Cluster analysis of multiplex ligation-dependent probe amplification data in choroidal melanoma.

PURPOSE: To determine underlying correlations in multiplex ligation-dependent probe amplification (MLPA) data and their significance regarding survival following treatment of choroidal melanoma (CM). METHODS: MLPA data were available for 31 loci across four chromosomes (1p, 3, 6, and 8) in tumor material obtained from 602 patients with CM treated at the Liverpool Ocular Oncology Center (LOOC) between 1993 and 2012. Data representing chromosomes 3 and 8q were analyzed in depth since their association with CM patient survival is well-known. Unsupervised k-means cluster analysis was performed to detect latent structure in the data set. Principal component analysis (PCA) was also performed to determine the intrinsic dimensionality of the data. Survival analyses of the identified clusters were performed using Kaplan-Meier (KM) and log-rank statistical tests. Correlation with largest basal tumor diameter (LTD) was investigated. RESULTS: Chromosome 3: A two-cluster (bimodal) solution was found in chromosome 3, characterized by centroids at unilaterally normal probe values and unilateral deletion. There was a large, significant difference in the survival characteristics of the two clusters (log-rank, p<0.001; 5-year survival: 80% versus 40%). Both clusters had a broad distribution in LTD, although larger tumors were characteristically in the poorer outcome group (Mann-Whitney, p<0.001). Threshold values of 0.85 for deletion and 1.15 for gain optimized the classification of the clusters. PCA showed that the first principal component (PC1) contained more than 80% of the data set variance and all of the bimodality, with uniform coefficients (0.28±0.03). Chromosome 8q: No clusters were found in chromosome 8q. Using a conventional threshold-based definition of 8q gain, and in conjunction with the chromosome 3 clusters, three prognostic groups were identified: chromosomes 3 and 8q both normal, either chromosome 3 or 8q abnormal, and both chromosomes 3 and 8q abnormal. KM analysis showed 5-year survival figures of approximately 97%, 80%, and 30% for these prognostic groups, respectively (log-rank, p<0.001). All MLPA probes within both chromosomes were significantly correlated with each other (Spearman, p<0.001). CONCLUSIONS: Within chromosome 3, the strong correlation between the MLPA variables and the uniform coefficients from the PCA indicates a lack of evidence for a signature gene that might account for the bimodality we observed. We hypothesize that the two clusters we found correspond to binary underlying states of complete monosomy or disomy 3 and that these states are sampled by the complete ensemble of probes. Consequently, we would expect a similar pattern to emerge in higher-resolution MLPA data sets. LTD may be a significant confounding factor. Considering chromosome 8q, we found that chromosome 3 cluster membership and 8q gain as traditionally defined have an indistinguishable impact on patient outcome.

Single nucleotide polymorphism array analysis of uveal melanomas reveals that amplification of CNKSR3 is correlated with improved patient survival.

Metastatic death from uveal melanoma occurs almost exclusively with tumors showing monosomy of chromosome 3. However, approximately 5% of patients with a disomy 3 uveal melanoma develop metastases, and a further 5% of monosomy 3 uveal melanoma patients exhibit disease-free survival for >5 years. In the present study, whole-genome microarrays were used to interrogate four clinically well-defined subgroups of uveal melanoma: i) disomy 3 uveal melanoma with long-term survival; ii) metastasizing monosomy 3 uveal melanoma; iii) metastasizing disomy 3 uveal melanoma; and iv) monosomy 3 uveal melanoma with long-term survival. Cox regression and Kaplan-Meier survival analysis identified that amplification of the CNKSR3 gene (log-rank, P = 0.022) with an associated increase in its protein expression (log-rank, P = 0.011) correlated with longer patient survival. Although little is known about CNKSR3, the correlation of protein expression with increased survival suggests a biological function in uveal melanoma, possibly working to limit metastatic progression of monosomy 3 uveal melanoma cells.

Sensitivity and Specificity of Different Prognostic Systems in Guiding Surveillance for Metastases in Uveal Melanoma.

