Abstract
The aim of this study was to evaluate the diagnostic performance of 18F-FDG PET/CT for the preoperative assessment of lymph node metastases (LNM) in endometrial cancer patients and for the assessment of endometrial cancer recurrence (ECR) after primary surgical treatment. Methods: A comprehensive search was performed on Pubmed/MEDLINE databases for studies reporting the diagnostic performance of 18F-FDG PET/CT for assessment of LNM and ECR published up to August 15, 2015. Twenty-one studies (13 for LNM and 8 for ECR) were included in the systematic review and meta-analysis. Pooled estimates of sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, and diagnostic odds ratio of the 18F-FDG PET/CT were calculated along with 95% confidence intervals (CIs). A summary receiver-operating-characteristics curve (SROC) was constructed, and the area under the SROC curve (AUC) was determined along with Q* index. Results: The overall pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic odds ratio, and AUC (with 95% CI) of 18F-FDG PET/CT for detection of LNM were 0.72 (95% CI, 0.63–0.80), 0.94 (95% CI, 0.93–0.96), 10.9 (95% CI, 7.9–15.1), 0.36 (95% CI, 0.27–0.48), 39.7 (95% CI, 21.4–73.6), and 0.94 (95% CI, 0.85–0.99), respectively, whereas the corresponding numbers for detection of ECR were 0.95 (95% CI, 0.91–0.98), 0.91 (95% CI, 0.86–0.94), 8.8 (95% CI, 6.0–12.7), 0.08 (95% CI, 0.05–0.15), 171.7 (95% CI, 67.9–434.3), and 0.97 (95% CI, 0.95–0.98), respectively. The overall diagnostic accuracy (Q* index) in LNM and ECR were 0.88 and 0.93, respectively. Conclusion: 18F-FDG PET/CT has an excellent diagnostic performance for detecting LNM preoperatively and disease recurrence postoperatively in endometrial cancer patients.
Endometrial cancer is the most common gynecologic malignancy in the developed countries (1). The prognosis is traditionally determined by clinical and histopathologic factors—that is, age, histologic type, grade, and stage of disease including assessment of cervical invasion, depth of myometrial invasion, lymph node spread, and distant metastases (2–4). The 5-y overall survival rate is generally favorable, around 80%. However, pelvic lymph node metastases (LNM) represents the most common site for extrauterine disease at primary treatment, and the 5-y survival rate is around 50% for this patient subgroup (5).
Currently, the final staging of endometrial cancer is based on histopathologic findings at primary surgery, which includes abdominal exploration, peritoneal cytology washing from the pelvis, hysterectomy, bilateral salpingo-oophorectomy, and lymphadenectomy in selected patients presumed to have a high risk of disease spread (6–8). Routine systemic pelvic lymphadenectomy for early-stage endometrial cancer disease, although not well defined as a surgical technique, improves detection of LNM, but the procedure showed no survival benefit in 2 randomized clinical trials (9,10). Valid preoperative identification of patients with LNM who may benefit from lymphadenectomy is thus essential if futile surgical staging and unnecessary postoperative staging–related complications are to be minimized. If a noninvasive imaging technique could accurately preclude LNM preoperatively, lymphadenectomy procedures, currently with unproven clinical benefit for survival, could be safely circumvented. Hence, the development of noninvasive imaging methods enabling more accurate preoperative staging of endometrial cancer may facilitate better-tailored surgical decision making based on the selection of appropriate risk groups for LNM.
Conventional diagnostic imaging by transvaginal ultrasound, MRI, and CT provide detailed anatomic information, whereas functional or metabolic tumor characteristics may remain undetected. However, vigorous debate has challenged the use of anatomic assessments solely relying on tumor morphologic information, not taking into account functional tumor characteristics that may prove highly relevant for the clinical phenotype (11–13). In this regard, to better understand the tumor microenvironment, metabolic PET tracers such as 18F-FDG, in combination with CT, can overcome the limitations of morphologic imaging alone, because functional changes possible to detect by 18F-FDG PET/CT often precede morphologic changes detectable by conventional MRI or CT (14,15).
