Prediction of molecular subclasses of uveal melanoma by deep learning using routine haematoxylin-eosin-stained tissue slides.

AIMS: Uveal melanoma has a high propensity to metastasize. Prognosis is associated with specific driver mutations and copy number variations, and these can only be obtained after genetic testing. In this study we evaluated the efficacy of patient outcome prediction using deep learning on haematoxylin and eosin (HE)-stained primary uveal melanoma slides in comparison to molecular testing. METHODS: In this retrospective study of patients with uveal melanoma, 113 patients from the Erasmus Medical Centre who underwent enucleation had tumour tissue analysed for molecular classification between 1993 and 2020. Routine HE-stained slides were scanned to obtain whole-slide images (WSI). After annotation of regions of interest, tiles of 1024 × 1024 pixels were extracted at a magnification of 40×. An ablation study to select the best-performing deep-learning model was carried out using three state-of-the-art deep-learning models (EfficientNet, Vision Transformer, and Swin Transformer). RESULTS: Deep-learning models were subjected to a training cohort (n = 40), followed by a validation cohort (n = 20), and finally underwent a test cohort (n = 48). A k-fold cross-validation (k = 3) of validation and test cohorts (n = 113 of three classes: BAP1, SF3B1, EIF1AX) demonstrated Swin Transformer as the best-performing deep-learning model to predict molecular subclasses based on HE stains. The model achieved an accuracy of 0.83 ± 0.09 on the validation cohort and 0.75 ± 0.04 on the test cohort. Within the subclasses, this model correctly predicted 70% BAP1-mutated, 61% SF3B1-mutated and 80% EIF1AX-mutated UM in the test set. CONCLUSIONS: This study showcases the potential of the deep-learning methodology for predicting molecular subclasses in a multiclass manner using HE-stained WSI. This development holds promise for advanced prognostication of UM patients without the need of molecular or immunohistochemical testing. Additionally, this study suggests there are distinct histopathological features per subclass; mainly utilizing epithelioid cellular morphology for BAP1-classification, but an unknown feature distinguishes EIF1AX and SF3B1.

Uveal Melanoma Patients Have a Distinct Metabolic Phenotype in Peripheral Blood.

Uveal melanomas (UM) are detected earlier. Consequently, tumors are smaller, allowing for novel eye-preserving treatments. This reduces tumor tissue available for genomic profiling. Additionally, these small tumors can be hard to differentiate from nevi, creating the need for minimally invasive detection and prognostication. Metabolites show promise as minimally invasive detection by resembling the biological phenotype. In this pilot study, we determined metabolite patterns in the peripheral blood of UM patients (n = 113) and controls (n = 46) using untargeted metabolomics. Using a random forest classifier (RFC) and leave-one-out cross-validation, we confirmed discriminatory metabolite patterns in UM patients compared to controls with an area under the curve of the receiver operating characteristic of 0.99 in both positive and negative ion modes. The RFC and leave-one-out cross-validation did not reveal discriminatory metabolite patterns in high-risk versus low-risk of metastasizing in UM patients. Ten-time repeated analyses of the RFC and LOOCV using 50% randomly distributed samples showed similar results for UM patients versus controls and prognostic groups. Pathway analysis using annotated metabolites indicated dysregulation of several processes associated with malignancies. Consequently, minimally invasive metabolomics could potentially allow for screening as it distinguishes metabolite patterns that are putatively associated with oncogenic processes in the peripheral blood plasma of UM patients from controls at the time of diagnosis.

Genetics of Ocular Melanoma: Insights into Genetics, Inheritance and Testing.

Ocular melanoma consists of posterior uveal melanoma, iris melanoma and conjunctival melanoma. These malignancies derive from melanocytes in the uveal tract or conjunctiva. The genetic profiles of these different entities differ from each other. In uveal melanoma, GNAQ and GNA11 gene mutations are frequently found and prognosis is based on mutation status of BAP1, SF3B1 and EIF1AX genes. Iris melanoma, also originating from the uvea, has similarities to the genetic makeups of both posterior uveal melanoma (UM) and conjunctival melanoma since mutations in GNAQ and GNA11 are less common and genes involved in conjunctival melanoma such as BRAF have been described. The genetic spectrum of conjunctival melanoma, however, includes frequent mutations in the BRAF, NRAS and TERT promoter genes, which are found in cutaneous melanoma as well. The BRAF status of the tumor is not correlated to prognosis, whereas the TERT promoter gene mutations are. Clinical presentation, histopathological characteristics and copy number alterations are associated with survival in ocular melanoma. Tissue material is needed to classify ocular melanoma in the different subgroups, which creates a need for the use of noninvasive techniques to prognosticate patients who underwent eye preserving treatment.