Uveal melanoma (UM) metastasises in ~50% of patients, most frequently to the liver. Surveillance imaging can provide early detection of hepatic metastases; however, guidance regarding UM patient risk stratification for surveillance is unclear. This study compared sensitivity and specificity of four current prognostic systems, when used for risk stratification for surveillance, on patients treated at the Liverpool Ocular Oncology Centre (LOOC) between 2007-2016 (n = 1047). It found that the Liverpool Uveal Melanoma Prognosticator Online III (LUMPOIII) or Liverpool Parsimonious Model (LPM) offered greater specificity at equal levels of sensitivity than the American Joint Committee on Cancer (AJCC) system or monosomy 3 alone, and suggests guidance to achieve 95% sensitivity and 51% specificity (i.e., how to detect the same number of patients with metastases, while reducing the number of negative scans). For example, 180 scans could be safely avoided over 5 years in 200 patients using the most specific approach. LUMPOIII also offered high sensitivity and improved specificity over the AJCC in the absence of genetic information, making the result relevant to centres that do not perform genetic testing, or where such testing is inappropriate or fails. This study provides valuable information for clinical guidelines for risk stratification for surveillance in UM.

A Retrospective Analysis of 10 Years of Liver Surveillance Undertaken in Uveal Melanoma Patients Treated at the Supraregional "Liverpool Ocular Oncology Centre", UK.

PURPOSE: To determine liver screening frequency and modality in UM patients following primary treatment, and the characteristics of detected metastases. METHODS: A 10-year retrospective study of 615 UM patients undergoing liver surveillance in Liverpool. Information was collected from liver scan reports of these patients. RESULTS: Of 615 UM patients analyzed, there were 337 men (55%) and 278 women (45%). Median age at primary treatment was 61 years (range, 22-94). At study end, median follow-up was 5.1 years, with 375 patients (61%) alive and 240 deceased (39%). Of the deceased patients, 187 (78%) died due to metastatic UM; 24 (10%) deaths were due to other causes; and 29 (12%) patients died of unknown conditions. In total, 3854 liver scans were performed in the 615 UM patients, with a median of 6.2 scans per patient (range, 1-40). Liver MRI was most frequently performed (62.8%). In total, 229 (37%) UM patients developed metastases during the study period: 150 were detected via liver surveillance and 79 were observed post-mortem. CONCLUSIONS: Metastatic UM onset is related to the size and genetic profiles of the primary UM, and can be predicted using the model LUMPO3. Regular liver surveillance allowed for timely detection of metastases, and through metastasectomy can lead to prolongation of life in some patients.

Small High-Risk Uveal Melanomas Have a Lower Mortality Rate.

Our aim was to determine whether size impacts on the difference in metastatic mortality of genetically high-risk (monosomy 3) uveal melanomas (UM). We undertook a retrospective analysis of data from a patient cohort with genetically characterized UM. All patients treated for UM in the Liverpool Ocular Oncology Centre between 2007 and 2014, who had a prognostic genetic tumor analysis. Patients were subdivided into those with small (≤2.5 mm thickness) and large (>2.5 mm thickness) tumors. Survival analyses were performed using Gray rank statistics to calculate absolute probabilities of dying as a result of metastatic UM. The 5-year absolute risk of metastatic mortality of those with small monosomy 3 UM was significantly lower (23%) compared to the larger tumor group (50%) (p = 0.003). Small disomy 3 UM also had a lower absolute risk of metastatic mortality (0.8%) than large disomy 3 UM (6.4%) (p = 0.007). Hazard rates showed similar differences even with lead time bias correction estimates. We therefore conclude that earlier treatment of all small UM, particularly monosomy 3 UM, reduces the risk of metastatic disease and death. Our results would support molecular studies of even small UM, rather than 'watch-and-wait strategies'.

Cost-utility analysis of a decade of liver screening for metastases using the Liverpool Uveal Melanoma Prognosticator Online (LUMPO).

