Cancer Risk Prediction Research Articles

Artificial intelligence methods applied to longitudinal data from electronic health records for prediction of cancer: a scoping review

BackgroundEarly detection and diagnosis of cancer are vital to improving outcomes for patients. Artificial intelligence (AI) models have shown promise in the early detection and diagnosis of cancer, but there is limited evidence on methods that fully exploit the longitudinal data stored within electronic health records (EHRs). This review aims to summarise methods currently utilised for prediction of cancer from longitudinal data and provides recommendations on how such models should be developed.MethodsThe review was conducted following PRISMA-ScR guidance. Six databases (MEDLINE, EMBASE, Web of Science, IEEE Xplore, PubMed and SCOPUS) were searched for relevant records published before 2/2/2024. Search terms related to the concepts “artificial intelligence”, “prediction”, “health records”, “longitudinal”, and “cancer”. Data were extracted relating to several areas of the articles: (1) publication details, (2) study characteristics, (3) input data, (4) model characteristics, (4) reproducibility, and (5) quality assessment using the PROBAST tool. Models were evaluated against a framework for terminology relating to reporting of cancer detection and risk prediction models.ResultsOf 653 records screened, 33 were included in the review; 10 predicted risk of cancer, 18 performed either cancer detection or early detection, 4 predicted recurrence, and 1 predicted metastasis. The most common cancers predicted in the studies were colorectal (n = 9) and pancreatic cancer (n = 9). 16 studies used feature engineering to represent temporal data, with the most common features representing trends. 18 used deep learning models which take a direct sequential input, most commonly recurrent neural networks, but also including convolutional neural networks and transformers. Prediction windows and lead times varied greatly between studies, even for models predicting the same cancer. High risk of bias was found in 90% of the studies. This risk was often introduced due to inappropriate study design (n = 26) and sample size (n = 26).ConclusionThis review highlights the breadth of approaches to cancer prediction from longitudinal data. We identify areas where reporting of methods could be improved, particularly regarding where in a patients’ trajectory the model is applied. The review shows opportunities for further work, including comparison of these approaches and their applications in other cancers.

Ensemble machine learning models for lung cancer incidence risk prediction in the elderly: a retrospective longitudinal study

BackgroundIdentifying high risk factors and predicting lung cancer incidence risk are essential to prevention and intervention of lung cancer for the elderly. We aim to develop lung cancer incidence risk prediction model in the elderly to facilitate early intervention and prevention of lung cancer.MethodsWe stratified the population into six subgroups according to age and gender. For each subgroup, random forest, extreme gradient boosting, deep neural networks, support vector machine, multiple logistic regression and deep Q network (DQN) models were developed and validated. Models were trained and tested using samples from 2000 to 2015 and independent external validated through those from 2016 to 2019. The suitable model for lung cancer risk prediction and high risk factors identification was chosen based on internal validation and independent external validation.ResultsThe DQN model achieved the optimal prediction performance in stratified subgroups, with AUROC ranging from 0.937 to 0.953, recall ranging from 0.932 to 0.943, F2-score ranging from 0.929 to 0.946, precision ranging from 0.926 to 0.952, F1-score ranging from 0.933 to 0.963 and RMSE ranging from 0.21 to 0.27. SHAP values were supplied for model interpretability. High risk factors of lung cancer incidence were identified in the elderly. Men ≥ 65 carrying C > A/G > T mutation had the highest lung cancer incidence decrease of 39.5% after five years quitting in stratified elderly groups, which were 1.83 times more than women ≥ 65 not carrying C > A/G > T mutation.ConclusionsThe DQN model may be suitable for identifying high risk factors and predicting lung cancer risk with high performance. The proposed intervention and diagnosis pathways could be used for early screening and intervention before the occurrence of lung cancer, which could help oncologists develop targeted intervention strategies for the stratified elderly to reduce lung cancer incidence and improve therapeutic effect. Proposed method could also be used in predicting the risk of other chronic diseases to help conduct intervention and reduce incidence.