Evaluation of Circulating Tumor DNA as a Liquid Biomarker in Uveal Melanoma.

Purpose: Uveal melanoma (UM) has a high propensity to metastasize. Prognosis is associated with specific driver mutations and copy number variations (CNVs), but limited primary tumor tissue is available for molecular characterization due to eye-sparing irradiation treatment. This study aimed to assess the rise in circulating tumor DNA (ctDNA) levels in UM and evaluate its efficacy for CNV-profiling of patients with UM. Methods: In a pilot study, we assessed ctDNA levels in the blood of patients with UM (n = 18) at various time points, including the time of diagnosis (n = 13), during fractionated stereotactic radiotherapy (fSRT) treatment (n = 6), and upon detection of metastatic disease (n = 13). Shallow whole-genome sequencing (sWGS) combined with in silico size-selection was used to identify prognostically relevant CNVs in patients with UM (n = 26) from peripheral blood retrieved at the time of diagnosis (n = 9), during fSRT (n = 5), during post-treatment follow-up (n = 4), metastasis detection (n = 6), and metastasis follow-up (n = 4). Results: A total of 34 patients had blood analyzed for ctDNA detection (n = 18) and/or CNV analysis (n = 26) at various time points. At the time of diagnosis, 5 of 13 patients (38%) had detectable ctDNA (median = 0 copies/mL). Upon detection of metastatic disease, ctDNA was detected in 10 of 13 patients (77%) and showed increased ctDNA levels (median = 24 copies/mL, P < 0.01). Among the six patients analyzed during fSRT, three (50%) patients had detectable ctDNA at baseline and three of six (50%) patients had undetectable levels of ctDNA. During the fSRT regimen, ctDNA levels remained unchanged (P > 0.05). The ctDNA fractions were undetectable to low in localized disease, and sWGS did not elucidate chromosome 3 status from blood samples. However, in 7 of 10 (70%) patients with metastases, the detection of chromosome 3 loss corresponded to the high metastatic-risk class. Conclusions: The rise in ctDNA levels observed in patients with UM harboring metastases suggests its potential utility for CNV profiling. These findings highlight the potential of using ctDNA for metastasis detection and patient inclusion in therapeutic studies targeting metastatic UM.

Detection of metastases using circulating tumour DNA in uveal melanoma.

BACKGROUND: Approximately 50% of uveal melanoma (UM) patients will develop metastatic disease depending on the genetic features of the primary tumour. Patients need 3-12 monthly scans, depending on their prognosis, which is costly and often non-specific. Circulating tumour DNA (ctDNA) quantification could serve as a test to detect and monitor patients for early signs of metastasis and therapeutic response. METHODS: We assessed ctDNA as a biomarker in three distinct UM cohorts using droplet-digital PCR: (A) a retrospective analysis of primary UM patients to predict metastases; (B) a prospective analysis of UM patients after resolution of their primary tumour for early detection of metastases; and (C) monitoring treatment response in metastatic UM patients. RESULTS: Cohort A: ctDNA levels were not associated with the development of metastases. Cohort B: ctDNA was detected in 17/25 (68%) with radiological diagnosis of metastases. ctDNA was the strongest predictor of overall survival in a multivariate analysis (HR = 15.8, 95% CI 1.7-151.2, p = 0.017). Cohort C: ctDNA monitoring of patients undergoing immunotherapy revealed a reduction in the levels of ctDNA in patients with combination immunotherapy. CONCLUSIONS: Our proof-of-concept study shows the biomarker feasibility potential of ctDNA monitoring in for the clinical management of uveal melanoma patients.

High C-Reactive Protein Levels Are Related to Better Survival in Patients with Uveal Melanoma.

