The clinical value of positron emission tomography (PET) with [18F]fluoro-2-deoxy-D-glucose (FDG) for primary staging in cervical carcinoma appears to be promising. The authors sought to evaluate the diagnostic efficacy and benefit of PET in restaging cervical carcinoma at the time of first recurrence.
Forty patients with cervical carcinoma who experienced confirmed treatment failure but who were feasible candidates for curative salvage therapy were enrolled prospectively in the current study. Restaging was performed with PET and with computed tomography and/or magnetic resonance imaging (CT/MRI). Dual-phase PET was performed by adding 3-hour-delayed images to the 40-minute scans. The results of the PET and CT/MRI scans were compared. Lesion status was determined by pathologic findings or by clinical follow-up. The receiver operating characteristic curve method with calculation of area under the curve (AUC) was used to evaluate diagnostic efficacy. The primary endpoint was percent improvement in restaging (with improvement indicated by treatment modification) after PET. The secondary endpoint was 2-year overall survival among study participants compared with comparable previously treated patients who did not undergo disease restaging with PET.
Twenty-two patients (55%) had their treatment modified due to PET findings. PET was significantly superior to CT/MRI (sensitivity: 92% vs. 60%; AUC: 0.962 vs. 0.771; P < 0.0001) in identifying metastatic lesions. For individuals receiving primary surgery, a significantly better 2-year overall survival rate was observed among study participants compared with patients who underwent disease restaging without PET (HR, 0.21 [95% confidence interval, 0.05–0.83]; P = 0.020).
Uterine cervical carcinoma is the third most prevalent female malignancy worldwide.1 Approximately 30–35% of patients with International Federation of Gynecology and Obstetrics (FIGO) Stage IB–IV disease have recurrent or persistent disease after primary definitive treatment. Outcome generally is poor when primary treatment has failed.2–4 For local and/or regional recurrences after primary surgery, curative salvage radiotherapy (RT) with or without chemotherapy can be administered if adjuvant pelvic RT has not already been administered. For those who have received primary or adjuvant RT to the pelvis, only central failure (cervical/vaginal involvement with or without bladder or rectal extension and with free pelvic sidewall) can be salvaged with surgery. In case of distant or multiple/disseminated recurrences, appropriate palliation is indicated.2–6 Once recurrence or persistent malignancy has been confirmed, a decision regarding curative or palliative treatment is crucial. Accurate restaging may improve survival, quality of life, and allocation of health care resources.
[18F]fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET) is a more accurate method of tumor detection than computed tomography (CT) and magnetic resonance imaging (MRI) for several types of malignancies, including cervical carcinoma.7–14 Several studies in which PET was incorporated into the primary staging of cervical carcinoma have reported promising results,10–13 although decreased sensitivity in detecting metastatic pelvic lymph nodes has been noted in other reports.14, 15 A limited number of retrospective studies have investigated the use of PET in the prediction of response to RT, in routine posttreatment surveillance, or in the determination of whether various clinical situations that are suggestive of recurrence actually are recurrences.16–20 To our knowledge, until the current report, a prospective study evaluating the role of PET in the restaging of recurrent cervical carcinoma has been lacking. We sought to evaluate the diagnostic efficacy and benefit of PET in restaging cervical carcinoma for patients who had experienced confirmed treatment failure but who potentially were curable. Trial participants were compared with comparable previously treated patients who had not undergone restaging of disease with PET at the time of first recurrence.
