Abstract
Clarifying the diagnosis of clinically suspected recurrence of cervical cancer can be challenging. The aim of this study was to investigate the clinical value of 18F-FDG PET in this context. Methods: The medical records of a cohort of 40 18F-FDG PET referrals in whom recurrence of cervical cancer was clinically suspected were reviewed. Two expert gynecologic oncologists assessed the level of pre-PET clinical doubt, quality of pre-PET work-up, and impact of 18F-FDG PET on diagnostic understanding and management using questionnaires. Results: In patients with clinically equivocal recurrence, 18F-FDG PET had a sensitivity of 92% and a specificity of 93% (prevalence, 65%). Before 18F-FDG PET, there was high disagreement about the adequacy of the conventional work-up (intraclass correlation coefficient [ICC], 0.25) and the presence of recurrence (ICC, 0.24). 18F-FDG PET increased experts' confidence (median increase, 14% and 25%; P < 0.0001) and diagnostic agreement (from 68% to 98%; ICC, from 0.24 to 0.95). When 18F-FDG PET was positive for recurrence, the median overall survival was 13 mo. For patients with negative 18F-FDG PET findings, the median survival was not reached (log rank, 15.50, P = 0.0001). When the treatment plan was categorized as local therapy, systemic therapy, and expectative management, 18F-FDG PET changed the treatment plan in half of all cases. The 2 experts reported that 18F-FDG PET led to a better diagnosis and a beneficial change in management in, respectively, 60% and 65% of cases. Conclusion: 18F-FDG PET can help to clarify the diagnosis of clinically suspected recurrence of cervical cancer. In this patient population, 18F-FDG PET had significant value in diagnostic understanding and management of recurrent cervical cancer, facilitating decision making and treatment planning. Therefore, 18F-FDG PET should be part of the diagnostic work-up in detection of recurrent cervical cancer. The high positive predictive value of 18F-FDG PET in these patients suggests that inclusion in intervention trials might be based on a positive 18F-FDG PET scan.
Cervical cancer is the second most frequently diagnosed cancer in women worldwide (1). Depending on the stage of the disease, primary treatment consists of surgery, radiotherapy, chemotherapy, or a combination of these. After treatment for stage IB or higher, approximately one third of patients will have a recurrence (2,3). For recurrent disease, curative and palliative therapies prevail (4–6); some patients with localized recurrence may be cured with appropriate aggressive therapy (4,5). However, curative therapy is futile in the presence of distant metastases but may induce unnecessary morbidity.
Clinical suspicion of recurrent disease most often arises from symptoms (pain, lymph edema), rising tumor markers (SCC or CA-125), or radiologic findings. Imaging with CT or MRI has limitations in distinguishing recurrences and iatrogenic changes (7,8). Taking representative biopsies of radiologically equivocal lesions proves to be difficult, so that diagnosis of recurrent disease can be a major clinical challenge. This obviously leads to prolonged anxiety in patients and delay of treatment. 18F-FDG PET seems to have an acceptable diagnostic performance in suspected recurrent cervical cancer (9–12), and it may affect patient management (13). Furthermore, 18F-FDG avidity may predict the prognosis of recurrent cervical cancer (13). Taken together, prognostic stratification on noninvasive tests would imply a major improvement. The aim of the present retrospective analysis was to explore the potential role of 18F-FDG PET in the context of clinically equivocal recurrent cervical cancer.
MATERIALS AND METHODS
Patients
In a retrospective analysis, we identified all consecutive patients referred for 18F-FDG PET because of equivocal clinical symptoms or findings or radiologic test results on the presence of recurrent cervical cancer. Patients with histologically confirmed cervical cancer and primary treatment with curative intent were eligible. The conventional work-up (CWU) was recorded and comprised physical examination, laboratory findings, and radiologic imaging.18F-FDG PET was performed at the Department of Nuclear Medicine and PET Research, VU University Medical Center, Amsterdam, The Netherlands. After 18F-FDG PET, clinical follow-up was at least 12 mo. We reviewed all available medical records to document the clinical follow-up.
