Presented at the American Society for Therapeutic Radiology and Oncology Translational Meeting, Arlington, Virginia, October 17-18, 2008.
This study aimed to evaluate the variation in cervical cancer glucose metabolism for different tumor histologies and levels of differentiation, as measured by the uptake of 18F-fluorodeoxyglucose (FDG) by positron emission tomography (PET).
The study population consisted of 240 patients with International Federation of Gynecology and Obstetrics stages Ib1 through IVb cervical cancer, who underwent a pretreatment FDG-PET. Tumor histology included 221 squamous cell (SC), 4 adenosquamous (AS), and 15 adenocarcinoma (AC) tumors. There were 14 well, 145 moderately, and 81 poorly differentiated tumors. The stage distribution was as follows: 70 stage I tumors (9 AC, 2 AS, and 59 SC), 102 stage II tumors (3 AC, 1 AS, and 98 SC), 64 stage III tumors (3 AC, 1 AS, and 60 SC), and 4 stage IV tumors (4 SC). From the FDG-PET, maximal standardized uptake value (SUVmax) was determined. The variation in SUVmax was analyzed for differences based on tumor histology and differentiation.
For all patients, the mean SUVmax was 11.62 (range, 2.50-50.39). The mean SUVmax by histology was as follows: SC, 11.91 (range, 2.50-50.39); AS, 8.85 (range, 6.53-11.26); and AC, 8.05 (range, 2.83-13.92). Squamous versus nonsquamous tumors demonstrated a significant difference in SUVmax (P = .0153). SUVmax and tumor volume were not found to be correlated (R2 = 0.013). The mean SUVmax was 8.58 for well-differentiated, 11.56 for moderately differentiated, and 12.23 for poorly differentiated tumors. The mean SUVmax was significantly different for well‒differentiated versus poorly differentiated cervical tumors (P = .0474).
Cervical cancer is among the top 3 cancer diagnoses in women worldwide.1 In the United States, cervical cancer is the third leading cause of cancer death in women ages 15 to 34 years and the fifth leading cause of cancer death in women ages 35 to 54 years and, therefore, is a leading cause of years of life lost.
Positron emission tomography (PET) with 18F-2-fluoro-2-deoxyglucose (FDG) is an important tool for diagnosing and staging cervical and many other cancers. In particular, FDG-PET aids in identifying involved lymph nodes, distant metastasis, and recurrent disease in patients with cervical cancer.3-6 In addition, our group has also demonstrated that the FDG uptake of the primary cervical tumor, as measured by the maximal standardized uptake value (SUVmax), provides valuable prognostic information for predicting lymph node involvement, treatment response, and overall survival.7, 8
Traditional prognostic factors for cervical cancer include patient age, performance status, clinical tumor stage, lymph node involvement, overall treatment time, radiation dose, and use of chemotherapy. Over the years, there has been some debate regarding the prognostic significance of tumor histology for cervical cancer, with some studies suggesting adenocarcinomas have a worse outcome,9-11 whereas other studies have shown no difference based on histology.12-14
FDG uptake has been shown to correlate with tumor proliferation rates, reflecting tumor aggressiveness in many types of cancer, including lymphoma, nonsmall cell lung cancer (NSCLC), esophageal cancer, and head and neck cancer.9, 15-19 For NSCLC in particular, FDG uptake has been shown to be significantly different for adenocarcinoma versus squamous cell carcinoma and for well-differentiated versus poorly differentiated tumors.20-23
The goal of the current study was to evaluate whether cervical tumor histology and/or the level of differentiation are related to primary tumor FDG uptake, as measured by SUVmax.
MATERIALS AND METHODS
Two hundred forty patients with newly diagnosed cervical cancer who underwent pretreatment FDG-PET from April 1998 to September 2007 at Washington University in St. Louis and had pathology evaluation including histology and tumor grade were included in this analysis. This study was approved by the Washington University Human Research Protection Office. All patients were staged clinically according to International Federation of Gynecology and Obstetrics staging, with the distribution as follows: 32 with stage Ib1 disease, 38 with stage Ib2 disease, 7 with stage IIa disease, 95 with stage IIb disease, 2 with stage IIIa disease, 62 with stage IIIb disease, 3 with stage IVa disease, and 1 with stage IVb disease. The pathologic diagnosis of all patients was reviewed by Washington University pathologists, with histologic designation as follows: 221 squamous cell (SC) carcinoma, 15 adenocarcinoma (AC), and 4 adenosquamous (AS). There were 14 well‒differentiated, 145 moderately differentiated, and 81 poorly differentiated tumors. See Table 1 for stage distribution by histology and differentiation. Patient ranged in age from 23 to 88 years (mean, 52 years).
Table 1. Patient and Tumor Characteristics
FIGO indicates International Federation of Gynecology and Obstetrics; SCV, supraclavicular.