MATERIALS AND METHODS
Patients and Study Design
Eligibility criteria were 1) biopsy-documented recurrent or persistent cervical carcinoma (including squamous cell carcinoma, adenocarcinoma, and adenosquamous carcinoma) after definitive RT or surgery; 2) no contraindications to and willingness to undergo contrast-enhanced CT/MRI and PET scanning; 3) willingness to undergo image-guided biopsy or surgical exploration if indicated; and 4) potentially curable disease and willingness to receive curative salvage therapy if restaging with PET confirmed the possibility of curing the disease. Patients were ineligible if they 1) experienced re-recurrence after salvage therapy; 2) had only a superficial lesion on the cervix or vaginal cuff; 3) had disseminated abdominal or pleural lesions with positive fluid cytology; 4) had more than two involved regions; 5) were medically or psychologically unfit to receive curative salvage therapy; or 6) had a history of other malignancy, excluding basal cell carcinoma of the skin. Potentially curable disease was defined as 1) confirmed recurrent disease confined to the pelvis, if the patient had not received previous primary or adjuvant pelvic RT; 2) disease confined to the central pelvis, without pelvic sidewall or extrapelvic involvement, if RT had been administered before recurrence; 3) distant recurrences at a single site (such as paraaortic, supraclavicular, or inguinal lymph nodes [LNs]) that could be completely resected or encompassed by a curative RT procedure; or 4) pulmonary metastasis that had not spread to mediastinal LNs, was solitary or multiple (< 4 nodules) but confined to 1 lobe, and could be completely excised or irradiated.6 Before study entry, at the time of assessment for eligibility, chest X-ray and abdominal and pelvic CT scanning (when central disease was not suspected; n = 21) or MRI scanning (when central disease was suspected; n = 19) were performed. All enrolled patients then underwent additional neck and chest CT scanning. The study was approved by the institutional review board of the Chang Gung Memorial Hospital (Taoyuan, Taiwan), and informed consent was obtained from each enrolled patient.
CT/MRI scans of patients who were referred after initial treatment at another hospital (n = 2) were acceptable if the images were of sufficient quality. CT images were obtained with a spiral CT scanner (Hi-Speed; General Electric Medical Systems, Milwaukee, WI) or with a multislice CT scanner (Somatom Plus 4, Version A40; Siemens AG Medical, Forscheim, Germany). For all patients, oral meglumine diatrizoate (Gastrografin; Schering Health Care, West Sussex, United Kingdom) was administered for bowel preparation. Craniocaudal scanning from the upper neck to the symphysis pubis was performed. Gadolinium–diethylenetriamine pentaacetic acid (Ultravist 300; Schering Health Care) in a 100 mL intravenous bolus was used for contrast-enhanced CT. With the Hi-Speed CT scanner, contiguous 5 mm slices were obtained from the pelvis, abdomen, chest, and neck. With the Somatom Plus 4 CT scanner, contiguous 10 mm slices were obtained from the pelvis and abdomen, and 10 mm reconstructed helical scans were obtained from the chest and neck. Venous-phase imaging of target sites yielded adequate vascular tissue differentiation.
MRI scans were obtained with a 1.5-tesla scanner (Magnetom Vision or Magnetom Expert; Siemens Medical Systems, Erlangen, Germany) using a phased-array body coil with a 50 cm transverse field of view. For the pelvis and abdomen, transaxial, sagittal, and coronal sections were acquired with both T2 spin-echo (time repetition [TR]/time echo [TE], 4000/99) and T1 spin-echo (TR/TE, 500/15) sequences. For the chest and neck, transaxial, sagittal, and coronal sections were acquired with only a T2 spin-echo (TR/TE, 4000/150) sequence. The matrix size was 256 × 256. For the pelvis and abdomen, the slice thickness was 5 mm (transaxial plane) or 2.5 mm (sagittal and coronal planes). For the chest and neck, the slice thickness was 7 mm (transaxial plane) or 6 mm (sagittal and coronal planes).
