Surgical versus radiographic determination of para-aortic lymph node metastases before chemoradiation for locally advanced cervical carcinoma†‡
A Gynecologic Oncology Group study†
Ms. Denise Mackey, Gynecologic Oncology Group Administrative Office, Four Penn Center, 1600 JFK Boulevard, Suite 1020, Philadelphia, PA 19103; Fax: (215); E-mail: [email protected]
See editorial on pages 1874–6, this issue
The following Gynecologic Oncology Group member institutions participated in the primary treatment studies: University of Alabama at Birmingham; Oregon Health Sciences University; Duke University Medical Center; Abington Memorial Hospital; University of Rochester Medical Center; Walter Reed Medical Center; University of Southern California at Los Angeles; University of Mississippi Medical Center; Colorado Gynecologic Oncology Group, P.C.; University of California at Los Angeles; University of Miami School of Medicine; Milton S. Hershey Medical Center; Georgetown University Hospital; University of Cincinnati; University of North Carolina School of Medicine; University of Iowa Hospitals and Clinics; University of Texas Southwestern Medical Center atDallas; Indiana University School of Medicine;Wake Forest University School ofMedicine; Albany Medical Center; University of California Medical Center at Irvine; Tufts-New England Medical Center; Rush-Presbyterian-St. Luke's Medical Center; State University of New York Downstate Medical Center; Eastern Virginia Medical School; Johns Hopkins Cancer Center; State University of New York at Stony Brook; Eastern Pennsylvania Gynecology/Oncology Center, P.C.; Southwest Oncology Group; Cooper Hospital/University Medical Center; Columbus Cancer Council; University of Mississippi Medical Center; University of Oklahoma; Indiana University Cancer Center; Duke University Medical Center; Wayne State University; SUNY-Downstate Medical Center; University of California at Irvine; Ellis Fischel Cancer Center; Walter Reed Army Medical Center; University of Southern California at Los Angeles; Tampa Bay/H. Lee Moffitt Cancer Center; University of Iowa Hospitals and Clinics; University of Kentucky; Medical University of South Carolina; Colorado Gynecologic Oncology Group, PC; M.S. Hershey Medical Center; University of Texas Southwestern Medical Center at Dallas; Columbus Cancer Council; Case Western Reserve University; University of Alabama at Birmingham; University of North Carolina School of Medicine; Wake Forest University School of Medicine; Stanford University Medical Center; Tacoma General Hospital; University of California at Los Angeles; University of Washington/Puget Sound Oncology Consortium; Tufts-New England Medical Center; Abington Memorial Hospital; Albany Medical Center; University of Miami School of Medicine; University of Cincinnati; Cleveland Clinic Foundation; SUNY-Stony Brook; Cooper Hospital/University Medical Center; Fox Chase Cancer Center; Women's Cancer Center; Oregon Health Sciences University; Rush-Presbyterian-St. Luke's Medical Center; Eastern Pennsylvania Gynecology/Oncology Center; University of Massachusetts Medical Center; Johns Hopkins Oncology Center, University of Chicago, University of Rochester Medical Center; University of Minnesota Medical School; Eastern Virginia Medical Center; Emory University Clinic; Georgetown University Hospital; The New York Hospital; Cornell Medical Center; Community Cancer Oncology Program; University of Arizona; University of Virginia; and Mayo Clinic.
Abstract
BACKGROUND
Patients with cervical cancer who had negative para-aortic lymph nodes (PALNs) identified by pretreatment surgical staging were compared with patients who had only radiographic exclusion of PALN metastases before they received treatment with pelvic radiation and brachytherapy (RT) plus cisplatin (C)-based chemotherapy.
METHODS
Patients who participated in 1 of 3 Phase III Gynecologic Oncology Group (GOG) trials (GOG 85, GOG 120, and GOG 165) and who were assigned randomly to receive either RT plus C or RT plus C combined with 5-fluorouracil with or without hydroxyurea comprised this retrospective analysis. Patients who had negative PALN status determined by surgical sampling (mandatory in GOG 85 and GOG 120 and optional in GOG 165) were compared with patients who had negative PALN status determined radiographically (GOG 165).