18F-FDG PET/CT has long been used successfully for evaluation of several malignancies including endometrial cancer (Fig. 1) (15,16). On the basis of a systematic review, here we report diagnostic indices of 18F-FDG PET/CT for the preoperative prediction of LNM and for the detection of disease recurrence after surgery with curative intent in endometrial cancer patients.
An 87-y-old patient with FIGO stage III C1 and endometrioid grade 2. (A) Contrast-enhanced CT depicting nonenlarged lymph node (arrow) close to right internal iliac vessels. (B) PET scan depicting 18F-FDG uptake (red cross hair lines) in region of same pelvic lymph node. (C) Coregistered 18F-FDG PET/CT scan depicting increased 18F-FDG PET uptake (red cross hair lines) corresponding to histologically confirmed metastatic internal iliac lymph node.
MATERIALS AND METHODS
Search Strategy
Because the study was not conducted on patients, no informed consent or ethical committee approval was needed. To identify all relevant publications, we performed systematic searches in the bibliographic databases PUBMED.com from inception to August 17, 2015. Search terms included controlled terms from Mesh in PUBMED.com using the following: ‘FDG PET’ in combination with ‘Endometrial neoplasms’. The references of the identified articles were also searched for relevant publications.
Selection Process
One physician and 1 statistician reviewed each published article independently to determine the eligibility for inclusion in the meta-analysis and to extract information regarding clinical patient data and PET/CT characteristics. From the studies selected, data on first author, year of publication, number of patients included, study design (prospective or retrospective), patient age (mean/median), results from surgical International Federation of Gynecology and Obstetrics (FIGO) staging, percentage with nodal metastases, percentage with endometrioid subtype, 18F-FDG PET/CT technical characteristics, and numbers for diagnostic performance of 18F-FDG PET/CT (i.e., true-negatives, false-negatives, true-positives, false-positives, positive predictive value, and negative predictive value) were extracted and recorded. Any differences were resolved by consensus.
PET/CT studies that met the following criteria were included: first, studies that reported the diagnostic performance of 18F-FDG PET/CT in detecting LNM preoperatively or disease recurrence in endometrial cancer patients after primary surgery; second, clinical studies that included at least 10 patients; third, studies that applied 18F-FDG as a tracer on a dedicated device and were published after peer review. Studies on animals or in vitro studies, studies not available in full text or not written in English, and nonoriginal articles (e.g., reviews, editorials, letters, legal cases, interviews, case reports) were not evaluated systematically in this review.
Statistical Analysis
We performed standard methods recommended for meta-analysis of diagnostic test evaluations (17). Statistical analyses were performed using Meta-DiSc 1.4 software (developed by the unit of Clinical Biostatistics team of the Ramon y Cajal Hospital in Madrid, Spain) (18). We computed pooled measures for the following test indices of each study: sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR). Further, the summary receiver-operating-characteristics curve (SROC) was constructed, and the area under the SROC curve (AUC) was determined.
A random-effects model was used for statistical pooling of the data. Pooled data were presented with 95% confidence intervals (CIs). The CI for diagnostic indices are exact—that is, they are based on the binomial distribution and hence are asymmetric. The I-square index was used to test for heterogeneity between studies. The AUC was calculated to measure the overall diagnostic performance of 18F-FDG PET/CT in detecting the LNM and endometrial cancer recurrence (ECR). The sensitivity and specificity for the single test threshold identified for each study were used to plot the SROC curve along with Q* index representing an overall measure of the test’s discriminatory power.
RESULTS
Literature Search Results
The literature search yielded 58 references in PUBMED.com. In addition, 3 relevant recent articles on LNM in endometrial cancer and 1 in ECR that we were aware of were included in our database. The flow chart of the search and selection process is presented in Figure 2. Of a total of 62 articles, only 21 were eligible according to the criteria. Tables 1 and 2 summarize details for the included endometrial cancer studies of LNM and ECR by 18F-FDG PET/CT imaging, respectively.