Purpose: To determine whether peripheral blood leukocyte numbers and serum markers of inflammation can be used to predict which patients with primary uveal melanoma will develop metastasis. Design: Retrospective study. Participants: Medical records of patients with uveal melanoma (UM) who received treatment for primary UM between February 1992 and December 2020 at the Erasmus University Medical Center (Rotterdam, The Netherlands) and the Rotterdam Eye Hospital (Rotterdam, The Netherlands) were reviewed. Methods: Inclusion criteria were the presence of a melanoma of the choroid or ciliary body and the availability of data from peripheral blood samples taken before treatment of the melanoma. Data including patient demographics, C-reactive protein (CRP) levels; erythrocyte sedimentation rate (ESR); number of leukocytes, neutrophils, monocytes, and lymphocytes; and histopathologic findings were obtained from medical records. Neutrophil-to-lymphocyte ratio (NLR) and lymphocyte-to-monocyte ratio (LMR) were calculated. Main Outcome Measures: Metastasis-free survival. Results: Of the 807 patients with UM, serum and leukocyte data were available for 183 of them at the time of primary tumor treatment. In the total group, no correlation was found between ESR before treatment; the number of leukocytes; percentages of neutrophils, monocytes, and lymphocytes; or NLR or LMR values and any of the clinical characteristics or metastasis-free survival. Among patients who underwent enucleation, those with negative BAP1 findings showed significantly lower numbers of leukocytes (P < 0.05). In the entire cohort, a significant association was found between high CRP levels and longer metastasis-free survival (MFS; P = 0.049). Conclusions: The total blood leukocyte number was related to loss of BAP1 staining in patients who underwent enucleation, with lower leukocyte counts correlating with absent BAP1 staining. Higher CRP levels were associated with a longer MFS in the entire cohort. Neither the NLR nor the LMR is a good predictor for metastasis developing in patients with UM.

Is Tissue Still the Issue? The Promise of Liquid Biopsy in Uveal Melanoma.

Uveal melanoma (UM) is the second most frequent type of melanoma. Therapeutic options for UM favor minimally invasive techniques such as irradiation for vision preservation. As a consequence, no tumor material is obtained. Without available tissue, molecular analyses for gene expression, mutation or copy number analysis cannot be performed. Thus, proper patient stratification is impossible and patients' uncertainty about their prognosis rises. Minimally invasive techniques have been studied for prognostication in UM. Blood-based biomarker analysis has become more common in recent years; however, no clinically standardized protocol exists. This review summarizes insights in biomarker analysis, addressing new insights in circulating tumor cells, circulating tumor DNA, extracellular vesicles, proteomics, and metabolomics. Additionally, medical imaging can play a significant role in staging, surveillance, and prognostication of UM and is addressed in this review. We propose that combining multiple minimally invasive modalities using tumor biomarkers should be the way forward and warrant more attention in the coming years.

Post-operative Refractive Prediction Error After Phacovitrectomy: A Retrospective Study.

INTRODUCTION: Many authors have reported on a myopic post-operative refractive prediction error when combining phacoemulsification with pars plana vitrectomy (phacovitrectomy). In this study we evaluate the amount of this error in our facility and try to elucidate the various factors involved. METHODS: This was a retrospective study which included 140 patients who underwent phacovitrectomy (39 with macular holes, 88 with puckers, and 13 with floaters). Post-operative refractive error was defined as the difference between the actual spherical equivalent (SEQ) and expected SEQ based on the SRK/T and Holladay-II formulas. Both univariate (paired t test, independent t test, one-way analysis of variance, or Mann-Whitney test) and multivariate (regression analysis) statistical analyses were performed. RESULTS: Overall, a refractive error of - 0.13 dpt (p = 0.033) and - 0.26 dpt (p < 0.01) were found in the SRK/T and Holladay-II formulas, respectively. For the independent diagnoses, only macular holes showed a myopic error with the SRK/T (- 0.31 dpt; p < 0.01) and Holladay-II (- 0.44 dpt; p < 0.01) formulas. In univariate analysis, significant factors involved in myopic refractive error were macular hole as diagnosis (p < 0.01 for SRK/T and Holladay-II), gas tamponade (SRK/T p = 0.024; Holladay-II p = 0.025), pre-operative myopia (p < 0.01 for SRK/T), and optical technique for axial length measurement (SRK/T and Holladay-II p < 0.01). In the multivariate analysis, pre-operative axial length (p = 0.026), optical technique for axial length measurement (p < 0.01), and pre-operative SEQ (p < 0.01) were independent predictors for myopic refractive error in the SRK/T formula. For the Holladay-II formula, optical technique for axial length measurement (p < 0.01) and pre-operative SEQ (p = 0.04) were predictive. CONCLUSION: Various factors are involved in determining the myopic refractive error after phacovitrectomy. Not every factor seems to be as important in each individual patient, suggesting a more tailored approach is warranted to overcome this problem.