- Email: [email protected]
Association of number of positive nodes and cervical stroma invasion with outcome of advanced endometrial cancer treated with chemotherapy or whole abdominal irradiation: A Gynecologic Oncology Group study
Gynecologic Oncology 125 (2012) 87–93
Contents lists available at SciVerse ScienceDirect
Gynecologic Oncology journal homepage: www.elsevier.com/locate/ygyno
Association of number of positive nodes and cervical stroma invasion with outcome of advanced endometrial cancer treated with chemotherapy or whole abdominal irradiation: A Gynecologic Oncology Group study ☆ ,☆☆ ,★ ,★★ Krishnansu S. Tewari a,⁎, Virginia L. Filiaci b, Nick M. Spirtos c, Robert S. Mannel d, J. Tate Thigpen e, Michael L. Cibull f, Bradley J. Monk g, Marcus E. Randall f a
University of California Medical Center, Irvine, CA 92868 USA GOG Statistical and Data Center, Buffalo, NY 14263, USA c Women's Cancer Center of Nevada, Las Vegas, NV 89109, USA d University of Oklahoma, Oklahoma City, OK 73190, USA e University of Mississippi Medical Center, Jackson, MS 39216, USA f University of Kentucky, Lexington, KY 40536, USA g Creighton University School of Medicine, Phoenix, AZ 85013, USA b
a r t i c l e
i n f o
Article history: Received 22 August 2011 Available online 8 December 2011 Keywords: Endometrial cancer Radiotherapy Chemotherapy Single positive lymph node Cervical stroma
a b s t r a c t Objective. To determine whether the number of positive pelvic nodes (PPN), cervical stromal involvement (CSI), and/or lymphovascular space involvement (LVSI) were prognostic factors among women with advanced endometrial carcinoma treated with adriamycin plus cisplatin (AP) or whole abdominal irradiation (WAI). Methods. Data were abstracted from records of patients treated with adjuvant WAI or AP in a GOG randomized trial. Cox proportional hazards models were used to estimate the association of CSI and PPN with differences in PFS and OS while adjusting for treatment and previously studied factors. Results. WAI was randomly allocated to 202 and AP to 194 eligible patients. CSI (n = 93 total) was associated with a 44% increase in risk of progression and a 33% increase in risk of death. There was a trend for increasing number PPN being associated with a 7% per positive node increase in risk of progression/death. For CSI, the estimated unadjusted treatment hazard ratios (HRs) were: PFS 0.85 (0.53, 1.38); OS 0.81 (0.50, 1.33). For metastatic disease limited to a single PPN (n = 25), the unadjusted HRs were: PFS 0.96 (0.34, 2.74); OS 0.73 (0.24, 2.18). The test of homogeneity of treatment effect (ie., AP vs WAI) across subgroups (CSI, number of positive pelvic nodes) was not statistically significant for either endpoint, thus supporting the superiority of chemotherapy as reported in the original manuscript. Conclusions. The presence of CSI and increasing number of PPN were associated with poor prognosis. On average, patients with CSI experienced improved PFS and OS when treated with AP relative to WAI. © 2011 Elsevier Inc. All rights reserved.
☆ This study was funded by the Gynecologic Oncology Group/Abraxis Oncology Young Investigator Award given to Dr. Tewari in 2007. ☆☆ This study was a plenary presentation at the 2010 Annual Meeting of the Society of Gynecologic Oncologists in San Francisco, CA and at the 2010 Biennial Meeting of the International Gynecologic Cancer Society in Prague. ★ This study was supported by National Cancer Institute grants to the Gynecologic Oncology Group (GOG) Administrative Office (CA 27469), the GOG Tissue Bank (CA 27469 and CA 11479) and the GOG Statistical and Data Center (CA 37517). ★★ The following GOG member institutions participated in this protocol: University of Alabama at Birmingham, Duke University Medical Center, Abington Memorial Hospital, Walter Reed Army Medical Center, Wayne State University, University of Minnesota Medical School, University of Mississippi, University of Colorado-Anschutz Cancer Pavilion, University of California at Los Angeles (UCLA), Fred Hutchinson Cancer Research Center, University of Pennsylvania Cancer Center, Penn State Milton S. Hershey Medical Center, University of Cincinnati Medical Center, University of North Carolina, University of Iowa Hospitals and Clinics, Southwestern Medical Center of Texas, Indiana University Cancer Center, Wake Forest University School of Medicine, Associates in Gynecologic Care, PC, University of California Medical Center at Irvine, Tufts-New England Medical Center, Rush University Medical Center, State University of New York at Brooklyn, Cleveland Clinic Foundation, State University of New York at Stony Brook, Washington University School of Medicine, Cooper Hospital University Medical Center, Columbus Cancer Council/Ohio State, University of Massachusetts Memorial Health Care, Fox Chase Cancer Center, Medical University of South Carolina, Women's Cancer Center of Nevada, University of Oklahoma, University of Virginia Health Sciences Center, University of Chicago, Tacoma General Hospital, Thomas Jefferson University Hospital, Mayo Clinic, University Hospitals Case Medical Center, Tampa Bay Cancer Consortium, Gynecologic Oncology Network (GON), and Ellis Fischel Cancer Center. ⁎ Corresponding author at: University of California Medical Center, Irvine, Chao Family Comprehensive Cancer Center, Department of Obstetrics & Gynecology, 101 The City Drive South, Bldg 56, Room 275, Orange, CA 92868, USA. Fax: + 1 714 456 7754. E-mail address: [email protected] (K.S. Tewari). 0090-8258/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2011.12.414
88
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
Table 1 A–C: Pathologic factors stratified by treatment. Frequency
WAI
ChemoRx
Total
A. Lymphovascular space invasion stratified by treatment LVSI not recorded 61 63 124 LVSI negative 36 43 79 LVSI positive 105 88 193 Total 202 194 396
PFS events
Deaths
78 52 132
72 48 125
B. Cervical stromal involvement PFS events and deaths stratified by treatment Cvx stroma not recorded 6 7 13 7 7 Cvx stroma negative 158 132 290 185 173 Cvx stroma positive 38 55 93 70 65 Total 202 194 396 C. Extra-uterine disease limited to positive pelvic lymph nodes PFS events and deaths stratified by treatment > 1 positive node 13 15 28 12 11 1 positive node 12 13 25 14 13 Total 25 28 53 LVSI: lymphovascular space involvement; WAI: whole abdominal irradiation; ChemoRx: chemotherapy; PFS: progression-free survival; Cvx: cervix.
Introduction In 2011, there will be an estimated 46,470 new cases of endometrial cancer in the United States and 8120 deaths [1]. In a study utilizing the Surveillance, Epidemiology, and End Results database from 1988 to 2001 containing 45,510 patients, Ueda et al. reported that during this 14-year period there was an increase in the proportion of patients dying from advanced cancers, attributing the trend to an increase in the incidence rate of advanced stage disease, high-risk histologic subtypes, and lack of adequate surgical staging [2,3]. For patients with recurrent, metastatic, or high-risk disease, the most recent National Comprehensive Cancer Network (NCCN) clinical practice guidelines list cisplatin plus doxorubicin (AP) with or without paclitaxel as category 1 regimens and also emphasize the importance of access to relevant clinical trials whenever possible [4]. In 2006, Randall et al. reported results from Gynecologic Oncology Group (GOG) Protocol 122 which was designed to compare AP to whole abdominal irradiation (WAI) for women with advanced stage endometrial carcinomas [5]. Over 400 patients with FIGO stage III or IV disease were enrolled on this phase III multi-center trial from 1992 to 2000 [5]. Importantly, the combination of AP was shown to significantly improve progression-free survival (PFS) (HR 0.67, 95% CI 0.52–0.87) and overall survival (OS) (HR 0.69, 95% CI 0.53–0.89) compared with whole abdominal irradiation [5]. Specifically, with a median follow-up of 74 months, at 5 years and adjusting for stage, 55% of patients treated with AP were predicted to be alive compared with 42% of those who received WAI [5]. Acute toxicity was observed
in the chemotherapy arm with greater frequency and severity than in the radiation treatment arm. The current study is a retrospective, exploratory analysis of data abstracted from reports prospectively collected for GOG 122. We hypothesized that certain uterine-specific surgicopathologic factors (i.e., lymphovascular space involvement (LVSI), cervical stromal involvement (CSI)) and the number of positive pelvic lymph nodes (PPN), would have a prognostic impact on PFS and OS among patients with advanced endometrial carcinoma. In addition, we also hypothesized that the superiority of AP reported in GOG 122 would be sustained among patients with LVSI, CSI, and/or a single PPN.
Patients and methods Patients providing written informed consent and enrolled on GOG 122 with International Federation of Gynecology and Obstetrics (FIGO) stage III or IV endometrial carcinoma [6] of any histology were eligible for this trial. Eligibility required total abdominal hysterectomy and bilateral salpingoophorectomy, surgical staging, tumor resection, and no single site of residual tumor more than 2 cm. Nodal sampling was optional. Negative scalene node biopsies and chest computed tomography scans were required for patients with positive para-aortic lymph nodes. Patients with recurrent disease, hematogenous metastasis, inguinal lymph node involvement, or a history of pelvic or abdominal radiation therapy or chemotherapy were ineligible. The WAI arm prescribed postoperative radiotherapy with an open field AP–PA technique at a dose of 30 Gy in 20 daily fractions. After WAI, patients were to receive a boost of 15 Gy in 8 fractions to the true pelvis or to an extended field encompassing pelvic nodes and para-aortic nodes. A boost to both areas was to be administered to patients with positive pelvic nodes and no para-aortic node sampling or patients with neither pelvic lymph nodes nor para-aortic lymph node sampling. Doxorubicin 60 mg/m 2 plus cisplatin 50 mg/m 2 was prescribed every 3 weeks for eight cycles on the AP arm. The maximum allowable cumulative dose of doxorubicin was 420 mg/m 2; therefore, only cisplatin was to be infused during cycle eight. Patient and tumor characteristics including age, race, performance status, cell type, histologic grade, FIGO surgical stage, and residual tumor size were reported in the original publication. The original study received local IRB approval. Diagnostic slides were reviewed by the GOG Pathology Committee for eligibility (primary site and histology). For the current study, retrospective review of hospital pathology reports was conducted to abstract the following data: lymphovascular space involvement (LVSI), cervical stromal involvement (endocervical canal involvement alone was not counted), and lymph node information (e.g., number of nodes obtained, number of
Table 2 Multiple regression analysis of surgicopathologic prognostic factors. Prognostic factor
No lymph nodes identified in pathology sample or none retrieved Lymph nodes retrieved but not counted (average affect) Number of pelvic lymph nodes identified Number of positive pelvic lymph nodes identified Number of para-aortic nodes identified Number of positive para-aortic lymph nodes LVSI negative or missing LVSI positive Cervical stromal involvement negative or missing Cervical stromal involvement positive
Progression-free survival
Overall survival
Hazard ratio
95% CL
Hazard ratio
95% CL
Referent group 0.60 0.99 1.07 1.00 0.94 Referent group 1.03 Referent group 1.44
– 0.33–1.11 0.98–1.01 1.00–1.15 0.98–1.02 0.85–1.04 – 0.79–1.35 – 1.07–1.93
Referent group 0.59 0.99 1.06 1.00 0.95 Referent group 1.26 Referent group 1.33
– 0.30–1.15 0.97–1.00 0.99–1.14 0.98–1.02 0.85–1.05 – 0.96–1.66 – 0.98–1.80
Models are adjusted for treatment, stage, age, race, cell type, and grade (PFS and OS), and residual disease (PFS) , all of which were statistically significant as reported in the original clinical manuscript; 134 patients are missing LVSI and 13 patients are missing cervical stromal involvement; nodal dissection was not required for entry onto GOG 0122. CI: confidence limit; LVSI: lymphovascular space involvement.