FDG was provided by the Institute of Nuclear Energy Research of Taiwan (Lungtan, Taiwan). The imaging instrument used was an ECAT EXACT HR + PET camera (CTI, Knoxville, TN) with a full-width at half-maximum of 4.5 mm and a 15 cm transaxial field of view. For optimal tumor identification, all patients were catheterized and given diuretics to reduce bladder activity. After patients had fasted for > 6 hours and received 370 megabecquerels of FDG administered intravenously, dual-phase PET images were acquired at 40 minutes (40–96 minutes after FDG injection) from the head to the upper thigh using a 2-dimensional mode. This was followed by a session of 3-hour scans (180–210 minutes after injection) from the 11th thoracic vertebra to the upper thigh using a 3-dimensional mode. For dual-phase PET, early and late images were combined to determine whether observed lesions were fixed and to identify changes in standard uptake values (SUVs). Data acquisition and reconstruction were performed according to our previous protocol.21
PET images were interpreted primarily on the basis of visual analysis using transaxial, sagittal, and coronal displays. Visual score ranged from 0 to 4 (0, normal; 1, probably normal; 2, equivocal; 3, probably abnormal; and 4, definitely abnormal).7, 21 PET findings were interpreted by three nuclear physicians, with at least two in consensus. CT/MRI findings also were graded using a 5-point system (0, normal; 1, probably normal, visible LNs, size < 0.5 cm, reactive; 2, equivocal, at least 1 visible LN with size ≤ 1 cm; 3, probably abnormal, LN > 1 cm along short axis and/or multiple paraaortic or bilateral pelvic LNs (n ≥ 3) with size 0.5–1 cm; and 4, definitely abnormal, confluent LNs with central necrosis or irregular contour). CT/MRI findings for images other than LNs were interpreted by visual analysis.21
Study Procedures and Determination of Lesion Status
CT/MRI and PET scans were performed within 2 weeks of each other, ≥ 2 weeks before operation or biopsy. Fusion of CT/MRI and PET images using a commercially available software program (Hermes; Nuclear Diagnostics AB, Hagersten, Sweden) was performed for abnormally raised FDG uptake regions or in case of discrepant results. The axes were adapted automatically by using iterative registration based on minimization techniques. The fitted images were visually assessed by checking topographic landmarks. Tissue-based verification, either via surgical exploration or CT- or ultrasound-guided biopsy, was performed whenever possible for suspicious lesions (≥ visual score 3) detected by CT/MRI or PET. If biopsy was not feasible, a second assessment (including CT/MRI and PET imaging) was performed 3–6 months later to avoid false-negative results. Disease status was determined based on pathologic findings or clinical follow-up.
Clinical Characteristics of Study Participants and Historical Controls
A cohort of consecutive patients who received primary treatment at Chang Gung Memorial Hospital with either radical surgery or definitive RT and who experienced disease recurrence between 1991 and 1995 was retrieved from our database. Patients considered to be potentially curable based on conventional restaging according to the enrollment criteria of the current study were selected as historical controls (n = 125). Information such as age at initial diagnosis and at recurrence; dates of initial diagnosis and recurrence; FIGO stage, histologic type, and grade of differentiation at initial diagnosis; type of primary treatment; adjuvant therapy use; conventional radiologic examination (e.g., chest X-ray, CT/MRI) findings; results of all biopsies; pattern of recurrence; evidence (histologic, radiographic, or clinical) and site(s) of disease progression after salvage therapy; and status (alive or dead, with or without recurrence) at the date of last follow-up was extracted from the previous database for the historical control group and prospectively recorded for the study cohort. PET results, such as SUV and visual score for each site, and information on the salvage treatment plan (curativeness/palliativeness, modality, and extent) before and after PET were recorded prospectively for the study cohort.
The current study was a safety and efficacy trial. The primary endpoint was percentage of patients with improved restaging (as indicated by a change in treatment plan) after PET imaging. We hypothesized that 35% of study participants would have their treatment plans modified after PET imaging. Based on a 95% confidence interval (CI) of ±15%, approximately 40 patients were to be accrued to provide a sufficient sample size for evaluating the efficacy of PET in restaging. The secondary endpoint was relative benefit in terms of 2-year overall survival between study patients (with PET restaging) and comparable previously treated patients (without PET restaging). At the time of protocol design, toxicity related to guided biopsy or exploratory surgery was to be measured using the National Cancer Institute Common Toxicity Criteria for Adverse Events (CTCAE) Version 2.0; however, because CTCAE Version 3.022 was available at the time of analysis, it was used instead.
