Predicting Bone Metastasis Using Gene Expression-Based Machine Learning Models

Artificial Intelligence in Detection, Management, and Prognosis of Bone Metastasis: A Systematic Review

BACKGROUND

Metastasis commonly occur in the bone tissue. Artificial intelligence (AI) has become increasingly prevalent in the medical sector as support in decision-making, diagnosis, and treatment processes. The objective of this systematic review was to assess the reliability of AI systems in clinical, radiological, and pathological aspects of bone metastases.

METHODS

We included studies that evaluated the use of AI applications in patients affected by bone metastases. Two reviewers performed a digital search on 31 December 2023 on PubMed, Scopus, and Cochrane library and extracted authors, AI method, interest area, main modalities used, and main objectives from the included studies.

RESULTS

We included 59 studies that analyzed the contribution of computational intelligence in diagnosing or forecasting outcomes in patients with bone metastasis. Six studies were specific for spine metastasis. The study involved nuclear medicine (44.1%), clinical research (28.8%), radiology (20.4%), or molecular biology (6.8%). When a primary tumor was reported, prostate cancer was the most common, followed by lung, breast, and kidney.

CONCLUSIONS

Appropriately trained AI models may be very useful in merging information to achieve an overall improved diagnostic accuracy and treatment for metastasis in the bone. Nevertheless, there are still concerns with the use of AI systems in medical settings. Ethical considerations and legal issues must be addressed to facilitate the safe and regulated adoption of AI technologies. The limitations of the study comprise a stronger emphasis on early detection rather than tumor management and prognosis as well as a high heterogeneity for type of tumor, AI technology and radiological techniques, pathology, or laboratory samples involved.

Gene expression in metastatic breast cancer-patterns in primary tumors and metastatic tissue with prognostic potential

Background: Metastatic breast cancer (MBC) is the main cause of breast cancer-related death. The outcome of MBC varies, and there is a lack of biomarkers to aid in prognostication. The primary aim of this study was to evaluate the prognostic value of gene expression (GEX) signatures in the primary tumor (PT) and distant metastasis (DM) for progression-free survival (PFS) and overall survival (OS). The secondary aim was to describe GEX changes through MBC evolution and to identify MBC subtypes. Methods: RNA was extracted from the PT, lymph node metastasis (LNM), and DM from MBC patients in a prospective observational study (n = 142; CTC-MBC NCT01322893) and was subjected to GEX analysis retrospectively using the NanoString Breast Cancer 360™ panel. 31 continuous GEX variables in DMs and PTs were analyzed for PFS and OS by Cox regression analysis and Kaplan-Meier estimates. Multivariable Cox regressions were adjusted for number of DM sites and CTCs, visceral metastasis, ECOG status, age at MBC diagnosis and, in additional analyses, PAM50 subtype. Differential GEX analyses and Euclidean distances were used to describe subgroup differences and visualize within-patient heterogeneity. Results: Compared to DM GEX, GEX of the PT was at least equally useful for predicting MBC outcome. The strongest marker for a favorable PFS, both when expressed in the PT and the DM was AR, even after adjustment for prognostic markers including PAM50. GEX signatures related to hormone responsiveness, including ESR1, FOXA1, PGR, and AR were favorable prognostic markers, and the p53 signature was unfavorable for PFS when expressed in PT or DM. The previously published PAM50MET signature was prognostic for both PFS and OS. We established five distinct DM GEX profiles where two associated with liver and bone metastases, respectively. Finally, we identified four DM GEX profiles able to identify MBCs with poor OS in this cohort. Conclusion: GEX of both DM and PT are useful in MBC prognostication. GEX of AR adds prognostic information for MBC. Our descriptive analyses illuminate the biological differences between MBCs in relation to outcome and metastatic site.

Reclassified the phenotypes of cancer types and construct a nomogram for predicting bone metastasis risk: A pan-cancer analysis

BACKGROUND

Numerous of models have been developed to predict the bone metastasis (BM) risk; however, due to the variety of cancer types, it is difficult for clinicians to use these models efficiently. We aimed to perform the pan-cancer analysis to create the cancer classification system for BM, and construct the nomogram for predicting the BM risk.

METHODS

Cancer patients diagnosed between 2010 and 2018 in the Surveillance, Epidemiology, and End Results (SEER) database were included. Unsupervised hierarchical clustering analysis was performed to create the BM prevalence-based cancer classification system (BM-CCS). Multivariable logistic regression was applied to investigate the possible associated factors for BM and construct a nomogram for BM risk prediction. The patients diagnosed between 2017 and 2018 were selected for validating the performance of the BM-CCS and the nomogram, respectively.

