Fax: (314) 747-9557
The standardized uptake value for F-18 fluorodeoxyglucose is a sensitive predictive biomarker for cervical cancer treatment response and survival†
Article first published online: 4 SEP 2007
Copyright © 2007 American Cancer Society
Volume 110, Issue 8, pages 1738–1744, 15 October 2007
How to Cite
Kidd, E. A., Siegel, B. A., Dehdashti, F. and Grigsby, P. W. (2007), The standardized uptake value for F-18 fluorodeoxyglucose is a sensitive predictive biomarker for cervical cancer treatment response and survival. Cancer, 110: 1738–1744. doi: 10.1002/cncr.22974
See related editorial on pages 000-000, this issue.
- Issue published online: 19 SEP 2007
- Article first published online: 4 SEP 2007
- Manuscript Accepted: 17 APR 2007
- Manuscript Revised: 9 APR 2007
- Manuscript Received: 20 FEB 2007
- F-18 fluorodeoxyglucose-positron emission tomography;
- standardized uptake value
The objective of this study was to evaluate cervical tumor uptake of F-18 fluorodeoxyglucose (FDG) measured as the maximal standardized uptake value (SUVmax) by positron emission tomography (PET) and its association with treatment response and prognosis in patients with cervical cancer.
The study population consisted of 287 patients with stage IA2 through IVB cervical cancer who underwent pretreatment FDG-PET studies. SUVmax, tumor volume, and sites of lymph node metastasis were recorded. Therapy included surgery, chemoradiation, or palliation.
The mean SUVmax was 11.4 (range, 1–50.4). The mean tumor volume by stage was 42.1 cm3 for stage I tumors (using International Federation of Gynecology and Obstetrics [FIGO] staging criteria), 63.7 cm3 for stage II tumors, 129.2 cm3 for stage III tumors, and 166.2 cm3 for stage IV tumors. There was no correlation between tumor volume and SUVmax (correlation coefficient [R2] = 0.01). No significant difference in SUVmax was observed between squamous histology (n = 247 patients) and nonsquamous histology (n = 40 patients; P = .089). Higher SUVmax was associated with an increased risk of lymph node metastasis at diagnosis (P = .0009). A Cox proportional-hazards model for death from cervical cancer was used to evaluate tumor histology, lymph node metastasis, tumor volume, and SUVmax. The results indicated that SUVmax was the only significant independent factor (P = .0027). Three prognostic groups were established using SUVmax. The overall survival rates at 5 years were 95% for an SUVmax ≤ 5.2, 70% for an SUVmax > 5.2 and ≤13.3, and 44% for an SUVmax > 13.3 (P < .0001). Increasing SUVmax was associated with persistent abnormal FDG uptake in the cervix on 3-month FDG-PET studies in 238 patients who received curative chemoradiation (P = .04).
The SUVmax of the cervical tumor at diagnosis was a sensitive biomarker of treatment response and prognosis for patients with cervical cancer. Cancer 2007. © 2007 American Cancer Society.
Positron emission tomography (PET) with the glucose analogue F-18 fluorodeoxyglucose (FDG) is an important tool for the diagnosis and staging of many cancers. Our group and others have demonstrated the value of PET in identifying lymph node involvement, distant disease, and recurrent disease in patients with cervical cancer.1–7 The results from recent studies in cancers of the head and neck, lung, and esophagus indicate that high FDG uptake in the primary tumor, typically characterized as a standardized uptake value (SUV) greater than the median, predicts a worse outcome.8–15 Our group has also demonstrated a similar association in cervical cancer, correlating a higher SUV with a shorter disease-free survival.16 To our knowledge, no other study has examined more specifically the association between cervical tumor FDG uptake and local tumor control, recurrence rate, risk of distant metastasis, and overall survival. In the current study, we investigated the association between the primary cervical tumor FDG uptake, which was measured as the maximum SUV (SUVmax) at initial diagnosis, and local tumor control, recurrence rate, risk of distant metastasis, and overall survival through an analysis of data from 287 patients with cervical cancer who underwent FDG-PET studies.