RESULTS
Five hundred fifty-five patients underwent surgical PALN sampling (the S group), and 130 patients underwent radiographic evaluation only (the R group). Age, race, histology, and tumor grade were similar. Patients in the R group had better performance status (P < .01), less advanced stage (P = .023), and smaller tumor size (P = .004) compared with patients in the S group, although patients with stage III and IV disease in the S group had better 4-year progression-free survival (48.9% vs 36.3%) and overall survival (54.3% vs 40%) compared with patients in the R group. In multivariate analysis, the R group was associated independently with a poorer prognosis compared with the S group (for disease progression: hazard ratio [HR], 1.35, 95% confidence interval [95% CI], 1.01–1.81; for death: HR, 1.46, 95% CI, 1.08–1.99).
CONCLUSIONS
Surgical exclusion (compared with radiographic exclusion) of positive PALNs in patients with cervical cancer who received chemoradiation (RT plus C-based chemotherapy) had a significant prognostic impact. Cancer 2008. © 2008 American Cancer Society.
Primary radiotherapy (RT) fails to control locoregional involvement in 20% to 85% of women with advanced cervical carcinoma, depending on disease stage, tumor bulk, and lymph node status. One reason for the high failure rate may be that cervical cancer represents the only remaining gynecologic malignancy that is staged clinically. Compared with surgical staging, clinical staging is only 60% accurate. In many instances, errors in clinical staging are related to undiagnosed lymph node metastases. It is accepted commonly that patients with para-aortic lymph node (PALN) metastases have lower progression-free survival (PFS), survival after recurrence, and overall survival (OS).1
Although they were not intended to replace International Federation of Gynecology and Obstetrics (FIGO) staging, several imaging modalities have been used to determine the extent of disease before initiating therapy. In a clinical-pathologic study that was conducted by the Gynecologic Oncology Group (GOG), the sensitivity and specificity of computerized tomography (CT) were only 34% and 96%, respectively.2 A more recent meta-analysis reported that the positive predictive value of CT was 60% and that the negative predictive value was 91%.3 Contemporary literature demonstrates that CT or magnetic resonance imaging (MRI) fails to detect macroscopic lymph node metastases in 20% to 50% of patients and fails to detect most, if not all, patients with microscopic disease.4-8 Detection of para-aortic metastases appears to be more successful with functional tumor-imaging studies, such as positron emission tomography (PET) and PET/CT, although reported sensitivities range widely between 38% and 86%.7, 9-13
Currently, extraperitoneal surgical staging before RT for patients with bulky or locally advanced disease remains controversial. Proponents of surgical staging cite the ability to detect and then treat metastatic disease beyond the standard radiation fields.14, 15 Recent evidence suggests a survival advantage both in patients with cervical cancer and PALN metastases who receive irradiation1, 16 and in patients who receive prophylactic para-aortic irradiation despite having negative or unevaluated PALNs.17 In addition to this is the potential for therapeutic benefit from debulking large lymph node metastases that are beyond the scope of standard RT doses.18-21 Those opposed to the routine use of surgical staging argue that only a small number of patients will benefit from extended-field RT based on their belief that patients with advanced cervical carcinoma fail both in the pelvis and at distant sites, making para-aortic failure irrelevant.22, 23 With multimodal treatments and improved brachytherapy, however, there is a trend toward improved local control, increasing the importance of incorporating measures that preferentially may improve control in the PALNs.24, 25
The GOG has conducted several prospective, randomized Phase III trials of cisplatin-based chemoradiation in patients with locoregionally advanced cervical cancer. Before 1997, protocols targeting this disease mandated surgical sampling of PALNs. In 1997, while comparing weekly cisplatin to protracted venous infusion 5-fluorouracil (5-FU) as radiation sensitizers (GOG 165), there was a paradigm shift within the GOG, and it was decided that para-aortic sampling should be left to the discretion of the primary surgeon. This approach not only allowed for quicker accrual and study completion, but it was more likely to represent the current standard-of-care approach to advanced cervical cancer, as indicated by the reluctance of investigators to perform surgical PALN sampling. Only 18% of patients on GOG 165 underwent pretreatment lymph node sampling.