AI image analysis as the basis for risk-stratified screening.

Artificial intelligence (AI) has emerged as a transformative tool in breast cancer screening, with two distinct applications: computer-aided cancer detection (CAD) and risk prediction. While AI CAD systems are slowly finding its way into clinical practice to assist radiologists or make independent reads, this review focuses on AI risk models, which aim to predict a patient's likelihood of being diagnosed with breast cancer within a few years after negative screening. Unlike AI CAD systems, AI risk models are mainly explored in research settings without widespread clinical adoption. This review synthesizes advances in AI-driven risk prediction models, from traditional imaging biomarkers to cutting-edge deep learning methodologies and multimodal approaches. Contributions by leading researchers are explored with critical appraisal of their methods and findings. Ethical, practical, and clinical challenges in implementing AI models are also discussed, with an emphasis on real-world applications. This review concludes by proposing future directions to optimize the adoption of AI tools in breast cancer screening and improve equity and outcomes for diverse populations.

A benchmark of deep learning approaches to predict lung cancer risk using national lung screening trial cohort

Deep learning (DL) methods have demonstrated remarkable effectiveness in assisting with lung cancer risk prediction tasks using computed tomography (CT) scans. However, the lack of comprehensive comparison and validation of state-of-the-art (SOTA) models in practical settings limits their clinical application. This study aims to review and analyze current SOTA deep learning models for lung cancer risk prediction (malignant-benign classification). To evaluate our model’s general performance, we selected 253 out of 467 patients from a subset of the National Lung Screening Trial (NLST) who had CT scans without contrast, which are the most commonly used, and divided them into training and test cohorts. The CT scans were preprocessed into 2D-image and 3D-volume formats according to their nodule annotations. We evaluated ten 3D and eleven 2D SOTA deep learning models, which were pretrained on large-scale general-purpose datasets (Kinetics and ImageNet) and radiological datasets (3DSeg-8, nnUnet and RadImageNet), for their lung cancer risk prediction performance. Our results showed that 3D-based deep learning models generally perform better than 2D models. On the test cohort, the best-performing 3D model achieved an AUROC of 0.86, while the best 2D model reached 0.79. The lowest AUROCs for the 3D and 2D models were 0.70 and 0.62, respectively. Furthermore, pretraining on large-scale radiological image datasets did not show the expected performance advantage over pretraining on general-purpose datasets. Both 2D and 3D deep learning models can handle lung cancer risk prediction tasks effectively, although 3D models generally have superior performance than their 2D competitors. Our findings highlight the importance of carefully selecting pretrained datasets and model architectures for lung cancer risk prediction. Overall, these results have important implications for the development and clinical integration of DL-based tools in lung cancer screening.

Selective adsorption of unmethylated DNA on ZnO nanowires for separation of methylated DNA.

DNA methylation is a crucial epigenetic modification used as a biomarker for early cancer progression. However, existing methods for DNA methylation analysis are complex, time-consuming, and prone to DNA degradation. This work demonstrates selective capture of unmethylated DNAs using ZnO nanowires without chemical or biological modifications, thereby concentrating methylated DNA, particularly those with high methylation levels that can predict cancer risk. We observe varying affinities between methylated and unmethylated DNA on ZnO nanowires, which may be influenced by differences in hydrogen bonding strength, potentially related to the effects of methylation on DNA strand behavior, including self-aggregation and stretching inhibition. As a result, the nanowire-based microfluidic device effectively collects unmethylated DNA, leading to a significantly increased ratio of methylated to unmethylated DNA, particularly for collecting low-concentration methylated DNA. This simplified microfluidic device, composed of ZnO nanowires, enables direct separation of specific methylated DNA, offering a potential approach for DNA methylation mapping in clinical disease diagnostics.