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
positive nodes, clinically positive vs. microscopically positive, and sites of positive nodes). A Cox proportional hazards model [7] was used to estimate the association of malignant lymph node counts with the log hazard of progression or death (PFS) and the log hazard of death (OS) while adjusting for treatment, stage, residual disease, age, race, cell type, grade and positive cytology as reported in the original clinical manuscript. The reference group for the analysis of lymph node counts is defined as patients with no lymph nodes retrieved or no lymph nodes identified in the surgical pathology specimen as reported by the local pathologist reviewing the case. An indicator for patients who
A
had nodes retrieved but for whom no counts were reported was used to estimate the average association of those patients who had nodes retrieved but not counted. When documented, total node counts, up to a maximum of 33 pelvic and 12 para-aortic lymph nodes (95th percentiles), and positive node counts were included in the model. The minimum value of the actual node count and the 95th percentile of counts were used in the model. A linear relationship between the numbers of nodes retrieved and the log hazard for both pelvic and para-aortic nodes was assumed; these were assessed using plots of the smoothed regression of lymph node counts on the martingale residuals from a model omitting the counts. Patients with
B
Progression-Free Survival
Survival by Lymph-Vascular Space Involvement 1.0
1.0 0.9
LVSI Negative Positive
0.8
0.9
Censor Event Total 27 52 79 61 132 193
0.8
Proportion Surviving
Proportion Surviving Progression-Free
by Lymph-Vascular Space Involvement
0.7 0.6 0.5 0.4 0.3
0.7 0.6 0.5 0.4 0.3
0.2
0.2
0.1
0.1
0.0
0.0
0
12
24
36
48
60
72
84
96
108
120
LVSI Negative Positive
0
12
24
Censor Event Total 31 48 79 68 125 193
36
Months on Study
72
84
96
108
120
96
108
120
by Cervical Stroma Involvement 1.0
1.0 0.9
Stroma Negative Positive
0.8
0.9
Censor Event Total 105 185 290 23 70 93
0.8
Proportion Surviving
Proportion Surviving Progression-Free
60
Survival
D
by Cervical Stroma Involvement
0.7 0.6 0.5 0.4 0.3
0.7 0.6 0.5 0.4 0.3
0.2
0.2
0.1
0.1
Stroma Negative Positive
0.0
0.0
0
12
24
36
48
60
72
84
96
108
120
0
12
24
Censor Event Total 117 173 290 28 65 93
36
Months on Study
48
60
72
84
Months on Study
Progression-Free Survival Among Patients with Metastasis limited to Pelvic Nodes
Survival
F
Among Patients with Metastasis limited to Pelvic Nodes 1.0
1.0 0.9
# Nodes + >1 Node + 1 Node +
0.8
0.9
Censor Event Total 16 12 28 11 14 25
0.8
Proportion Surviving
Proportion Surviving Progression-Free
48
Months on Study
Progression-Free Survival
C
E
89
0.7 0.6 0.5 0.4 0.3
0.7 0.6 0.5 0.4 0.3
0.2
0.2
0.1
0.1
# Nodes + >1 Node + 1 Node +
0.0
0.0
0
12
24
36
48
60
72
Months on Study
84
96
108
120
0
12
24
Censor Event Total 17 11 28 12 13 25
36
48
60
72
84
96
108
120
Months on Study
Fig. 1. A–F: Kaplan Meier survival curves for pathologic factors. A. Progression-free survival stratified by lymphovascular space involvement. B. Overall survival stratified by lymphovascular space involvement. C. Progression-free survival stratified by cervical stromal involvement. D. Overall survival stratified by cervical stromal involvement. E. Progressionfree survival stratified by metastases to pelvic lymph nodes. F. Overall survival stratified by metastases to pelvic lymph nodes.
90
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
Results
for PFS and OS treatment hazard ratios all included 1. The estimated treatment hazard ratios in patients with metastasis limited to 2 or more positive pelvic nodes favored WAI for PFS (HR = 1.25) and favored neither arm for OS (HR = 1.04). For patients with 1 single positive pelvic node, the estimate for OS (HR = .731) identified a trend which favored chemotherapy. The test for homogeneity of treatment effect across subgroups was not statistically significant for either endpoint (PFS: p = 0.72 OS: p = 0.66). The association of the number of PPN with PFS and OS in terms of treatment allocation appears in Table 3E–F. Figs. 3A and 3B depict forest plots of AP relative to WAI for PFS and OS, respectively for all three surgico-pathologic factors studied. The diamonds reflect the unadjusted treatment effect. The stage-adjusted 95% confidence intervals for both endpoints are: PFS HR 0.735, 95% CI 0.575, 0.939; OS HR 0.752, 95% CI 0.584, 0.968. The placement of the diamond or HR estimates to the left of the point of equal hazards would indicate a trend favoring superiority of AP over WAI in the study population.
Clinical material
Discussion
A total of 396 eligible patients were enrolled and treatment with either WAI (n = 202) or AP (n = 194) was randomly assigned. LVSI information was missing in 124 (31%) patients and cervical stroma data were missing in 13 (3%). For those in whom LVSI was identified, 105 were treated with WAI and 88 were treated with AP (Table 1A). For 202 women treated with WAI, 38 had CSI and 158 did not. For those who received AP, 55 had CSI and 132 did not (Table 1B). On the WAI arm, PPN were identified in 91 women and positive para-aortic nodes were identified in 40. On the AP arm, PPN were identified in 113 women and positive para-aortic nodes were identified in 49. Therefore there were at least 204 lymph node-positive subjects. In the absence of other metastatic disease, 25 patients were found to have only one PPN (n= 12 WAI, n = 13 AP) and 28 women found to have greater than one PPN (n= 13 WAI, n = 15 AP) (Table 1C).
In the past, most patients with endometrial cancer received radiotherapy either pre-operatively or in the adjuvant setting following surgery. This paradigm was challenged, first with the recognition that adjuvant radiotherapy only improved local control, and secondly with the identification of recurrence risk factors in early stage disease to identify who would benefit from localized adjuvant treatment with radiation. After many years, the pendulum has swung in the opposite direction. GOG 122, reported by Randall et al., represents one of those few clinical studies that results in a change in practice patterns for treatment of advanced disease [5,11–20]. In this study, AP significantly
Analysis of prognostic factors (entire study population)
Table 3 A–F. Unadjusted and stage-adjusted treatment hazard ratios and confidence limits for pathologic factor subgroups.
missing and negative LVSI were combined into one category. Therefore the hazard ratio reported for LVSI estimates the association relative to those with negative or missing values and not a clear association relative to a true negative reference group. The same was done for cervical stromal involvement but there were very few with missing data. The Kaplan–Meier method [8] was used to estimate the distribution of PFS and OS event times. Forest plots [9] were generated to display the consistency of treatment effect within subgroups and the original study's conclusions. Tests of treatment effect homogeneity [10] were also performed to identify potential signals of interest in the subgroups under investigation. All analyses were exploratory with no adjustment to account for multiple testing and control of type I error. The precision in the estimated treatment effects within small subgroups was limited and thus only confidence intervals are reported.
A multiple regression analysis of the surgicopathologic prognostic factors was performed (Table 2). CSI was associated with poorer PFS and OS, with a 44% increase in risk of progression or death and a 33% increase in risk of death. There was a trend for increasing number of PPN removed and identified being associated with a 7% increased risk of progression or death. Kaplan–Meier estimates are displayed in Figs. 1A–F. Analysis of surgicopathologic factors on endpoints of AP relative to WAI There was a trend for patients with LVSI and CSI to experience improved PFS and OS when treated with AP relative to WAI. Confidence intervals for PFS and OS treatment HRs in each subgroup defined by LVSI and CSI included 1. However the estimated unadjusted (and stage-adjusted) treatment hazard ratios favored AP for both endpoints. These estimates ranged between 0.75 and 0.95 for LVSI and 0.75 and 0.85 for CSI. Furthermore, the tests for homogeneity of treatment effect across subgroups were not statistically significant for either endpoint or factor (LVSI unadjusted PFS: p = 0.90; OS: p = 0.81 and CSI unadjusted PFS: p = 0.79; OS: p = 0.79) whether or not the analysis was adjusted for stage (Tables 3A–D). Patients who did not have CSI (n= 13) reported were excluded from each treatment effect analysis of CSI. The Kaplan–Meier survival curves for PFS and OS stratified by LVSI and treatment assignment appear in Figs. 2A and 2B, respectively, and those stratified by CSI and treatment assignment appear in Figs. 2C and 2D, respectively. There was a trend for patients with a single PPN to experience improved OS when treated with AP relative to WAI. Confidence intervals
Subgroup
Unadjusted Treatment hazard ratio
Stage adjusted 95% HR CL
Treatment hazard ratio
95% CL
A. Lymphovascular space invasion, PFS. LVSI not recorded 0.777 LVSI negative 0.855 LVSI positive 0.886
0.498–1.212 0.492–1.485 0.629–1.250
0.671 0.663 0.792
0.427–1.054 0.376–1.170 0.560–1.120
B. Lymphovascular space invasion, OS LVSI not recorded 0.752 LVSI negative 0.953 LVSI positive 0.791
0.473–1.197 0.536–1.696 0.555–1.128
0.712 0.946 0.730
0.446–1.135 0.531–1.686 0.511–1.043
C. Cervical stromal involvement, PFS Cvx stroma negative 0.790 Cvx stroma positive 0.852
0.590–1.058 0.528–1.376
0.672 0.835
0.500–0.905 0.517–1.348
D. Cervical stromal involvement, OS Cvx stroma negative 0.750 Cvx stroma positive 0.812
0.553–1.016 0.496–1.330
0.679 0.852
0.506–0.933 0.519–1.397
Subgroup
95% CL
Treatment HR
E. Metastases limited to pelvic lymph nodes, PFS > 1 positive node 1.273 0.403–4.019 1 positive node 0.959 0.336–2.741 F. Metastases limited to pelvic lymph nodes, OS > 1 positive node 1.049 0.319–3.444 1 positive node 0.731 0.245–2.183 LVSI: lymphovascular space involvement; Cx: cervix; HR: hazard ratio; CL: confidence limit; Rx: treatment.