The objective of the current study was to compare PFS and OS in patients with locally advanced cervical cancer who underwent pre-RT PALN surgical sampling versus patients whose negative lymph node status was determined radiographically.
MATERIALS AND METHODS
Data from patients who participated in 1 of 3 GOG randomized Phase III trials (GOG 85, GOG 120, and GOG 165) were reviewed retrospectively. Random treatment allocations were as follows: GOG 85,26 RT and either hydroxyurea or cisplatin plus 5-FU; GOG 120,27 RT and either cisplatin alone, hydroxyurea alone, or cisplatin plus 5-FU and hydroxyurea; and GOG 165,28 RT and either cisplatin or protracted venous infusion 5-FU. The patients who were included in this analysis had received cisplatin-based chemoradiation. All patients had FIGO stage IIB through IVA disease (95% had stage IIB and IIIB disease) without known PALN metastasis. In GOG 85 and GOG 120, pretreatment extraperitoneal surgical sampling of the PALNs was mandatory, whereas GOG 165 required radiographically negative PALN involvement with optional surgical sampling. For purposes of the extraperitoneal surgical procedure, the boundaries of the PALN dissection were defined as the middle of the common iliac artery caudally and the inferior mesenteric artery cranially. Dissection cephalad to the inferior mesenteric artery was restricted to those patients who had palpably suspicious lymph nodes above this level. Boundaries of the PALNs were not defined for patients who were staged radiographically but were assumed to be cephalad to the bifurcation of the aorta as observed on lymphangiogram, CT, or MRI studies. Detailed information regarding patient eligibility, treatment administration, and outcome was published previously.26-28 For all 3 trials, the primary endpoints were PFS and OS. PFS was calculated from the date of study enrollment to the date of reappearance or progression of disease, death, or most recent follow-up visit. OS was calculated from the date of study enrollment to the date of death or last contact.
Patients who underwent pretreatment PALN surgical sampling (the S group) were compared with patients who underwent only radiographic determination (the R group) of PALN metastases. Patient demographic and clinical characteristics were compared by using the Student t test (for continuous variables) or the Pearson chi-square method (for categorical variables). Survival curves were estimated according to Kaplan-Meier procedure, and the relative risks with 95% confidence intervals (95% CIs) were estimated by using a Cox proportional-hazards model adjusted for covariates (patient age, race, GOG performance status, histology, disease stage, tumor grade, tumor size, and extent of parametrial involvement). Sites of first recurrence also were compared between groups by using the Pearson chi-square test according to 2 patterns: 1) pelvic only versus extrapelvic (defined as any site outside the pelvis including lung, abdomen, vagina, PALNs and others) and 2) any PALN involvement versus any other site of recurrence without PALNs. Multiple sites (pelvis plus any site outside the pelvis) were considered distant recurrence. All statistical analyses were 2-tailed and were performed on Statistical Analysis Software (SAS) version 9.1 (SAS Institute, Cary, NC). The original 3 GOG studies were presented to and approved by local institutional review boards before activation, and all patients provided written consent before enrollment on them.
RESULTS
Data from total of 685 patients were included in this analysis as follows: GOG 85, RT and cisplatin plus 5-FU (n = 177); GOG 120, RT and cisplatin (n = 176) or RT and cisplatin with 5-FU plus hydroxyurea (n = 173); and GOG 165, RT plus cisplatin (n = 159). Staging was completed surgically in 555 patients (including 29 patients on GOG 165, in which the procedure was optional) and radiographically in 130 patients. Patient characteristics, including age, histology, and tumor grade, were similar between the 2 groups. Patients in the R group had better baseline performance status (P < .001), less advanced stage (P = .023), and smaller tumor size (P = .004) compared with patients in the S group. There also was a suggestion that bilateral parametrial involvement was observed less commonly in the R patients (33.1% vs 42.5%; P = .059) (Table 1).