18F-FDG PET and Analysis
After patients had fasted for at least 6 h, whole-body 18F-FDG PET scans (vertex to mid thighs) were performed using an ECAT HR+ PET scanner (Siemens/CTI). Two-dimensional emission scans of 5 min per bed position were performed, starting about 60 min after intravenous administration of 18F-FDG (370 MBq). Serum glucose levels were within the reference range in all patients. Patients were orally hydrated in advance and instructed to void before scanning started, to reduce physiologic urinary tract activity. All scans were corrected for decay, scatter, and randoms. Scan data were reconstructed by ordered-subset expectation maximization with 2 iterations and 16 subsets, followed by smoothing of the reconstructed image using a Hanning filter at Nyquist frequency. Attenuation correction was not performed (14).
An independent, experienced (>10 y reading PET results) nuclear medicine physician, who was unaware of the patients' histories and outcomes, analyzed the 18F-FDG PET images. Focally enhanced 18F-FDG uptake other than physiologic uptake was interpreted as potentially malignant. Previously obtained CT and MRI scans were available for localization of PET abnormalities. 18F-FDG PET and CWU results were compared with the true lesion status obtained by histopathology or clinical follow-up and were classified as true-positive, true-negative, false-positive, or false-negative. In the setting of the accuracy determination, we classified a negative clinical follow-up at 12 mo after 18F-FDG PET as absence of recurrence.
Data Analysis
Two independent gynecologic oncologists (14 and 19 y of experience in gynecology), from 2 different academic centers, were asked for an objective expert opinion. The 2 experts were unaware of the patients' outcome and completed questionnaires consisting of 3 parts (Appendix). In the first part, questions were based on the medical and radiologic reports before 18F-FDG PET. Physicians were asked to provide their opinion on the adequacy of the CWU, the diagnosis and extent of recurrence, their confidence about the diagnosis (using a visual analog scale), and the management plan. The second part presented the 18F-FDG PET of the masked nuclear medicine physicians and the same questions were repeated. In the third part, the actual treatment and follow-up of the individual patients were provided, and the 2 experts were requested to rate the usefulness of 18F-FDG PET retrospectively. For this purpose, the impact on diagnostic understanding and treatment plan was graded by a method adapted from Wittenberg et al. (15–17). This method involves the following 5-point ordinal scale, with higher scores representing an increase in positive impact: 18F-FDG PET led me to choose therapy that, in retrospect, was not in the best interest of the patient (1), 18F-FDG PET led to investigations I would not otherwise have performed, but this did not harm the patient (2), 18F-FDG PET was of no influence in my diagnosis and choice of treatment (3), 18F-FDG PET provided information that substantially improved my understanding of this patient's disease but did not alter my choice of treatment (4), and 18F-FDG PET led to a better diagnosis and led to a beneficial change in management (5).
Statistical Analysis
Statistical analysis was performed using SPSS software (version 15.0 for Windows; SPSS, Inc.). The level of interobserver agreement on the presence or absence of recurrent disease was assessed with intraclass correlation coefficients (ICCs) (18). In this setting, ICC is the proportion of total variance that can be ascribed to “true” differences. Values for the ICC range from 0 to 1. Values close to 0 indicate poor agreement, whereas values close to 1 indicate high agreement. ICC values obtained from the fixed-effects ANOVA tables are comparable to weighted κ-scores, with the weighted κ calculated using quadratic weights. The Wilcoxon signed rank test was performed to compare the expert's confidence about the diagnosis before and after 18F-FDG PET. To explore the prognostic value of CWU and 18F-FDG PET, the Kaplan–Meier method was used. Survival was the time between the date of 18F-FDG PET and the date of death or the date at which patients were last known to be alive.