Lymph node involvement
Pelvic and Para-aortic
Pelvic, para-aortic, and SCV
Before receiving any treatment, all patients underwent FDG-PET or PET/computed tomography (CT) for lymph node staging. Before November 2002, FDG-PET was performed with a conventional PET scanner (Siemens ECAT Exact) and interpreted in a routine manner as previously described.4 Thereafter, all FDG-PET studies were performed with a hybrid PET/CT scanner (Siemens Biograph LSO).24 The PET/CT images were interpreted in a standard clinical manner, both separately and in fused mode. Seventy-four patients underwent FDG-PET, and 166 patients underwent hybrid PET/CT. The PET and PET/CT were calibrated to ensure accuracy and reproducibility of SUV between the 2 scanners. The PET study was deferred if blood glucose concentration exceeded 200 mg/dL. The mean blood glucose was 102 mg/dL (range, 68-187 mg/dL; standard deviation [SD], 22.05 mg/dL). Urinary tract activity was minimized by the placement of a Foley catheter before the injection of FDG and by the administration of furosemide and intravenous fluids after the injection of FDG. The mean time between FDG injection and the start of imaging was 67 minutes (range, 42 minutes-120 minutes; SD, 15.81 minutes).
For each PET study, the SUVmax of the primary cervical cancer was measured. SUV is a semiquantitative measure of radiotracer uptake calculated by the following formula: SUV = tissue radioactivity concentration (nCi/mL)/ (injected dose (nCi)/patient weight (g)).
In this study for images performed on the PET/CT scanner, the volumetric tool from Siemens e.soft software was used to create a region of interest encompassing the tumor and to determine the maximum SUV. For images performed on the dedicated PET scanner, a series of regions of interest were placed around the entire primary tumor on every slice that contained the tumor to determine the maximum SUV. Sites of lymph node involvement, as observed on FDG-PET, were recorded (Table 1). In addition, PET tumor volume of the cervical cancer was measured on pretreatment PET with the 40% threshold volume method, as described previously.25
The mean SUVmax was calculated for the 3 histologies (SC, AC, and AS) and for the 3 levels of differentiation (well, moderate, and poor). Logistic regression was used for comparing SUVmax and tumor size. A Student t test for unpaired data was used for comparing the mean SUVmax for squamous versus nonsquamous histology, and for comparing the mean SUVmax for the 3 levels of tumor differentiation with each other. The risk of lymph node involvement by histology was analyzed using a chi-square test.
Tumor characteristics by histology are shown in Table 1. The majority of cervical tumors were of SC histology. The mean cervical tumor size was 63.2 cc (range, 2.2 cc-535.7 cc; SD, 70.25 cc). Although the SC histology group had a smaller proportion of early stage disease (Table 1), there was no correlation noted between SUVmax and tumor volume (R2 = 0.013) (Fig. 1). In comparing SUVmax and tumor volume for squamous (R2 = 0.008) and nonsquamous (R2 = 0.163) tumors separately, no association between SUVmax and tumor volume was found.
Patients with SC cervical tumors were also more likely to present with lymph node metastasis at the time of diagnosis (58% for SC, 47% for AC, and 50% for AS) (Table 1). Higher SUVmax was associated with an increased risk of lymph node metastasis (P = .0015). In a logistic likelihood ratio test for the risk of lymph node involvement including SUVmax, stage, histology, and differentiation, only SUVmax was found to be a significant predictor (P = .0042). Patients with an SUVmax of <5 had a 24% chance of lymph node metastasis at diagnosis, versus 59% for those whose SUVmax was between 5 and 14, and 72% for those with an SUVmax >14.
For all patients, the mean SUVmax was 11.62 (range, 2.50-50.39) (Figure 2). The mean SUVmax by histology was as follows: SC, 11.91 (range, 2.50-50.39); AS, 8.85 (range, 6.53-11.26); and AC, 8.05 (range, 2.83-13.92) (Fig. 3A-C). There was a significant difference in the SUVmax for squamous tumors versus nonsquamous (P = .0153).
The mean SUVmax was 8.58 for well-differentiated, 11.56 for moderately differentiated, and 12.23 for poorly differentiated tumors. There was a significant difference in the SUVmax of well versus poorly differentiated cervical tumors (P = .0474).
For SC tumors alone, the mean SUVmax based on the level of differentiation was 11.64 for well-differentiated, 11.65 for moderately differentiated, and 12.40 for poorly differentiated SC. Using a Student t test for unpaired data, there was no significant difference in the mean SUVmax for the different levels of differentiation of SC tumors.