Prediction of pancreatic cancer risk in patients with new-onset diabetes

Prediction of pancreatic cancer risk in patients with new-onset diabetes

A web-based tool for cancer risk prediction for middle-aged and elderly adults using machine learning algorithms and self-reported questions.

From a global perspective, China is one of the countries with higher incidence and mortality rates for cancer. Our objective is to create an online cancer risk prediction tool for middle-aged and elderly Chinese adults by leveraging machine learning algorithms and self-reported data. Drawing from a cohort of 19,798 participants aged 45 and above from the China Health and Retirement Longitudinal Study (2011 - 2018), we employed nine machine learning algorithms (LR: Logistic Regression, Adaboost: Adaptive Boosting, SVM: Support Vector Machine, RF: Random Forest, GNB: Gaussian Naive Bayes, GBM: Gradient Boosting Machine, LGBM: Light Gradient Boosting Machine, XGBoost: eXtreme Gradient Boosting, KNN: K - Nearest Neighbors), which are mainly used for classification and regression tasks, to construct predictive models for various cancers. Utilizing non-invasive self-reported predictors encompassing demographic, educational, marital, lifestyle, health history, and other factors, we focused on predicting "Cancer or Malignant Tumour" outcomes. The types of cancers that can be predicted mainly include lung cancer, breast cancer, cervical cancer, colorectal cancer, gastric cancer, esophageal cancer, and other rare cancers. The developed tool, MyCancerRisk, demonstrated significant performance, with the Random Forest algorithm achieving an AUC of 0.75 and ACC of 0.99 using self-reported variables. Key predictors identified include age, self-rated health, sleep patterns, household heating sources, childhood health status, living conditions, and smoking habits. MyCancerRisk aims to serve as a preventative screening tool, encouraging individuals to undergo testing and adopt healthier behaviours to mitigate the public health impact of cancer. Our study also sheds light on unconventional predictors, such as housing conditions, offering valuable insights for refining cancer prediction models.

Comparison of PREMM5 and PREMMplus Risk Assessment Models to Identify Lynch Syndrome.

Clinical risk assessment models can identify patients with hereditary cancer susceptibility, but it is unknown how multigene cancer syndrome prediction models compare with syndrome-specific models in assessing risk for individual syndromes such as Lynch syndrome (LS). Our aim was to compare PREMMplus (a 19-gene cancer risk prediction model) with PREMM5 (a LS gene-specific model) for LS identification. We analyzed data from two cohorts of patients undergoing germline testing from a commercial laboratory (n = 12,020) and genetics clinic (n = 6,232) with personal and/or family histories of LS-associated cancer. Individual PREMMplus and PREMM5 scores were calculated for all patients. Using a score cutoff of 2.5%, we calculated the sensitivity, specificity, positive predictive value, and negative predictive value (NPV) for identifying LS with each model. Overall ability to discriminate LS carriers from noncarriers was measured using receiver operating characteristic (ROC)-AUC. PREMMplus had higher sensitivity than PREMM5 in the laboratory- (63.7% [95% CI, 57.0 to 70.0] v 89.2% [95% CI, 84.4 to 93.0]) and clinic-based cohorts (60.8% [95% CI, 52.7 to 68.4] v 90.5% [95% CI, 84.8 to 94.6]). NPV was ≥98.8% for both models in both cohorts. PREMM5 had superior discriminatory capacity to PREMMplus in the laboratory- (ROC-AUC, 0.81 [95% CI, 0.77 to 0.84] v 0.71 [95% CI, 0.67 to 0.75]) and clinic-based cohorts (ROC-AUC, 0.79 [95% CI, 0.75 to 0.84] v 0.68 [95% CI, 0.64 to 0.73]). Both PREMM5 and PREMMplus demonstrated high NPVs (>98%) in LS discrimination across all patient cohorts, and both models may be used to identify individuals at risk of LS. The choice of which model to use can be based on the goals of risk assessment and patient population.