This paper outlines a method for cost-utility analysis of liver screening for metastases in patients with posterior uveal melanoma (UM). A semiparametric model of the cumulative incidence of onset of liver metastases was fitted to a retrospective data set of 615 subjects with clinical follow-up with respect to liver surveillance imaging and outcome. The model was internally validated via bootstrap resampling in terms of its discrimination and calibration performance. Receiver operating characteristics (ROC) were derived at different time points. The discrimination performances are consistent across time. The area under the ROC curve at 5 years post treatment was 0.85 [95% CI: 0.81-0.88]. A goodness-of-fit test gives χ2(10)=5.3,p=0.9 demonstrating no evidence against the null hypothesis of zero difference between observed and expected onset of metastatic events. Results showed that at 80% sensitivity, 87% of UM patients will avoid unnecessary radiological scans. This provides potential cost savings of between £46,000 and £97,000 per year to the National Health Service assuming 600 new cases per year.

Artificial neural networks estimating survival probability after treatment of choroidal melanoma.

PURPOSE: To describe neural networks predicting survival from choroidal melanoma (i.e., any uveal melanoma involving choroid) and to demonstrate the value of entering age, sex, clinical stage, cytogenetic type, and histologic grade into the predictive model. DESIGN: Nonrandomized case series. PARTICIPANTS: Patients resident in mainland Britain treated by the first author for choroidal melanoma between 1984 and 2006. METHODS: A conditional hazard estimating neural network (CHENN) was trained according to the Bayesian formalism with a training set of 1780 patients and evaluated with a test set of another 874 patients. Conditional hazard estimating neural network-generated survival curves were compared with those obtained with Kaplan-Meier analyses. A second model was created with information on chromosome 3 loss, using training and test sets of 211 and 140 patients, respectively. MAIN OUTCOME MEASURES: Comparison of CHENN survival curves with Kaplan-Meier analyses. Representative results showing all-cause survival and inferred melanoma-specific mortality, according to age, sex, clinical stage, cytogenetic type, and histologic grade. RESULTS: The predictive model plotted a survival curve with 95% credibility intervals for patients with melanoma according to relevant risk factors: age, sex, largest basal tumor diameter, ciliary body involvement, extraocular extension, tumor cell type, closed loops, mitotic rate, and chromosome 3 loss (i.e., monosomy 3). A survival curve for the age-matched general population of the same sex allowed estimation of the melanoma-related 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. Largest basal tumor diameter was most predictive of mortality in tumors showing histologic and cytogenetic features of high-grade malignancy. Ciliary body involvement and extraocular extension lost significance when cytogenetic and histologic data were included in the model. Patients with a monosomy 3 melanoma of a particular size were predicted to have shorter survival if their tumor showed epithelioid cells and closed loops. CONCLUSIONS: 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.

A web-based tool for the assessment of discrimination and calibration properties of prognostic models.

Prognostic models are developed to assist clinicians in making decisions regarding treatment and follow-up management. The accuracy of these models is often assessed either in terms of their discrimination performance or calibration but rarely both. In this paper, we describe the development of an online tool for discrimination using Harrell C index and calibration using a Hosmer-Lemeshow type analysis (http://clinengnhs.liv.ac.uk/AADP/AADP_Welcome.htm). We show examples of using the tool on real data. We highlight situations where the model performed well in terms of either discrimination or calibration but not both depending on the sample size of the test set. We conclude that prognostic models should be assessed both in terms of discrimination and calibration and that calibration analysis should be carried out numerically and graphically.

Prognostication of metastatic death in uveal melanoma patients: A Markov multi-state model.

BACKGROUND/AIMS: Uveal melanoma is fatal in almost 50% of patients. We previously developed a prognostic model to predict all-cause mortality. The aim of this study was to improve our model by predicting metastatic death as a cause-specific event distinct from other causes of death. METHODS: Patients treated in Liverpool were included if they resided in England, Scotland or Wales and if their uveal melanoma involved the choroid. They were flagged at the National Health Service Cancer Registry, which automatically informed us of the date and cause of death of any deceased patients. A semiparametric Markov multi-state model was fitted. Two different baseline hazard rates were assumed, with state transition-specific covariates. For both failure types, age at treatment and sex were used. For the metastatic death case, these factors were added: anterior margin position, largest basal tumour diameter, tumour thickness, extra-ocular extension, presence of epithelioid melanoma cells, presence of closed connective tissue loops, increased mitotic count, chromosome 3 loss, and chromosome 8q gain. Missing data required a multiple-imputation procedure. RESULTS: The cohort comprised 4161 patients, 893 of whom died of metastastic disease with another 772 dying of other causes. The optimism-corrected, bootstrapped C-index for metastatic death prediction was 0.86, denoting very good discriminative performance. Bootstrapped calibration curves at two and five years also showed very good performance. CONCLUSIONS: Our improved model provides reliable, personalised metastatic death prognostication using clinical, histological and genetic information, and it can be used as a decision support tool to individualize patient care in a clinical environment.

