Presented at the 2005 meeting of the Society of Gynecologic Oncologists, Miami, Florida, March 18–23, 2005.
Increasing evidence has documented the value of positron emission tomography (PET) in oncology, but only limited data are available comparing PET findings with the pathologic status of regional lymph nodes in patients with cervical carcinoma. The objective of this study was to determine the sensitivity and specificity of PET in detecting lymph node metastasis in women with early-stage cervical carcinoma.
The authors performed a retrospective review of all patients with Stage IA–IIA cervical carcinoma who underwent PET before surgery from 1999 to 2004. The status of the regional lymph nodes was correlated with lymph node pathology.
Fifty-nine patients were identified. Pelvic lymph node metastases were present in 32% of the patients and were detected by PET with a sensitivity of 53%, a specificity of 90%, a positive predictive value (PPV) of 71%, and a negative predictive value (NPV) of 80%. Paraaortic lymph node disease was present in 9% of patients and was detected by PET with a sensitivity of 25%, a specificity 98%, a PPV of 50%, and an NPV of 93%. The mean size of the tumor deposits was larger in the PET-positive pelvic nodes (15.2 mm; range, 2–35 mm) than in the PET-negative lymph nodes (7.3 mm; range, 0.3–20 mm; P = 0.002). Computed tomography (CT) scans were obtained before surgery in 42 patients. The combined sensitivity of PET and CT in these patients was 75%. PET alone detected 9 (36%) of the positive lymph node groups, whereas CT alone detected 3 (12%) of the positive lymph node groups. Neither PET nor CT detected the positive lymph node groups in 8 patients (32%).
In 2005, 10,370 women will be newly diagnosed with cervical carcinoma in the United States, and 3710 women will die from the disease.1 Treatment for women with cervical carcinoma consists of surgery or radiation therapy, usually in combination with chemotherapy. Women with early-stage cervical carcinoma are amenable to treatment with either surgery or radiation therapy whereas chemoradiation is the treatment of choice for patients with advanced-stage disease. The current International Federation of Gynecology and Obstetrics (FIGO) staging system for cervical carcinoma is based primarily on physical examination. Although the status of the regional lymph nodes is not part of the FIGO staging system, the presence of lymph node metastases is a strong prognostic factor and is used to guide treatment.2
A variety of imaging modalities have been used to detect lymph node disease in women with cervical carcinoma. Historically, lymphangiography (LAG) was one of the first procedures used to evaluate the pelvic and paraaortic lymph nodes. More recently, computed tomography (CT), magnetic resonance imaging (MRI), and ultrasonography have been used for the evaluation of the regional lymph nodes. Although LAG depends on the presence of lymph node filling defects for the detection of metastatic lymph node involvement, MRI and CT rely on lymph node enlargement. A meta-analysis of 17 studies comparing LAG, MRI, and CT revealed that all 3 tests are only moderately sensitive and specific for women with cervical carcinoma.3 The diagnostic performance of anatomic imaging procedures, such as CT and MRI, is particularly limited in women with early-stage disease, in whom the involved lymph nodes often contain only microscopic metastases.4
Because it overcomes some of the limitations of anatomic imaging, positron emission tomography (PET) using the glucose analogue [18F]-fluoro-2-deoxy-D-glucose (FDG) has gained increased utility in clinical oncology.5 FDG-PET relies on increased glucose uptake and metabolism by tumor cells. Thus, FDG-PET imaging is not dependent on the morphologic characteristics of the lymph nodes alone. Although pretreatment FDG-PET has been used for staging women with cervical carcinoma, a paucity of data exists to correlate the findings of FDG-PET with the pathologic status of the pelvic and paraaortic lymph nodes, particularly for women with early-stage cervical carcinoma. The objective of the current study was to determine the performance characteristics and utility of FDG-PET in women with early-stage cervical carcinoma.
MATERIALS AND METHODS
This study was approved by the Washington University School of Medicine Human Studies Committee. We retrospectively reviewed the medical records of all patients with invasive, Stage IA2–IIA cervical carcinoma who underwent surgery at Barnes-Jewish Hospital between January 1, 1999 and September 30, 2004. At the time of initial diagnosis, all patients underwent clinical evaluation, including pelvic and rectovaginal examination, routine laboratory testing, and chest radiography. A subset of the patients underwent CT of the abdomen and pelvis as part of the initial staging evaluation. At the discretion of the attending gynecologic oncologist, a subset of the cohort also underwent imaging evaluation by PET-FDG.