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
Progression-Free Survival
A
Survival
B
by Lymph-Vascular Space Involvement 1.0
1.0 0.9
LVSI/Regimen Negative/WAI Negative/AP Positive/WAI Positive/AP Missing/WAI Missing/AP
0.8 0.7 0.6
Alive Event Total 13 23 36 14 29 43 32 73 105 29 59 88 20 41 61 26 37 63
0.5 0.4 0.3
0.9 0.8
Proportion Surviving
Proportion Surviving Progression-Free
by Lymph-Vascular Space Involvement
0.7 0.6 0.5 0.4
LVSI/Regimen Negative/WAI Negative/AP Positive/WAI Positive/AP Missing/WAI Missing/AP
0.3
0.2
0.2
0.1
0.1
Alive Event Total 15 21 36 16 27 43 33 72 105 35 53 88 22 39 61 30 33 63
0.0
0.0 0
12
24
36
48
60
72
84
96
108
120
0
12
24
36
Months on Study
C
48
60
72
84
96
108
120
108
120
Months on Study
D
Progression-Free Survival
Survival
by Cervical Stroma Involvement
by Cervical Stroma Involvement
1.0
1.0
0.9
Stroma/Regimen Censor Event Total Negative/WAI 52 106 158 Negative/AP 53 79 132 Positive/WAI 10 28 38 Positive/AP 13 42 55
0.8 0.7 0.6 0.5 0.4 0.3
0.9 0.8
Proportion Surviving
Proportion Surviving Progression-Free
91
0.7 0.6 0.5 0.4 0.3
0.2
0.2
0.1
0.1
Stroma/Regimen Censor Event Total Negative/WAI 56 102 158 Negative/AP 61 71 132 Positive/WAI 11 27 38 Positive/AP 17 38 55
0.0
0.0 0
12
24
36
48
60
72
84
96
108
120
Months on Study
0
12
24
36
48
60
72
84
96
Months on Study
Fig. 2. A–D. Kaplan Meier survival curves stratified by pathologic factors and treatment allocation. A: Progression-free survival stratified by lymphovascular space involvement and treatment allocation. B: Overall survival stratified by lymphovascular space involvement and treatment allocation. C: Progression-free survival stratified by cervical stromal involvement and treatment allocation. D: Overall survival stratified by cervical stromal involvement and treatment allocation.
improved both PFS (HR 0.67, 95% CI 0.52–0.87) and OS (HR 0.69, 95% CI 0.53–0.89) [5]. With a median follow-up of 74 months, at 5 years and adjusting for stage, 55% of patients treated with AP were predicted to be alive compared with 42% of those who received WAI [5]. The protocol effectively transformed the underpinnings of adjuvant therapy for metastatic disease. However, the concern over more frequent and severe acute toxicity [5] prompted the current subgroup analyses of the treatment effect in patients with a single PPN. Several retrospective studies have evaluated outcomes of advanced disease treated with radiation (regional or whole abdomen) with conflicting results. Klopp et al. noted that stage IIIC disease had a high rate of locoregional recurrence in patients not treated with regional radiotherapy, whereas those who underwent surgical staging followed by external beam irradiation had a high rate of cure [21]. Although their report is limited by small numbers, non-random treatment policy confounded with time and absence of longitudinal tracking, several hypotheses for differences between the conclusions reached in their study and in GOG 122. The latter trial included patients (14%) with up to 2 cm of residual tumor after debulking. They suggest that the dose delivered (30 Gy to the whole abdomen followed by a boost of 15 Gy to the pelvis) was inadequate to treat gross residual disease. Also, whole abdominal irradiation resulted in treatment interruption and early discontinuation in 16% of patients. Finally, GOG 122 allowed serous and clear cell tumors for which the risk of intraperitoneal dissemination is high and the need of systemic therapy is implicit. Conversely, stage migration due to improvements in radiologic imaging in the later years of Klopp's study was likely responsible for exclusion of patients found to have carcinomatosis.
Nelson et al. reported outcomes in 17 fully staged patients with extrauterine disease limited to pelvic nodes and pathologically negative para-aortic nodes treated post-operatively with pelvic (n = 13) or whole abdomen (n = 4) irradiation [22]. Their report suggested that adjuvant RT alone can be a viable treatment option in this welldefined patient population and that extended field RT might further improve results [22]. Gibbons et al. observed 7-year disease-free survivorship rates of 57.8% (stage III) and 25.0% (stage IV) following adjuvant WAI in 56 patients [23]. Acute toxicity was common but mild, and chronic toxicity remained almost entirely subclinical. Finally, Smith et al. also recognized a role for WAI in some patients with optimally debulked advanced-stage uterine cancer, citing 79% and 89% PFS and OS, respectively [24]. Important questions pertaining to lymphatic spread have not been addressed adequately [25–32]. The clinical significance of the extent of nodal dissection (commonly measured by the total number of lymph nodes identified by the pathologist) and the role of adjuvant therapy (WAI or AP) in the setting of a single positive node are debated. There are reports of patients with one or two positive nodes in a background of multiple negative nodes (10–12 or greater) being treated successfully with postoperative pelvic RT alone, however the percentage of those with minimal, residual advanced stage disease is not consistently detailed. Onda et al. reported on 30 node-positive patients including 10 with pelvic node metastases alone and 20 with para-aortic metastases; all of which underwent complete lymphadenectomy followed by adjuvant radiation for those with pelvic metastases alone or by pelvic plus extended field radiation plus chemotherapy for those with aortic nodal disease [25]. Five year survival
92
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
A
Hazard Ratios for Progression - Free Survival Rel Haz Var(In(HR)) Chemotherapy is better
LVSI
Missing Negative Positive
0.777 0.855 0.886
0.052 0.079 0.031
Pelvic Nodes
>1+ 1+
1.273 0.959
0.344 0.287
Negative Positive
0.790 0.852
0.022 0.060
Cervical stroma
0.33
0.5
0.67
WAI is better
1.0
1.5
2.0
3.
Relative Hazard Hazard Ratio Progression-Free Survival 0.735
B
95% Hazard Ratio Confidence Limits 0.575, 0.939
Hazard Ratios for Overall Survival Rel Haz Var(In(HR))
LVSI
Missing 0.752 Negative 0.953 Positive 0.791
0.056 0.036 0.033
Pelvic Nodes
>1+ 1+
1.049 0.731
0.368 0.312
Cervical Stroma
Negative 0.750 Positive 0.812
0.024 0.063
WAI is better
Chemotherapy is better
0.5
0.33
0.67
1.0
1.5
2.0
3.
Relative Hazard Hazard Ratio 95% Hazard Ratio Confidence Limits Overall Survival
0.752
0.584, 0.968
Fig. 3. A–B. Forest plots of treatment effect for progression-free survival and overall survival based on surgicopathologic factors. A: Progression-free survival. B: Overall survival. The red diamond reflects the overall unadjusted treatment effect with 95% confidence interval. LVSI: lymphovascular space involvement.
was 100% for those with positive pelvic nodes only and 75% for those with para-aortic disease [25]. Finally, in GOG 184 sequential adjuvant therapy (chemotherapy following volume-directed radiotherapy) was associated with increased toxicity without significant improvement in recurrence-free survival in women with advanced disease, including those with positive para-aortic nodes [20]. Takeshima et al. reported on 26 women with positive para-aortic nodal involvement who received postoperative ifosfamide, epiadriamycin, and cisplatin [27]. Their results are consistent with the findings of GOG 122, although the natural history of a single positive pelvic node following AP vs WAI remains obscure [27]. The results of GOG 122 suggest that systemic therapy with AP is superior in this setting, although acute toxicity, on average is increased but relatively manageable. The primary analysis of protocol 122 did not include separate analysis of the treatment effect on PFS and OS by either the number of lymph nodes removed or the setting of a single PPN. The sample size for such subgroup analyses in the current study is admittedly too small for definitive conclusions; with low precision, the confidence intervals are wide and do not rule out important differences. Additionally, the extent of lymph node dissection and node counting was not standardized for the original study.