RESULTS
Patient Population
A total of 40 eligible patients were identified (median age at initial presentation, 41 y; range, 27–61 y). One patient was excluded because of insufficient clinical data. Patients had presented with cervical cancer of at least Fédération Internationale de Gynécologie et d'Obstétrique stage IB (Table 1). Primary treatment had consisted of surgery (n = 11), radiotherapy (n = 7), chemoradiation (n = 13), surgery and radiotherapy (n = 3), surgery and chemotherapy (n = 1), or surgery and chemoradiation (n = 5).
Patient Characteristics
In most patients (58%), clinical symptoms (e.g., pain) were the first symptoms bringing on a suspicion of recurrence, but equivocal lesions on CT or MRI were the most frequently reported indications for 18F-FDG PET (60%) (Table 2). The median time to 18F-FDG PET was 16 mo (interquartile range [IQR], 10–34 mo) from primary presentation of cervical cancer (Table 1) and 3 mo (IQR, 1–5 mo) from first suspicion of recurrence. The median follow-up after 18F-FDG PET was 18 mo (IQR, 10–37 mo) for the total patient population and 36 mo (IQR, 19–50 mo) for patients with a negative 18F-FDG PET result.
Indications for 18F-FDG PET
In these patients with clinically equivocal recurrences, 18F-FDG PET had a sensitivity of 92% (95% confidence interval [CI], 81%−96%), a specificity of 93% (95% CI, 71%−100%), a positive predictive value of 96% (95% CI, 84%−100%), and a negative predictive value of 87% (95% CI, 66%−93%). Two patients had a false-negative 18F-FDG PET result. One had lung, mediastinal, and paraaortic lymph node metastases, and the other had a pelvic recurrence, clinically detected 6 and 10 mo, respectively, after a 18F-FDG PET examination with negative results. There was 1 false-positive result, showing a retrovesical 18F-FDG focus, also suggestive at CT.
For staging of recurrent disease, 18F-FDG PET accurately detected local recurrence in 9 cases (sensitivity, 100%; specificity, 97%), regional recurrence in 13 cases (sensitivity, 87%; specificity, 100%), and distant metastases in 6 cases (sensitivity, 75%; specificity, 100%). The lower sensitivity for distant metastases was caused by 2 false-negative results in 2 patients with lung metastases found 5 and 6 mo, respectively, after 18F-FDG PET. The 18F-FDG PET readings of the masked nuclear medicine physician were highly congruent with the original 18F-FDG PET reports (95% for the presence of recurrence, 93% for the extent of recurrence; Table 3).
Reading by Experienced Nuclear Medicine Physician and Original 18F-FDG PET Report
In retrospect, 26 patients (65%) had recurrent disease at the time of 18F-FDG PET. Six of them were treated with curative intent. In 2 patients (1 with exenterative surgery and chemotherapy and 1 with chemoradiation), continued follow-up for more than 35 mo showed no evidence of disease. Three patients died within 36 mo, and 1 patient had an incurable rerecurrence within 30 mo after exenterative surgery. In addition, 2 patients were diagnosed with recurrent disease at 35 and 46 mo after a 18F-FDG PET examination with negative results. At the end of follow-up, 23 patients had died of their disease, 3 were alive with disease, and 14 had no evidence of disease.
Clinical Expert Opinion
In 29 cases (73%), at least 1 expert reported inadequate CWU; in 14 cases (35%), both experts reported inadequacy. Considerable disagreement existed between the 2 observers about the adequacy of the CWU (ICC, 0.25). Additional imaging (e.g., bone scintigraphy, CT, and MRI) and biopsies were most frequently reported to be missing. On the basis of the clinical data before 18F-FDG PET, the 2 experts had a median confidence of 65% and 50% about their diagnoses (Table 4), with poor agreement between them (ICC, 0.24). After 18F-FDG PET, the experts changed their diagnoses on the presence of recurrence in 25% and 30% of cases and concerning the extent of recurrence in 38% and 53% of cases. In 5 and 3 cases, the 2 experts correctly changed the status of recurrent disease from absence to presence of distant metastases. 18F-FDG PET results increased the diagnostic confidence of the experts by a median of 14% and 25% (Wilcoxon signed rank test, P < 0.0001). Furthermore, 18F-FDG PET improved the diagnostic agreement on the presence of recurrence by 30% (from 68% to 98%; ICC, from 0.24 to 0.95).