MATERIALS AND METHODS
Two hundred eighty-seven patients with cervical cancer who underwent pretreatment FDG-PET studies between January 1998 and September 2006 at Washington University in St. Louis were included in this analysis. This study was approved by the Washington University Human Studies Committee. All patients were staged clinically according to International Federation of Gynecology and Obstetrics (FIGO) staging criteria, with the distribution as follows: Two patients had stage IA2 disease, 45 patients had stage IB1 disease, 44 patients had stage IB2 disease, 7 patients had stage IIA disease, 100 patients had stage IIB disease, 3 patients had stage IIIA disease, 64 patients had stage IIIB disease, 6 patients had stage IVA disease, and 16 patients had stage IVB disease. The pathologic diagnosis of each patient was reviewed by Washington University pathologists, and the distribution according to histology was as follows: 247 squamous cell carcinomas, 23 adenocarcinomas, 8 adenosquamous carcinoma, 4 clear cell carcinomas, 2 small cell carcinomas, 2 undifferentiated carcinomas, and 1 poorly differentiated carcinoma. Patients ranged in age from 23 years to 93 years (mean age, 53 years).
Patients were treated in the following manner: 238 patients received a combination of external irradiation and intracavitary brachytherapy with curative intent, 21 patients underwent surgery only, 6 patients underwent surgery and received postoperative irradiation, and 22 patients received palliative treatment only. Radiation treatment for cervical cancer was based on standard treatment practices for cervical cancer at Washington University in St. Louis.17 The majority of patients who received irradiation also received concurrent cisplatin chemotherapy (40 mg/m2 weekly for 6 cycles).
Before treatment, all patients underwent FDG-PET or PET/computed tomography (CT) studies for lymph node staging. A repeat PET or PET/CT scan was performed 3 months after patients completed radiation treatment to evaluate response and to determine the presence of any residual or progressive disease. Before November 2002, FDG-PET studies were performed with a conventional PET scanner and were interpreted in a routine manner as described previously.2 Thereafter, essentially all FDG-PET studies were performed with a hybrid PET/CT scanner as previously described by Wright et al.18 The PET/CT images were interpreted in a standard clinical fashion, both separately and in fused mode. The PET study was deferred if blood glucose concentrations exceeded 200 mg/dL. The mean blood glucose concentration was 102 mg/dL (range, 68–187 mg/dL).
For each PET study, the SUVmax of the primary cervical cancer was measured. SUV is a semiquantitative measure of radiotracer uptake and is calculated according to the following formula: SUV = tissue radioactivity concentration [nCi/mL]/[injected dose (mCi)/patient weight (g)].
In this study, the maximum-pixel SUV within a region of interest encompassing the tumor was used. In addition, the PET tumor volume of the cervical cancer was measured from the pretreatment PET image as described previously.4
Patients had follow-up examinations approximately every 2 months for the first 6 months, every 3 months for the next 2 years, and every 6 months thereafter. FDG-PET studies were was repeated 3 months after patients completed treatment and then yearly or when warranted by clinical examination or symptoms. The mean follow-up for all patients who were alive at the time of last follow-up was 32 months (range, 1–89 months).
Clinical-pathologic factors and outcome data were analyzed for correlation with SUVmax. SUVmax and PET tumor volume were compared using regression analysis. We analyzed local control rates only for patients who received irradiation with curative intent, whereas we used data from all patients when analyzing overall survival and progression-free survival. The logistic likelihood-ratio test was used to compare the SUVmax of different subgroups. The Cox proportional-hazards model was used for multivariate analyses.19 Disease-free and overall survival were calculated by using the Kaplan-Meier method.20
Patient characteristics, including age at diagnosis, FIGO clinical stage, histology, and lymph node status at diagnosis, are listed in Table 1. The average SUVmax was 11.4 (range, 1–50) for all patients and 11.4 (range, 2–50) for patients who were treated with curative intent using radiation. Figure 1 shows the distribution of primary cervical tumor SUVmax for all patients. The average cervical tumor SUVmax was9.6 (range, 2–23) for patients with stage I disease, 12.1(range, 2–38) for patients with stage II disease, 12.4 (range, 3–50) for patients with stage III disease, and 9 (range, 5–13) for patients with stage IV disease. The mean tumor volume ± standard error by stage was 45 ± 5.9 cm3 for stage I tumors, 60.2 ± 9.1 cm3 for stage II tumors, 115.8 ± 14.9 cm3 for stage III tumors, and 166.9 ± 41.6 cm3 for stage IV tumors. We plotted the cervical tumor volume on PET studies versus the cervical tumor SUVmax and, as shown in Figure 2, there was no association. The average primary tumor SUVmax was 11.6 for squamous histology and 9.7 for nonsquamous histology (P = .0894).
|Characteristic||All patients||Curative intent|
|Mean age at diagnosis (range), y||52.9 (23–93)||52.9 (23–88)|
|Lymph node status at diagnosis|
|Pelvic lymph nodes||107||99|
|Para-aortic lymph nodes||35||33|
|Supraclavicular lymph nodes||15||4|
In addition, tumor-specific characteristics were analyzed for their association with primary tumor SUVmax. One hundred fifty-seven patients had disease that involved lymph nodes at the time of diagnosis. The correlation between lymph node status and cervical SUVmax was analyzed by using the Student t test for unpaired data, and the results indicated that the primary tumor SUVmax was significantly greater in patients with FDG-PET evidence of lymph node metastasis than in patients without metastases (P = .0009).