The presence of LVSI in the uterus may also have prognostic implications and therefore could be used to guide clinical management in the adjuvant setting. As all of the subjects in GOG 122 had advanced disease, the presence or absence of LVSI in the uterus was not initially viewed to be an important entity in the original design of that protocol. Unfortunately, LVSI data for over 30% of the study subjects were not available. Interpretation of results in the presence of such a high proportion of missing data is problematic. Orezzoli et al. recently reported on 55 patients with CSI and 26 with endocervical gland involvement treated from 1993 to 2003 [33]. The five-year survival rate was 83% for patients with endocervical gland involvement, and 71% for those with CSI. Substaging into IIA vs IIB or depth of stromal invasion had no significant association with survival. Age (p= 0.001), LVSI (p= 0.017), and type of treatment (p= 0.022) were jointly associated with survival among these patients [33]. In a recent Surveillance Epidemiology and End Results database study by Schmid et al., the association of adjuvant radiotherapy and outcome for 943 women with stage IIIC endometrial cancer treated from 1988 to 2001 was evaluated [34]. The median number of lymph nodes removed was 11, and 54.9% had a single positive lymph node and 45.1% had 2–5 positive nodes. For those that received adjuvant
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
radiotherapy (67.3%), the 5-year disease-specific survival was 67.9% compared to 53.4% who did not receive radiotherapy (pb 0.001) [34]. For those with a single positive node, adjuvant radiotherapy was associated with improved survival; from 54.4% to 74.3% (pb 0.001) and for those with 2–5 positive nodes adjuvant radiation was also associated with improved survival from 52.4% to 59.7% (p= 0.089) [34]. There are a number of scientific issues concerning “node counting” studies. The actual number of nodes removed from an individual patient may be a function of the patient's body habitus, age, extent of disease based on gross surgical inspection, the surgeon's assessment of the value of thorough lymphadenectomy, the surgeon's skill level, and possibly the technique through which lymph node dissection is performed (i.e. laparotomy, laparoscopy, robotics). In addition, the total number of lymph nodes reported is also dependent on how the laboratory technician and/or pathologist dissect, submit and process the tissues retrieved from the operating room [35,36]. Often when evaluating the same patient the surgeon and the pathologist often do not agree about how many nodes were actually removed. While each of these features may determine the number of nodes removed, any or all of them may be prognostic by themselves. The current study cannot unravel causes from effects. For example, if a surgeon's skill level is related to the number of nodes that can be removed, his/her skill level may hypothetically have a direct effect of the patient's probability of recurrence, which biologically does not involve number of nodes removed, but is only correlated with it. Perhaps it is the unknown status of the nodes left behind, not those which are removed, that is prognostic. In this ancillary data analysis, CSI was determined to be a prognostic factor for both PFS and OS for the entire study population. Similarly, there was a trend for increasing number of PPN removed to be associated with a 7% increase in progression and/or death for all patients. In evaluating our hypothesis that the superiority of AP over WAI as reported in the original publication of GOG 122 [5] would be sustained in the presence of these surgicopathologic factors, tests of homogeneity were performed. Although the confidence intervals for the estimated unadjusted and stage-adjusted treatment HRs for subgroups included 1, because the tests of homogeneity of treatment effect were not statistically significant, the HRs are interpreted to favor AP for both endpoints among women with LVSI, CSI, and/or a single PPN. Importantly, in this study population, the superiority of AP over WAI for all three surgicopathologic factors is reflected in the forest plot diamonds being situated to the left of the point of equal hazards for PFS and for OS (Figs. 3A–B). Due to the limited number of patients in each subgroup there is limited precision in the estimated treatment effects. Nevertheless, there is no significant evidence that the improved PFS (HR 0.67, 95% CI 0.52–0.87) and OS (HR 0.69, 95% CI 0.53–0.89) conferred by AP as reported in GOG 122 [5] depends on LVSI, CSI, or PPN status/count. Conflict of interest statement The authors wish to report that they have no conflicts of interest with the exception of Dr. Nick Spirtos who wishes to disclose relevant financial relationships with Genzyme — Speaker's Bureau, Ethicon — Speaker's Bureau and Olympus — Speaker's Bureau.
References [1] Siegel R, Ward E, Brawley O, Jemal Ahmedin J. The impact of eliminating socioeconomic and racial disparities on premature cancer deaths. CA Cancer J Clin 2011;61:212–36. [2] Ueda SM, Kapp DS, Cheung MK, et al. Trends in demographic and clinical characteristics in women diagnosed with corpus cancer and their potential impact on the increasing number of deaths. Am J Obstet Gynecol 2008;198(218):e1–6. [3] Creasman WT. Revised FIGO staging for carcinoma of the endometrium. Int J Gynaecol Obstet 2009:105–9. [4] NCCN Practice Guidelines in Oncology — v.1.2010. www.nccn.org. [5] Randall ME, Filiaci VI, Muss H, et al. Randomized phase III trial of whole-abdominal irradiation versus doxorubicin and cisplatin chemotherapy in advanced endometrial carcinoma: a Gynecologic Oncology Group study. J Clin Oncol 2006;24:36–44. [6] Mikuta JJ. International Federation of Gynecology and Obstetrics staging of endometrial cancer 1988. Cancer 1993;71(4 Suppl):1460–3.
93
[7] Cox DR. Regression model and life tables (with discussion). J R Stat Soc B 1972;34: 187–219. [8] Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457–81. [9] Duchateau L, Sylvester R. Meta-analysis. In: Crowley John, editor. Handbook of statistics in clinical oncology. New York: Marcel Dekker, Inc.; 2001. p. 525–43. [10] Normand ST. Tutorial in biostatistics meta-analysis: formulating, evaluating, combining, and reporting. Stat Med 1999;18:321–59. [11] Randall ME, Reisinger S. Radiation therapy and combined chemo-irradiation in advanced and recurrent endometrial carcinoma. Semin Oncol 1994;1004(21):91–9. [12] Thigpen JT, Buchsbaum HJ, Mangan C, et al. Phase II trial of Adriamycin in the treatment of advanced or recurrent endometrial carcinoma: a Gynecologic Oncology Group study. Cancer Treat Rep 1979;63:21–7. [13] Thigpen JT, Blessing JA, Homesley H, et al. Phase II trial of cisplatin as first-line chemotherapy in patients with advanced or recurrent endometrial carcinoma. A Gynecologic Oncology Group study. Gynecol Oncol 1989;33:68–70. [14] Thigpen JT, Blessing JA, DiSaia PJ, et al. A randomized comparison of doxorubicin alone versus doxorubicin plus cyclophosphamide in the management of advanced or recurrent endometrial carcinoma: a Gynecologic Oncology Group study. J Clin Oncol 1994;12:1408–14. [15] Thigpen JT, Brady MF, Homesley HD, et al. Phase III trial of doxorubicin with or without cisplatin in advanced endometrial carcinoma: a Gynecologic Oncology Group study. J Clin Oncol 2004;22:3902–8. [16] Aapro MS, van Wijk FH, Bolis G, et al. Doxorubicin versus doxorubicin and cisplatin in endometrial carcinoma: definitive results of a randomized study (55872) by the EORTC Gynaecological Cancer Group. Ann Oncol 2003;14:441–8. [17] Homesley HD, Filiaci V, Gibbons SK, et al. A randomized phase III trial in advanced endometrial carcinoma of surgery and volume directed radiation followed by cisplatin and doxorubicin with or without paclitaxel: a Gynecologic Oncology Group study. Gynecol Oncol 2009;112:543–52. [18] Miller DS (study chair). GOG protocol 209: a randomized phase III trial of doxorubicin/ cisplatin/paclitaxel and G-CSF versus carboplatin/paclitaxel in patients with stage III & IV or recurrent endometrial cancer. http://www.cancer.gov/search/ViewClinicalTrials. aspx?cdrid=305940&version=HealthProfessional&protocolsearchid=8603435 [19] Matei D (study chair). GOG protocol 258: a randomized phase III trial of cisplatin and tumor volume directed irradiation followed by carboplatin and paclitaxel vs carboplatin and paclitaxel for optimally debulked, advanced endometrial carcinoma. http://www.cancer.gov/search/ViewClinicalTrials.aspx?cdrid=649079&version= HealthProfessional&protocolsearchid=7990461 [20] Homesley HD, Filiaci V, Gibbons SK, et al. A randomized phase III trial in advanced endometrial carcinoma of surgery and volume directed radiation followed by cisplatin and doxorubicin with or without paclitaxel: a Gynecologic Oncology Group study. Gynecol Oncol 2009;112:543–52. [21] Klopp AH, Jhingran A, Ramondetta L, et al. Node-positive adenocarcinoma of the endometrium: outcome and patterns of recurrence with and without external beam irradiation. Gynecol Oncol 2009;115:6–11. [22] Nelson G, Randall M, Sutton G, et al. FIGO stage III endometrial carcinoma with metastases confined to pelvic lymph nodes: analysis of treatment outcomes, prognostic variables, and failure patterns following adjuvant RT. Gynecol Oncol 1999;75:211–4. [23] Gibbons S, Martinez A, Schray M, et al. Adjuvant whole abdominopelvic irradiation for high risk endometrial carcinoma. Int J Radiat Oncol Biol Phys 1991;21:1019–25. [24] Smith RS, Kap DS, Chen Q, et al. Treatment of high-risk uterine cancer with whole abdominopelvic RT. Int J Radiat Oncol Biol Phys 2000;48:767–78. [25] Onda T, Yoshikawa H, Mizutani K, et al. Treatment of node-positive endometrial cancer with complete node dissection, chemotherapy and radiation therapy. Br J Cancer 1997;5:1836–41. [26] Lutman CV, Havrilesky LJ, Cragun JM, et al. Pelvic lymph node count is an important prognostic variable for FIGO stage I and II endometrial carcinoma with highrisk histology. Gynecol Oncol 2006;102:92–7. [27] Takeshima N, Umayahara K, Fujiwara K, et al. Effectiveness of postoperative chemotherapy for para-aortic lymph node metastasis of endometrial cancer. Gynecol Oncol 2006;102:214–7. [28] Bristow RE, Zahurak ML, Alexander CJ, et al. FIGO stage IIIC endometrial carcinoma: resection of macroscopic nodal disease and other determinants of survival. Int J Gynecol Cancer 2003;13:664–72. [29] Havrilesky LJ, Cragun JM, Calingaert B, et al. Resection of lymph node metastases influences survival in stage IIIC endometrial cancer. Gynecol Oncol 2005;99:689–95. [30] McMeekin DS, Tillmanns T. Endometrial cancer: treatment of nodal metastases. Curr Treat Options Oncol 2003;4:121–30. [31] Katz LA, Andrews SJ, Fanning J. Survival after multimodality treatment for stage IIIC endometrial cancer. Am J Obstet Gynecol 2001;184:1071–3. [32] Umesaki N, Tanaka T, Miyama M, et al. Postoperative adjuvant chemotherapy with cisplatin, etoposide, and pirarubicin for endometrial carcinoma patients with lymph node metastasis: a pilot study. Oncol Rep 2000;7:1083–6. [33] Orezzoli JP, Sioletic S, Olawaiye A, et al. Stage II endometrioid adenocarcinoma of the endometrium. Clinical implications of cervical stromal invasion. Gynecol Oncol 2009;113:316–23. [34] Schmid S, Hsu IC, Hu JM, et al. Adjuvant radiation therapy in stage III nodepositive uterine cancer. Gynecol Oncol 2009;115:239–43. [35] Ricciardi R, Baxter N. Association versus causation versus quality improvement: setting benchmarks for lymph node evaluation in colon cancer. J Natl Cancer Inst 2007;99:414–5. [36] Rieger NA, Barnett FS, Moore JW, et al. Quality of pathology reporting impacts on lymph node yield in colon cancer. J Clin Oncol 2007;25:463.
