Aims To evaluate the prognostic value of positron emission tomography (PET) imaging in patients with choroidal melanoma.
Methods We undertook a retrospective review of 40 consecutive patients with choroidal melanoma who underwent pretreatment whole-body PET, received either brachytherapy using ruthenium-106 plaque, enucleation or gamma knife radiotherapy, and had 1 year of follow-up. Metabolic activity of choroidal melanoma measured as standardised uptake value (SUV) by PET imaging was evaluated with respect to the survival of patients.
Results SUV (p=0.003) and the largest basal diameter of the tumour (p=0.003) were significantly correlated with metastatic death (Cox proportional hazards regression). There was an inverse correlation between tumour metabolic activity and time to metastasis (p=0.049; linear regression).
Conclusion Metabolic activity by PET imaging significantly predicted the survival of patients with choroidal melanoma.
- uveal melanoma
Statistics from Altmetric.com
Choroidal melanoma is the most common primary intraocular malignancy in adults and approximately 50% of patients develop fatal metastatic disease, which usually involves the liver.1 2 In recent years, 2-Deoxy-2-[18F]-D-glucose (FDG) positron emission tomography (PET) has gained interest as an investigative tool for initial staging, recurrence or metastasis for choroidal melanoma.3–8 Different tumours have different rates of inherent metabolism and therefore differ in their rate of [18F]FDG uptake, which is represented as standardised uptake values (SUV). Finger et al9 has suggested SUV as a non-invasive biomarker for metastatic choroidal melanoma, as two of their 14 patients with highest SUV developed metastatic disease and increased SUV appeared to correlate with known clinical and pathological poor prognostic factors, including larger size of the tumour, epithelioid-cell type and extrascleral extension. Recently, SUV has been shown to be associated with chromosome 3 defect,10 which is most strongly associated with metastatic death for choroidal melanoma.11
We hypothesise that high SUV should be associated with metastatic death for choroidal melanoma. We also studied whether SUV correlated with time to metastasis from initial evaluation.
The medical records of all patients diagnosed with choroidal melanoma at the Department of Ophthalmology at Yonsei University College of Medicine, Seoul, Korea, between December 2003 and June 2009 were reviewed. Inclusion criteria included: (1) available pretreatment PET; (2) treatment either with brachytherapy using ruthenium (Ru)-106 plaque, gamma knife radiotherapy (GKR) or enucleation; (3) follow-up of 1 year or more. Patients who underwent PET/CT as a baseline study were excluded due to potential bias in comparing SUV from different PET scanners. Largest basal tumour diameter (LBD) and tumour height were measured by B-scan ultrasonography.
Brachytherapy using Ru-106 plaque (BEBIG, Berlin, Germany) has been performed in our institute since October 2006 and is the first choice of treatment. All patients with brachytherapy received adjunctive transpupillary thermotherapy (TTT) of the tumour apex. Exposure time was 1 min per application and laser energy was adjusted stepwise until the tumour surface turned slightly grey at the end of 1 min exposure. On follow-up, additional TTT sessions were performed with intervals of 3 months (up to three sessions) in an office setting for cases with insufficient tumour regression.
GKR was performed using a Leksell gamma knife (Elekta Instruments AB, Stockholm, Sweden) as an eye-conserving therapy until October 2006, after which time radiation therapy was primarily replaced by brachytherapy. Enucleation was performed if the patient was reluctant to undergo eye-conserving therapy or if the physician deemed that such therapy was unlikely to succeed.
Whole body dual-modality PET/CT
All patients underwent routine FDG PET with the GE advance PET scanner (GE advance, Milwaukee, Wisconsin, USA). All patients fasted for at least 6 h and glucose levels in peripheral blood in all patients were confirmed to be ≤7.7 mmol/l (140 mg/dl) before FDG injection. Approximately 5.5 MBq/kg body weight of FDG were administered intravenously 1 h before image acquisition. Transmission scans were taken with 68Ge. A standard PET imaging protocol from neck to the proximal thighs with acquisition time of 5 min/bed in two-dimensional mode was acquired. Images were then reconstructed using the ordered subset expectation maximisation (two iterations, 20 subsets).
For evaluation of the FDG uptake of the ocular lesion at the initial PET before operation, an experienced nuclear medicine specialist (AC) interpreted the PET images by maximum SUV. A region of interest was drawn at the ocular lesion seen in the corresponding CT or MRI and maximum SUV was measured (figures 1 and 2).
For evaluation of the FDG uptake at the postoperative follow-up, any nodular FDG uptake higher than the liver was considered positive, and the corresponding lesion was verified either by follow-up FDG PET, FDG PET/CT or other imaging studies.
We contacted the National Health Insurance Corporation to verify the vital status of each patient up to 30 June 2010. The date and cause of death were confirmed by the National Health Statistical Office and the National Registry Program. Survival duration was determined as the time elapsed from the date of initial PET scan to the date of death or to 30 June 2010.
Patients were divided into three groups according to the treatment modalities they received. Demographic and clinical features were compared among groups using the Kruskal–Wallis test. Four-year survival rates of three groups were analysed using the logrank test. Receiver operating characteristic (ROC) curve analysis of SUV measurements was performed to define the best cut-off threshold for predicting metastatic death. Univariate Cox proportional hazards regression was performed to evaluate which factors among age, sex, ciliary body involvement of tumour, tumour LBD, tumour height and SUV predicted metastatic death. The variables that were significant on a univariate level (p<0.05) were entered into multivariate Cox regression analysis using the stepwise method.