RESULTS

A total of 50 cancer types with 2,438,680 patients were included in the construction model. Unsupervised hierarchical clustering analysis classified the 50 cancer types into three main phenotypes, namely, categories A, B, and C. The pooled BM prevalence in category A (17.7%; 95% CI: 17.5%-17.8%) was significantly higher than that in category B (5.0%; 95% CI: 4.5%-5.6%), and category C (1.2%; 95% CI: 1.1%-1.4%) (p < 0.001). Advanced age, male gender, race, poorly differentiated grade, higher T, N stage, and brain, lung, liver metastasis were significantly associated with BM risk, but the results were not consistent across all cancers. Based on these factors and BM-CCS, we constructed a nomogram for predicting the BM risk. The nomogram showed good calibration and discrimination ability (AUC in validation cohort = 88%,95% CI: 87.4%-88.5%; AUC in construction cohort = 86.9%,95% CI: 86.8%-87.1%). The decision curve analysis also demonstrated the clinical usefulness.

CONCLUSION

The classification system and prediction nomogram may guide the cancer management and individualized BM screening, thus allocating the medical resources to cancer patients. Moreover, it may also have important implications for studying the etiology of BM.

Artificial Intelligence in Bone Metastasis Imaging: Recent Progresses from Diagnosis to Treatment - A Narrative Review

The introduction of artificial intelligence (AI) represents an actual revolution in the radiological field, including bone lesion imaging. Bone lesions are often detected both in healthy and oncological patients and the differential diagnosis can be challenging but decisive, because it affects the diagnostic and therapeutic process, especially in case of metastases. Several studies have already demonstrated how the integration of AI-based tools in the current clinical workflow could bring benefits to patients and to healthcare workers. AI technologies could help radiologists in early bone metastases detection, increasing the diagnostic accuracy and reducing the overdiagnosis and the number of unnecessary deeper investigations. In addition, radiomics and radiogenomics approaches could go beyond the qualitative features, visible to the human eyes, extrapolating cancer genomic and behavior information from imaging, in order to plan a targeted and personalized treatment. In this article, we want to provide a comprehensive summary of the most promising AI applications in bone metastasis imaging and their role from diagnosis to treatment and prognosis, including the analysis of future challenges and new perspectives.

An Explainable AI Approach for Breast Cancer Metastasis Prediction Based on Clinicopathological Data

OBJECTIVE

Breast Cancer is the most prevalent cancer and the first cause of cancer deaths among women worldwide. In 90% of the cases, mortality is related to distant metastasis. Computer-aided prognosis systems using machine learning models have been widely used to predict breast cancer metastasis. Despite that, these systems still face several challenges. First, the models are generally biased toward the majority class due to datasets unbalance. Second, their increased complexity is associated with decreased interpretability which causes clinicians to distrust their prognosis.

METHODS

To tackle these issues, we have proposed an explainable approach for predicting breast cancer metastasis using clinicopathological data. Our approach is based on cost-sensitive CatBoost classifier and utilises LIME explainer to provide patient-level explanations.

RESULTS

We used a public dataset of 716 breast cancer patients to assess our approach. The results demonstrate the superiority of cost-sensitive CatBoost in precision (76.5%), recall (79.5%), and f1-score (77%) over classical and boosting models. The LIME explainer was used to quantify the impact of patient and treatment characteristics on breast cancer metastasis, revealing that they have different impacts ranging from high impact like the non-use of adjuvant chemotherapy, and moderate impact including carcinoma with medullary features histological type, to low impact like oral contraception use. The code is available at https://github.com/IkramMaouche/CS-CatBoost Conclusion: Our approach serves as a first step toward introducing more efficient and explainable computer-aided prognosis systems for breast cancer metastasis prediction.

SIGNIFICANCE

This approach could help clinicians understand the factors behind metastasis and assist them in proposing more patient-specific therapeutic decisions.

A deep learning model predicts the presence of diverse cancer types using circulating tumor cells

Circulating tumor cells (CTCs) are cancer cells that detach from the primary tumor and intravasate into the bloodstream. Thus, non-invasive liquid biopsies are being used to analyze CTC-expressed genes to identify potential cancer biomarkers. In this regard, several studies have used gene expression changes in blood to predict the presence of CTC and, consequently, cancer. However, the CTC mRNA data has not been used to develop a generic approach that indicates the presence of multiple cancer types. In this study, we developed such a generic approach. Briefly, we designed two computational workflows, one using the raw mRNA data and deep learning (DL) and the other exploiting five hub gene ranking algorithms (Degree, Maximum Neighborhood Component, Betweenness Centrality, Closeness Centrality, and Stress Centrality) with machine learning (ML). Both workflows aim to determine the top genes that best distinguish cancer types based on the CTC mRNA data. We demonstrate that our automated, robust DL framework (DNNraw) more accurately indicates the presence of multiple cancer types using the CTC gene expression data than multiple ML approaches. The DL approach achieved average precision of 0.9652, recall of 0.9640, f1-score of 0.9638 and overall accuracy of 0.9640. Furthermore, since we designed multiple approaches, we also provide a bioinformatics analysis of the gene commonly identified as top-ranked by the different methods. To our knowledge, this is the first study wherein a generic approach has been developed to predict the presence of multiple cancer types using raw CTC mRNA data, as opposed to other models that require a feature selection step.