Contents lists available at SciVerse ScienceDirect
Gynecologic Oncology journal homepage: www.elsevier.com/locate/ygyno
Association of number of positive nodes and cervical stroma invasion with outcome of advanced endometrial cancer treated with chemotherapy or whole abdominal irradiation: A Gynecologic Oncology Group study ☆ ,☆☆ ,★ ,★★ Krishnansu S. Tewari a,⁎, Virginia L. Filiaci b, Nick M. Spirtos c, Robert S. Mannel d, J. Tate Thigpen e, Michael L. Cibull f, Bradley J. Monk g, Marcus E. Randall f a
University of California Medical Center, Irvine, CA 92868 USA GOG Statistical and Data Center, Buffalo, NY 14263, USA c Women's Cancer Center of Nevada, Las Vegas, NV 89109, USA d University of Oklahoma, Oklahoma City, OK 73190, USA e University of Mississippi Medical Center, Jackson, MS 39216, USA f University of Kentucky, Lexington, KY 40536, USA g Creighton University School of Medicine, Phoenix, AZ 85013, USA b
a r t i c l e
i n f o
Article history: Received 22 August 2011 Available online 8 December 2011 Keywords: Endometrial cancer Radiotherapy Chemotherapy Single positive lymph node Cervical stroma
a b s t r a c t Objective. To determine whether the number of positive pelvic nodes (PPN), cervical stromal involvement (CSI), and/or lymphovascular space involvement (LVSI) were prognostic factors among women with advanced endometrial carcinoma treated with adriamycin plus cisplatin (AP) or whole abdominal irradiation (WAI). Methods. Data were abstracted from records of patients treated with adjuvant WAI or AP in a GOG randomized trial. Cox proportional hazards models were used to estimate the association of CSI and PPN with differences in PFS and OS while adjusting for treatment and previously studied factors. Results. WAI was randomly allocated to 202 and AP to 194 eligible patients. CSI (n = 93 total) was associated with a 44% increase in risk of progression and a 33% increase in risk of death. There was a trend for increasing number PPN being associated with a 7% per positive node increase in risk of progression/death. For CSI, the estimated unadjusted treatment hazard ratios (HRs) were: PFS 0.85 (0.53, 1.38); OS 0.81 (0.50, 1.33). For metastatic disease limited to a single PPN (n = 25), the unadjusted HRs were: PFS 0.96 (0.34, 2.74); OS 0.73 (0.24, 2.18). The test of homogeneity of treatment effect (ie., AP vs WAI) across subgroups (CSI, number of positive pelvic nodes) was not statistically significant for either endpoint, thus supporting the superiority of chemotherapy as reported in the original manuscript. Conclusions. The presence of CSI and increasing number of PPN were associated with poor prognosis. On average, patients with CSI experienced improved PFS and OS when treated with AP relative to WAI. © 2011 Elsevier Inc. All rights reserved.
☆ This study was funded by the Gynecologic Oncology Group/Abraxis Oncology Young Investigator Award given to Dr. Tewari in 2007. ☆☆ This study was a plenary presentation at the 2010 Annual Meeting of the Society of Gynecologic Oncologists in San Francisco, CA and at the 2010 Biennial Meeting of the International Gynecologic Cancer Society in Prague. ★ This study was supported by National Cancer Institute grants to the Gynecologic Oncology Group (GOG) Administrative Office (CA 27469), the GOG Tissue Bank (CA 27469 and CA 11479) and the GOG Statistical and Data Center (CA 37517). ★★ The following GOG member institutions participated in this protocol: University of Alabama at Birmingham, Duke University Medical Center, Abington Memorial Hospital, Walter Reed Army Medical Center, Wayne State University, University of Minnesota Medical School, University of Mississippi, University of Colorado-Anschutz Cancer Pavilion, University of California at Los Angeles (UCLA), Fred Hutchinson Cancer Research Center, University of Pennsylvania Cancer Center, Penn State Milton S. Hershey Medical Center, University of Cincinnati Medical Center, University of North Carolina, University of Iowa Hospitals and Clinics, Southwestern Medical Center of Texas, Indiana University Cancer Center, Wake Forest University School of Medicine, Associates in Gynecologic Care, PC, University of California Medical Center at Irvine, Tufts-New England Medical Center, Rush University Medical Center, State University of New York at Brooklyn, Cleveland Clinic Foundation, State University of New York at Stony Brook, Washington University School of Medicine, Cooper Hospital University Medical Center, Columbus Cancer Council/Ohio State, University of Massachusetts Memorial Health Care, Fox Chase Cancer Center, Medical University of South Carolina, Women's Cancer Center of Nevada, University of Oklahoma, University of Virginia Health Sciences Center, University of Chicago, Tacoma General Hospital, Thomas Jefferson University Hospital, Mayo Clinic, University Hospitals Case Medical Center, Tampa Bay Cancer Consortium, Gynecologic Oncology Network (GON), and Ellis Fischel Cancer Center. ⁎ Corresponding author at: University of California Medical Center, Irvine, Chao Family Comprehensive Cancer Center, Department of Obstetrics & Gynecology, 101 The City Drive South, Bldg 56, Room 275, Orange, CA 92868, USA. Fax: + 1 714 456 7754. E-mail address: [email protected] (K.S. Tewari). 0090-8258/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2011.12.414
88
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
Table 1 A–C: Pathologic factors stratified by treatment. Frequency
WAI
ChemoRx
Total
A. Lymphovascular space invasion stratified by treatment LVSI not recorded 61 63 124 LVSI negative 36 43 79 LVSI positive 105 88 193 Total 202 194 396
PFS events
Deaths
78 52 132
72 48 125
B. Cervical stromal involvement PFS events and deaths stratified by treatment Cvx stroma not recorded 6 7 13 7 7 Cvx stroma negative 158 132 290 185 173 Cvx stroma positive 38 55 93 70 65 Total 202 194 396 C. Extra-uterine disease limited to positive pelvic lymph nodes PFS events and deaths stratified by treatment > 1 positive node 13 15 28 12 11 1 positive node 12 13 25 14 13 Total 25 28 53 LVSI: lymphovascular space involvement; WAI: whole abdominal irradiation; ChemoRx: chemotherapy; PFS: progression-free survival; Cvx: cervix.
Introduction In 2011, there will be an estimated 46,470 new cases of endometrial cancer in the United States and 8120 deaths [1]. In a study utilizing the Surveillance, Epidemiology, and End Results database from 1988 to 2001 containing 45,510 patients, Ueda et al. reported that during this 14-year period there was an increase in the proportion of patients dying from advanced cancers, attributing the trend to an increase in the incidence rate of advanced stage disease, high-risk histologic subtypes, and lack of adequate surgical staging [2,3]. For patients with recurrent, metastatic, or high-risk disease, the most recent National Comprehensive Cancer Network (NCCN) clinical practice guidelines list cisplatin plus doxorubicin (AP) with or without paclitaxel as category 1 regimens and also emphasize the importance of access to relevant clinical trials whenever possible [4]. In 2006, Randall et al. reported results from Gynecologic Oncology Group (GOG) Protocol 122 which was designed to compare AP to whole abdominal irradiation (WAI) for women with advanced stage endometrial carcinomas [5]. Over 400 patients with FIGO stage III or IV disease were enrolled on this phase III multi-center trial from 1992 to 2000 [5]. Importantly, the combination of AP was shown to significantly improve progression-free survival (PFS) (HR 0.67, 95% CI 0.52–0.87) and overall survival (OS) (HR 0.69, 95% CI 0.53–0.89) compared with whole abdominal irradiation [5]. Specifically, with a median follow-up of 74 months, at 5 years and adjusting for stage, 55% of patients treated with AP were predicted to be alive compared with 42% of those who received WAI [5]. Acute toxicity was observed
in the chemotherapy arm with greater frequency and severity than in the radiation treatment arm. The current study is a retrospective, exploratory analysis of data abstracted from reports prospectively collected for GOG 122. We hypothesized that certain uterine-specific surgicopathologic factors (i.e., lymphovascular space involvement (LVSI), cervical stromal involvement (CSI)) and the number of positive pelvic lymph nodes (PPN), would have a prognostic impact on PFS and OS among patients with advanced endometrial carcinoma. In addition, we also hypothesized that the superiority of AP reported in GOG 122 would be sustained among patients with LVSI, CSI, and/or a single PPN.
Patients and methods Patients providing written informed consent and enrolled on GOG 122 with International Federation of Gynecology and Obstetrics (FIGO) stage III or IV endometrial carcinoma [6] of any histology were eligible for this trial. Eligibility required total abdominal hysterectomy and bilateral salpingoophorectomy, surgical staging, tumor resection, and no single site of residual tumor more than 2 cm. Nodal sampling was optional. Negative scalene node biopsies and chest computed tomography scans were required for patients with positive para-aortic lymph nodes. Patients with recurrent disease, hematogenous metastasis, inguinal lymph node involvement, or a history of pelvic or abdominal radiation therapy or chemotherapy were ineligible. The WAI arm prescribed postoperative radiotherapy with an open field AP–PA technique at a dose of 30 Gy in 20 daily fractions. After WAI, patients were to receive a boost of 15 Gy in 8 fractions to the true pelvis or to an extended field encompassing pelvic nodes and para-aortic nodes. A boost to both areas was to be administered to patients with positive pelvic nodes and no para-aortic node sampling or patients with neither pelvic lymph nodes nor para-aortic lymph node sampling. Doxorubicin 60 mg/m 2 plus cisplatin 50 mg/m 2 was prescribed every 3 weeks for eight cycles on the AP arm. The maximum allowable cumulative dose of doxorubicin was 420 mg/m 2; therefore, only cisplatin was to be infused during cycle eight. Patient and tumor characteristics including age, race, performance status, cell type, histologic grade, FIGO surgical stage, and residual tumor size were reported in the original publication. The original study received local IRB approval. Diagnostic slides were reviewed by the GOG Pathology Committee for eligibility (primary site and histology). For the current study, retrospective review of hospital pathology reports was conducted to abstract the following data: lymphovascular space involvement (LVSI), cervical stromal involvement (endocervical canal involvement alone was not counted), and lymph node information (e.g., number of nodes obtained, number of
Table 2 Multiple regression analysis of surgicopathologic prognostic factors. Prognostic factor
No lymph nodes identified in pathology sample or none retrieved Lymph nodes retrieved but not counted (average affect) Number of pelvic lymph nodes identified Number of positive pelvic lymph nodes identified Number of para-aortic nodes identified Number of positive para-aortic lymph nodes LVSI negative or missing LVSI positive Cervical stromal involvement negative or missing Cervical stromal involvement positive
Progression-free survival
Overall survival
Hazard ratio
95% CL
Hazard ratio
95% CL
Referent group 0.60 0.99 1.07 1.00 0.94 Referent group 1.03 Referent group 1.44
– 0.33–1.11 0.98–1.01 1.00–1.15 0.98–1.02 0.85–1.04 – 0.79–1.35 – 1.07–1.93
Referent group 0.59 0.99 1.06 1.00 0.95 Referent group 1.26 Referent group 1.33
– 0.30–1.15 0.97–1.00 0.99–1.14 0.98–1.02 0.85–1.05 – 0.96–1.66 – 0.98–1.80
Models are adjusted for treatment, stage, age, race, cell type, and grade (PFS and OS), and residual disease (PFS) , all of which were statistically significant as reported in the original clinical manuscript; 134 patients are missing LVSI and 13 patients are missing cervical stromal involvement; nodal dissection was not required for entry onto GOG 0122. CI: confidence limit; LVSI: lymphovascular space involvement.