A visual score of 3 or 4 was considered to be positive in calculations of sensitivity, specificity, and accuracy, and scores of 0–2 were considered to be negative.21 The area under the receiver operating characteristic (ROC) curve was calculated. Area under the curve (AUC) was compared between PET results and CT/MRI results using the method of Metz.23 Survival curves were plotted according to the Kaplan–Meier method, and the log-rank test was used to evaluate the significance of the differences between the curves.24 For malignancy-related mortality, hazard ratios (HR) with 95% CIs were calculated using the Cox regression model.25 When applicable, the chi-square test or the Mann–Whitney U test was used to examine differences in clinical covariates between groups. All P values presented are two-sided.
A profile of the current trial is shown in Figure 1. Between May 3, 2001, and September 30, 2002, 45 patients were assessed for eligibility. Five were excluded, three due to the absence of tissue-based evidence, one due to concurrent lung carcinoma, and one due to the patient's refusal of further management. A total of 40 patients were eligible. Their median age at recurrence was 51 years (range, 25–87 years). Follow-up was continued until July 31, 2003. The median follow-up period until recurrence was 16 months (range, 10–27 months). The initial FIGO stage was IB in 13 cases, IIA in 4, IIB in 16, IIIB in 3, and IVA in 4. Fifteen patients underwent primary radical surgery, and 25 received primary RT. The interval between initial diagnosis of cervical carcinoma and first documented recurrence ranged from 2 to 273 months (mean ± standard deviation, 34 ± 58 months). The histologic disease type was squamous cell carcinoma in 30 cases and adenocarcinoma or adenosquamous carcinoma in 10.
Diagnostic Efficacy of PET
On a lesion-by-lesion basis, PET and CT/MRI results were compared with each other in 400 regions of interest (ROIs) (Table 1). Sixty-seven ROIs were recognized as having true-positive results based on either biopsy proof (n = 43) or positive PET or CT/MRI findings (without biopsy proof) and clinically obvious development of disease within 6 months (n = 24). Three hundred thirty-three ROIs were considered to have true-negative results. Of these 333 ROIs, 14 were found on PET or CT/MRI to contain ≥ score 3 lesions but exhibited negative biopsy findings, and 319 were negative on both PET and CT/MRI scans, without biopsy-proven negativity. The remaining 333 lesions were disease free at least 6 months after the imaging studies. The sensitivity of PET was greater than that of CT/MRI in terms of overall lesion detection (91% vs. 67%; P = 0.001), and the difference was even greater for detection of metastatic lesions (92% vs. 60%; P = 0.0003); sensitivity was similar for detection of recurrent/persistent local tumors (90% vs. 84%; P = 0.631). The ROC curves indicated that PET was superior to CT/MRI in terms of overall lesion detection (AUC: 0.971 vs. 0.854; P = 0.0004), and this superiority was most prominent in the detection of metastatic sites (AUC: 0.962 vs. 0.771; P < 0.0001) (Fig. 2).
Table 1. Comparison of the Evaluation of 400 Regions of Interest Using Positron Emission Tomography and Computed Tomography/Magnetic Resonance Imaging in 40 Patients with Recurrent Cervical Carcinoma
Central site/pelvis (recurrent or persistent tumors)
Treatment Modification Based on PET Imaging
After PET imaging, 22 (55%) of the 40 study patients had their treatment changed; 7 continued to be treated with curative intent but had their treatment field or modality changed, and 15 shifted to palliative treatment (Fig. 1). Of the 25 patients treated with curative intent (7 with treatment modification and 18 without modification), 13 underwent surgery with or without intraoperative RT, and 12 received concurrent chemotherapy and RT (CCRT) only. Table 2 summarizes the differential distributions of pattern of recurrence and salvage treatment plan before and after PET imaging among the 40 trial participants. Of the five patients in the primary surgery group who received palliative treatment, one experienced recurrence at a previously adjunctively irradiated pelvic sidewall, one had proven mediastinal LN metastasis, and the remaining three had unresectable recurrences at previously irradiated central/pelvic areas, along with distant metastases. Of the 10 patients in the primary RT group who received palliative treatment, 1 experienced treatment failure at an in-field paraaortic LN, 2 had mediastinal LN metastases, and the remaining 7 had unresectable central/pelvic disease and distant metastases.