| Characteristic | No. of patients (%) | Radiographic group | P |
|---|---|---|---|
| Surgical group | |||
| Age, y | |||
| Mean ± SD, y | 49.1 ± 12.0 | 49.6 ± 11.4 | .670 |
| <40 | 135 (24.3) | 31 (23.9) | |
| 40–49 | 169 (30.5) | 42 (32.3) | |
| 50–59 | 141 (25.4) | 26 (20) | |
| ≥60 | 110 (19.8) | 31 (23.9) | |
| Race | .596 | ||
| White | 338 (60.9) | 80 (61.5) | |
| Black | 126 (22.7) | 25 (19.2) | |
| Other | 91 (16.4) | 25 (19.2) | |
| Performance status | <.001 | ||
| 0 | 333 (60) | 108 (83.1) | |
| 1 | 197 (35.5) | 19 (14.6) | |
| 2 | 24 (4.3) | 2 (1.5) | |
| 3 | 1 (0.2) | 1 (0.8) | |
| Histology | .220 | ||
| Squamous | 488 (87.9) | 112 (86.2) | |
| Adenocarcinoma | 40 (7.2) | 7 (5.4) | |
| Adenosquamous | 23 (4.1) | 8 (6.2) | |
| Others | 4 (0.7) | 3 (2.3) | |
| Tumor grade | .524 | ||
| 1 | 40 (7.2) | 10 (7.7) | |
| 2 | 328 (59.1) | 70 (53.9) | |
| 3/Unspecified | 187 (33.7) | 50 (38.5) | |
| FIGO stage | .023 | ||
| IIB | 310 (55.9) | 86 (66.2) | |
| IIIA | 13 (2.3) | 0 (0) | |
| IIIB | 217 (39.1) | 38 (29.2) | |
| IVA | 15 (2.7) | 6 (4.6) | |
| Tumor size, cm | .004 | ||
| Mean ± SD, cm | 6.5 ± 2.2 | 5.9 ± 1.8 | |
| ≤4 | 101 (18.2) | 27 (20.8) | |
| 4.1–5 | 71 (12.8) | 27 (20.8) | |
| 5.1–6 | 135 (24.3) | 28 (21.5) | |
| 6.1–7 | 82 (14.8) | 22 (16.9) | |
| 7.1–8 | 100 (18) | 16 (12.3) | |
| >8 | 66 (11.9) | 10 (7.7) | |
| Parametrial involvement | .059 | ||
| Unilateral | 319 (57.5) | 87 (66.9) | |
| Bilateral | 236 (42.5) | 43 (33.1) |
- SD indicates standard deviation; FIGO, International Federation of Gynecology and Obstetrics.
The results from multivariate analysis suggested that patients who were staged without surgery had a poorer prognosis than patients who were staged surgically. Adjusted for age, race, performance status, histology, tumor size, and extent of parametrial involvement, the relative risk of disease progression in the R group compared with the S group was 1.35 (95% CI, 1.01–1.81; P = .043) (Table 2), and the relative risk for death was 1.46 (95% CI, 1.08–1.99; P = .014) (Table 3). The difference in prognosis between the 2 groups was most evident in patients with advanced-stage disease. Patients with stage III/IV disease had an estimated 4-year PFS of 36.2% for the R group and 48.9% for the S group, and the adjusted relative risk was 1.51 (95% CI, 0.99–2.31; P = .055) (Fig. 1 [top]); the estimated 4-year OS was 40% and 54.3%, respectively (adjusted relative risk, 1.60; 95% CI, 1.03–2.48; P = .038) (Fig. 1 [bottom]). The differences in PFS and OS between the 2 groups were not significant in patients with stage II disease (relative risk, 1.22; 95% CI, 0.81–1.84; P = .346) for disease progression (Fig. 2 [top]), and the relative risk for death was 1.38 (95% CI, 0.90–2.12; P = .143) (Fig. 2 [bottom]).
