survex: an R package for explaining machine learning survival models.

SUMMARY: Due to their flexibility and superior performance, machine learning models frequently complement and outperform traditional statistical survival models. However, their widespread adoption is hindered by a lack of user-friendly tools to explain their internal operations and prediction rationales. To tackle this issue, we introduce the survex R package, which provides a cohesive framework for explaining any survival model by applying explainable artificial intelligence techniques. The capabilities of the proposed software encompass understanding and diagnosing survival models, which can lead to their improvement. By revealing insights into the decision-making process, such as variable effects and importances, survex enables the assessment of model reliability and the detection of biases. Thus, transparency and responsibility may be promoted in sensitive areas, such as biomedical research and healthcare applications. AVAILABILITY AND IMPLEMENTATION: survex is available under the GPL3 public license at https://github.com/modeloriented/survex and on CRAN with documentation available at https://modeloriented.github.io/survex.

A novel radiomics approach for predicting TACE outcomes in hepatocellular carcinoma patients using deep learning for multi-organ segmentation.

Transarterial chemoembolization (TACE) represent the standard of therapy for non-operative hepatocellular carcinoma (HCC), while prediction of long term treatment outcomes is a complex and multifactorial task. In this study, we present a novel machine learning approach utilizing radiomics features from multiple organ volumes of interest (VOIs) to predict TACE outcomes for 252 HCC patients. Unlike conventional radiomics models requiring laborious manual segmentation limited to tumoral regions, our approach captures information comprehensively across various VOIs using a fully automated, pretrained deep learning model applied to pre-TACE CT images. Evaluation of radiomics random survival forest models against clinical ones using Cox proportional hazard demonstrated comparable performance in predicting overall survival. However, radiomics outperformed clinical models in predicting progression-free survival. Explainable analysis highlighted the significance of non-tumoral VOI features, with their cumulative importance superior to features from the largest liver tumor. The proposed approach overcomes the limitations of manual VOI segmentation, requires no radiologist input and highlight the clinical relevance of features beyond tumor regions. Our findings suggest the potential of this radiomics models in predicting TACE outcomes, with possible implications for other clinical scenarios.

SATB2, CKAE1/AE3, and synaptophysin as a sensitive immunohistochemical panel for the detection of lymph node metastases of Merkel cell carcinoma.

Histopathological evaluation of lymph nodes in Merkel cell carcinoma has become crucial in progression estimation and treatment modification. This study was undertaken to determine the most sensitive immunohistochemical panel for detecting MCC nodal metastases. We included 56 patients with 102 metastatic MCC lymph nodes, which were tested with seven antibodies: cytokeratin (CKAE1/AE3), CK20, chromogranin A, synaptophysin, INSM1, SATB2, and neurofilament (NF). Tissue microarrays (TMA) composed of 2-mm tissue cores from each nodal metastasis were constructed. A semiquantitative 5-tier scoring system (0%, < 25%, 25-74%, 75-99%, 100% positive MCC cells with moderate to strong reactivity) was implemented. In the statistical assessment, we included Merkel cell polyomavirus (MCPyV) status and expression heterogeneity between lymph nodes from one patient. A cumulative percentage of moderate to strong expression ≥ 75% of tumoral cells was observed for single cell markers as follows: 91/102 (89.2%) SATB2, 85/102 (83%) CKAE1/AE3, 80/102 (78.4%) synaptophysin, 75/102 (75.5%) INSM1, 68/102 (66.7%) chromogranin A, 60/102 cases (58.8%) CK20, and 0/102 (0%) NF. Three markers presented a complete lack of immunoreactivity: 8/102 (7.8%) CK20, 7/102 (6.9%) chromogranin A, and 6/102 (5.9%) NF. All markers showed expression heterogeneity in lymph nodes from one patient; however, the most homogenous was INSM1. The probability of detecting nodal MCC metastases was the highest while using SATB2 as a first-line marker (89.2%) with subsequential adding CKAE1/AE3 (99%); these results were independent of MCPyV status. Synaptophysin showed a superior significance in confirming the neuroendocrine origin of metastatic cells. This comprehensive analysis allows us to recommend simultaneous evaluation of SATB2, CKAE1/AE3, and synaptophysin in the routine pathologic MCC lymph node protocol.

Ki67 is a better marker than PRAME in risk stratification of BAP1-positive and BAP1-loss uveal melanomas.