OncoRTT: Predicting novel oncology-related therapeutic targets using BERT embeddings and omics features

Late-stage drug development failures are usually a consequence of ineffective targets. Thus, proper target identification is needed, which may be possible using computational approaches. The reason being, effective targets have disease-relevant biological functions, and omics data unveil the proteins involved in these functions. Also, properties that favor the existence of binding between drug and target are deducible from the protein’s amino acid sequence. In this work, we developed OncoRTT, a deep learning (DL)-based method for predicting novel therapeutic targets. OncoRTT is designed to reduce suboptimal target selection by identifying novel targets based on features of known effective targets using DL approaches. First, we created the “OncologyTT” datasets, which include genes/proteins associated with ten prevalent cancer types. Then, we generated three sets of features for all genes: omics features, the proteins’ amino-acid sequence BERT embeddings, and the integrated features to train and test the DL classifiers separately. The models achieved high prediction performances in terms of area under the curve (AUC), i.e., AUC greater than 0.88 for all cancer types, with a maximum of 0.95 for leukemia. Also, OncoRTT outperformed the state-of-the-art method using their data in five out of seven cancer types commonly assessed by both methods. Furthermore, OncoRTT predicts novel therapeutic targets using new test data related to the seven cancer types. We further corroborated these results with other validation evidence using the Open Targets Platform and a case study focused on the top-10 predicted therapeutic targets for lung cancer.

Rates of metastatic prostate cancer in newly diagnosed patients: Numbers needed to image according to risk level

BACKGROUND

The numbers needed to image to identify pelvic lymph node and/or distant metastases in newly diagnosed prostate cancer (PCa) patients according to risk level are unknown.

METHODS

Relying on Surveillance, Epidemiology, and End Results (2010-2016), we tabulated rates and proportions of patients with (a) lymph node or (b) distant metastases according to National Comprehensive Cancer Network (NCCN) risk level and calculated the number needed to image (NNI) for both endpoints. Multivariable logistic regression analyses were performed.

RESULTS

Of 145,939 newly diagnosed PCa patients assessable for analyses of pelvic lymph node metastases (cN1), 4559 (3.1%) harbored cN1 stage: 13 (0.02%), 18 (0.08%), 63 (0.3%), 512 (2.8%), and 3954 (14.9%) in low, intermediate favorable, intermediate unfavorable, high, and very high-risk levels. These resulted in NNI of 4619, 1182, 319, 35, and 7, respectively. Of 181,109 newly diagnosed PCa patients assessable for analyses of distant metastases (M1_a-c ), 8920 (4.9%) harbored M1_a-c stage: 50 (0.07%), 45 (0.1%), 161 (0.5%), 1290 (5.1%), and 7374 (22.0%) in low, intermediate favorable, intermediate unfavorable, high, and very high-risk. These resulted in NNI of 1347, 602, 174, 20, and 5, respectively.

CONCLUSIONS

Our observations perfectly validated the NCCN recommendations for imaging in newly diagnosed high and very high-risk PCa patients. However, in unfavorable intermediate-risk PCa patients, in whom bone and soft tissue imaging is recommended, the NNI might be somewhat elevated to support routine imaging in clinical practice.

MetastaSite: Predicting metastasis to different sites using deep learning with gene expression data

Deep learning has massive potential in predicting phenotype from different omics profiles. However, deep neural networks are viewed as black boxes, providing predictions without explanation. Therefore, the requirements for these models to become interpretable are increasing, especially in the medical field. Here we propose a computational framework that takes the gene expression profile of any primary cancer sample and predicts whether patients' samples are primary (localized) or metastasized to the brain, bone, lung, or liver based on deep learning architecture. Specifically, we first constructed an AutoEncoder framework to learn the non-linear relationship between genes, and then DeepLIFT was applied to calculate genes' importance scores. Next, to mine the top essential genes that can distinguish the primary and metastasized tumors, we iteratively added ten top-ranked genes based upon their importance score to train a DNN model. Then we trained a final multi-class DNN that uses the output from the previous part as an input and predicts whether samples are primary or metastasized to the brain, bone, lung, or liver. The prediction performances ranged from AUC of 0.93-0.82. We further designed the model's workflow to provide a second functionality beyond metastasis site prediction, i.e., to identify the biological functions that the DL model uses to perform the prediction. To our knowledge, this is the first multi-class DNN model developed for the generic prediction of metastasis to various sites.

Identification of Signature Genes and Construction of an Artificial Neural Network Model of Prostate Cancer

This study aimed to establish an artificial neural network (ANN) model based on prostate cancer signature genes (PCaSGs) to predict the patients with prostate cancer (PCa). In the present study, 270 differentially expressed genes (DEGs) were identified between PCa and normal prostate (NP) groups by differential gene expression analysis. Next, we performed Metascape gene annotation, pathway and process enrichment analysis, and PPI enrichment analysis on all 270 DEGs. Then, we identified and screened out 30 PCaSGs based on the random forest analysis and constructed an ANN model based on the gene score matrix consisting of 30 PCaSGs. Lastly, analysis of microarray dataset GSE46602 showed that the accuracy of this model for predicating PCa and NP samples was 88.9 and 78.6%, respectively. Our results suggested that the ANN model based on PCaSGs can be used for effectively predicting the patients with PCa and will be helpful for early PCa diagnosis and treatment.