After evaluation, the patients were taken to the operating room with the intention of undergoing a Type II or Type III radical hysterectomy.6 Based on the clinical likelihood of lymph node disease and surgeon preferences, several patients underwent retroperitoneal pelvic and paraaortic lymphadenectomy before they underwent radical hysterectomy.7 The pelvic lymph node dissection in all patients consisted of removal of the lymph node-bearing tissue surrounding the external and internal iliac vessels as well as the lymph node tissue within the obturator fossa superior to the obturator nerve. The paraaortic dissection in all patients entailed removal of the lymph node pads from the bifurcation of the common iliac artery to the inferior mesenteric artery. Suspicious lymph nodes were sent for frozen section, and, if they were positive, the surgical procedure was aborted. Those patients in whom the radical hysterectomy was aborted were treated with definitive chemoradiation. Adjuvant treatment of the surgically staged patients was individualized by each clinician and was based on pathologic risk factors.
Before November 2002, FDG-PET was performed with a conventional PET scanner, and the images were interpreted in a routine manner, as described previously.8 Thereafter, essentially all FDG-PET studies were performed with a hybrid PET-CT scanner (Biograph LSO 2; Siemens Medical Solutions, Malvern, PA) (Fig. 1). The CT component of the PET-CT studies was performed without administration of either oral or intravenous contrast agents. CT images (5-mm slices) typically were obtained from the base of the skull through the proximal thighs at 130 kVp and 110 mA. Emission PET images were obtained over the same anatomic extent beginning 45–60 minutes after the administration of 15–20 mCi FDG, with imaging times of 2–4 minutes per bed position, depending on patient weight. Initially, PET-CT studies were obtained without placement of a Foley catheter and without administration of intravenous fluids or furosemide (which was done routinely for conventional PET studies). Beginning in May 2004, these interventions to minimize urinary tract activity were reinstituted. PET images were scatter-corrected and reconstructed using ordered subset expectation maximization with the use of a postreconstruction gaussian filter at 8 mm full width at half maximum (FWHM). The PET-CT images were interpreted in a standard clinical fashion, both separately and in fused mode. Conventional CT was performed using fourth-generation CT units. Before the scan, patients received both oral and intravenous contrast agents. Axial images (8–10 mm thick) were obtained of the abdomen and pelvis from the base of the lungs to the pubic symphysis. All CT scans were interpreted in a standard clinical fashion. Lymph nodes that measured > 10 mm in greatest dimension were considered abnormal. Initial pathologic analysis was performed in a routine clinical fashion. Lymph nodes with metastatic tumor deposits were rereviewed by a single pathologist to assess the total greatest dimension of the involved lymph nodes as well as the size of the tumor deposits within the lymph nodes.
For statistical analysis, the lymph nodes were divided into the following lymph node groups: right and left paraaortic lymph nodes and right and left pelvic lymph nodes. Sensitivity, specificity, and the positive predictive value (PPV) and negative predictive value (NPV) of FDG-PET were calculated for the pelvic and paraaortic lymph nodes in each patient (patient-based analysis). To further evaluate the performance of FDG-PET, a lymph node-based analysis of the ability of FDG-PET to detect metastatic spread in each of the lymph node groups was performed (pelvic and paraaortic lymph node-based analysis). Categorical variables were compared using the Fisher exact test, and continuous variables were compared using the Student t test. A P value of < 0.05 was considered statistically significant.
During the study period, 198 women underwent radical hysterectomy and pelvic/paraaortic lymphadenectomy or lymphadenectomy with aborted radical hysterectomy. Fifty-nine of those patients (30%) underwent pretreatment FDG-PET. These 59 women comprise the study cohort. The demographic characteristics of the study population and of the women who underwent routine clinical staging are displayed in Table 1. The patients within the FDG-PET cohort more commonly presented with advanced-stage, high-grade tumors. The mean age of women in the PET group was 43 years. At presentation, the stage of their disease was Stage IA2 in 1 woman (1.7%), Stage IB1 in 49 women (83.1%), Stage IB2 in 4 women (6.8%), and Stage IIA in 5 women (8.5%). Forty-nine patients underwent a radical or modified radical hysterectomy and lymph node dissection, whereas 10 women underwent a retroperitoneal lymphadenectomy and an aborted hysterectomy because of the finding of lymph node metastases. Squamous cell carcinomas were found in 39 women (66.1%) in the cohort.