Flow chart of selection process of eligible studies.
Summary of Clinical 18F-FDG PET/CT Studies on LNM in Endometrial Cancer
Summary of Clinical 18F-FDG PET/CT Studies on ECR in Endometrial Cancer
Preoperative Detection of LNM
In our meta-analysis of LNM, 13 studies were included, comprising a total of 861 endometrial cancer patients. The overall pooled diagnostic indices of preoperative 18F-FDG PET/CT for detecting LNM are calculated on a patient basis. The pooled sensitivity and specificity values were 0.72 (95% CI, 0.63–0.80) (Fig. 3A) and 0.94 (95% CI, 0.93–0.96) (Fig. 3B), respectively. The pooled PLR was 10.9 (95% CI, 7.9–15.1), the pooled NLR was 0.36 (95% CI, 0.27–0.48), and the DOR was 39.7 (95% CI, 21.4–73.6).
LNM: forest plot of sensitivity pooling (A); forest plot of specificity pooling (B); SROC curve (C). Individual study estimates of sensitivity and specificity of 18F-FDG PET/CT for identifying LNM in endometrial cancer. In detection of LNM, 18F-FDG PET/CT has moderate sensitivity of 0.72, high specificity of 0.94, and AUC of 0.94, demonstrating good diagnostic performance.
The SROC representing a global summary score for the test performance yielded an AUC of 0.94 and a Q* value of 0.88 (Fig. 3C), indicating a relatively high level of overall accuracy.
Detection of ECR
In the present meta-analysis for ECR, 8 studies comprising 378 patients have been included. The pooled sensitivity was 0.95 (95% CI, 0.91–0.98) (Fig. 4A), and the pooled specificity was 0.91 (95% CI,0.86–0.94) (Fig. 4B). The pooled PLR was 8.8 (95% CI, 6.0–12.7), NLR was 0.08 (95% CI, 0.05–0.15), and DOR was 171.7 (95% CI, 67.9–434.3). The SROC curve for the 18F-FDG PET/CT in the detection of ECR yielded an AUC and Q* values of 0.97 and 0.93 (Fig. 4C), respectively, suggesting that the level of overall accuracy was high.
ECR: forest plot of sensitivity pooling (A); forest plot of specificity pooling (B); SROC curve (C). Individual study estimates of sensitivity and specificity of 18F-FDG PET/CT for identifying disease recurrence in endometrial cancer. In detection of ECR, 18F-FDG PET/CT has high sensitivity of 0.95, high specificity of 0.91, and AUC of 0.97, demonstrating excellent diagnostic performance.
DISCUSSION
Lymphadenectomy is currently commonly applied for lymph node staging in endometrial carcinoma as part of the surgical FIGO staging systems. However, noninvasive accurate lymph node staging in endometrial cancer by preoperative imaging seems advantageous compared with the more invasive nature of surgical lymph node staging, also with an unproven survival benefit from the procedure (9,10). Similar to other tumors, endometrial cancer has an increased tumor glucose metabolism and glycolysis rate, which makes it suitable for 18F-FDG PET/CT imaging (19–21). The present meta-analysis yields high diagnostic performances of 18F-FDG PET/CT for diagnosing LNM preoperatively. High diagnostic accuracy was also demonstrated for the procedure detecting ECR after primary surgical treatment. A high diagnostic performance clearly supports a role for 18F-FDG PET/CT to enable more accurately tailored primary surgical endometrial cancer treatment and subsequent patient care.