NOTIFICATION: Development of a Novel Approach for Breast Cancer Prediction and Early Detection Using Minimally Invasive Procedures and Molecular Analysis: How Cytomorphology Became a Breast Cancer Risk Predictor

NOTIFICATION: Development of a Novel Approach for Breast Cancer Prediction and Early Detection Using Minimally Invasive Procedures and Molecular Analysis: How Cytomorphology Became a Breast Cancer Risk Predictor

Polygenic score distribution differences across European ancestry populations: implications for breast cancer risk prediction

BackgroundThe 313-variant polygenic risk score (PRS313) provides a promising tool for clinical breast cancer risk prediction. However, evaluation of the PRS313 across different European populations which could influence risk estimation has not been performed.MethodsWe explored the distribution of PRS313 across European populations using genotype data from 94,072 females without breast cancer diagnosis, of European-ancestry from 21 countries participating in the Breast Cancer Association Consortium (BCAC) and 223,316 females without breast cancer diagnosis from the UK Biobank. The mean PRS was calculated by country in the BCAC dataset and by country of birth in the UK Biobank. We explored different approaches to reduce the observed heterogeneity in the mean PRS across the countries, and investigated the implications of the distribution variability in risk prediction.ResultsThe mean PRS313 differed markedly across European countries, being highest in individuals from Greece and Italy and lowest in individuals from Ireland. Using the overall European PRS313 distribution to define risk categories, leads to overestimation and underestimation of risk in some individuals from these countries. Adjustment for principal components explained most of the observed heterogeneity in the mean PRS. The mean estimates derived when using an empirical Bayes approach were similar to the predicted means after principal component adjustment.ConclusionsOur results demonstrate that PRS distribution differs even within European ancestry populations leading to underestimation or overestimation of risk in specific European countries, which could potentially influence clinical management of some individuals if is not appropriately accounted for. Population-specific PRS distributions may be used in breast cancer risk estimation to ensure predicted risks are correctly calibrated across risk categories.

Investigating the added value of incorporatingmammographic density to an integrated breastcancer risk model with questionnaire-based riskfactors and polygenic risk score.

Incorporation of mammographic density to breast cancer risk models could improve risk stratification to tailor screening and prevention strategies according to risk. Robust evaluation of the value of adding mammographic density to models with comprehensive information on questionnaire-based risk factors and polygenic risk score is needed to determine its effectiveness in improving risk stratification of such models. We used the Individualized Coherent Absolute Risk Estimator (iCARE) tool for risk model building and validation to incorporate density to a previously validated literature-based model with questionnaire-based risk factors and a 313-variant polygenic risk score (PRS). The model was evaluated for calibration and discrimination in three prospective cohorts of European-ancestry women (1,468 cases, 19,104 controls): US-based Nurses' Health Study (NHS I and II) and Mayo Mammography Health Study (MMHS); and Sweden-based Karolinska Mammography Project for Risk Prediction of Breast Cancer (KARMA) study. Analyses were done separately for women younger (NHS II, KARMA) and older than 50 years (NHS I, MMHS, KARMA). Improvements in terms of risk stratification and reclassification proportions were assessed among European-ancestry women aged 50-70 years in US and Sweden. For women younger and older than 50 years, the model with questionnaire-based risk factors, PRS and density was generally well calibrated across risk with some evidence of miscalibration at the extremes of the risk distribution. Incorporation of density led to modest improvements risk discrimination beyond the model with questionnaire-based risk factors and PRS: the area under the curve (AUC) among younger women was 67.0% (95% CI: 63.5-70.6%) vs. 65.6% (95% CI: 61.9-69.3%) for models with and without density; and 66.1% (95% CI 64.4-67.8%) vs. 65.5% (95% CI: 63.8-67.2%) among older women. The model with density identified 18.4% of US women 50-70 years old ≥ 3% 5-year predicted risk (threshold used for recommending risk-reducing medication in the US), with 42.4% of future cases expected to occur in this group. At this threshold, 7.9% of US women were reclassified by adding density to the model, resulting in the identification of 2.8% of additional future cases. The model with density identified 10.3% of Swedish women ≥ 3% 5-year predicted risk, with 29.4% of future cases expected to occur in this group. At this threshold, 5.3% of women were reclassified with the addition of density, leading to the identification of an additional 4.4% of future cases. Integrating density with questionnaire-based risk factors and PRS could potentially identify more women of European-ancestry with elevated risk of breast cancer in the United States and Sweden. Further investigations of the integrated model in non-European ancestry populations are needed prior to considering clinical applications.

