Surveillance following treatment of primary ocular melanoma
Introduction 50% of UM patients relapse with predominantly liver metastases Risk of metastatic disease can be predicted relatively accurately through use of clinicopathological features & molecular genetics Surgical management of liver metastases offers the only real likelihood of long-term disease control at present (Frenkel et al. 2009; Mariani et al., 2009; Marshall et al., 2013). Introduction of surveillance programmes for patients with a highrisk of developing disease with hope of demonstrating survival benefit. Optimal screening method, timing, patient selection and overall advantage of surveillance remain under debate
Aims of surveillance in UM patients To identify patients who have relapsed. i.e. who have developed metastatic disease. The GDG agreed that the aim is to detect small volume, pre-clinical disease rather than first identifying large volume metastatic disease, clinically detectable. Aim to detect metastatic disease as early as possible. The GDG agreed that there is a need to qualify whether early detection makes a difference. There is evidence in Section 7 that small volume treatment has better outcome. Early treatment increases benefit and lines of treatment
PICO Questions 1. Should all patients be offered surveillance? 2. Should there be a risk-adapted strategy for surveillance? 3. If so, what is a high-risk and or low-risk uveal melanoma? 4. What is the optimal imaging modality for surveillance? 5. What is the interval? 6. What is the duration of surveillance?
Inclusion & exclusion criteria for selecting evidence Inclusion criteria Case control studies Case series > 3 patients Review articles combined with case reports Expert opinion Adult patients only Exclusion criteria Preclinical and animal studies Case reports (1-3 cases) Review articles without any original case information
Inclusion & exclusion criteria for selecting evidence: Results (preamble) No randomized or nonrandomized comparative clinical trial e.g. of total post-treatment survival in subgroups assigned to regular periodic surveillance for metastasis versus no surveillance testing. No studies were found that addressed Question 6: i.e. the duration of surveillance review.
1. Should all patients be offered surveillance? Authors Year Method Studies Surveillance tools Outcome Comment Augsbsurger et al. Am J Ophthal 2011 Literature review; expert opinion 31/4222, Small, retrospective Variable Differing strategies Levels of specific biomarkers reported when metastasis was first confirmed (n = 14) % of patients with abnormal results on surveillance testing (n=13) Failed to find a survival benefit assoc. with regular surveillance No RCT performed Values of diagnostic markers assoc. with surveillance regimen (7) Survival time after metastasis detection (7) Total survival time after initial diagnosis/tx (3) % of patients whose metastatic tumors were detected by presymptomatic testing (5) Surveillance regimens employed by different groups (1) Relationship with other prognostic factors for UM metastasis (1)
Authors Yr Method No. of pts Surveillance tools Outcome Comment Eskelin et al. 1999 Retrospective 46 1985-1996 LFT CXR Abdom US 59% with mets asymptomatic 80% hepatic mets only Recommends biannual screening Combination of US and LFTs CXR of no use Maeda 2007 Retrospective 159 2000-2005 MRI 5.2 + 1.7 years (range, 1.2 6.6 yrs) Biopsy/resection 13% with asymptomatic abnormalities; 9% proven mets Resulted in initiation of Tx No improved survival Kim et al 2010 Retrospectiv e 90 asympt. 259 symptoms LFTs -> CT of liver, if abn LFTs Median time from primary UM diagnosis to met diagnosis was 31.4 mo in asymptomatic group vs 40.3 mo in symptomatic group Presymptomatic detection of metastatic UM by routine surveillance testing seems to confer a survival advantage only in the first year after diagnosis of metastasis.