(Top) Kaplan-Meier estimate of disease progression-free survival (PFS) according to the method of staging for patients with International Federation of Gynecology and Obstetrics (FIGO) stage III/IV cervical cancer. The adjusted relative risk of disease progression in patients who underwent radiographic staging, compared with patients who underwent surgical staging, was 1.51 (95% confidence interval [95% CI], 0.99–2.31; P = .055). PG indicates disease progression; NPG, no disease progression. (Bottom) Kaplan-Meier estimate of overall survival (OS) according to the method of staging for patients with FIGO stage III/IV cervical cancer. The adjusted relative risk of death in patients who underwent radiographic staging, compared with patients who underwent surgical staging, was 1.60 (95% CI, 1.03–2.48; P = .038).
(Top) Kaplan-Meier estimate of disease progression-free survival (PFS) according to the method of staging for patients with International Federation of Gynecology and Obstetrics (FIGO) stage II cervical cancer. The adjusted relative risk of disease progression in patients who underwent radiographic staging, compared with patients who underwent surgical staging, was 1.22 (95% confidence interval [95% CI, 0.81–1.84; P = .346). PG indicates disease progression; NPG, no disease progression. (Bottom) Kaplan-Meier estimate of overall survival (OS) according to the method of staging for patients with FIGO stage II cervical cancer. The adjusted relative risk of death in patients who underwent radiographic staging, compared with patients who underwent surgical staging, was 1.38 (95% CI, 0.90–2.12; P = .143).
| Factor | Adjusted HR (95% CI) | P |
|---|---|---|
| Age, y | ||
| <40 | Referent | |
| 40–49 | 0.89 (0.66–1.19) | .432 |
| 50–59 | 0.66 (0.47–0.91) | .011 |
| ≥60 | 0.942 (0.66–1.27) | .597 |
| Race | ||
| White | Referent | |
| Black | 1.17 (0.90–1.52) | .244 |
| Other | 0.78 (0.56–1.09) | .147 |
| Performance status | ||
| 0 | Referent | |
| ≥1 | 1.33 (1.06–1.67) | .014 |
| Histology | ||
| Squamous | Referent | |
| Others | 1.48 (1.09–2.02) | .013 |
| Stage | ||
| II | Referent | |
| III/IV | 1.81 (1.44–2.27) | <.001 |
| Tumor grade | ||
| 1 or 2 | Referent | |
| 3 | 1.33 (1.06–1.68) | .014 |
| Tumor size, cm | ||
| <6 | Referent | |
| ≥6 | 1.18 (0.92–1.52) | .196 |
| Parametrial involvement | ||
| Unilateral | Referent | |
| Bilateral | 1.18 (0.94–1.49) | .152 |
| Para-aortic lymph node staging | ||
| Surgical | Referent | |
| Radiographic | 1.35 (1.01–1.81) | .043 |
- HR indicates hazard ratio; 95% CI, 95% confidence interval.
| Factor | Adjusted HR (95% CI) | P |
|---|---|---|
| Age, y | ||
| <40 | Referent | |
| 40–49 | 0.79 (0.58–1.07) | .129 |
| 50–59 | 0.66 (0.47–0.93) | .017 |
| ≥60 | 0.96 (0.69–1.34) | .813 |
| Race | ||
| White | Referent | |
| Black | 1.16 (0.89–1.53) | .280 |
| Other | 0.82 (0.58–1.16) | .258 |
| Performance status | ||
| 0 | Referent | |
| ≥1 | 1.43 (1.13–1.81) | .003 |
| Histology | ||
| Squamous | Referent | |
| Others | 1.39 (1.00–1.93) | .052 |
| Stage | ||
| II | Referent | |
| III/IV | 1.88 (1.48–2.38) | <.001 |
| Tumor grade | ||
| 1 or 2 | Referent | |
| 3 | 1.37 (1.08–1.73) | .010 |
| Tumor size, cm | ||
| <6 | Referent | |
| ≥6 | 1.11 (0.85–1.44) | .454 |
| Parametrial involvement | ||
| Unilateral | Referent | |
| Bilateral | 1.32 (1.04–1.67) | .021 |
| Para-aortic lymph node staging | ||
| Surgical | Referent | |
| Radiographic | 1.46 (1.08–1.99) | .014 |
- HR indicates hazard ratio; 95% CI, 95% confidence interval.