On the use of multi-objective evolutionary algorithms for survival analysis.

This paper proposes and evaluates a multi-objective evolutionary algorithm for survival analysis. One aim of survival analysis is the extraction of models from data that approximate lifetime/failure time distributions. These models can be used to estimate the time that an event takes to happen to an object. To use of multi-objective evolutionary algorithms for survival analysis has several advantages. They can cope with feature interactions, noisy data, and are capable of optimising several objectives. This is important, as model extraction is a multi-objective problem. It has at least two objectives, which are the extraction of accurate and simple models. Accurate models are required to achieve good predictions. Simple models are important to prevent overfitting, improve the transparency of the models, and to save computational resources. Although there is a plethora of evolutionary approaches to extract models for classification and regression, the presented approach is one of the first applied to survival analysis. The approach is evaluated on several artificial datasets and one medical dataset. It is shown that the approach is capable of producing accurate models, even for problems that violate some of the assumptions made by classical approaches.

The use of artificial neural networks in decision support in cancer: a systematic review.

Artificial neural networks have featured in a wide range of medical journals, often with promising results. This paper reports on a systematic review that was conducted to assess the benefit of artificial neural networks (ANNs) as decision making tools in the field of cancer. The number of clinical trials (CTs) and randomised controlled trials (RCTs) involving the use of ANNs in diagnosis and prognosis increased from 1 to 38 in the last decade. However, out of 396 studies involving the use of ANNs in cancer, only 27 were either CTs or RCTs. Out of these trials, 21 showed an increase in benefit to healthcare provision and 6 did not. None of these studies however showed a decrease in benefit. This paper reviews the clinical fields where neural network methods figure most prominently, the main algorithms featured, methodologies for model selection and the need for rigorous evaluation of results.

Modelling survival after treatment of intraocular melanoma using artificial neural networks and Bayes theorem.

This paper describes the development of an artificial intelligence (AI) system for survival prediction from intraocular melanoma. The system used artificial neural networks (ANNs) with five input parameters: coronal and sagittal tumour location, anterior tumour margin, largest basal tumour diameter and the cell type. After excluding records with missing data, 2331 patients were included in the study. These were split randomly into training and test sets. Date censorship was applied to the records to deal with patients who were lost to follow-up and patients who died from general causes. Bayes theorem was then applied to the ANN output to construct survival probability curves. A validation set with 34 patients unseen to both training and test sets was used to compare the AI system with Cox's regression (CR) and Kaplan-Meier (KM) analyses. Results showed large differences in the mean 5 year survival probability figures when the number of records with matching characteristics was small. However, as the number of matches increased to > 100 the system tended to agree with CR and KM. The validation set was also used to compare the system with a clinical expert in predicting time to metastatic death. The rms error was 3.7 years for the system and 4.3 years for the clinical expert for 15 years survival. For < 10 years survival, these figures were 2.7 and 4.2, respectively. We concluded that the AI system can match if not better the clinical expert's prediction. There were significant differences with CR and KM analyses when the number of records was small, but it was not known which model is more accurate.

Estimating prognosis for survival after treatment of choroidal melanoma.

Choroidal melanoma is fatal in about 50% of patients. This is because of metastatic disease, which usually involves the liver. Kaplan-Meier survival curves based only on tumor size and extent do not give a true indication of prognosis. This is because the survival prognosis of choroidal melanoma correlates not only with clinical stage but also with histologic grade, genetic type, and competing causes of death. We have developed an online tool that predicts survival using all these data also taking normal life-expectancy into account. The estimated prognosis is accurate enough to be relevant to individual patients. Such personalized prognostication improves the well-being of patients having an excellent survival probability, not least because it spares them from unnecessary screening tests. Such screening can be targeted at high-risk patients, so that metastases are detected sooner, thereby enhancing any opportunities for treatment. Concerns about psychological harm have proved exaggerated. At least in Britain, patients want to know their prognosis, even if this is poor. The ability to select patients with a high risk of metastasis improves prospects for randomised studies evaluating systemic adjuvant therapy aimed at preventing or delaying metastatic disease. Furthermore, categorization of tissue samples according to survival prognosis enables laboratory studies to be undertaken without waiting many years for survival to be measured. As a result of advances in histologic and genetic studies, biopsy techniques and statistics, prognostication has become established as a routine procedure in our clinical practice, thereby enhancing the care of patients with uveal melanoma.

