Role of fluorodeoxyglucose positron emission tomography (FDG PET)-computed tomography (CT) in the staging of bladder cancer Henry Goodfellow, Zaid Viney*, Paul Hughes*, Sheila Rankin*, Giles Rottenberg*, Simon Hughes*, Felicity Evison**, Prokar Dasgupta†, Timothy O'Brien* and Muhammad Shamim Khan* Guy's, King's and St Thomas' Medical School, Kings College London, *Urology Centre, Guy's Hospital, Guy's and St Thomas' NHS Trust, **University Hospitals Birmingham NHS Foundation Trust, and †Medical Research Council (MRC) Centre for Transplantation & National, Institute for Health Research (NIHR) Comprehensive Biomedical Research Centre, King's College London, King's Health Partners, Guy's Hospital, London, UK
Objective 18
To determine whether to use F-fluorodeoxyglucose positron emission tomography (FDG PET) scans in the preoperative staging of bladder cancer (BC).
Patients and Methods In all, 233 patients with muscle-invasive BC (MIBC) or high-risk non-MIBC being considered for radical cystectomy (RC) between 2005 and 2011 had FDG-PET and computed tomography (CT) of the chest, abdomen and pelvis to assess for pelvic lymph node (LN) involvement or distant metastases. Sensitivity and specificity for detecting pelvic LN involvement was determined by comparing the results of the scans to the histopathology reports in patients undergoing RC. These parameters for distant metastases were determined from biopsy results or follow-up imaging. In patients who did not undergo RC, follow-up imaging was used to evaluate the sensitivity and specificity. Patients were excluded from analysis if they either had neoadjuvant chemotherapy or had 10% of the patients then you would recommend that PET scans be used routinely. If the PET scan only changed management in 5–10% of patients then PET scans should be used in selected patients. Finally, if 30 days) between the scans, the CT component of the PET study was used instead of the staging CT to determine the preoperative staging in these patients. All the PET and staging CT scans were reviewed by the reporting radiologist at the time of the scan and re-reviewed by a second radiologist to confirm the findings. The average time between the PET scan and RC was 50 days. Assessment of Pelvic LNs RCs were performed by two uro-oncology surgeons. Pelvic lymphadenectomy was performed up to the bifurcation of the aorta. In patients where no positive LNs were detected on histopathology, a minimum of 10 dissected LNs were required to consider the patient negative for metastatic pelvic LN involvement [17]. The LNs were placed in two separate containers and labelled as ‘right/left pelvic nodes’ for histopathological assessment. Peri-vesical LNs were identified by the histopathologists in the RC specimens. The results of histology were compared with staging CT and PET scans in this subgroup. All LNs that had increased FDG uptake were considered positive on PET scan. Pelvic LNs with a maximum short axis diameter of >8 mm on CT were considered positive according to the consensus of the radiologists at our institution. Using combined imaging, LN
FDG pet in the staging of bladder cancer
status was regarded as negative if both imaging methods failed to show any abnormality. Conversely, a LN was regarded as positive if either the PET or CT scan or both showed an abnormality. To establish a difference in sensitivity and specificity between the two types of scan and the combined scan a receiver operating characteristic (ROC) curve analysis was completed. As data on the number of malignant LNs detected by the PET/CT scans was not recorded, but rather only the determination of malignancy, an expected number of LNs for each scan was calculated. This expected number was based on the probability of a scan missing malignant LNs and the numbers seen during biopsy. The sensitivity of the ROC curves to the specific probability calculated was then tested by varying the probability (±0.2).
corresponding CT scan (Fig. 2). The CT scan showed three abnormalities not evident on the PET scan. Of the 13 patients with PET avid lesions, six were distant metastases (three skeletal, one pulmonary and two LN) (Table 3). The remaining seven patients with abnormal PET scans were found to have an adenocarcinoma of the colon (one), colonic adenoma (one), a parotid basal cell tumour (one) and inflammatory lesions (four) (Table 3). Fig. 1 Flow chart of study. In all, 233 patients were considered for RC and had a PET and CT scan for the preoperative staging of BC. The PET and CT scans were compared for metastatic disease outside of the pelvis in 207 patients. A RC was performed on 159 patients of whom 66 had neoadjuvant therapy or did not have an adequate lymphadenectomy. The histopathology report of the pelvic LNs could be compared with the original CT and PET scans in 93 patients.
Results In all, 233 patients were studied in this cohort. Of these 175 (75%) were men and 58 (25%) were women. The mean (range) age was 69 (23–85) years (Table 1). Most (221 patients, 95%) had TCC and 12 (5%) had other histological types (Squamous, five; adenocarcinoma, four; small cell, one; paraganglioma, one; neuroendocrine, one) (Table 1). In all, 26 patients were excluded from the study as they were either followed up at another hospital or died from a cause other than metastatic BC within a year of presentation (Fig. 1).