We analyzed the associations between overall survival and primary tumor SUVmax, tumor volume (as a surrogate for stage), tumor histology, and lymph node status using the Cox proportional-hazards model and observed that SUVmax was the only significant independent predictor of death from cervical cancer (P = .0027) (Table 2). To analyze the association between primary tumor SUVmax and overall survival further, we divided patients into 3 groups based on their primary tumor SUVmax. The division points for the 3 groups were based on the Cox model. The 3 risk groups were 1) patients with an SUVmax ≤ 5.2, 2) patients with an SUVmax from >5.2 to ≤13.3, and 3) patients with an SUVmax > 13.3. The overall survival rate at 5 years was 95% for patients who had an SUVmax ≤ 5.2, 70% for patients who had an SUVmax from >5.2 to ≤13.3, and 44% for patients who had an SUVmax >13.3 (P < .0001) (Fig. 3). Kaplan-Meier progression-free survival curves for the 3 SUVmax groups are shown in Figure 4 (P < .0001).
|Pelvic lymph nodes on PET||−0.657||0.453||−1.451||2.105||.1469||0.519|
|PET tumor volume||0.003||0.002||1.600||2.559||.1097||1.003|
|Primary tumor SUVmax||0.075||0.024||3.095||9.576||.0020||1.078|
Patients Who Received Curative Chemoradiation
Most patients who received treatment with curative intent (chemoradiation) underwent a repeat FDG-PET study 3 months after they completed treatment (175 of 238 patients). Thirty-three of those 175 patients (19%) had persistent disease evident on their posttreatment PET study. A higher primary cervical tumor SUVmax at diagnosis was predictive of persistent FDG uptake in the cervix (persistent tumor) on the 3-month posttherapy FDG-PET study (P = .0472).
By using the Cox proportional-hazards model, we analyzed the association between cervical tumor SUVmax and first recurrence in the pelvis or distantly. We observed that a higher primary tumor SUVmax significantly increased the risk of a pelvic recurrence (P = .0232) and had some association with the development of distant metastasis (P = .0965). The Cox proportional-hazards model also demonstrated that the SUVmax was significant for cause-specific survival (P = .0126) and for overall survival (P = .0119).
In this study, we observed that the primary tumor SUVmax was not related to patient-specific factors, such as histology, tumor stage, patient age, or tumor volume, but the primary tumor SUVmax was a predictive biomarker of lymph node status, persistent disease after treatment, pelvic recurrence, and overall survival. In particular, the primary tumor SUVmax at the time of diagnosis was a more significant predictor of outcome in cervical cancer than FIGO stage, tumor volume, histology, or lymph node involvement. In dividing patients into 3 groups based on SUVmax, 3 distinct, nonoverlapping survival curves emerged. This means that the primary cervical tumor SUVmax assessed by FDG-PET studies at the time of diagnosis is a significant biomarker for disease prognosis, treatment response, and overall outcome in patients with cervical cancer.
Historically, patient- and tumor-related factors, such as age, histology, tumor volume, and stage, have been critical attributes for predicting patient outcome and overall survival for cervical cancer.21–24 In our study, these attributes had a less significant association with prognosis and outcome than the primary tumor SUVmax. For lung cancer, some investigators have reported an association between primary tumor SUV and tumor stage, volume, and histology.11 In our analysis, tumor stage, volume, and histology were not associated with the primary tumor SUVmax, suggesting that the primary tumor SUVmax is an independent prognostic factor.
Previous studies in lung and head and neck cancers have suggested that a higher SUV is correlated with a worse prognosis.9, 11–13, 25 Other studies also demonstrated that patients who had a primary tumor SUV greater than the median tended to have poorer local control and disease-free survival.8, 10, 14, 15 In earlier work, our group also demonstrated that FDG uptake in the primary cervical cancer is associated with disease-free survival for patients who receive radiation.16 Our current study, which included more patients and longer follow-up, has taken this analysis to the next level by investigating the association between the SUVmax and a greater number of outcome measures, comparing the predictive value of SUVmax with previous standards, and examining outcomes for specific SUV groups other than those dichotomized around the median.