All 59 patients underwent pelvic lymphadenectomy. Pelvic lymph node metastases were documented histologically in 19 patients and in 26 pelvic lymph node groups. FDG-PET correctly identified 10 of 19 patients with lymph node disease. Four FDG-PET scans were falsely positive for pelvic lymph node disease. The patient-based sensitivity of FDG-PET for pelvic lymph node disease was 53% with a specificity of 90%, a PPV of 71%, and an NPV of 80% (Table 2). Twelve of 26 pelvic lymph node groups were identified correctly by FDG-PET. Eight false-positive lymph node groups were noted. The pelvic lymph node-based sensitivity was 46% with specificity of 91%, a PPV of 60%, and an NPV of 86%.
Table 2. The Accuracy of [18F]-Fluoro-2-Deoxy-D-Glucose–Positron Emission Tomography in the Detection of Pelvic and Paraaortic Lymph Node Metastases
Positive predictive value
Negative predictive value
FDG-PET: [18F]-fluoro-2-deoxy-D-glucose–positron emission tomography; 95% CI: 95% confidence intervals (for the analysis of patient-based and lymph node-based pelvic lymph nodes).
Pelvic lymph nodes (n = 59)
Lymph node based
Paraaortic lymph nodes (n = 45)
Lymph node based
Paraaortic lymphadenectomy was undergone by 45 of 59 patients (76%). Pathologically confirmed paraaortic metastasis occurred in four patients and in five paraaortic lymph node groups. FDG-PET correctly identified one of the four patients with positive paraaortic lymph nodes and produced one false-positive result. All four patients with paraaortic disease also had pathologically documented pelvic lymph node metastases. PET successfully identified the pelvic lymph node metastases in three of those four patients. The patient-based paraaortic sensitivity was 25%, the specificity was 98%, the PPV was 50%, and the NPV was 93% (Table 2). The paraaortic lymph node-based sensitivity was 40%, the specificity was 99%, the PPV was 67%, and the NPV was 97%. Distant and visceral metastases were not identified by FDG-PET or detected at the time of surgery in any of the patients.
The metastatic implants within the involved lymph nodes were then examined. The mean histologic size of the 12 PET-positive pelvic lymph nodes was 16.7 mm, compared with 15.4 mm for the PET-negative pelvic lymph nodes (P = 0.66) (Table 3). However, the mean histologic size of the metastatic tumor implants within the involved lymph nodes was greater in the PET-positive pelvic lymph nodes than in the PET-negative pelvic lymph nodes (15.2 mm vs. 7.3 mm; P = 0.0017). The histologic size of the tumor implants within the PET-negative lymph nodes ranged from 0.3 mm to 20 mm. The total paraaortic node size (11.5 mm vs. 11.7 mm) and the histologic size of the tumor implants (6 mm vs. 8.4 mm) also are presented in Table 3. No statistical comparisons were made because of the small sample sizes.
Table 3. Histologic Findings of Involved Pelvic and Paraaortic Lymph Nodes
PET-positive lymph nodes
PET-negative lymph nodes
PET: positron emission tomography.
The mean size of the entire involved lymph node.
The mean size of the tumor implant within involved lymph nodes.
Preoperative contrast CT scans were obtained in 42 of the patients. Fourteen of those patients had pathologically confirmed positive lymph nodes (20 positive pelvic lymph node groups and 5 positive paraaortic lymph node groups). Of the 25 involved lymph node groups, both PET and CT identified 5 (20%) of the involved lymph node groups, whereas both modalities failed to detect the lymph node metastases in 8 groups (32%). FDG-PET alone was able to detect metastatic disease in 9 patients (36%), whereas CT alone was positive in 3 patients (12%). Overall, the combination of FDG-PET and CT correctly identified 17 (68%) of the involved lymph node basins (Table 4).