Authors Yr Method No. of pts Surveillance tools Outcome Comment Eskelin et al. 1999 Retrospective 46 1985-1996 LFT CXR Abdom US 59% with mets asymptomatic 80% hepatic mets only Recommends biannual screening Combination of US and LFTs CXR of no use Maeda 2007 Retrospective 159 2000-2005 MRI 5.2 + 1.7 years (range, 1.2 6.6 yrs) Biopsy/resection 13% with asymptomatic abnormalities; 9% proven mets Resulted in initiation of Tx No improved survival Kim et al 2010 Retrospectiv e 90 asympt. 259 symptoms LFTs -> CT of liver, if abn LFTs Median time from UM diag to met diag was 31.4 mo in asymptomatic group vs 40.3 mo in symptom group Presymptomatic detection of metastatic UM by routine surveillance testing seems to confer a survival advantage only in the first year after diagnosis of metastasis. Marshall et al 2013 Prospective 2000-2010 188 pts Median FU of 28.8 months Study closure @ 5yrs MRI of high risk pts LFTs 90 pts with mets 92% asymptom. 38 operable on MRI -> R0 resection in 12 pts Hepatic MRI has made it possible to evaluate treatments for metastases at an earlier stage of disease. 35% pts
Question 2: Should there be a risk-adapted strategy for UM metastasis surveillance? Little evidence in literature only one study to date: Marshall et al. BJO 2013 Targeted approach: Is cost effective in NHS setting Enables studies for comparison of screening techniques Comparative studies required Marshall et al 2013 Prospective 2000-2010 188 pts Median FU of 28.8 months Study closure @ 5yrs MRI of high risk pts LFTs 90 pts with mets 92% asymptom. 38 operable on MRI -> R0 resection in 12 pts Hepatic MRI has made it possible to evaluate treatments for metastases at an earlier stage of disease. 35% pts survived 5 yrs
Question 3: If so, what is a high-risk and or low-risk uveal melanoma? No consensus on definition in literature (see prognostication Chapter) The definition of high risk uveal melanoma poses difficulties since not all Centres apply the molecular genetic testing, or only in very few selected cases, e.g. enucleation samples. A definition of high risk cannot be based only on molecular genetic abnormalities, but must include clinical and histomorphological features of the tumours, when assessable.
Prognostic factor combinations Authors Year Method Pts Patients/Outcome Comment Damato et al. 2008 Ophthalm ology conditional hazard estimating neural network (CHENN) Taktak A et al., 2004 Artificial Neural Networks Kaisermann et al. Training set of 1780, test set of another 874 patients 2331, split randomly into training and test sets 2005 ANN 153 pts predicting 5-year mortality All-cause survival curves generated by the CHENN matched those produced with Kaplan-Meier analysis (Kolmogorov-Smirnov, P<0.05). In older patients, however, the estimated melanoma-related mortality was lower with the CHENN, which accounted for competing risks, unlike Kaplan-Meier analysis. AI system can match if not better the clinical expert's prediction logistic regression reached 86% forecasting accuracy, with a very low LR (0.8), whereas the human expert forecasting ability was <70% (LR, 1.85) Estimation of survival prognosis in patients with choroidal melanoma requires multivariate assessment of age, sex, clinical tumor stage, cytogenetic melanoma type, and histologic grade of malignancy No genetics included
Question 3: If so, what is a high-risk and or low-risk uveal melanoma? Combination of parameters is basis of Liverpool Uveal Melanoma Prognostication Online (LUMPO) model. 2654 UM patients: Training set = 1780 Test set = 874 3653 UM patients: Bootstrap re-sampling (200x) Bayesian regularization www.ocularmelanomaonline.com
Question 3: If so, what is a high-risk and or low-risk uveal melanoma? Following discussions at OOG Krakow (March 2014) High risk for metastasis could be defined by one or more of the following criteria: Large tumour size on USS (>TNM Stage IIIA) with risk of metastasis at 5 years being Stage IIIA 34% Stage IIIB 55% Stage IIIB 74% Ciliary body location Epithelioid cell morphology (when histology available) Monosomy 3 GEP Class 2
Question 4: What is the optimal imaging modality for surveillance, overall and of the liver? Overlap with Staging chapter Differing methodologies - USS, CT or MRI (with or without contrast enhancement) or body imaging with CT or PET-CT Principal hypothesis behind screening in the surveillance of UM patients is the detection of resectable liver metastases -> liver imaging as the primary modality to detect early disease Note: in autopsy studies, 87-90% of UM (advanced) cases with non-liver metastases.