BACKGROUND: Accurate risk stratification of uveal melanoma (UM) patients is important for determining the interval and frequency of surveillance. Loss of BAP1 expression has been shown to be strongly associated with UM-related death and metastasis. METHODS: In this study of 164 enucleated UMs, we assessed the prognostic role of preferentially expressed antigen in melanoma (PRAME) expression and Ki67 proliferation index measured by digital quantitation using QuPath programme in patients with BAP1-positive and BAP1-loss UMs. RESULTS: In univariate analyses with log-rank tests and Kaplan-Meier curves, PRAME further stratified only overall survival (OS) in BAP1-positive and BAP1-loss tumour groups. However, Ki67 further stratified both OS and disease-free survival (DFS) in BAP1-positive and BAP1-loss tumour groups. In multivariate analyses, Ki67 percentage and BAP1 were independent survival predictors for both OS and DFS, whereas PRAME was not a significant covariate. In model comparisons, combined Ki67 and BAP1 performed better than combined PRAME and BAP1 in risk-stratifying patients for both OS and DFS. Ki67 was better than PRAME in risk stratification of BAP1-positive UMs. Low Ki67 index correlated with significantly prolonged DFS in BAP1-loss UMs. CONCLUSION: A panel of Ki67 and BAP1 could be a helpful risk stratification strategy for UM.

Machine learning models demonstrate that clinicopathologic variables are comparable to gene expression prognostic signature in predicting survival in uveal melanoma.

PURPOSE: Since molecular assays are not accessible to all uveal melanoma patients, we aim to identify cost-effective prognostic tool in risk stratification using machine learning models based on routine histologic and clinical variables. EXPERIMENTAL DESIGN: We identified important prognostic parameters in a discovery cohort of 164 enucleated primary uveal melanomas from 164 patients without prior therapies. We then validated the prognostic prediction of top important parameters identified in the discovery cohort using 80 uveal melanomas from the Tumor Cancer Genome Atlas database with available gene expression prognostic signature (GEPS). The performance of three different survival analysis models (Cox proportional hazards (CPH), random survival forest (RSF), and survival gradient boosting (SGB)) was compared against GEPS using receiver operating curves (ROC). RESULTS: In all three selection methods, BAP1 status, nucleoli size, age, mitotic rate per 1 mm2, and ciliary body infiltration were identified as significant overall survival (OS) predictors; and BAP1 status, nucleoli size, largest basal tumor diameter, tumor-infiltrating lymphocyte density, and tumor-associated macrophage density were identified as significant progression-free survival (PFS) predictors. ROC plots for the median survival time point showed that significant parameters in SGB studied model can predict OS better than GEPS. For PFS, SGB model performed similarly to GEPS. The time-dependent area under the curve (AUC) showed SGB model performing better than GEPS in predicting OS and metastatic risk. CONCLUSIONS: Our study shows that routine histologic and clinical variables are adequate for patient risk stratification in comparison with not readily accessible GEPS.

Amelanotic Uveal Melanomas Evaluated by Indirect Ophthalmoscopy Reveal Better Long-Term Prognosis Than Pigmented Primary Tumours-A Single Centre Experience.

(1) Background: There is a constant search for new prognostic factors that would allow us to accurately determine the prognosis, select the type of treatment, and monitor the patient diagnosed with uveal melanoma in a minimally invasive and easily accessible way. Therefore, we decided to evaluate the prognostic role of its pigmentation in a clinical assessment. (2) Methods: The pigmentation of 154 uveal melanomas was assessed by indirect ophthalmoscopy. Two groups of tumours were identified: amelanotic and pigmented. The statistical relationships between these two groups and clinical, pathological parameters and the long-term survival rate were analyzed. (3) Results: There were 16.9% amelanotic tumours among all and they occurred in younger patients (p = 0.022). In pigmented melanomas, unfavourable prognostic features such as: epithelioid cells (p = 0.0013), extrascleral extension (p = 0.027), macronucleoli (p = 0.0065), and the absence of BAP1 expression (p = 0.029) were statistically more frequently observed. Kaplan−Meier analysis demonstrated significantly better overall (p = 0.017) and disease-free (p < 0.001) survival rates for patients with amelanotic tumours. However, this relationship was statistically significant for lower stage tumours (AJCC stage II), and was not present in larger and more advanced stages (AJCC stage III). (4) Conclusions: The results obtained suggested that the presence of pigmentation in uveal melanoma by indirect ophthalmoscopy was associated with a worse prognosis, compared to amelanotic lesions. These findings could be useful in the choice of therapeutic and follow-up options in the future.