Longitudinal interpretability of deep learning based breast cancer risk prediction.

Deep-learning-based models have achieved state-of-the-art breast cancer risk (BCR) prediction performance. However, these models are highly complex, and the underlying mechanisms of BCR prediction are not fully understood. Key questions include whether these models can detect breast morphologic changes that lead to cancer. These findings would boost confidence in utilizing BCR models in practice and provide clinicians with new perspectives. In this work, we aimed to determine when oncogenic processes in the breast provide sufficient signal for the models to detect these changes.&#xD;&#xD;Approach: In total, 1210 screening mammograms were collected for patients screened at different times before the cancer was screen-detected and 2400 mammograms for patients with at least ten years of follow-up. MIRAI, a BCR risk prediction model, was used to estimate the BCR. Attribution Heterogeneity was defined as the relative difference between the attributions obtained from the right and left breasts using one of the eight interpretability techniques. Model reliance on the side of the breast with cancer was quantified with AUC. The Mann-Whitney U test was used to check for significant differences in median absolute Attribution Heterogeneity between cancer patients and healthy individuals.&#xD;&#xD;Results: All tested attribution methods showed a similar longitudinal trend, where the model reliance on the side of the breast with cancer was the highest for the 0-1 Years-To-Cancer interval (AUC=0.85-0.95), dropped for the 1-3 Years-To-Cancer interval (AUC=0.64-0.71), and remained above the threshold for random performance for the 3-5 Years-To-Cancer interval (AUC=0.51-0.58). For all eight attribution methods, the median values of absolute attribution heterogeneity were significantly larger for patients diagnosed with cancer at one point (p<0.01).&#xD;&#xD;Significance: Interpretability of BCR prediction has revealed that long-term predictions (beyond three years) are most likely based on typical breast characteristics, such as breast density; for mid-term predictions (one to three years), the model appears to detect early signs of tumor development, while for short-term predictions (up to a year), the BCR model essentially functions as a breast cancer detection model.&#xD.

Cancer Risk in Thyroid Nodules: An Analysis of Over 1000 Consecutive FNA Biopsies Performed in a Single Canadian Institution

Objective: To determine the cancer risk in thyroid nodules using ACR TI-RADS. Methods: A retrospective analysis of all thyroid biopsies was performed over a 3-year period (2021 to 2023). Variables including gender, age, history of thyroid cancer or neck irradiation, nodule size and location, TR level, and sonographic features such as punctate echogenic foci (PEF), a very hypoechoic appearance, taller-than-wide shape, and suspected extrathyroidal extension were analyzed. Results: A total of 1140 nodules were assessed in 993 patients, including 740 females (74.5%) and 253 males (25.5%). The mean patient age was 57.1 ± 15.4 years. Variables significantly associated with nodule malignancy included (1) younger age, (2) a prior history of thyroid cancer or neck irradiation, (3) a higher TR level, (4) a taller-than-wide shape in nodules &lt;1 cm, (5) PEF, (6) a very hypoechoic appearance, and (5) suspected extrathyroidal extension (p &lt; 0.05). Gender, nodule location and size were not associated with a higher cancer risk (p &gt; 0.05). Malignancy was found in 40.7% of TR5, 4.8% of TR4, 0.3% of TR3, and 0% of TR1 and 2 nodules. The odds ratios (ORs) for cancer were as follows: TR4 or 5, OR = 19; PEF, OR = 11; a very hypoechoic appearance, OR = 13.3; and suspected extrathyroidal extension, OR = 27.2 (p &lt; 0.01). Conclusions: Higher TR levels, PEF, a very hypoechoic appearance, and suspected extrathyroidal extension are important features for predicting cancer risk. These findings affirm the effectiveness of ACR TI-RADS in nodule risk stratification.