Table 2. Differential Distributions of Pattern of Recurrence and Salvage Treatment Plan before and after Positron Emission Tomography Scanning of the 40 Trial Participants
Clinical Characteristics of Study Participants and Historical Controls
Table 3 lists detailed clinical characteristics according to primary treatment and use of PET imaging. In the primary surgery group, there were no significant differences between study participants (n = 15) and historical control patients (n = 40) in terms of age at initial diagnosis or first recurrence, FIGO stage, histologic disease type, grade of differentiation, time from first diagnosis to recurrence, or adjuvant treatment. In the primary RT group, study participants (n = 25) were marginally significantly younger at initial diagnosis (P = 0.020) and at recurrence (P = 0.039) and had a marginally longer time to recurrence (P = 0.073) compared with control patients (n = 85); in addition, CCRT was used more frequently in the study group (88%) than in the historical control group (15%; P < 0.0001).
Table 3. Clinical Characteristics According to Primary Treatment and Restaging with (n = 40) or without (n = 125) Positron Emission Tomography for Patients with Recurrent Cervical Carcinoma
With PET (n = 15) (%)
Without PET (n = 40) (%)
With PET (n = 25) (%)
Without PET (n = 85) (%)
PET: positron emission tomography; FIGO: International Federation of Gynecology and Obstetrics; SCC: squamous cell carcinoma; AD: adenocarcinoma; AS: adenosquamous carcinoma.
Early Stage IIB disease.
Grade of differentiation at initial diagnosis was not available for 22 patients in the primary radiotherapy group who did not undergo disease staging with positron emission tomography.
Adjuvant chemotherapy or radiotherapy with or without chemotherapy for the primary surgery group and concurrent chemotherapy for the primary radiotherapy group.
Benefits and Adverse Events Associated with the Use of PET for Restaging
Fifteen patients died of disease within 2 years, and 25 remained alive (14 with disease present). All seven patients who continued to receive curative treatment but who had their treatment field or modality altered currently are alive. In the primary surgery group, a significantly better 2-year overall survival rate was noted among study participants compared with patients whose disease was restaged without PET (HR, 0.21; 95% CI, 0.05–0.83; P = 0.020) (Fig. 3). Among patients receiving primary RT or CCRT, there was no difference between trial participants and patients whose disease was not restaged with PET (HR, 0.99; 95% CI, 0.53–1.85; P = 0.996).
Of the 40 study participants, 14 underwent an additional guided biopsy (n = 11) or exploratory surgery (n = 3) because of PET findings. Grade 1 or 2 pain (22.5%), gastrointestinal symptoms (5%), and constitutional symptoms (7.5%) lasting less than 7 days were observed in association with additional guided biopsy or exploratory surgery. Only one patient had Grade 3 pain due to exploratory surgery. No Grade 4 adverse events were encountered.
Metabolic PET imaging is a promising technique for primary staging in the management of cervical carcinoma.10–13 The efficacy of FDG-PET in evaluating locally advanced cervical carcinoma was demonstrated in a prospective surgicopathologic study in which PET scanning exhibited a sensitivity of 75%, a specificity of 92%, a positive predictive value of 75%, and a negative predictive value of 92% for paraaortic LN metastases.10 A small number of retrospective studies also have used PET for posttreatment surveillance or for confirmation of suspected recurrences.16–20 These studies are difficult to evaluate, because of the variability in their definitions of high-risk groups and suspicious for recurrence. In a study involving PET imaging for routine posttreatment surveillance of 249 patients with cervical carcinoma, 80 patients had positive lesions on PET scanning, but only 28 had confirmed recurrences (false-positive rate, 65%).18 In another study, involving 76 patients with cervical carcinoma who underwent pretreatment and posttreatment PET imaging, there were no survivors at 2 years among the 11 patients who developed new sites of abnormal FDG uptake.17 For the sake of improving the allocation of medical resources, sensible use of PET should involve consideration of cost-benefit issues, as PET remains relatively expensive.