Table 4. [18F]-Fluoro-2-Deoxy-D-Glucose–Positron Emission Tomography and Computed Tomography Correlations
Pelvic lymph nodes
Paraaortic lymph nodes
Pelvic and paraaortic lymph nodes
PET: positron emission tomography.
Positive lymph node groups of patients underwent both PET and computed tomography (CT) tests.
Values shown are the combined sensitivity for both PET and CT (lymph node detected by either FDG-PET or CT.
At last follow-up (median, 13.5 mos), recurrent disease had been documented in 9 patients (15.3%), including 6 recurrences at distant sites and 3 recurrences in the pelvis. Fifty-one patients (86.4%) were without evidence of disease, 5 women (8.5%) were alive with disease, and 3 women (5.1%) had died of disease.
The value of molecular imaging with FDG-PET has been demonstrated for a variety of solid tumors and has led to the increased application of PET imaging in oncology.5 For women with cervical carcinoma, FDG-PET has been used for initial staging and treatment planning, posttreatment surveillance and follow-up, and in prognostication.8–14 The prognostic value of FDG-PET was demonstrated in an evaluation of 152 patients with cervical carcinoma who were treated with primary radiation. In this cohort, posttreatment FDG-PET was a powerful predictor of survival: The 5-year, cervical carcinoma-specific survival rate among 114 women with no FDG uptake on posttherapy PET was 80%. The cervical carcinoma-specific survival rate dropped to 32% for patients with persistent uptake, whereas there were no survivors among the cohort with new sites of uptake.12 Havrilesky et al. noted that FDG-PET had high sensitivity and specificity for the detection of recurrent cervical carcinoma.15
Despite the widespread incorporation of FDG-PET imaging into the management of women with cervical carcinoma, a paucity of data exists to correlate histopathologic findings with the results of FDG-PET, particularly for women with early-stage disease.9, 10, 16–20 Rose and coworkers performed FDG-PET before lymphadenectomy in 32 patients with advanced-stage (Stage IIB–IVA) tumors. The sensitivity of FDG-PET for the detection of paraaortic metastases was 75%, whereas it was 100% for pelvic lymph node disease.20 Among patients with early-stage (Stage I–IIA) cervical tumors who are potential candidates for surgical management, the sensitivity of FDG-PET has been reported as 20–91%, whereas the reported specificity ranged from 77% to 100%.9, 17–19 Many of those reports were limited by the inclusion of patients with more advanced tumors who underwent pretreatment lymphadenectomy before definitive chemoradiation. Reinhardt and colleagues evaluated FDG-PET before radical hysterectomy in women with Stage IB–II disease. Among their 35 patients, the sensitivity was 91%, and the specificity was 100%.9 Belhocine et al. reported on 22 patients who underwent FDG-PET before planned radical hysterectomy and lymphadenectomy. Those authors found that 5 of their 18 patients who underwent lymphadenectomy had lymph node metastasis, with a total of 10 involved lymph nodes. Three false-positive results were noted in that series.17 Our findings are in accord with those previously published results. In our cohort of 59 women with Stage IA–IIA neoplasms, the sensitivity for pelvic lymph node involvement was 53%, and the specificity was 90%. For the detection of paraaortic disease, the sensitivity was 25%, and the specificity was 98%.
For women with Stage IB–IIA cervical neoplasms, treatment consists of either radical hysterectomy or primary radiation with or without chemotherapy; the 5-year survival rate among these women is equivalent for the 2 modalities.21 For those women who undergo surgical treatment, approximately 8–26% will have lymph node metastases.4, 21 In this subset of lymph node-positive, early-stage patients, survival is improved by the addition of postoperative adjuvant chemoradiation.22 Although complications are associated with both surgical treatment and primary radiotherapy, the combination of surgery and adjuvant postoperative radiation is associated with increased morbidity, especially for late complications.21, 23–26 Thus, an important objective of pretreatment evaluation for women with early-stage cervical carcinoma is the accurate assessment of the regional lymph nodes. If a reliable imaging modality is available before surgery, then those patients with known positive lymph nodes may be allocated to primary radiation to avoid the added morbidity of multimodality treatment. Currently available imaging modalities, such as CT and MRI, lack the required sensitivity to replace surgical lymphadenectomy.3 In our series, FDG-PET accurately identified 10 of 19 women with pelvic lymph node metastases and 12 of the 26 involved pelvic lymph node groups. In addition to nine false-negative FDG-PET scans of the pelvic lymph nodes, there were four false-positive scans. The patient-based specificity of FDG-PET for the identification of pelvic lymph node disease was 90%, and the NPV was 80%. Although FDG-PET demonstrated a relatively high specificity, it lacked the sensitivity and predictive value needed to replace surgical lymphadenectomy.