A multimodal machine learning model for the stratification of breast cancer risk.

Machine learning models for the diagnosis of breast cancer can facilitate the prediction of cancer risk and subsequent patient management among other clinical tasks. For the models to impact clinical practice, they ought to follow standard workflows, help interpret mammography and ultrasound data, evaluate clinical contextual information, handle incomplete data and be validated in prospective settings. Here we report the development and testing of a multimodal model leveraging mammography and ultrasound modules for the stratification of breast cancer risk based on clinical metadata, mammography and trimodal ultrasound (19,360 images of 5,216 breasts) from 5,025 patients with surgically confirmed pathology across medical centres and scanner manufacturers. Compared with the performance of experienced radiologists, the model performed similarly at classifying tumours as benign or malignant and was superior at pathology-level differential diagnosis. With a prospectively collected dataset of 191 breasts from 187 patients, the overall accuracies of the multimodal model and of preliminary pathologist-level assessments of biopsied breast specimens were similar (90.1% vs 92.7%, respectively). Multimodal models may assist diagnosis in oncology.

Using New Technologies to Analyze Gut Microbiota and Predict Cancer Risk.

The gut microbiota significantly impacts human health, influencing metabolism, immunological responses, and disease prevention. Dysbiosis, or microbial imbalance, is linked to various diseases, including cancer. It is crucial to preserve a healthy microbiome since pathogenic bacteria, such as Escherichia coli and Fusobacterium nucleatum, can cause inflammation and cancer. These pathways can lead to the formation of tumors. Recent advancements in high-throughput sequencing, metagenomics, and machine learning have revolutionized our understanding of the role of gut microbiota in cancer risk prediction. Early detection is made easier by machine learning algorithms that improve the categorization of cancer kinds based on microbiological data. Additionally, the investigation of the microbiome has been transformed by next-generation sequencing (NGS), which has made it possible to fully profile both cultivable and non-cultivable bacteria and to understand their roles in connection with cancer. Among the uses of NGS are the detection of microbial fingerprints connected to treatment results and the investigation of metabolic pathways implicated in the development of cancer. The combination of NGS with machine learning opens up new possibilities for creating customized medicine by enabling the development of diagnostic tools and treatments that are specific to each patient's microbiome profile, even in the face of obstacles like data complexity. Multi-omics studies reveal microbial interactions, biomarkers for cancer detection, and gut microbiota's impact on cancer progression, underscoring the need for further research on microbiome-based cancer prevention and therapy.

Liver Function Biomarkers and Lung Cancer Risk: A Prospective Cohort Study in the UK Biobank.