Treatment options for recurrent cervical carcinoma after radical surgery or radiotherapy remain limited and controversial. Recurrences that are not isolated and centrally located usually are considered incurable, except for isolated lung metastases2, 6 and isolated paraaortic LN metastases.5, 26 Even for patients with isolated central recurrences, pelvic exenteration usually is necessary as secondary therapy if adjuvant or primary RT/CCRT was administered initially. Up to 40% of exenterations are abandoned due to the discovery of peritoneal, lymph nodal, or pelvic sidewall disease on exploratory laparotomy.27 It is clear that early detection and accurate assessment of recurrent disease are crucial for choosing the appropriate salvage treatment.
Differentiation of malignant lesions from postsurgical or RT-related inflammation, edema, scarring, and fibrosis often is difficult.28–30 Most scans in previous PET studies were performed approximately 1 hour after tracer injection.10–12, 14–16, 18–20 Inflammatory or benign proliferative lesions accumulate FDG more intensely but less durably than do malignancies. The utility of dual–time point PET scanning in differentiating malignant lesions from inflammatory and reparative changes has been reported in head and neck and lung malignancies.31, 32 A pilot study conducted by our group on the use of dual-phase PET (adding 3-hour-delayed images to scans obtained at 40 minutes) in cervical carcinoma indicated that combined analysis of early and late images was significantly superior to analysis of 40-minute scans alone in detecting metastases.21 In the current study, the difference in sensitivity of recurrent/persistent local tumor detection between FDG-PET and CT/MRI was not significant, but PET was superior to CT/MRI in the overall detection of lesions (P = 0.0004) and in the detection of metastases (P < 0.0001). Furthermore, 55% of study participants had their treatment modified due to PET findings; this outcome exceeded the expectations of the trial design.
Treatment failure may result from unrecognized treatable metastases located outside the target site(s). In contrast, unproven suspicious lesions could prohibit viable curative action for some patients. Typical manifestations involve the vagina, cervix, bladder, rectum, and parametrium and are more easily recognized by CT/MRI, pelvic examination, or other clinical examinations. Most atypical manifestations; which involve the solid abdominal organs, peritoneum, mesentery, gastrointestinal tract, pleura, and the mediastinal, inguinal, and supraclavicular LNs,33 in addition to the pelvic and paraaortic LNs; were better detected by PET imaging in the current study and in our previous study,21 as well as in another study.18 Identification of incurable metastases eliminates unnecessary salvage procedures and suffering, while more accurate delineation of tumor extent increases the probability of successful treatment. Although the follow-up period in the current study is short, all seven patients who continued to be treated with curative intent but whose treatment field or modality was changed are alive. In addition, 35% (14 of 40) of the study participants underwent an additional guided biopsy or exploratory surgery because of PET findings; the associated morbidity was acceptable.
Better restaging does not guarantee improved outcome. In the primary surgery group in the current study, we observed significant benefit in terms of 2-year overall survival among study participants relative to comparable historical patients who did not undergo disease restaging with PET (P = 0.020); however, 2-year overall survival rates were similar for the corresponding cohorts in the primary RT group (P = 0.996). Clinical characteristics were similar for study participants and historical control patients in the primary surgery group. Therefore, the observed benefit in terms of overall survival probably is not due to background prognostic factors. In the primary RT group, CCRT use was more common among study participants than among historical control patients (P < 0.0001), because CCRT has been the standard of care for locally advanced cervical carcinoma since 1999. It is conceivable that tumors that recur after more aggressive treatment are more difficult to treat.
The current prospective study indicates that PET is superior to CT/MRI in the restaging of recurrent cervical carcinoma. For patients with proven treatment failure who are candidates for curative salvage therapy based on conventional assessment, PET findings may significantly cut down on unnecessary salvage attempts. Restaging with PET provides benefit by allowing the physician to offer optimal management of recurrent cervical carcinoma.