Diagnosis Before and After 18F-FDG PET
18F-FDG PET was strongly predictive of overall survival (log rank, 15.50; P = 0.0001), and this was significantly better than the predictive value of the CWU results of the individual experts (Fig. 1). When 18F-FDG PET was positive for recurrence, the median overall survival was 13 mo. For patients with 18F-FDG PET findings negative for recurrence, the median survival was not reached.
Kaplan–Meier curves for overall survival of patients with clinically suspected recurrence of cervical cancer according to CWU of experts 1 (A) and 2 (B) and according to 18F-FDG PET (C).
When the treatment plan was categorized as local therapy (surgery, radiotherapy, hyperthermia), systemic therapy (chemotherapy or palliative treatment consisting of supportive care), and expectative management, 18F-FDG PET changed the treatment plans of the 2 experts in 19 (48%) and 20 (50%) cases (Table 5). Local therapy was the most frequently changed treatment modality (7 and 10 cases), followed by systemic therapy (both experts, 6 cases). Radiotherapy was the most frequently changed local therapy (both experts, 4 cases), whereas chemotherapy was the most frequently changed systemic treatment (both experts, 3 cases). In 3 and 5 cases, 18F-FDG PET revealed unexpected distant metastases, preventing extensive locoregional therapy in 1 and 2 cases, respectively. In 6 cases (15%), both experts changed the intent of the management plan (curative vs. palliative intent), mainly because of a change in the extent of the disease. In 80% of the cases, the expert management plan was similar to what happened in clinical practice (Table 6). Finally, the experts reported that 18F-FDG PET led to a better diagnosis and a beneficial change in management in 60% and 65% of cases, respectively (Table 7).
Change in Management
Actual Management, Compared with Management of Expert
Impact of 18F-FDG PET on Diagnostic Understanding and Management*
DISCUSSION
The current study is the first, to our knowledge, to report on the added value of 18F-FDG PET on a measured and externally validated level of uncertainty about the presence of recurrent cervical cancer. Before 18F-FDG PET, the expert panel reported inadequacy of the CWU and high disagreement about the diagnosis with low ICC values. We found considerable added value for 18F-FDG PET not only in terms of expert opinion but also in terms of substantiated accuracy measures for detection (sensitivity or specificity > 90%) and staging. Furthermore, 18F-FDG PET affected treatment planning, and 18F-FDG PET results had a higher prognostic value than did CWU. A high agreement between physicians facilitates decision making on treatment planning. The low rate of false-positive results and high prognostic value suggest that 18F-FDG PET is an important diagnostic tool for detecting recurrent cervical cancer in clinically equivocal patients.
For treatment planning, we have extended the study by Chung et al. (13), which reported that 18F-FDG PET/CT changed management in 23.1% of cases in suspected recurrent cervical cancer. However, in the current study the experts reported a potential change in management in almost half of all cases (48% and 50%). In 5 and 9 of these cases, respectively, 18F-FDG PET might have saved patients from the morbidity of active treatment regimens; treatment plans consisting of surgery, radiotherapy, hyperthermia, or chemotherapy might have been changed to expectative management as a result of an improved assessment of dissemination.