For patients who initially were rendered free of disease by their primary therapy, first sites of disease recurrence also were compared. Among 47 patients in the R group who developed recurrent disease, 63.8% of recurrences included sites outside the pelvis compared with only 51.6% among 219 patients from the S group (P = .127). However, there were significantly more recurrences that involved the PALNs (isolated PALNs or PALNs in addition to any other site) in the R group than in the S group (31.9% vs 15.1%; P = .006) (Table 4).
| Site of recurrence | % Surgical group, n = 219 | % Radiographic group, n = 47 | P |
|---|---|---|---|
| Out of pelvis | .127 | ||
| Yes | 51.6 | 63.8 | |
| No | 48.4 | 36.2 | |
| PALN | .006 | ||
| Yes | 15.1 | 31.9 | |
| No | 84.9 | 68.1 |
- PALN indicates para-aortic lymph nodes.
DISCUSSION
It is anticipated that between 15% and 30% of patients with locoregionally advanced cervical cancer have metastatic disease to the PALNs. The GOG demonstrated this in 2 prospective trials, which indicated positivity rates of 14.6%29 and 16.9%.30 PALN involvement was associated with disease stage: Positive PALNs were reported in 16% to 21% of patients with stage IIB disease and in 25% to 31% of patients with stage IIIB disease.2, 31, 32
Identification of positive PALNs by radiographic techniques has proven to be difficult. The sensitivities of CT, MRI, and even PET, as described above, have their limitations and may result in patients with unidentified PALN metastases receiving treatment with whole pelvic RT. For example, among individuals with stage IIIB disease, positive PALNs would be expected to occur in 25% to 30%31, 32 of patients. Heller et al. posited that, radiographically, only approximately 33% of these patients will be identified correctly. If this is to be believed, then an estimated 16% to 20% of patients undergoing radiographic staging will have extrapelvic disease but will fail to receive treatment with extended-field RT. A common criticism of the article by Heller et al. is that, although it was true at the time of publication, improvements in CT scanning have made the results invalid. Abnormal lymph nodes were defined in their study as exceeding 1.5 cm in greatest dimension. Because the resolution of CT scanners has improved, the criteria for reporting a lymph node as abnormal have changed. Currently, definitions vary from 0.5 cm to 1.0 cm, but lymph nodes >1.0 cm in the short axis generally are considered abnormal. Even with more conventional estimates of the sensitivity of cross-sectional imaging, however, 50% of PALN metastases will be missed, resulting in the under-treatment of an estimated 12% to 15% of the patients in our example.
One of the strengths of our analysis is the recent period during which patients in the R group were treated. All of the patients who underwent radiographic exclusion of the PALNs participated in GOG 165, a trial that was initiated in October 1997 and was closed in July 2000. Thus, the R group represents a contemporary assessment of the accuracy of available imaging. The estimated 12% to 15% difference in survival in our example above approximates the 12.7% difference in 4-year PFS and the 14.3% difference in OS for patients with stage IIIB disease in our study. This observation is strengthened further by the 17% increase in para-aortic recurrences detected in our study among patients who underwent radiographic exclusion of PALN metastases. In relative terms, this 17% absolute increase translates to a 111% increase in the relative risk of PALN recurrence.
Another explanation for improved survival among patients undergoing surgical staging may be that selection bias favored the surgical group. It is not unreasonable to believe that a healthier group of patients with better disease characteristics would have been selected to undergo surgical staging before chemotherapy and RT. In our series, however, this was not the case. Although patient age, histologic type, and tumor grade were similar between the 2 groups, patients in the R group had better baseline performance status, less advanced stage, smaller tumor size, and a trend toward less bilateral parametrial involvement compared with patients in the S group. The difference in performance status may be explained by the observation that patients in the S group were immediately postoperative and, thus, temporarily had a reduction in their performance status compared with patients in the R group, who had not undergone recent surgery. Differences in age and pathology remain, however, and cannot be explained away.