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
positive nodes, clinically positive vs. microscopically positive, and sites of positive nodes). A Cox proportional hazards model [7] was used to estimate the association of malignant lymph node counts with the log hazard of progression or death (PFS) and the log hazard of death (OS) while adjusting for treatment, stage, residual disease, age, race, cell type, grade and positive cytology as reported in the original clinical manuscript. The reference group for the analysis of lymph node counts is defined as patients with no lymph nodes retrieved or no lymph nodes identified in the surgical pathology specimen as reported by the local pathologist reviewing the case. An indicator for patients who
A
had nodes retrieved but for whom no counts were reported was used to estimate the average association of those patients who had nodes retrieved but not counted. When documented, total node counts, up to a maximum of 33 pelvic and 12 para-aortic lymph nodes (95th percentiles), and positive node counts were included in the model. The minimum value of the actual node count and the 95th percentile of counts were used in the model. A linear relationship between the numbers of nodes retrieved and the log hazard for both pelvic and para-aortic nodes was assumed; these were assessed using plots of the smoothed regression of lymph node counts on the martingale residuals from a model omitting the counts. Patients with
B
Progression-Free Survival
Survival by Lymph-Vascular Space Involvement 1.0
1.0 0.9
LVSI Negative Positive
0.8
0.9
Censor Event Total 27 52 79 61 132 193
0.8
Proportion Surviving
Proportion Surviving Progression-Free
by Lymph-Vascular Space Involvement
0.7 0.6 0.5 0.4 0.3
0.7 0.6 0.5 0.4 0.3
0.2
0.2
0.1
0.1
0.0
0.0
0
12
24
36
48
60
72
84
96
108
120
LVSI Negative Positive
0
12
24
Censor Event Total 31 48 79 68 125 193
36
Months on Study
72
84
96
108
120
96
108
120
by Cervical Stroma Involvement 1.0
1.0 0.9
Stroma Negative Positive
0.8
0.9
Censor Event Total 105 185 290 23 70 93
0.8
Proportion Surviving
Proportion Surviving Progression-Free
60
Survival
D
by Cervical Stroma Involvement
0.7 0.6 0.5 0.4 0.3
0.7 0.6 0.5 0.4 0.3
0.2
0.2
0.1
0.1
Stroma Negative Positive
0.0
0.0
0
12
24
36
48
60
72
84
96
108
120
0
12
24
Censor Event Total 117 173 290 28 65 93
36
Months on Study
48
60
72
84
Months on Study
Progression-Free Survival Among Patients with Metastasis limited to Pelvic Nodes
Survival
F
Among Patients with Metastasis limited to Pelvic Nodes 1.0
1.0 0.9
# Nodes + >1 Node + 1 Node +
0.8
0.9
Censor Event Total 16 12 28 11 14 25
0.8
Proportion Surviving
Proportion Surviving Progression-Free
48
Months on Study
Progression-Free Survival
C
E
89
0.7 0.6 0.5 0.4 0.3
0.7 0.6 0.5 0.4 0.3
0.2
0.2
0.1
0.1
# Nodes + >1 Node + 1 Node +
0.0
0.0
0
12
24
36
48
60
72
Months on Study
84
96
108
120
0
12
24
Censor Event Total 17 11 28 12 13 25
36
48
60
72
84
96
108
120
Months on Study
Fig. 1. A–F: Kaplan Meier survival curves for pathologic factors. A. Progression-free survival stratified by lymphovascular space involvement. B. Overall survival stratified by lymphovascular space involvement. C. Progression-free survival stratified by cervical stromal involvement. D. Overall survival stratified by cervical stromal involvement. E. Progressionfree survival stratified by metastases to pelvic lymph nodes. F. Overall survival stratified by metastases to pelvic lymph nodes.
90
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
Results
for PFS and OS treatment hazard ratios all included 1. The estimated treatment hazard ratios in patients with metastasis limited to 2 or more positive pelvic nodes favored WAI for PFS (HR = 1.25) and favored neither arm for OS (HR = 1.04). For patients with 1 single positive pelvic node, the estimate for OS (HR = .731) identified a trend which favored chemotherapy. The test for homogeneity of treatment effect across subgroups was not statistically significant for either endpoint (PFS: p = 0.72 OS: p = 0.66). The association of the number of PPN with PFS and OS in terms of treatment allocation appears in Table 3E–F. Figs. 3A and 3B depict forest plots of AP relative to WAI for PFS and OS, respectively for all three surgico-pathologic factors studied. The diamonds reflect the unadjusted treatment effect. The stage-adjusted 95% confidence intervals for both endpoints are: PFS HR 0.735, 95% CI 0.575, 0.939; OS HR 0.752, 95% CI 0.584, 0.968. The placement of the diamond or HR estimates to the left of the point of equal hazards would indicate a trend favoring superiority of AP over WAI in the study population.
Clinical material
Discussion
A total of 396 eligible patients were enrolled and treatment with either WAI (n = 202) or AP (n = 194) was randomly assigned. LVSI information was missing in 124 (31%) patients and cervical stroma data were missing in 13 (3%). For those in whom LVSI was identified, 105 were treated with WAI and 88 were treated with AP (Table 1A). For 202 women treated with WAI, 38 had CSI and 158 did not. For those who received AP, 55 had CSI and 132 did not (Table 1B). On the WAI arm, PPN were identified in 91 women and positive para-aortic nodes were identified in 40. On the AP arm, PPN were identified in 113 women and positive para-aortic nodes were identified in 49. Therefore there were at least 204 lymph node-positive subjects. In the absence of other metastatic disease, 25 patients were found to have only one PPN (n= 12 WAI, n = 13 AP) and 28 women found to have greater than one PPN (n= 13 WAI, n = 15 AP) (Table 1C).
In the past, most patients with endometrial cancer received radiotherapy either pre-operatively or in the adjuvant setting following surgery. This paradigm was challenged, first with the recognition that adjuvant radiotherapy only improved local control, and secondly with the identification of recurrence risk factors in early stage disease to identify who would benefit from localized adjuvant treatment with radiation. After many years, the pendulum has swung in the opposite direction. GOG 122, reported by Randall et al., represents one of those few clinical studies that results in a change in practice patterns for treatment of advanced disease [5,11–20]. In this study, AP significantly
Analysis of prognostic factors (entire study population)
Table 3 A–F. Unadjusted and stage-adjusted treatment hazard ratios and confidence limits for pathologic factor subgroups.
missing and negative LVSI were combined into one category. Therefore the hazard ratio reported for LVSI estimates the association relative to those with negative or missing values and not a clear association relative to a true negative reference group. The same was done for cervical stromal involvement but there were very few with missing data. The Kaplan–Meier method [8] was used to estimate the distribution of PFS and OS event times. Forest plots [9] were generated to display the consistency of treatment effect within subgroups and the original study's conclusions. Tests of treatment effect homogeneity [10] were also performed to identify potential signals of interest in the subgroups under investigation. All analyses were exploratory with no adjustment to account for multiple testing and control of type I error. The precision in the estimated treatment effects within small subgroups was limited and thus only confidence intervals are reported.
A multiple regression analysis of the surgicopathologic prognostic factors was performed (Table 2). CSI was associated with poorer PFS and OS, with a 44% increase in risk of progression or death and a 33% increase in risk of death. There was a trend for increasing number of PPN removed and identified being associated with a 7% increased risk of progression or death. Kaplan–Meier estimates are displayed in Figs. 1A–F. Analysis of surgicopathologic factors on endpoints of AP relative to WAI There was a trend for patients with LVSI and CSI to experience improved PFS and OS when treated with AP relative to WAI. Confidence intervals for PFS and OS treatment HRs in each subgroup defined by LVSI and CSI included 1. However the estimated unadjusted (and stage-adjusted) treatment hazard ratios favored AP for both endpoints. These estimates ranged between 0.75 and 0.95 for LVSI and 0.75 and 0.85 for CSI. Furthermore, the tests for homogeneity of treatment effect across subgroups were not statistically significant for either endpoint or factor (LVSI unadjusted PFS: p = 0.90; OS: p = 0.81 and CSI unadjusted PFS: p = 0.79; OS: p = 0.79) whether or not the analysis was adjusted for stage (Tables 3A–D). Patients who did not have CSI (n= 13) reported were excluded from each treatment effect analysis of CSI. The Kaplan–Meier survival curves for PFS and OS stratified by LVSI and treatment assignment appear in Figs. 2A and 2B, respectively, and those stratified by CSI and treatment assignment appear in Figs. 2C and 2D, respectively. There was a trend for patients with a single PPN to experience improved OS when treated with AP relative to WAI. Confidence intervals
Subgroup
Unadjusted Treatment hazard ratio
Stage adjusted 95% HR CL
Treatment hazard ratio
95% CL
A. Lymphovascular space invasion, PFS. LVSI not recorded 0.777 LVSI negative 0.855 LVSI positive 0.886
0.498–1.212 0.492–1.485 0.629–1.250
0.671 0.663 0.792
0.427–1.054 0.376–1.170 0.560–1.120
B. Lymphovascular space invasion, OS LVSI not recorded 0.752 LVSI negative 0.953 LVSI positive 0.791
0.473–1.197 0.536–1.696 0.555–1.128
0.712 0.946 0.730
0.446–1.135 0.531–1.686 0.511–1.043
C. Cervical stromal involvement, PFS Cvx stroma negative 0.790 Cvx stroma positive 0.852
0.590–1.058 0.528–1.376
0.672 0.835
0.500–0.905 0.517–1.348
D. Cervical stromal involvement, OS Cvx stroma negative 0.750 Cvx stroma positive 0.812
0.553–1.016 0.496–1.330
0.679 0.852
0.506–0.933 0.519–1.397
Subgroup
95% CL
Treatment HR
E. Metastases limited to pelvic lymph nodes, PFS > 1 positive node 1.273 0.403–4.019 1 positive node 0.959 0.336–2.741 F. Metastases limited to pelvic lymph nodes, OS > 1 positive node 1.049 0.319–3.444 1 positive node 0.731 0.245–2.183 LVSI: lymphovascular space involvement; Cx: cervix; HR: hazard ratio; CL: confidence limit; Rx: treatment.