Total patients N = 233 Excluded • Follow up at another hospital • Died from cause other than BC
PET and CT Scans N = 207
Excluded • Did not have a radical cystectomy
Staging for Distant Metastases The PET/CT (PET) scan and staging CT scan were compared to determine concordance for the detection of metastatic disease outside of the pelvis in 207 patients. Overall, the PET scan was more sensitive than the CT for detecting metastatic disease outside of the pelvis (54% vs 41%; Table 2). The specificity on the contrary was high in both PET (97%) and CT (98%). There was discordance in findings between the PET and CT scans in 16 patients (4.2%). The PET scan identified 13 patients with FDG avid lesions with no abnormality on the
Number of patients in study Mean (range) age, years N (%): Men Women Ta Tis T1 T2 T3 T4 TCC Other cell type Neoadjuvant treatment Metastatic lesions
Excluded • Inadequate lymphadectomy • Neoadjuvant treatment
Histopathology compared to PET and CT scans N = 93
Table 1 Demographics of study. Variable
Radical Cystectomy N = 159
Value 233 69 (23–85) 175 (75) 58 (25) 17 (7) 6 (3) 60 (26) 82 (35) 50 (21) 18 (8) 221 (95) 12 (5) 41 (18) 56
Table 2 The detection of metastatic disease outside of the pelvis by the PET and staging CT scans. Metastatic disease
PET
CT
Total False positives False negative True positive True negative Sensitivity Specificity Positive predictive value Negative predictive value Accuracy
207 4 26 30 147 0.54 0.97 0.88 0.85 0.86
207 3 33 23 148 0.41 0.98 0.88 0.82 0.83
© 2013 The Authors BJU International © 2013 BJU International
391
Goodfellow et al.
Fig. 2 PET scan and CT component of the PET scan for two patients with metastatic disease. A, B and C is the PET, CT and coloured overlay for a patient who had a metastatic BC lesion in the right femur. A′, B′ and C′ is the PET, CT and coloured overlay for a patient who had a metastatic pelvic LNs.
Table 3 Lesions identified by the PET scan with no corresponding pathology on the CT scan. Disease
Location
Pathology
Confirmation*
Metastatic
Femur Ilium Cervical LN Vertebrae Lung Para aortic LN Colon Colon Colon Parotid Aortocaval LN Axilla LN
TCC TCC TCC TCC TCC TCC Carcinoma Adenoma inflammatory Basal cell inflammatory inflammatory
Biopsy Biopsy Biopsy CT scan CT scan CT scan Biopsy Biopsy CT scan Biopsy CT scan Biopsy
Primary cancer Benign
The CT scan detected lesions in three patients not shown on PET scan. One patient had enlarged para-aortic and iliac LNs not detected on the PET scan. The other two were small (
Objective 18
To determine whether to use F-fluorodeoxyglucose positron emission tomography (FDG PET) scans in the preoperative staging of bladder cancer (BC).
Patients and Methods In all, 233 patients with muscle-invasive BC (MIBC) or high-risk non-MIBC being considered for radical cystectomy (RC) between 2005 and 2011 had FDG-PET and computed tomography (CT) of the chest, abdomen and pelvis to assess for pelvic lymph node (LN) involvement or distant metastases. Sensitivity and specificity for detecting pelvic LN involvement was determined by comparing the results of the scans to the histopathology reports in patients undergoing RC. These parameters for distant metastases were determined from biopsy results or follow-up imaging. In patients who did not undergo RC, follow-up imaging was used to evaluate the sensitivity and specificity. Patients were excluded from analysis if they either had neoadjuvant chemotherapy or had 10% of the patients then you would recommend that PET scans be used routinely. If the PET scan only changed management in 5–10% of patients then PET scans should be used in selected patients. Finally, if 30 days) between the scans, the CT component of the PET study was used instead of the staging CT to determine the preoperative staging in these patients. All the PET and staging CT scans were reviewed by the reporting radiologist at the time of the scan and re-reviewed by a second radiologist to confirm the findings. The average time between the PET scan and RC was 50 days. Assessment of Pelvic LNs RCs were performed by two uro-oncology surgeons. Pelvic lymphadenectomy was performed up to the bifurcation of the aorta. In patients where no positive LNs were detected on histopathology, a minimum of 10 dissected LNs were required to consider the patient negative for metastatic pelvic LN involvement [17]. The LNs were placed in two separate containers and labelled as ‘right/left pelvic nodes’ for histopathological assessment. Peri-vesical LNs were identified by the histopathologists in the RC specimens. The results of histology were compared with staging CT and PET scans in this subgroup. All LNs that had increased FDG uptake were considered positive on PET scan. Pelvic LNs with a maximum short axis diameter of >8 mm on CT were considered positive according to the consensus of the radiologists at our institution. Using combined imaging, LN
FDG pet in the staging of bladder cancer
status was regarded as negative if both imaging methods failed to show any abnormality. Conversely, a LN was regarded as positive if either the PET or CT scan or both showed an abnormality. To establish a difference in sensitivity and specificity between the two types of scan and the combined scan a receiver operating characteristic (ROC) curve analysis was completed. As data on the number of malignant LNs detected by the PET/CT scans was not recorded, but rather only the determination of malignancy, an expected number of LNs for each scan was calculated. This expected number was based on the probability of a scan missing malignant LNs and the numbers seen during biopsy. The sensitivity of the ROC curves to the specific probability calculated was then tested by varying the probability (±0.2).