As the primary organ of metabolism and detoxification, the liver may contribute to the pathogenesis of lung cancer. We aimed to illuminate the intricate link between liver function biomarkers and lung cancer risk, as well as delineate the role of smoking behavior within this association. We investigated the associations of seven liver function biomarkers levels (alkaline phosphatase [ALP], alanine transaminase [ALT], total bilirubin [TBIL], albumin [ALB], gamma-glutamyltransferase [GGT], aspartate transaminase [AST], and total protein [TP]) with lung cancer risk across the UK Biobank (N = 337 499) through restricted cubic splines and Cox proportional hazards models. Moreover, Mendelian randomization (MR) was utilized to evaluate the causal effect of smoking behavior on these biomarkers. Then a lung cancer risk prediction model was developed among smokers by backward stepwise logistic regression. During a median follow-up of 13.3 years, 3003 lung cancer cases were identified. We found ALP levels positively associated with lung cancer risk, whereas ALT, TBIL, ALB, and AST were inversely correlated; TP exhibited a U-shaped association, whereas GGT displayed a mirrored J-shaped relationship. These associations were amplified among smokers. MR analysis indicated that smoking behavior could increase ALP (odds ratio [OR]: 1.05) and GGT (OR: 1.15) levels while decreasing TBIL (OR: 0.92), ALB (OR: 0.92), and TP (OR: 0.96) levels. The lung cancer risk model incorporating these biomarkers in smokers demonstrated robust discrimination. Our finding provides perspectives and evidences towards the intricate crosstalk between the hepatic and pulmonary systems, as well as the processes through which tobacco catalyzes lung carcinogenesis.

Genetic Variants and Haplotype Structures in the CASC Gene Family to Predict Cancer Risk: A Bioinformatics Study.

The cancer susceptibility (CASC) gene family of long noncoding RNAs (lncRNAs) plays an important role in cancer. The aim of this study was to identify genetic variants and haplotype structures of CASC genes associated with cancer risk. Genome-wide association studies (GWAS) significant variants (p ≤ 5 × 10-8) on CASC family genes were identified from the GWAS Catalog-EMBL-EBI, and then cancer-associated variants on CASC genes were extracted. These variants were functionally analyzed, including lncRNA:miRNA binding sites, Regulomedb scores, and eQTL. The 1000 Genome Project genotyping data Phase III were used to identify haplotypic blocks. Finally, the genes associated with them were examined for expression and gene-gene correlation analyses using OncoDB. There were six haplotypic blocks in four genes. The GC, TA, and AGAC haplotypes are located in the CASC8 gene and increase the risk of prostate cancer, breast cancer, and colorectal cancer, respectively. The CA haplotype in the CASC15 gene increases the risk of neuroblastoma, AA haplotype in the CASC16 gene increases the risk of breast cancer, and ACGATG haplotype in the CASC17 gene increases the risk of prostate cancer (p ≤ 5 × 10-8). Their genes are interrelated and their expression is increased in these cancers. The rs2294214 is associated with skin cancer and has positive effects on five CASC15:miRNA binding sites. The rs3803662 is located in CASC16:miRNA binding sites, which has positive effects on hsa-miR-4475 and hsa-miR-7845-5p and negative effects on hsa-miR-4524a-3p and hsa-miR-4524b-3p. These haplotypic structures and lncRNA:miRNA:SNP interactions on CASC family lncRNAs reveal novel genetic associations between CASC genes and various cancers.

Plasma prolactin and postmenopausal breast cancer risk: a pooled analysis of four prospective cohort studies.

Prolactin, a hormone produced by the pituitary gland, regulates breast development and may contribute to breast cancer etiology. However, most epidemiologic studies of prolactin and breast cancer have been restricted to single, often small, study samples with limited exploration of effect modification. The Biomarkers in Breast Cancer Risk Prediction consortium includes 8,279 postmenopausal women sampled from four prospective cohort studies, of whom 3,441 were diagnosed with invasive breast cancer after enrollment. Prolactin concentrations were measured for all study participants on plasma samples collected when all women were postmenopausal and before any breast cancer diagnosis using ELISA assays. Pooled, unconditional logistic regression models, adjusted for confounders, estimated odd ratios (OR) for associations of prolactin and postmenopausal breast cancer risk overall and stratified by breast cancer risk factors. Higher plasma prolactin concentrations were positively associated with postmenopausal breast cancer risk (> 13.2 ng/mL vs. < 7.9 ng/mL, OR: 1.20, 95% CI: 1.06, 1.36; P-trend < 0.001). Although associations did not appear to vary by time since blood draw or most breast cancer risk factors, associations were primarily observed in current users of postmenopausal hormones at blood draw (> 13.2 ng/mL vs. < 7.9 ng/mL, current users, OR: 1.58, 95% CI: 1.27, 1.96, P-trend < 0.001; non-current users, OR: 1.08, 95% CI: 0.93, 1.27, P-trend = 0.11; P-heterogeneity = 0.06). Prolactin may be a risk factor for postmenopausal breast cancer, particularly in the context of postmenopausal hormone use. Investigations of prolactin interactions with other hormonal factors may further inform breast cancer etiology.