Multiplex ligation-dependent probe amplification analysis of uveal melanoma with extraocular extension demonstrates heterogeneity of gross chromosomal abnormalities.

PURPOSE: To determine whether biopsy of extraocular extension of uveal melanoma (UM) is representative of the intraocular tumor with respect to copy number of chromosomes 1p, 3, 6, and 8. METHODS: Multiplex ligation-dependent probe amplification (MLPA) using the P027 assay was performed on formalin-fixed, paraffin-embedded sections from 10 UMs. The intraocular and extraocular parts of the tumor were microdissected and analyzed separately. RESULTS: Of the 10 UMs analyzed, seven showed heterogeneity for at least one chromosome arm; the most frequently heterogeneous chromosome arm was 6p. No heterogeneity of 8p was observed between the intraocular and extraocular areas of the tumor. One tumor showed monosomy 3 in the intraocular area of the tumor but loss of the 3q arm only for the extraocular area. CONCLUSIONS: Biopsy of an extraocular tumor extension may not be representative of the underlying UM with respect to chromosome 1p, 3, 6, and 8q abnormalities detectable by MLPA. These results suggest that for UM with extraocular extension, both the intraocular and the extraocular parts of the tumor should be sampled for accurate genetic prognostic testing.

Multiplex ligation-dependent probe amplification of conjunctival melanoma reveals common BRAF V600E gene mutation and gene copy number changes.

PURPOSE: To determine the occurrence of BRAF V600E gene mutations and copy number changes of all autosome arms and genes known to be frequently altered in tumorigenesis in primary and metastatic conjunctival melanomas (CoMs). METHODS: DNA (200 ng) was analyzed by three multiplex ligation-dependent probe amplification assays (P027 uveal melanoma, P036 human telomere, and P206 spitzoid melanoma). RESULTS: Eight of 16 primary tumor samples and 4 of 6 metastatic samples showed BRAF V600E gene mutations. CDKN1A and RUNX2 (both 6p21.2) were amplified in 11 and 16 of 21 primary CoMs, respectively. In metastatic CoMs, MLH1 (3p22.1) and TIMP2 (17q25.3) were frequently amplified, and MGMT (20q26.3) and ECHS1 (10q26.3) were frequently deleted. The BDH (3q), FLJ20265 (4p), OPRL1 (20q), and PAO (10q) genes, representing the telomeres of their respective chromosome arms in the P036 assay, were frequently amplified in metastatic CoMs. No statistically significant associations were identified between BRAF mutation or CDKN1A or RUNX2 amplification and sex, age, histologic cell type, or patient survival. CONCLUSIONS: No copy number changes were associated exclusively with metastatic CoMs. However, further investigation of the role of CDKN1A and RUNX2 in CoMs development and that of MLH1, TIMP2, MGMT, and ECHS1 in metastatic CoMs is warranted. Validation of the observed gene and chromosome arm copy number changes in a larger cohort of primary and metastatic CoMs is necessary to identify the patients at highest risk for CoMs metastasis.

Genetic heterogeneity in uveal melanoma assessed by multiplex ligation-dependent probe amplification.