Question 4: What is the optimal imaging modality for surveillance of the liver? Liver imaging in metastatic UM Authors Year Method Pts Patients/Outcome Comment Servois et al. 2010 Eur J Surg Oncol. Retrospective Comparison of MRI and FDG- PET 15 R0 resection performed in 12 patients MRI was superior to FDG-PET for staging of liver metastases from UM Preoperative confirmation remains imperfect Orcurto et al. 2009 Melanoma Res Retrospective Comparison of MRI and FDG- PET
Question 4: What is the optimal imaging modality for surveillance outside the liver? Extrahepatic imaging in metastatic UM Authors Year Method Pts Patients/Outcome Comment Lorigan et al. 1991 Am J Roentgen ol Retrospective; descriptive CT MR 110 92% patients with hepatic mets 10 patients developed extrahepatic disease, detected using both technologies None re superiority Klingenstein et al. 2010 Melanoma Res FDG-PET CT 11 Hepatic metastases (83%), Osseous (42%), LN (33%), Pulmonary (17%) Adrenal (8%) Muscular metastases (8%). Staging Six patients showed multiple organ involvement (50%).
Question 4: What is the optimal imaging modality for surveillance outside the liver? Extrahepatic imaging in metastatic UM Authors Year Method Pts Patients/Outcome Comment Lorigan et al. 1991 Am J Roentgen ol Retrospective; descriptive CT MR 110 92% patients with hepatic mets 10 patients developed extrahepatic disease, detected using both technologies None re superiority Klingenstein et al. 2010 Melanoma Res FDG-PET CT 11. Staging Kurli et al. Am J Ophthalm ol 2005 Interventional nonrandomized clinical study PET-CT 20 Mean time from initial diagnosis to metastasis was 47 months (range 0 to 154) Liver (100%), bone (50%), lung (25%), lymph nodes (25%), and subcutaneous tissue (25%). Staging Cardiac, brain, thyroid, and posterior abdominal wall lesions (12.5%). LFTs abnormal in only 13% pts
Question 5: What is the optimal surveillance interval? There is very little evidence on which to base decisions regarding either frequency or duration of follow-up. General consensus in the field is that 6-monthly imaging is preferable.
Question 6: What is the optimal duration of surveillance? UM may continue to relapse for many decades following primary diagnosis, with 20-33% of deaths attributed to metastatic recurrence even at 15-42 yrs. Anecdotal reports that resection rates in later relapsing patients more favourable: tumour biology? No studies in literature, to support or discount lifelong surveillance.
Evidence Statements -1 1. To date, a survival benefit to screening of uveal melanoma patients has not been proven. 2. If a substantial and clinically meaningful survival benefit were truly associated with periodic surveillance testing for UM metastases, such benefit would be demonstrated most convincingly by means of a prospective comparative clinical trial in which subgroups of patients with UM (after treatment of their primary intraocular tumor) were subjected to either regular periodic surveillance testing by some consistent regimen or no surveillance testing at all and then followed until death from any cause. 3. Despite the lack of evidence there is general consensus that surveillance testing is not completely worthless, and indeed is performed in virtually all centres in a periodic manner using differing methods for differing lengths of periods. 4. Surveillance clearly identifies many patients with metastasis at a substantially less advanced metastatic substage than would occur if only postsymptomatic testing were employed.
Evidence Statements - 2 5. Targetted surveillance is likely to bring more benefit. A consensus definition of high-risk UM is required, incorporating clinical, histomorphological and genetic features of the tumours. 6. Most surveillance testing for metastatic UM concentrates on the liver, with the effect that highly-sensitive modalities for liver imaging are chosen. 7. The role of extrahepatic imaging in surveillance is unclear, particularly as the frequency of extrahepatic metastatic relapse remains unknown. 8. Hepatic surveillance of UM has resulted in an increased detection rate of metastases in the liver, resulting in increased loco-regional treatment in some centres and trial recruitment.
Recommendations 1. Surveillance is intuitively advantageous, allowing locoregional management of liver-only metastases, and facilitating early systemic treatment and particularly trial enrolment before the disease burden causes deteriorations in general health and performance status. 2. Surveillance facilitates patient follow-up, provides a link with oncology services and allows a more holistic approach to cancer patients that includes early access to cancer nurse specialists and smooth transition to services such as palliative care at an appropriate stage. 3. Prognostication linked with targeted liver screening of high-risk patients should be the preferred option. This should be carried out within a specialist multidisciplinary team that incorporates expertise from ophthalmology, oncology, cancer nursing and hepatic services. 4. Target high-risk patients and perform 6-monthly surveillance incorporating a clinical review, nurse specialist support, blood for putative circulating biomarkers and non-contrast liver MRI for the first 5 years. 5. Beyond 5 years, patients should be counselled with the option to continue lifelong surveillance with annual follow-up thereafter.