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
Progression-Free Survival
A
Survival
B
by Lymph-Vascular Space Involvement 1.0
1.0 0.9
LVSI/Regimen Negative/WAI Negative/AP Positive/WAI Positive/AP Missing/WAI Missing/AP
0.8 0.7 0.6
Alive Event Total 13 23 36 14 29 43 32 73 105 29 59 88 20 41 61 26 37 63
0.5 0.4 0.3
0.9 0.8
Proportion Surviving
Proportion Surviving Progression-Free
by Lymph-Vascular Space Involvement
0.7 0.6 0.5 0.4
LVSI/Regimen Negative/WAI Negative/AP Positive/WAI Positive/AP Missing/WAI Missing/AP
0.3
0.2
0.2
0.1
0.1
Alive Event Total 15 21 36 16 27 43 33 72 105 35 53 88 22 39 61 30 33 63
0.0
0.0 0
12
24
36
48
60
72
84
96
108
120
0
12
24
36
Months on Study
C
48
60
72
84
96
108
120
108
120
Months on Study
D
Progression-Free Survival
Survival
by Cervical Stroma Involvement
by Cervical Stroma Involvement
1.0
1.0
0.9
Stroma/Regimen Censor Event Total Negative/WAI 52 106 158 Negative/AP 53 79 132 Positive/WAI 10 28 38 Positive/AP 13 42 55
0.8 0.7 0.6 0.5 0.4 0.3
0.9 0.8
Proportion Surviving
Proportion Surviving Progression-Free
91
0.7 0.6 0.5 0.4 0.3
0.2
0.2
0.1
0.1
Stroma/Regimen Censor Event Total Negative/WAI 56 102 158 Negative/AP 61 71 132 Positive/WAI 11 27 38 Positive/AP 17 38 55
0.0
0.0 0
12
24
36
48
60
72
84
96
108
120
Months on Study
0
12
24
36
48
60
72
84
96
Months on Study
Fig. 2. A–D. Kaplan Meier survival curves stratified by pathologic factors and treatment allocation. A: Progression-free survival stratified by lymphovascular space involvement and treatment allocation. B: Overall survival stratified by lymphovascular space involvement and treatment allocation. C: Progression-free survival stratified by cervical stromal involvement and treatment allocation. D: Overall survival stratified by cervical stromal involvement and treatment allocation.
improved both PFS (HR 0.67, 95% CI 0.52–0.87) and OS (HR 0.69, 95% CI 0.53–0.89) [5]. With a median follow-up of 74 months, at 5 years and adjusting for stage, 55% of patients treated with AP were predicted to be alive compared with 42% of those who received WAI [5]. The protocol effectively transformed the underpinnings of adjuvant therapy for metastatic disease. However, the concern over more frequent and severe acute toxicity [5] prompted the current subgroup analyses of the treatment effect in patients with a single PPN. Several retrospective studies have evaluated outcomes of advanced disease treated with radiation (regional or whole abdomen) with conflicting results. Klopp et al. noted that stage IIIC disease had a high rate of locoregional recurrence in patients not treated with regional radiotherapy, whereas those who underwent surgical staging followed by external beam irradiation had a high rate of cure [21]. Although their report is limited by small numbers, non-random treatment policy confounded with time and absence of longitudinal tracking, several hypotheses for differences between the conclusions reached in their study and in GOG 122. The latter trial included patients (14%) with up to 2 cm of residual tumor after debulking. They suggest that the dose delivered (30 Gy to the whole abdomen followed by a boost of 15 Gy to the pelvis) was inadequate to treat gross residual disease. Also, whole abdominal irradiation resulted in treatment interruption and early discontinuation in 16% of patients. Finally, GOG 122 allowed serous and clear cell tumors for which the risk of intraperitoneal dissemination is high and the need of systemic therapy is implicit. Conversely, stage migration due to improvements in radiologic imaging in the later years of Klopp's study was likely responsible for exclusion of patients found to have carcinomatosis.
Nelson et al. reported outcomes in 17 fully staged patients with extrauterine disease limited to pelvic nodes and pathologically negative para-aortic nodes treated post-operatively with pelvic (n = 13) or whole abdomen (n = 4) irradiation [22]. Their report suggested that adjuvant RT alone can be a viable treatment option in this welldefined patient population and that extended field RT might further improve results [22]. Gibbons et al. observed 7-year disease-free survivorship rates of 57.8% (stage III) and 25.0% (stage IV) following adjuvant WAI in 56 patients [23]. Acute toxicity was common but mild, and chronic toxicity remained almost entirely subclinical. Finally, Smith et al. also recognized a role for WAI in some patients with optimally debulked advanced-stage uterine cancer, citing 79% and 89% PFS and OS, respectively [24]. Important questions pertaining to lymphatic spread have not been addressed adequately [25–32]. The clinical significance of the extent of nodal dissection (commonly measured by the total number of lymph nodes identified by the pathologist) and the role of adjuvant therapy (WAI or AP) in the setting of a single positive node are debated. There are reports of patients with one or two positive nodes in a background of multiple negative nodes (10–12 or greater) being treated successfully with postoperative pelvic RT alone, however the percentage of those with minimal, residual advanced stage disease is not consistently detailed. Onda et al. reported on 30 node-positive patients including 10 with pelvic node metastases alone and 20 with para-aortic metastases; all of which underwent complete lymphadenectomy followed by adjuvant radiation for those with pelvic metastases alone or by pelvic plus extended field radiation plus chemotherapy for those with aortic nodal disease [25]. Five year survival
92
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
A
Hazard Ratios for Progression - Free Survival Rel Haz Var(In(HR)) Chemotherapy is better
LVSI
Missing Negative Positive
0.777 0.855 0.886
0.052 0.079 0.031
Pelvic Nodes
>1+ 1+
1.273 0.959
0.344 0.287
Negative Positive
0.790 0.852
0.022 0.060
Cervical stroma
0.33
0.5
0.67
WAI is better
1.0
1.5
2.0
3.
Relative Hazard Hazard Ratio Progression-Free Survival 0.735
B
95% Hazard Ratio Confidence Limits 0.575, 0.939
Hazard Ratios for Overall Survival Rel Haz Var(In(HR))
LVSI
Missing 0.752 Negative 0.953 Positive 0.791
0.056 0.036 0.033
Pelvic Nodes
>1+ 1+
1.049 0.731
0.368 0.312
Cervical Stroma
Negative 0.750 Positive 0.812
0.024 0.063
WAI is better
Chemotherapy is better
0.5
0.33
0.67
1.0
1.5
2.0
3.
Relative Hazard Hazard Ratio 95% Hazard Ratio Confidence Limits Overall Survival
0.752
0.584, 0.968
Fig. 3. A–B. Forest plots of treatment effect for progression-free survival and overall survival based on surgicopathologic factors. A: Progression-free survival. B: Overall survival. The red diamond reflects the overall unadjusted treatment effect with 95% confidence interval. LVSI: lymphovascular space involvement.
was 100% for those with positive pelvic nodes only and 75% for those with para-aortic disease [25]. Finally, in GOG 184 sequential adjuvant therapy (chemotherapy following volume-directed radiotherapy) was associated with increased toxicity without significant improvement in recurrence-free survival in women with advanced disease, including those with positive para-aortic nodes [20]. Takeshima et al. reported on 26 women with positive para-aortic nodal involvement who received postoperative ifosfamide, epiadriamycin, and cisplatin [27]. Their results are consistent with the findings of GOG 122, although the natural history of a single positive pelvic node following AP vs WAI remains obscure [27]. The results of GOG 122 suggest that systemic therapy with AP is superior in this setting, although acute toxicity, on average is increased but relatively manageable. The primary analysis of protocol 122 did not include separate analysis of the treatment effect on PFS and OS by either the number of lymph nodes removed or the setting of a single PPN. The sample size for such subgroup analyses in the current study is admittedly too small for definitive conclusions; with low precision, the confidence intervals are wide and do not rule out important differences. Additionally, the extent of lymph node dissection and node counting was not standardized for the original study.