To analyze the lack of FDG uptake further in histologically positive lymph nodes, we reviewed pathologic samples from the patients with positive lymph nodes (26 positive pelvic lymph nodes, 5 positive paraaortic lymph nodes). We noted that the overall lymph node size was similar for the PET-detected lymph nodes and the PET false-negative lymph nodes. However, the metastatic implants within the pelvic lymph nodes were larger in the true-positive lymph nodes than in the false-negative pelvic lymph nodes (Table 3). Previous evaluations of anatomic imaging modalities have used the guideline of a lymph node dimension of 1.0 cm in the short axis as a cut-off size for lymph node metastasis.27, 28 In women with lymph node-positive cervical carcinoma, 80% of the involved lymph nodes measure < 1.0 cm in greatest dimension.4 Belhocine and coworkers noted the importance of lymph node size for FDG-PET detection. In their series, only 2 of 10 pathologically positive lymph nodes were detected. The 2 PET-positive lymph nodes measured > 1.0 cm in greatest dimension, whereas the 8 false-negative lymph nodes were micrometastases.17 In our series, the total greatest lymph node dimension was < 1.0 cm in 3 of the 17 (18%) PET false-negative lymph nodes, whereas the greatest dimension of the metastatic tumor deposits was < 1.0 cm in 12 (71%) of those false-negative lymph nodes. The mean size of the metastatic tumor implants within the involved pelvic lymph nodes was greater in the PET-positive pelvic lymph nodes than in the PET-negative pelvic lymph nodes (15.2 mm vs. 7.3 mm). Thus, it appears that the greatest dimension of the metastatic implant is more important than total lymph node size for FDG-PET detection. It is noteworthy that FDG-PET correctly identified micrometastatic tumor implants in 4 of 14 (29%) true-positive lymph nodes. However, it is of concern that FDG-PET false-negative results occurred even in lymph nodes with large tumor implants (up to 2 cm in greatest diameter). Thus, although total lymph node size and greatest metastatic implant dimension likely are important determinants for PET detection, even micrometastases with FDG uptake may be detectable by FDG-PET.
The current study had the advantage of including a large number of women with early-stage cervical carcinoma patients who were evaluated with whole-body FDG-PET before surgical staging. However, as in any retrospective study, an inherent selection bias likely exists. During the period of study, large numbers of women with cervical carcinoma underwent radical hysterectomy and lymphadenectomy without FDG-PET evaluation. Given the imbalance in the stage distribution and survival of the women in our series, it appears likely that women with small, early-stage tumors were not selected for PET imaging, whereas those women with larger neoplasms who were at higher risk for lymph node disease were more likely to undergo FDG-PET imaging. Therefore, our findings reflect the accuracy of FDG-PET in a selected cohort of women with cervical carcinoma, and not in the entire population of women with Stage IA–IIA tumors. In addition, during the final years of the study, imaging was performed with combined PET-CT. The influence of combined PET-CT on our findings is unknown; and, because of limited patient numbers, subgroup analysis was not possible. Future studies will rely on combined PET-CT.
The current findings revealed that FDG-PET imaging is a useful adjunct in the evaluation of women with Stage IA–IIA cervical carcinoma. Although the sensitivity of FDG-PET for the detection of lymph node disease was limited, the specificity of PET was high. These findings will aid in the preoperative evaluation and operative planning for patients with early-stage cervical carcinoma. A prospective trial to evaluate the utility of preoperative FDG-PET in women with cervical carcinoma may refine our estimates of sensitivity and specificity in those with early-stage disease.