A bibliometric review of predictive modelling for cervical cancer risk.

Cervical cancer represents a significant public health challenge, particularly affecting women's health globally. This study aims to advance the understanding of cervical cancer risk prediction research through a bibliometric analysis. The study identified 800 records from Scopus and Web of Science databases, which were reduced to 142 unique records after removing duplicates. Out of 100 abstracts assessed, 42 were excluded based on specific criteria, resulting in 58 studies included in the bibliometric review. Multiple scoping methods such as thematic analysis, citation analysis, bibliographic coupling, natural language processing, Latent Dirichlet Allocation and other visualisation techniques were used to analyse related publications between 2013 and 2024. The key findings revealed the importance of interdisciplinary collaboration in cervical cancer risk prediction, integrating expertise from mathematical disciplines, biomedical health, healthcare practitioners, public health, and policy. This approach significantly enhanced the accuracy and efficiency of cervical cancer detection and predictive modelling by adopting advanced machine learning algorithms, such as random forests and support vector machines. The main challenges were the lack of external validation on independent datasets and the need to address model interpretability to ensure healthcare providers understand and trust the predictive models. The study revealed the importance of interdisciplinary collaboration in cervical cancer risk prediction. It made recommendations for future research to focus on increasing the external validation of models, improving model interpretability, and promoting global research collaborations to enhance the comprehensiveness and applicability of cervical cancer risk prediction models.

Prostate Cancer Risk Prediction Model Using Clinical and Magnetic Resonance Imaging–Related Findings: Impact of Combining Lesions' Locations and Apparent Diffusion Coefficient Values

Objectives The aims of the study are to develop a prostate cancer risk prediction model that combines clinical and magnetic resonance imaging (MRI)–related findings and to assess the impact of adding Prostate Imaging-Reporting and Data System (PI-RADS) ≥3 lesions-level findings on its diagnostic performance. Methods This 3-center retrospective study included prostate MRI examinations performed with clinical suspicion of clinically significant prostate cancer (csPCa) between 2018 and 2022. Pathological diagnosis within 1 year after the MRI was used to diagnose csPCa. Seven clinical, 3 patient-level MRI-related, and 4 lesion-level MRI-related findings were extracted. After feature selection, 2 logistic regression models with and without lesions-level findings were created using data from facility I and II (development cohort). The area under the receiver operating characteristic curve (AUC) between the 2 models was compared in the PI-RADS ≥3 population in the development cohort and Facility III (validation cohort) using the Delong test. Interfacility differences of the selected predictive variables were evaluated using the Kruskal-Wallis test or chi-squared test. Results Selected lesion-level features included the peripheral zone involvement and apparent diffusion coefficient (ADC) values. The model with lesions-level findings had significantly higher AUC than the model without in 655 examinations in the development cohort (0.81 vs 0.79, respectively, P = 0.005), but not in 553 examinations in the validation cohort (0.77 vs 0.76, respectively). Large interfacility differences were seen in the ADC distribution (P &lt; 0.001) and csPCa proportion in PI-RADS 3–5 (P &lt; 0.001). Conclusions Adding lesions-level findings improved the csPCa discrimination in the development but not the validation cohort. Interfacility differences impeded model generalization, including the distribution of reported ADC values and PI-RADS score-level csPCa proportion.