The presence of LVSI in the uterus may also have prognostic implications and therefore could be used to guide clinical management in the adjuvant setting. As all of the subjects in GOG 122 had advanced disease, the presence or absence of LVSI in the uterus was not initially viewed to be an important entity in the original design of that protocol. Unfortunately, LVSI data for over 30% of the study subjects were not available. Interpretation of results in the presence of such a high proportion of missing data is problematic. Orezzoli et al. recently reported on 55 patients with CSI and 26 with endocervical gland involvement treated from 1993 to 2003 [33]. The five-year survival rate was 83% for patients with endocervical gland involvement, and 71% for those with CSI. Substaging into IIA vs IIB or depth of stromal invasion had no significant association with survival. Age (p= 0.001), LVSI (p= 0.017), and type of treatment (p= 0.022) were jointly associated with survival among these patients [33]. In a recent Surveillance Epidemiology and End Results database study by Schmid et al., the association of adjuvant radiotherapy and outcome for 943 women with stage IIIC endometrial cancer treated from 1988 to 2001 was evaluated [34]. The median number of lymph nodes removed was 11, and 54.9% had a single positive lymph node and 45.1% had 2–5 positive nodes. For those that received adjuvant
K.S. Tewari et al. / Gynecologic Oncology 125 (2012) 87–93
radiotherapy (67.3%), the 5-year disease-specific survival was 67.9% compared to 53.4% who did not receive radiotherapy (pb 0.001) [34]. For those with a single positive node, adjuvant radiotherapy was associated with improved survival; from 54.4% to 74.3% (pb 0.001) and for those with 2–5 positive nodes adjuvant radiation was also associated with improved survival from 52.4% to 59.7% (p= 0.089) [34]. There are a number of scientific issues concerning “node counting” studies. The actual number of nodes removed from an individual patient may be a function of the patient's body habitus, age, extent of disease based on gross surgical inspection, the surgeon's assessment of the value of thorough lymphadenectomy, the surgeon's skill level, and possibly the technique through which lymph node dissection is performed (i.e. laparotomy, laparoscopy, robotics). In addition, the total number of lymph nodes reported is also dependent on how the laboratory technician and/or pathologist dissect, submit and process the tissues retrieved from the operating room [35,36]. Often when evaluating the same patient the surgeon and the pathologist often do not agree about how many nodes were actually removed. While each of these features may determine the number of nodes removed, any or all of them may be prognostic by themselves. The current study cannot unravel causes from effects. For example, if a surgeon's skill level is related to the number of nodes that can be removed, his/her skill level may hypothetically have a direct effect of the patient's probability of recurrence, which biologically does not involve number of nodes removed, but is only correlated with it. Perhaps it is the unknown status of the nodes left behind, not those which are removed, that is prognostic. In this ancillary data analysis, CSI was determined to be a prognostic factor for both PFS and OS for the entire study population. Similarly, there was a trend for increasing number of PPN removed to be associated with a 7% increase in progression and/or death for all patients. In evaluating our hypothesis that the superiority of AP over WAI as reported in the original publication of GOG 122 [5] would be sustained in the presence of these surgicopathologic factors, tests of homogeneity were performed. Although the confidence intervals for the estimated unadjusted and stage-adjusted treatment HRs for subgroups included 1, because the tests of homogeneity of treatment effect were not statistically significant, the HRs are interpreted to favor AP for both endpoints among women with LVSI, CSI, and/or a single PPN. Importantly, in this study population, the superiority of AP over WAI for all three surgicopathologic factors is reflected in the forest plot diamonds being situated to the left of the point of equal hazards for PFS and for OS (Figs. 3A–B). Due to the limited number of patients in each subgroup there is limited precision in the estimated treatment effects. Nevertheless, there is no significant evidence that the improved PFS (HR 0.67, 95% CI 0.52–0.87) and OS (HR 0.69, 95% CI 0.53–0.89) conferred by AP as reported in GOG 122 [5] depends on LVSI, CSI, or PPN status/count. Conflict of interest statement The authors wish to report that they have no conflicts of interest with the exception of Dr. Nick Spirtos who wishes to disclose relevant financial relationships with Genzyme — Speaker's Bureau, Ethicon — Speaker's Bureau and Olympus — Speaker's Bureau.
References [1] Siegel R, Ward E, Brawley O, Jemal Ahmedin J. The impact of eliminating socioeconomic and racial disparities on premature cancer deaths. CA Cancer J Clin 2011;61:212–36. [2] Ueda SM, Kapp DS, Cheung MK, et al. Trends in demographic and clinical characteristics in women diagnosed with corpus cancer and their potential impact on the increasing number of deaths. Am J Obstet Gynecol 2008;198(218):e1–6. [3] Creasman WT. Revised FIGO staging for carcinoma of the endometrium. Int J Gynaecol Obstet 2009:105–9. [4] NCCN Practice Guidelines in Oncology — v.1.2010. www.nccn.org. [5] Randall ME, Filiaci VI, Muss H, et al. Randomized phase III trial of whole-abdominal irradiation versus doxorubicin and cisplatin chemotherapy in advanced endometrial carcinoma: a Gynecologic Oncology Group study. J Clin Oncol 2006;24:36–44. [6] Mikuta JJ. International Federation of Gynecology and Obstetrics staging of endometrial cancer 1988. Cancer 1993;71(4 Suppl):1460–3.
93
[7] Cox DR. Regression model and life tables (with discussion). J R Stat Soc B 1972;34: 187–219. [8] Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457–81. [9] Duchateau L, Sylvester R. Meta-analysis. In: Crowley John, editor. Handbook of statistics in clinical oncology. New York: Marcel Dekker, Inc.; 2001. p. 525–43. [10] Normand ST. Tutorial in biostatistics meta-analysis: formulating, evaluating, combining, and reporting. Stat Med 1999;18:321–59. [11] Randall ME, Reisinger S. Radiation therapy and combined chemo-irradiation in advanced and recurrent endometrial carcinoma. Semin Oncol 1994;1004(21):91–9. [12] Thigpen JT, Buchsbaum HJ, Mangan C, et al. Phase II trial of Adriamycin in the treatment of advanced or recurrent endometrial carcinoma: a Gynecologic Oncology Group study. Cancer Treat Rep 1979;63:21–7. [13] Thigpen JT, Blessing JA, Homesley H, et al. Phase II trial of cisplatin as first-line chemotherapy in patients with advanced or recurrent endometrial carcinoma. A Gynecologic Oncology Group study. Gynecol Oncol 1989;33:68–70. [14] Thigpen JT, Blessing JA, DiSaia PJ, et al. A randomized comparison of doxorubicin alone versus doxorubicin plus cyclophosphamide in the management of advanced or recurrent endometrial carcinoma: a Gynecologic Oncology Group study. J Clin Oncol 1994;12:1408–14. [15] Thigpen JT, Brady MF, Homesley HD, et al. Phase III trial of doxorubicin with or without cisplatin in advanced endometrial carcinoma: a Gynecologic Oncology Group study. J Clin Oncol 2004;22:3902–8. [16] Aapro MS, van Wijk FH, Bolis G, et al. Doxorubicin versus doxorubicin and cisplatin in endometrial carcinoma: definitive results of a randomized study (55872) by the EORTC Gynaecological Cancer Group. Ann Oncol 2003;14:441–8. [17] Homesley HD, Filiaci V, Gibbons SK, et al. A randomized phase III trial in advanced endometrial carcinoma of surgery and volume directed radiation followed by cisplatin and doxorubicin with or without paclitaxel: a Gynecologic Oncology Group study. Gynecol Oncol 2009;112:543–52. [18] Miller DS (study chair). GOG protocol 209: a randomized phase III trial of doxorubicin/ cisplatin/paclitaxel and G-CSF versus carboplatin/paclitaxel in patients with stage III & IV or recurrent endometrial cancer. http://www.cancer.gov/search/ViewClinicalTrials. aspx?cdrid=305940&version=HealthProfessional&protocolsearchid=8603435 [19] Matei D (study chair). GOG protocol 258: a randomized phase III trial of cisplatin and tumor volume directed irradiation followed by carboplatin and paclitaxel vs carboplatin and paclitaxel for optimally debulked, advanced endometrial carcinoma. http://www.cancer.gov/search/ViewClinicalTrials.aspx?cdrid=649079&version= HealthProfessional&protocolsearchid=7990461 [20] Homesley HD, Filiaci V, Gibbons SK, et al. A randomized phase III trial in advanced endometrial carcinoma of surgery and volume directed radiation followed by cisplatin and doxorubicin with or without paclitaxel: a Gynecologic Oncology Group study. Gynecol Oncol 2009;112:543–52. [21] Klopp AH, Jhingran A, Ramondetta L, et al. Node-positive adenocarcinoma of the endometrium: outcome and patterns of recurrence with and without external beam irradiation. Gynecol Oncol 2009;115:6–11. [22] Nelson G, Randall M, Sutton G, et al. FIGO stage III endometrial carcinoma with metastases confined to pelvic lymph nodes: analysis of treatment outcomes, prognostic variables, and failure patterns following adjuvant RT. Gynecol Oncol 1999;75:211–4. [23] Gibbons S, Martinez A, Schray M, et al. Adjuvant whole abdominopelvic irradiation for high risk endometrial carcinoma. Int J Radiat Oncol Biol Phys 1991;21:1019–25. [24] Smith RS, Kap DS, Chen Q, et al. Treatment of high-risk uterine cancer with whole abdominopelvic RT. Int J Radiat Oncol Biol Phys 2000;48:767–78. [25] Onda T, Yoshikawa H, Mizutani K, et al. Treatment of node-positive endometrial cancer with complete node dissection, chemotherapy and radiation therapy. Br J Cancer 1997;5:1836–41. [26] Lutman CV, Havrilesky LJ, Cragun JM, et al. Pelvic lymph node count is an important prognostic variable for FIGO stage I and II endometrial carcinoma with highrisk histology. Gynecol Oncol 2006;102:92–7. [27] Takeshima N, Umayahara K, Fujiwara K, et al. Effectiveness of postoperative chemotherapy for para-aortic lymph node metastasis of endometrial cancer. Gynecol Oncol 2006;102:214–7. [28] Bristow RE, Zahurak ML, Alexander CJ, et al. FIGO stage IIIC endometrial carcinoma: resection of macroscopic nodal disease and other determinants of survival. Int J Gynecol Cancer 2003;13:664–72. [29] Havrilesky LJ, Cragun JM, Calingaert B, et al. Resection of lymph node metastases influences survival in stage IIIC endometrial cancer. Gynecol Oncol 2005;99:689–95. [30] McMeekin DS, Tillmanns T. Endometrial cancer: treatment of nodal metastases. Curr Treat Options Oncol 2003;4:121–30. [31] Katz LA, Andrews SJ, Fanning J. Survival after multimodality treatment for stage IIIC endometrial cancer. Am J Obstet Gynecol 2001;184:1071–3. [32] Umesaki N, Tanaka T, Miyama M, et al. Postoperative adjuvant chemotherapy with cisplatin, etoposide, and pirarubicin for endometrial carcinoma patients with lymph node metastasis: a pilot study. Oncol Rep 2000;7:1083–6. [33] Orezzoli JP, Sioletic S, Olawaiye A, et al. Stage II endometrioid adenocarcinoma of the endometrium. Clinical implications of cervical stromal invasion. Gynecol Oncol 2009;113:316–23. [34] Schmid S, Hsu IC, Hu JM, et al. Adjuvant radiation therapy in stage III nodepositive uterine cancer. Gynecol Oncol 2009;115:239–43. [35] Ricciardi R, Baxter N. Association versus causation versus quality improvement: setting benchmarks for lymph node evaluation in colon cancer. J Natl Cancer Inst 2007;99:414–5. [36] Rieger NA, Barnett FS, Moore JW, et al. Quality of pathology reporting impacts on lymph node yield in colon cancer. J Clin Oncol 2007;25:463.