Similarly, a temporal bias may have been introduced, because all patients in the R group were enrolled in the most recent trial, whereas 95% of patients in the S group came from 2 earlier trials. Significant differences between trial design included the use of multiagent chemotherapy regimens in GOG 85 and GOG 120, small changes in the prescribed pelvic RT, and changes in surgical techniques that have occurred over time. One of the conclusions of GOG 85 and GOG 120 was that the addition of 5-FU with or without hydroxyurea did not improve survival compared with chemoradiation with cisplatin alone,24, 25 Therefore, the treatment regimens used in the S and R groups were considered equal and do not explain the lower survival observed in the R group. With each subsequent trial, minor improvements in radiation technique and stricter dose and duration limits occurred and were incorporated into GOG 165. If anything, this should have contributed to improved survival in the R group. Finally, improvements in surgical technique, including the use of laparoscopic staging, have occurred over time and were allowed on GOG 165. Because 95% of the patients in the S group came from earlier trials and, thus, did not receive the benefit of improvements in surgical evaluation, the differences observed in the current analysis may underestimate the actual impact of surgical evaluation in this patient population. Systematic factors that were introduced by the design of this analysis, therefore, either should have had no effect or actually should have favored the R group. Figure 3 illustrates the OS of patients on GOG 165 stratified only by mode of staging. Although it was not statistically significant because of the small numbers, again, the trend toward improved survival with surgical staging can be observed.
Kaplan-Meier estimate of overall survival according to the method of staging for patients on Gynecologic Oncology Group Protocol 165. The adjusted relative risk of death in patients who underwent radiographic staging, compared with patients who underwent surgical staging, was 1.74 (95% confidence interval, 0.75–4.03; P = .198).
Differences in the rates of positive pelvic lymph nodes between groups also would explain differences in survival. Information regarding the rates of positive pelvic lymph node status is incomplete. Surgical evaluation of the pelvic lymph nodes always has been considered optional. Overall, of 555 patients in the S group, only 58.4% underwent surgical assessment of their pelvic lymph nodes, and 25% (81 of 324 patients) were positive. Among 130 patients in the R group from GOG 165, only 21 patients (16.2%) reportedly had positive radiographic results. Either the patients who were chosen to undergo pelvic lymph node dissection were selected because of a suspicion of positive pelvic lymph nodes, or the rates were similar but the imaging failed to detect pelvic lymph node metastases. The alternate explanation is that, again, the S group represented the higher risk group.
A likely explanation for the improved survival in the S group is “stage migration.” If patients with PALN metastases were more likely to be excluded by surgical sampling rather than by radiographic evaluation alone, then the S group would have included a lower risk group of patients (ie, lower “stage”) than the R group. Patients with identified PALN metastases then could go on to be treated with potentially curative, extended-field RT. Before the widespread use of radiosensitizing chemotherapy, PALN-positive patients who were treated with RT alone had a 1-year survival rate of only 33%33 and had a 5-year survival rate between 17% and 29%.34-36 The Radiation Therapy Oncology Group reported an improved 2-year OS of 47% when extended-field RT was combined with chemotherapy,37 similar to the 44% 2-year OS rate reported by Odunsi et al.8 Varia et al., reporting for the GOG, demonstrated 3-year PFS and OS rates of 34% and 39%, respectively,38 when patients with PALN metastases were treated with extended-field RT and concomitant 5-FU/cisplatin. It is noteworthy that the Varia study also demonstrated 3-year PFS and OS rates of 38% and 59%, respectively, for a subset of patients who had PALN-negative radiographic results but who subsequently had PALN-positive surgical results. Finally, Goff et al.39 reported one of the best OS rates in this population: With a median follow-up of 33 months, 15 women who were identified with positive common iliac lymph nodes or PALNs had a 5-year survival rate of 52%.
Several studies actually suggested a survival advantage for patients who underwent extraperitoneal staging laparotomy before definitive RT. Downey et al. reported similar 5-year disease-free survival rates for patients with completely resected lymph nodes whether the lymph nodes were involved microscopically (57%) or macroscopically (51%).40 In a subsequent report, Cosin et al. again demonstrated equivalent 5- and 10-year disease-free survival for patients with completely resected microscopically involved lymph nodes (43% and 35%, respectively) or macroscopically involved lymph nodes (50% and 46%, respectively). Those authors also reported a significant improvement in disease-free survival for patients with macroscopically involved lymph nodes who underwent complete resection compared with patients who had unresectable lymph nodes (0% at 3 years).20 Marnitz et al.21 reported similar findings among 32 patients with positive PALNs. A significant survival advantage was observed among patients who underwent surgical debulking (defined as having ≥5 PALNs removed) compared with patients who had <5 PALNs resected, with 3-year OS rates approximating 55% and 13%, respectively (P = .01).