We found that FDG uptake in the primary cervical tumor varied by histology and differentiation. SC tumors demonstrated a significantly higher SUVmax than nonsquamous tumors. Although the nonsquamous cell carcinomas tended to be of lower stage, this does not provide an explanation for the higher mean SUVmax observed for the SC carcinoma, because tumor volume and SUVmax were not correlated. This work is significant because to our knowledge, no study to date has previously demonstrated how cervical cancer histology and level of differentiation correlate with SUVmax. For NSCLC, researchers have shown that SC carcinomas have significantly higher FDG uptake compared with nonsquamous tumors.22
The current study also found that poorly differentiated cervical tumors demonstrated a higher SUVmax than well or moderately differentiated tumors, with a significant difference in the SUVmax for well versus poorly differentiated tumors. The level of tumor differentiation has also been shown to correlate with NSCLC SUV, with well-differentiated tumors similarly demonstrating lower levels of FDG uptake.20, 21, 23 Vesselle et al suggested that the correlation of FDG uptake in NSCLC and tumor proliferation could allow FDG-PET to serve as a noninvasive means of grading tumors.21 The correlation between SUVmax and tumor grade for cervical cancer could be valuable and serve as another prognostic tool, especially because tumor grade is not always provided in pathology assessment.
Just as high FDG uptake has been shown to correlate with tumor proliferation, it has also been shown to be associated with other signs of aggressive tumor behavior, including lymph node involvement and metastasis risk.7, 26 For gliomas, high FDG uptake also was found to be predictive of more aggressive tumor behavior and an increased risk of a high‒grade glioma.27 It has also been found that FDG uptake is lower in indolent compared with aggressive non-Hodgkin lymphoma.28 In this current study and in previous research, we found that higher SUVmax was correlated with an increased risk of lymph node metastasis at diagnosis. For lung cancer, others have also found an association between primary tumor SUV and presence of lymph node involvement.29 Because cervical cancer lymph node status is significantly correlated with prognosis,4 and higher SUVmax predicts lymph node involvement and more aggressive tumor behavior, the increased FDG uptake observed for squamous and poorly differentiated cervical tumors has potential to be quite valuable, especially in the pretherapy setting.
Pretreatment prognostic information concerning the aggressiveness of cervical tumors can contribute to optimizing and individualizing patient therapy. In particular, from this current study, we know that cervical tumors with a higher SUVmax are more likely to be poorly differentiated and have an increased risk of lymph node involvement. This type of prognostic information could be valuable in a pretreatment setting.
There has been continuing debate regarding the prognostic significance of tumor histology for cervical cancer, with some studies suggesting ACs have worse outcome,9-11 whereas other studies have shown no difference based on histology.12-14 Peters et al, in Gynecologic Oncology Group (GOG) 109/Southwest Oncology Group [SWOG] 8797, found that patients with AC or AS tumors treated with radiation alone had a worse progression-free survival, but that with the addition of chemotherapy, this difference in prognosis for patients with squamous versus nonsquamous cervical tumors disappeared.30 Perhaps some of the underlying differences between squamous and nonsquamous tumors contributes to the differences observed for mean SUVmax among the cervical tumor histologies. Although the results of this study add to the information we have regarding some of the differences between squamous and nonsquamous cervical tumors, the differences between squamous and AC are most likely multifactorial and much still remains to be understood.
The cause of higher FDG uptake in squamous and poorly differentiated tumors remains uncertain, although there has been some research into glucose transporter gene expression. In a study of breast cancer, Bos et al found a correlation between FDG uptake and Glut-1 expression, mitotic activity index, amount of necrosis, tumor cell volume, and microvessel density, among other factors.31 In patients with NSCLC, FDG accumulation was found to be correlated with Glut-1 and Glut-3 expression and was significantly higher in SC carcinomas.22 Glut-1 expression in human breast cancer correlated with markers of proliferation, which could correlate with the higher FDG uptake observed with poorly differentiated tumors.32 For cervical cancer, Mendez et al33 found a correlation between Glut-1 expression and tumor grade and Yen et al34 found a correlation between Glut-1 expression and SUV in SC cervical cancer. In contrast, other researchers found no correlation between Glut-1 and Glut-3 expression and tumor differentiation or FDG SUV for oral SC carcinoma.35 Given the mixed results, additional research is still needed to better understand the molecular or biologic mechanism leading to increased SUV.
This current study offers some novel and unique findings regarding the correlation between cervical tumor histology and differentiation and SUV, but there are some limitations. This was a retrospective study with unequal patient numbers and stage distribution between groups, although the proportions reflect those observed in other studies and the general population. The large number of patients analyzed adds to the strength of our data, and further confirmation by a prospective study could reinforce our findings.
Overall, the results of the current study offer valuable new information concerning cervical cancer and FDG-PET, in particular that SC and poorly differentiated cervical tumors have higher FDG uptake. In future studies, it would be interesting to learn how these differences in SUVmax observed for cervical tumor histology and differentiation translate into patient outcome and to better understand the biologic basis for these differences.
Conflict of Interest Disclosures
Dr. Spencer is the recipient of the RSNA R&E Foundation Medical Student Departmental Program Grant #MSD0404.
Dr. Siegel reported stock ownership in Radiology Corporation of America (RCOA); being a medical advisory board member of RCOA; and receiving lecture honoraria from PETNET Pharmaceuticals, Inc; GE Healthcare, Inc; and Philips Medical Systems Puerto Rico, Inc.