Linear regression analysis was performed to study correlation between SUV and time to metastasis, which was defined as the interval of time between the initial PET scan to the date of metastasis confirmed by imaging studies including abdominal sonography, CT, MRI, PET or PET/CT. Statistical analysis was performed using SPSS 15.0 (SPSS, Inc.) and p<0.05 was considered significant for all analyses.
Forty eyes of 40 patients were included and the mean age of patients was 51±14 (range 24–82) years. There were 22 (55%) male patients. Tumours had a mean LBD of 10.7±3.1 (range 6.0–18.1) mm and a mean height of 6.5±2.7 (range 2.6–13.1) mm. The mean SUV was 2.0±1.0 (range 1.2–6.9). There were no sign of metastasis in the initial PET for all patients. Based on 7th edition of the American Joint Cancer Committee/Union Internationale Contre le Cancer (AJCC/UICC) classification,12 there were 12 (30.0%) patients for stage I, 12 (30.0%) for stage IIA, six (15.0%) for stage IIB, three (7.5%) for stage IIIA and seven (17.5%) for stage IV (table 1).
To study whether a particular treatment modality was associated with higher risk of metastatic death, patients were divided into three groups according to their treatments (table 1). Mean age, LBD of the tumour and tumour height were not significantly different among three groups. Estimated 4-year survival using the Kaplan–Meier method were 78% for both the GKR and enucleation groups, and 80% for the brachytherapy group. This difference was not statistically different (logrank test; p=0.769). Univariate analysis showed that SUV (p=0.003; HR 3.3; 95% CI 1.5 to 7.2) and LBD (p=0.003; HR 1.7; 95% CI 1.2 to 2.4) was significantly correlated with metastatic death. When these factors were entered for multivariate analysis, neither SUV (p=0.109) nor LBD (p=0.060) correlated with metastatic death.
ROC curve analysis showed that SUV >2.2 was 71% sensitive and 88% specific in predicting metastatic death. There was an inverse correlation between initial SUV of the tumour and time to metastasis; higher tumour SUV was correlated with shorter time to metastasis with borderline statistical significance (p=0.049; R2=0.57) (figure 3).
We have identified a statistically significant association between tumour metabolic activity measured by PET and metastatic death.
Increasing tumour size is known to be related to SUV positivity.4 6 Reddy et al4 found that 33% of medium and 75% of large choroidal melanomas had metabolic activity identified by PET/CT, which was defined as SUV >2.5. However, they suggested that factors other than tumour size may account for choroidal melanoma that were detectable with PET/CT, since 63% of medium-sized tumours were larger than the minimal resolution of their PET/CT scanner. In this study, both SUV and LBD were strongly associated with metastatic death on univariate analysis, but neither one was significant on multivariate level. Although we could not demonstrate statistically that SUV was still significant after LBD was controlled because of the limitation of small sample sizes, we believe that tumour size alone may not account solely for the metabolic activity of choroidal melanoma by PET since not all sizeable tumours showed high SUV. For example, out of 31 medium-sized tumours by the Collaborative Ocular Melanoma Study (COMS)13 definition in our study, 23 (74%) tumours showed SUV <2, being barely detectable by PET. All 31 cases had basal diameter >6 mm, being higher than the minimal resolution of our PET.
Currently, there is no set clinical standard for SUV that can define metabolically active versus inactive tumours. Finger et al9 used an SUV of 4.0 as the cut-off to define positive metabolic activity, as it appeared as a natural division in their data. Many previous publications have used 2.5 as a cut-off value for SUV.4 10 14 This is in agreement with our finding, as we statistically determined 2.2 as the threshold value for our study cohort.
Higher initial SUV not only predicted metastatic death, but also predicted shorter interval to detection of metastasis. This may have practical implications as it can affect the follow-up plan for patients. It has been suggested that metabolically active tumours determined by PET/CT may have different histopathological and molecular characteristics, as they were associated with mixed or epithelioid cell type, enlarged blood vessels and monosomy 3.9 10 14 Tumours with high metabolic activity may also exhibit more robust metastatic activity as well.
Despite the relatively low FDG uptake in the ocular lesion in the initial PET, metastatic lesions showed intense FDG uptake. All seven patients who developed metastatic disease had liver metastasis and four of them also had additional extra-hepatic metastasis including lung, bone and lymph nodes. Metastatic lesions, especially in liver and bone, tended to show intense FDG uptake; mean SUV for metastatic lesion was 9.6 (range 4.4–13.6) for liver and 6.0 (range 2.3–17.6) for bone (data not shown), whereas mean initial SUV for ocular lesions of metastatic diseases was 3.2 (range 1.2–6.9).
There are limitations of our study that should be considered. First, it was a retrospective study with limited number of patients, which may affect the statistical significance of the findings. Second, there may be potential bias in assessing risks for metastatic disease due to multiple treatment variables. Estimated survival rates, however, were comparable among treatment groups in current study. A multi-institutional, prospective randomised clinical trial by the COMS Group has demonstrated comparable efficacy between enucleation and brachytherapy for medium-sized choroidal melanomas.15
In summary, metabolic activity measured by PET was associated with metastatic death for choroidal melanoma with statistical significance. Increased FDG uptake also correlated with reduced interval to development of metastatic disease. A larger scale, prospective study of the prognostic value of initial PET imaging for choroidal melanoma is warranted.
Presented in part at the World Molecular Imaging Conference, 8–11 September 2010, Kyoto, Japan.
Funding This work was supported by a National Research Foundation of Korea Grant funded by the Korean Government (2009-0077504).
Competing interests None declared.
Ethics approval This study was conducted with the approval of the Yonsei University College of Medicine.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.