In this study, we demonstrated that the primary tumor SUVmax at diagnosis is predictive of lymph node involvement. For lung cancer, others also have reported an association between the primary tumor SUV and the presence of lymph node involvement.26 The finding that the primary tumor SUVmax can predict lymph node involvement in cervical cancer is significant, because our group previously demonstrated that lymph node status is related significantly to disease-free and overall survival,2 suggesting that the SUVmax may correlate with disease aggression.
We also observed that high primary tumor SUVmax at diagnosis was predictive of subsequent local recurrence and was correlated with an increased risk of persistent cervical disease, in particular, as evidenced by persistent disease on the 3-month posttreatment PET scan. Previous groups have reported that a lack of cervical tumor regression is associated with an inferior outcome.3, 27–29 For lung cancer, it has also been reported that the maximal tumor SUV was related to treatment response.9 Achieving local control is critical to prognosis and overall survival.
In cervical cancer, as in all cancers, overall survival is a key endpoint measure. Traditionally, tumor stage has been a predictor of overall survival. Recent studies have demonstrated that in cervical cancer, lymph node involvement is more predictive of outcome than stage.2 In the current study, we demonstrated that the SUVmax is more predictive of overall survival than tumor volume (as a surrogate for stage) or lymph node status. In particular, we observed that grouping patients by SUVmax created 3 distinct outcome strata for overall survival. To our knowledge, no one has demonstrated previously that, for cervical or other cancers, the SUVmax is more predictive of outcome than lymph node status, FIGO stage, or tumor volume. Because the primary tumor SUVmax is better at predicting outcomes than previous standards, this helps define SUVmax as a sensitive measure for prognosis. In addition, the significance of being able to stratify patients according to their SUVmax at diagnosis into 3 distinct outcome groupings suggests that its specific value is significant and that SUVmax also is a quantitative biomarker for predicting which patients will have a worse outcome before treatment is initiated. This defines the primary tumor SUVmax at diagnosis as a sensitive and quantitative biomarker for cervical cancer outcome.
Because our data indicate that the SUV of the primary tumor is an important predictor of prognosis, treatment response, and overall survival, this leads to the question of how glucose metabolism varies among cervical tumors and how that correlates with SUV and patient outcome. In an attempt to gain a better understanding of the biologic mechanism that causes increased FDG uptake in tumors, others have examined glucose transporter (Glut) gene expression and have reported mixed results. For example, with breast cancer, 1 group observed a correlation between Glut-1 expression and FDG uptake, whereas another group did not observe any association.30, 31 With regard to cervical cancer, Mendez et al.32 observed a correlation between Glut-1 expression and tumor grade, whereas Airley et al.33 did not observe an association between Glut-1 expression and disease-free or recurrence-free survival. Yen et al.34 observed a correlation between Glut-1 expression and SUV in cervical cancer, whereas Tian et al. did not report a correlation between the FDG SUV and Glut-1 or Glut-3 expression for oral squamous cell carcinoma.35 More types of glucose transporters are being discovered as more groups investigate this topic, and different methods of measuring expression also are being used, suggesting this is a complex issue that may involve multiple factors and that additional research is needed to investigate the cellular or biologic mechanism leading to increased SUV.
From this study, we have established 1) that the primary tumor SUVmax at diagnosis is correlated with the presence of lymph node involvement at diagnosis, 2) how well the primary tumor responds to treatment, and 3) the likelihood of disease recurrence and overall survival. These results suggest that the next step is to undertake prospective trials that will target those patients who present with elevated primary tumor SUVmax with more aggressive treatment. FDG uptake is not related to clinical factors, such as age, histology, stage, or tumor volume; therefore, another direction for further investigation may be the biomarker pathways that cause some tumors to have intense glucose uptake. Overall, the current study demonstrates that the primary tumor SUVmax at diagnosis is a valuable tool for defining the prognosis for patients with cervical cancer and has implications for translational research.
- 1FDG-PET evaluation of carcinoma of the cervix. Clin Pos Imaging. 1999; 2: 105–109., , .
- 17Uterine cervix. In: PerezCA,BradyLW,HalperinEC, et al., editors. Principles and Practice of Radiation Oncology.4th ed. Philadelphia: Lippincott, Williams & Williams; 2003: 1800–1915., .
- 19Regression models and life tables. J R Stat Soc. 1972; 34: 187–220..