that radiotherapy combined with chemotherapy might provide survival results for the general population that equal or exceed those of surgery. The current study suggests that surgery as the primary treatment modality for women with smaller stage Ib (ie, stage Ib1) and IIa cervical cancer as delivered in the general United States population can provide a significantly better 5-year survival than radiotherapy.
References 1. Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics. CA Cancer J Clin 1999;49:12–3. 2. National Cancer Institute. Surveillance, Epidemiology, and End Results Cancer statistics review 1973–1995. Bethesda, Maryland: National Cancer Institute, 1995. 3. Delgado G, Bundy BN, Zaino R, Sevin BU, Creasman WT, Major F, et al. A prospective surgical pathological study of stage I squamous carcinoma of the cervix: A Gynecologic Oncology Group study. Gynecol Oncol 1990;38:352–7. 4. Piver MS, Marchetti DL, Patton T, Halpern J, Blumenson L, Driscoll DL. Radical hysterectomy and pelvic lymphadenectomy versus radiation therapy for small (⬍3 cm) stage IB cervical carcinoma. Am J Clin Oncol 1988;11:21– 4. 5. Hopkins MP, Morley GW. Radical hysterectomy versus radiation therapy for stage IB squamous cell cancer of the cervix. Cancer 1991;98:272–7. 6. Morley GW, Seski JC. Radical pelvic surgery versus radiation therapy for stage I carcinoma of the cervix (exclusive of microinvasion). Am J Obstet Gynecol 1976;126:785–98. 7. Landoni F, Maneo A, Colombo A, Placa F, Milani R, Perego P, et al.
Brewster et al
Treatment of Cervical Cancer
253
8.
9.
10.
11.
12.
13.
Randomised study of radical surgery versus radiotherapy for stage Ib-IIa cervical cancer. Lancet 1997;350:535– 40. Piver MS, Rutledge F, Smith JP. Five classes of extended hysterectomy for women with cervical cancer. Obstet Gynecol 1974;44:265– 72. Photopulos GJ, Zwaag RV. Class II radical hysterectomy shows less morbidity and good treatment efficacy compared to class III. Gynecol Oncol 1991;40:21– 4. Eifel PJ, Moughan J, Owen J, Katz A, Mahon I, Hanks GE. Patterns of radiotherapy practice for patients with squamous carcinoma of the uterine cervix: Patterns of care study. Int J Radiat Oncol Biol Phys 1999;43:352– 8. Perez CA, Grigsby PW, Nene SM, Camel HM, Galakatos A, Kao MS, et al. Effect of tumor size on the prognosis of carcinoma of the uterine cervix treated with irradiation alone. Cancer 1992;69:2796 – 806. Bergmark K, Avall-Lundquvist E, Dickman PW, Henningson L, Steineck G. Vaginal changes and sexuality with a history of cervical cancer. N Engl J Med 1999;340:1383–9. Morris M, Eifel P, Lu J, Grigsby PW, Levenback C, Stevens RE, et al. Pelvic radiation with concurrent chemotherapy compared with pelvic and para-aortic radiation for high risk cancer. N Engl J Med 1991;325:1371–2.
NEW,
Address reprint requests to:
Wendy R. Brewster, MD, PhD University of California, Irvine Medical Center The Chao Family Comprehensive Cancer Center Department of Obstetrics and Gynecology Division of Gynecologic Oncology 101 The City Drive, Building 23, Room 107 Orange, CA 92868 E-mail: [email protected]
Received March 23, 2000. Received in revised form July 21, 2000. Accepted August 10, 2000.
Copyright © 2001 by The American College of Obstetricians and Gynecologists. Published by Elsevier Science Inc.
THIRD EDITION OF WRITING GUIDE AVAILABLE
To help prospective authors, especially those who are beginning in medical journal writing, to produce better papers, this journal has developed “A Guide to Writing for Obstetrics & Gynecology.” Copies of the third edition of this booklet are available, free of charge, from: Editorial Office, Obstetrics & Gynecology, PO Box 70410, Washington, DC 20024-0410. Requests may also be submitted by FAX: (202) 479-0830, or E-mail: [email protected].
254 Brewster et al
Treatment of Cervical Cancer
Obstetrics & Gynecology