The pooled sensitivity for preoperative detection of LNM by 18F-FDG PET/CT in this meta-analysis was 72%, highlighting that as much as about one fourth of the metastatic lymph nodes are still missed by 18F-FDG PET/CT. One possible explanation for this finding is that 18F-FDG avidity relies on the presence of a sufficient number of malignant cells exhibiting increased glucose metabolism. Furthermore, the spatial resolution of PET/CT is not good enough to reliably detect small tumors or micrometastatic disease. There is no documented threshold for lymph node size allowing PET/CT to correctly identify metastatic lymph nodes in endometrial cancer, although 1 study reported node-based sensitivities of 17% (4/24) for nodes of 4 mm or smaller, 67% (14/24) for nodes measuring 5–9 mm, and 93% (14/15) for nodes of 10 mm or larger (22). Similar figures with node-based sensitivities of 13%, 67%, and 100% in metastatic lymph nodes of 4 mm or smaller, 5–9 mm, and of 10 mm or larger, respectively, in endometrial cancer were reported in another study (23). It should, however, be kept in mind that although this meta-analysis found the overall sensitivity of 18F-FDG PET/CT to be moderate for the detection of LNM in endometrial cancer, it compares favorably with the reported sensitivities for LNM detection by conventional MRI and CT (24).
A high pooled specificity of 0.94 for metastatic lymph node detection by 18F-FDG PET/CT was found in this study, and it may be argued that this specificity is sufficiently high to safely omit a major surgical procedure in patients with low risk based on results from preoperative endometrial biopsy and preoperative imaging, reducing operative and postsurgical complications and costs (25). Furthermore, the present meta-analysis, showed that 18F-FDG PET/CT has a high PLR (10.9), pinpointing that 18F-FDG PET/CT findings suggesting metastatic lymph nodes are likely to be confirmed at surgical staging. The high diagnostic performance of 18F-FDG PET/CT for detecting endometrial cancer LNM is also justified by a high AUC of 0.94 in this meta-analysis. Interestingly, Kang et al. (26) reported almost identical figures for the diagnostic performance of 18F-FDG PET/CT for detecting LNM in cervical cancer, with a reported sensitivity of 0.73 (95% CI, 0.53–0.87) and specificity of 0.93 (95% CI, 0.86–0.97). Thus, 18F-FDG PET/CT seems to be equally feasible in endometrial and cervical cancer for lymph node staging, and 18F-FDG PET/CT may be particularly justified in endometrial and cervical cancer patients with high risk for disease spread, to identify metastatic lymph nodes preoperatively.
Several recent studies in endometrial cancer have demonstrated that preoperative primary tumor metabolic parameters have been associated with the presence of LNM. In a prospective study, Antonsen et al. (27) found a significantly higher SUVmax in patients with LNM than in those with no LNM (P = 0.04). Additionally, they found that SUVmax was significantly higher in patients with high FIGO stage, myometrial invasion, and cervical invasion. Furthermore, Crivellaro et al. (28) found strong association between the presence of LNM and metabolic tumor volume in endometrial cancer. Recently, we demonstrated that the preoperative metabolic tumor volume cutoff value of 30 mL yielded sensitivity and specificity of 85% and 76% for LNM, respectively, suggesting that the metabolic tumor volume is a promising marker for LNM (25). In this regard, preoperative 18F-FDG PET/CT imaging of primary endometrial carcinomas may provide an adequate tool for prognostication and LNM detection that facilitate personalized patient care. However, additional prospective studies are required to define optimal cutoff values for predicting LNM based on 18F-FDG PET/CT metabolic parameters. Earlier studies describe measures for SUVs from a single region of interest, which does not represent the overall tumor profile. Therefore, advanced techniques such as the whole tumor voxel-by-voxel analysis may be a preferable approach to reduce operator dependence and capture more relevant and comprehensive measures for tumor microenvironment and heterogeneity.