The potential impact of 18F-FDG PET on the treatment plan might be even higher, because the current analysis did not discriminate between local therapy and its adjuvant treatment (e.g., surgery with or without subsequent chemotherapy; Table 5). The higher frequency of change in treatment plan management found in our study, compared with the findings in the study of Chung et al. (13), may be explained by differences in the actual level of uncertainty. However, the retrospective design of the current study might pose a possible limitation because unknown factors not described in the medical records may have contributed to the original decision-making process. Nevertheless, in 80% of the cases, the expert management plan was similar to what happened in clinical practice. The observed discrepancies were random and few.
Some other studies have documented that 18F-FDG PET can change the management of recurrent cervical cancer (10,19). 18F-FDG PET seemed important in determining whether patients can be treated with curative intent (20,21).
Six of 26 patients who were diagnosed after signs and symptoms became apparent received curative treatment. However, only 2 of them remained disease-free for more than 35 mo, as is typically observed in symptomatic patients (22). It is yet unclear whether earlier diagnosis of recurrence will improve patient outcome, that is, whether our patients would have benefited from an earlier 18F-FDG PET scan.
If 18F-FDG PET were to be applied as a screening procedure in asymptomatic patients, the presently observed predictive values cannot be directly extrapolated. The much lower prevalence in such patients will result in higher false-positivity rates (23). Projecting our sensitivity and specificity on a prevalence of 12% (23), the false-positive rate would be about 40%, at a negative predictive value of 99%.
Recurrent cervical cancer continues to be a therapeutic dilemma, and prospective randomized clinical trials are necessary to evaluate new treatment strategies to prolong survival. Intervention trials require objective diagnostic procedures to detect and evaluate recurrent disease. However, detecting clinically suspected cervical cancer can be challenging, as illustrated by the reported inadequacy of the CWU, the rather low confidence of the experts, and the low diagnostic agreement among them. In the majority of cases, a suspicion of recurrence was first raised by clinical symptoms and elevated tumor markers, but these were not the direct indications for 18F-FDG PET (Table 2). Equivocal results of CT, MRI, and pathologic examination almost completely covered the spectrum of indications to apply for an 18F-FDG PET examination, indicating that these diagnostic tests do not provide conclusive evidence in confirmation of recurrence. When a recurrence of cervical cancer was clinically suspected, 18F-FDG PET had high accuracy in detecting and staging recurrence and was of major additional value in diagnostic understanding and treatment planning. Furthermore, the use of 18F-FDG PET/CT (rather than 18F-FDG PET) in these patients will likely reduce the number of diagnostic tests and help to identify biopsy sites, avoiding interpretation problems due to poor anatomic localization at 18F-FDG PET alone (13,24).
In current practice, 18F-FDG PET or PET/CT is not part of the CWU of recurrent cervical cancer and is usually performed only as a problem solver at the end of the CWU. We believe that this policy should be changed in such a way that 18F-FDG PET or PET/CT would be performed routinely in cases of a suspected recurrence of cervical cancer. Only then will 18F-FDG PET or PET/CT decrease the time lag between diagnosis and treatment of recurrent cervical cancer, potentially improving patient outcomes. Prospective studies are warranted to investigate if early detection of recurrence by 18F-FDG PET or PET/CT can prolong the survival or will reveal only the tip of the iceberg. The high positive predictive value of 18F-FDG PET in patients with a suspected recurrence of cervical cancer suggests that inclusion in prospective intervention trials might be based on a positive 18F-FDG PET scan.
CONCLUSION
18F-FDG PET can help to clarify the diagnosis of clinically suspected recurrent cervical cancer. In this patient population, 18F-FDG PET had significant value in diagnostic understanding and management of recurrent cervical cancer, facilitating decision making and treatment planning.
APPENDIX
Questionnaire on Impact of 18F-FDG PET
Acknowledgments
We thank Lotty Hooft for help with data analysis and interpretation.
Footnotes
-
COPYRIGHT © 2008 by the Society of Nuclear Medicine, Inc.
References
- Received for publication June 9, 2008.
- Accepted for publication September 2, 2008.
Jump to section
Related Articles
Cited By...
- No citing articles found.