In addition to surgical debulking, Goff et al.39 suggested that surgical lymphadenectomy provided more accurate diagnostic information than CT, resulting in a 43% treatment modification rate to allow for better individualization of therapy. Similar findings have been reported by others, with 18% to 45% of patients undergoing modification in their prescribed therapy based on the results of lymphadenectomy.8, 21, 41-43
Direct comparison between surgical staging and radiographic staging is limited. Holcomb et al.43 retrospectively evaluated 274 women with stage IIB through IVA cervical cancer who were treated with primary RT. Eighty-nine of those patients underwent pretreatment staging laparotomy, and 172 underwent standard clinical staging. In that study, PALN metastases were identified in 12.3% of the surgical group, and those patients were treated with extended-field RT and/or systemic chemotherapy. The median survival of patients in the surgical group was 29 months compared with 19 months among the clinically staged patients (P = .01). A multivariate analysis that controlled for stage and patient age confirmed that staging laparotomy was a significant predictor of survival.
The only randomized trial to our knowledge that compared surgical staging (extraperitoneal or laparoscopic) with clinical staging was performed by Lai et al.44 That study included 61 patients with IIB through IVA cervical cancer who were randomized between surgical staging and clinical staging. All patients were to receive whole pelvic RT with or without concurrent chemotherapy, and those with pathologically documented PALN metastases were to receive extended-field RT chemotherapy. PALN metastases were documented in 8 of the 32 surgically staged patients (25%). The surgical group was more likely to have stage IIIB disease and adenosquamous histology and was less likely to have received concurrent chemotherapy. Two patients with documented PALN metastases did not receive extended-field RT. With a median follow-up of 58 months, the 4-year OS rate was approximately 58% in the clinically staged group and 35% in the surgically staged group (P = .024). Of the 8 patients who had surgically documented PALN metastases, 6 patients died of disease, and the other 2 patients died of treatment-related causes without evidence of disease. Thus, the conclusions from that study were limited because of the small sample size, unequal distribution of unfavorable characteristics between groups, and deaths of all patients with surgically proven PALN metastases from treatment failures or complications.
Strengths of our study comparing surgical versus clinical exclusion of PALN metastases are, first, that it remains the largest study of its kind to our knowledge. Although this study was retrospective, patients were matched fairly well for both patient and tumor characteristics. Where differences were observed, they tended to favor the R group. Finally, patients in the R group underwent evaluation with contemporary imaging (1997–2000). Despite these findings, patients who underwent surgical exclusion of the PALNs demonstrated an improvement in survival and a decrease in PALN recurrence.
Although it is not a weakness, 1 limitation of the current study is that it did not address the impact of 18F-fluorodeoxyglucose (FDG)-PET or PET/CT imaging studies on the identification of PALN metastases. These functional imaging techniques are being used more frequently in the initial evaluation of patients with cervical cancer. A recent analysis of 15 published PET studies in cervical cancer by Havrilesky et al.45 reported that the pooled sensitivity and specificity of FDG-PET for PALN metastases were 84% (95% CI, 68%–94%) and 95% (95% CI, 89%–98%), respectively (the pooled sensitivity and specificity for CT or MRI were not available). The single-institution nature of these series, along with reports of false-positive FDG uptake associated with infection or inflammation and false-negative scans associated with small metastases, should limit the reliance on such techniques until results are confirmed in a multi-institutional, prospective trial.
There appears to be a significant prognostic impact associated with surgically versus radiographically determined positive PALNs in patients with locally advanced disease. The current results demonstrate that a significant number of these patients who are assessed only radiographically may have unidentified PALN metastases, resulting in an increased rate of regional recurrence and decreased PFS and OS. This is most dramatic in patients with stage III disease.