PURPOSE: To determine intratumor genetic heterogeneity in uveal melanoma (UM) by multiplex ligation-dependent probe amplification (MLPA) in formalin-fixed, paraffin-embedded (FFPE) tumor tissues. METHODS: DNA was extracted from whole tumor sections and from two to nine different areas microdissected from 32 FFPE UMs. Thirty-one loci on chromosomes 1, 3, 6, and 8 were tested with MLPA for copy number changes. The tumor was considered heterogeneous at a locus if (1) the difference in dosage quotients (DQs) of any two areas was 0.2 or more, and (2) the DQs of the areas belonged to different ranges. RESULTS: Comparison of MLPA data obtained from microdissected areas of the UMs showed heterogeneity in 1 to 26 examined loci in 24 (75%) tumors, with only 25% of the tumors being homogeneous. Intratumor heterogeneity of 3p12.2, 6p21.2, and 8q11.23 was most common, occurring in >30% of the UMs. Gains of chromosome 3 were observed in four UMs, with three of these tumors showing the highest degree of heterogeneity. Copy number variation was associated with differences in tumor cell type, but not with differences in tumor pigmentation or reactive inflammation. UMs with genetic heterogeneity across multiple sample sites showed equivocal MLPA results when the whole tumor section was examined. These results suggest that different clones dilute MLPA results. CONCLUSIONS: Heterogeneity of chromosomal abnormalities of chromosomes 1, 3, 6, and 8 is present in most UMs. This heterogeneity causes equivocal MLPA results. One random tumor sample may not be representative of the whole tumor and, therefore, may be insufficient for prognostic testing.

Whole-genome microarray detects deletions and loss of heterozygosity of chromosome 3 occurring exclusively in metastasizing uveal melanoma.

PURPOSE: To detect deletions and loss of heterozygosity of chromosome 3 in a rare subset of fatal, disomy 3 uveal melanoma (UM), undetectable by fluorescence in situ hybridization (FISH). METHODS: Multiplex ligation-dependent probe amplification (MLPA) with the P027 UM assay was performed on formalin-fixed, paraffin-embedded (FFPE) whole tumor sections from 19 disomy 3 metastasizing UMs. Whole-genome microarray analyses using a single-nucleotide polymorphism microarray (aSNP) were performed on frozen tissue samples from four fatal disomy 3 metastasizing UMs and three disomy 3 tumors with >5 years' metastasis-free survival. RESULTS: Two metastasizing UMs that had been classified as disomy 3 by FISH analysis of a small tumor sample were found on MLPA analysis to show monosomy 3. No ubiquitous gene deletions of chromosome 3 were seen in the remaining 17 metastasizing disomy 3 UMs by MLPA. aSNP analysis revealed 95 deleted genes and 16 genes with loss of heterozygosity (LOH) on chromosome 3 in the disomy 3 metastasizing UMs that were not deleted or showing LOH in the nonmetastatic tumors. CONCLUSIONS: MLPA can detect monosomy 3 cell populations in FFPE whole tumor sections previously missed by FISH performed on small tumor samples. Consistent deletion and LOH of genes on chromosome 3 occur in metastasizing disomy 3 UM and are detectable by aSNP analysis. Ninety-five genes were found to be deleted, and 16 genes showed LOH exclusively in disomy 3 metastasizing UM, suggesting a potential role for these genes in UM metastasis.

Partial logistic artificial neural network for competing risks regularized with automatic relevance determination.

Time-to-event analysis is important in a wide range of applications from clinical prognosis to risk modeling for credit scoring and insurance. In risk modeling, it is sometimes required to make a simultaneous assessment of the hazard arising from two or more mutually exclusive factors. This paper applies to an existing neural network model for competing risks (PLANNCR), a Bayesian regularization with the standard approximation of the evidence to implement automatic relevance determination (PLANNCR-ARD). The theoretical framework for the model is described and its application is illustrated with reference to local and distal recurrence of breast cancer, using the data set of Veronesi (1995).

Automated post hoc removal of power-line and CRT frame pulse contamination from retinal and cortical evoked potentials (EPs).

Recordings of the ERG, PERG, VEP and their multi-focal variants are occasionally contaminated with harmonic noise arising from the mains supply and CRT monitors. These noise contributions can be modelled as distorted sinusoids and identified by means of non-linear multiple regression and removed: no a priori estimates of number or frequency of noise sources are required. This approach is termed noise cancellation and does not constitute any form of notch filter: the fidelity of the underlying waveform is preserved. Here the simple theory is illustrated in artificial datasets and then applied to clinical examples of PERG and VEP. The programming language used throughout is MatLab R13SP3 (Mathworks UK Ltd.).