The pooled sensitivity and specificity of 18F-FDG PET/CT for the detection ECR were 0.95 and 0.91, respectively, with an AUC in ROC analysis of 0.97 (95% CI, 0.95–0.99), all supporting a high level of overall diagnostic accuracy. Again, similar 18F-FDG PET/CT diagnostic performance indices were reported for detecting recurrent uterine cervical carcinomas, with a reported pooled sensitivity, specificity, and AUC of 0.92 (95% CI, 0.91–0.94), 0.84 (95% CI, 0.74–0.91), and 0.95, respectively (29). Thus, 18F-FDG PET/CT seems to perform equally well in the diagnosis of endometrial and cervical cancer recurrences, supporting a promising role for 18F-FDG PET/CT as a diagnostic tool for patients with suspected recurrence.
The findings in this study regarding 18F-FDG PET/CT and ECR must, however, be interpreted with care, considering that the studies report lack of histologic confirmation of all putative metastases based on 18F-FDG PET/CT, and they report variable follow-up of the cases considered nonmetastatic based on 18F-FDG PET/CT. Thus, some of the cases classified as correctly staged for ECR by 18F-FDG PET/CT may have been erroneously classified. This limitation shared by most published studies, including the studies on cervical cancer recurrence, is, however, hard to circumvent, because it seems unethical to perform biopsies of all suspected metastatic lesions in patients due to risk of complications. Furthermore, frequent 18F-FDG PET/CT follow-up scans are very expensive and imply unwanted radiation exposure for the patients.
Because both LNM and ECR studies exhibit interstudy heterogeneity, the SROC curve should be asymmetric (Supplemental Fig. 1 [supplemental materials are available at http://jnm.snmjournals.org] for both symmetric and asymmetric SROCs). Because all possible curves with the same true odds ratio and different degrees of heterogeneity would pass through the same point on the antidiagonal, the heterogeneity does not affect the Q* estimate but rather the shape of the curve and its standard errors. Walter et al. (30) noted that the AUC standard errors calculated under the homogeneity assumption provide a good approximation for heterogeneous studies. The approximation may be poor for extremely high DOR values (higher than 20), as is the case in both meta-analyses presented here (37.5 and 171.7 for LNM and recurrence, respectively). However, the bias in the homogeneity-based standard errors is mostly positive, and hence conservative, that is, can be overestimated, but rarely underestimated. Supplemental Figure 1 illustrates that the CIs of the asymmetric SROC are much narrower than those of the symmetric SROC, whereas the difference between the AUC estimates is negligible.
This meta-analysis has several limitations. First, positive result publication bias is a major concern, because nonsignificant or unfavorable study results tend to be discarded. However, we evaluated publication bias in our meta-analysis using funnel plot asymmetry, finding the funnel plots to be symmetric for both sensitivity and specificity pooling, implying no large bias in our study. Second, the current meta-analysis did not include region-by-region or node-by-node evaluation because this was not reported in most studies; however, this could have provided additional information. Third, not all included studies had a prospective study design. Fourth, the gold standard for confirmation of LNM or ECR, being histopathologic examination from biopsies, was not obtained from all the lesions reported in the studies. However, clinical follow-up data and results from renewed diagnostic imaging were recorded, and clinically putative lymph nodes metastases or ECR was used as a gold standard when histologic confirmation was missing.
CONCLUSION
Overall, 18F-FDG PET/CT demonstrated a high diagnostic performance in identifying LNM preoperatively and in detecting recurrence after endometrial carcinoma surgery with curative intent. Larger prospective studies are needed to validate this high diagnostic performance of 18F-FDG PET/CT in endometrial cancer and further assess patient subgroups with particular clinical benefit from applying this advanced imaging procedure.
DISCLOSURE
The costs of publication of this article were defrayed in part by the payment of page charges. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734. This study was supported by funding from the Norwegian Cancer Society. No other potential conflict of interest relevant to this article was reported.
Footnotes
Published online Jan. 28, 2016.
- © 2016 by the Society of Nuclear Medicine and Molecular Imaging, Inc.
REFERENCES
- Received for publication December 6, 2015.
- Accepted for publication December 30, 2015.