corresponding CT scan (Fig. 2). The CT scan showed three abnormalities not evident on the PET scan. Of the 13 patients with PET avid lesions, six were distant metastases (three skeletal, one pulmonary and two LN) (Table 3). The remaining seven patients with abnormal PET scans were found to have an adenocarcinoma of the colon (one), colonic adenoma (one), a parotid basal cell tumour (one) and inflammatory lesions (four) (Table 3). Fig. 1 Flow chart of study. In all, 233 patients were considered for RC and had a PET and CT scan for the preoperative staging of BC. The PET and CT scans were compared for metastatic disease outside of the pelvis in 207 patients. A RC was performed on 159 patients of whom 66 had neoadjuvant therapy or did not have an adequate lymphadenectomy. The histopathology report of the pelvic LNs could be compared with the original CT and PET scans in 93 patients.
Results In all, 233 patients were studied in this cohort. Of these 175 (75%) were men and 58 (25%) were women. The mean (range) age was 69 (23–85) years (Table 1). Most (221 patients, 95%) had TCC and 12 (5%) had other histological types (Squamous, five; adenocarcinoma, four; small cell, one; paraganglioma, one; neuroendocrine, one) (Table 1). In all, 26 patients were excluded from the study as they were either followed up at another hospital or died from a cause other than metastatic BC within a year of presentation (Fig. 1).
Total patients N = 233 Excluded • Follow up at another hospital • Died from cause other than BC
PET and CT Scans N = 207
Excluded • Did not have a radical cystectomy
Staging for Distant Metastases The PET/CT (PET) scan and staging CT scan were compared to determine concordance for the detection of metastatic disease outside of the pelvis in 207 patients. Overall, the PET scan was more sensitive than the CT for detecting metastatic disease outside of the pelvis (54% vs 41%; Table 2). The specificity on the contrary was high in both PET (97%) and CT (98%). There was discordance in findings between the PET and CT scans in 16 patients (4.2%). The PET scan identified 13 patients with FDG avid lesions with no abnormality on the
Number of patients in study Mean (range) age, years N (%): Men Women Ta Tis T1 T2 T3 T4 TCC Other cell type Neoadjuvant treatment Metastatic lesions
Excluded • Inadequate lymphadectomy • Neoadjuvant treatment
Histopathology compared to PET and CT scans N = 93
Table 1 Demographics of study. Variable
Radical Cystectomy N = 159
Value 233 69 (23–85) 175 (75) 58 (25) 17 (7) 6 (3) 60 (26) 82 (35) 50 (21) 18 (8) 221 (95) 12 (5) 41 (18) 56
Table 2 The detection of metastatic disease outside of the pelvis by the PET and staging CT scans. Metastatic disease
PET
CT
Total False positives False negative True positive True negative Sensitivity Specificity Positive predictive value Negative predictive value Accuracy
207 4 26 30 147 0.54 0.97 0.88 0.85 0.86
207 3 33 23 148 0.41 0.98 0.88 0.82 0.83
© 2013 The Authors BJU International © 2013 BJU International
391
Goodfellow et al.
Fig. 2 PET scan and CT component of the PET scan for two patients with metastatic disease. A, B and C is the PET, CT and coloured overlay for a patient who had a metastatic BC lesion in the right femur. A′, B′ and C′ is the PET, CT and coloured overlay for a patient who had a metastatic pelvic LNs.
Table 3 Lesions identified by the PET scan with no corresponding pathology on the CT scan. Disease
Location
Pathology
Confirmation*
Metastatic
Femur Ilium Cervical LN Vertebrae Lung Para aortic LN Colon Colon Colon Parotid Aortocaval LN Axilla LN
TCC TCC TCC TCC TCC TCC Carcinoma Adenoma inflammatory Basal cell inflammatory inflammatory
Biopsy Biopsy Biopsy CT scan CT scan CT scan Biopsy Biopsy CT scan Biopsy CT scan Biopsy
Primary cancer Benign
The CT scan detected lesions in three patients not shown on PET scan. One patient had enlarged para-aortic and iliac LNs not detected on the PET scan. The other two were small (
