Cancer screening articles from across Nature Portfolio

A recent study reported the development and validation of the Liver Artificial Intelligence Diagnosis System (LiAIDS), a fully automated system that integrates deep learning for the diagnosis of liver lesions on the basis of contrast-enhanced CT scans and clinical information. This tool improved diagnostic precision, surpassed the accuracy of junior radiologists (and equalled that of senior radiologists) and streamlined patient triage. These advances underscore the potential of artificial intelligence to enhance hepatology care, although challenges to widespread clinical implementation remain.

Implementation of organized low-dose computed tomography screening in over 4,000 individuals with high risk for lung cancer as part of the Ontario Lung Cancer Screening Pilot reported high cancer detection rates, early detection of cancer and low serious harms.

Developed on cytology images of hydrothorax and ascites from 57,220 cases at four hospitals, a deep-learning model shows high accuracy in tumor origin prediction and presents prognostic value when patient treatment is consistent with the cancer origin predicted by the model.

A recent study reported the development and validation of the Liver Artificial Intelligence Diagnosis System (LiAIDS), a fully automated system that integrates deep learning for the diagnosis of liver lesions on the basis of contrast-enhanced CT scans and clinical information. This tool improved diagnostic precision, surpassed the accuracy of junior radiologists (and equalled that of senior radiologists) and streamlined patient triage. These advances underscore the potential of artificial intelligence to enhance hepatology care, although challenges to widespread clinical implementation remain.

The benefits and potential harms of mammography-based screening for breast cancer are often a matter of debate. Here, I discuss the promises and limitations of a recent study that tested an artificial intelligence-based tool for the detection of breast cancer in digital mammograms in a large, prospective screening setting.

In a head-to-head study, a multitarget version of the widely used FIT (fecal immunochemical test) showed a higher cancer detection rate than the standard test.

The largest whole-genome sequencing study thus far has revealed myriad actionable alterations and potential biomarkers for 33 cancer types, but various logistical, technical and economic challenges must be overcome before this technique can become standard of care.

Prostate-Specific Antigen (PSA) is a first-line recommended serum biomarker used for prostate cancer (PCa) screening. However, this test has critical limitations in specificity within the gray zone of PSA from 4–10 ng/ml. The low specificity for clinically significant Gleason grade group 2–5 (GG2-5) PCa leads to unnecessary negative biopsies, over-detection of indolent cancers, and over-treatment, especially in Black men with favorable risk PCa. The urological community relies on secondary screening tools to augment specificity and increase sensitivity for GG2-5 PCa such as multiparametric magnetic resonance imaging of the prostate (MRI). Current National Comprehensive Cancer Network (NCCN) Early Detection guidelines specifically call for biomarkers and MRI tests to improve the specificity of PSA. Urologists rely on the negative predictive value (NPV) of the secondary screening tools for GG2-5 PCa to defer biopsies for men with elevated PSA. However, the NPV of any threshold chosen decreases in populations with a high prevalence of GG2-5 PCa. For most of these popular secondary biomarkers, specificity and NPV have not been assessed in Black men despite their known high risk. It is necessary to focus on other aspects of accuracy like specificity, NPV and the correlation of predicted risk versus observed risk (i.e., calibration) in populations with both high and low PCa risk (e.g., Black and Asian men). The impact of the new NCCN guideline-supported PSA ≥ 3.0 ng/ml threshold and the currently recommended thresholds of the secondary screening tools should be modeled or prospectively evaluated to see if they provide an acceptable risk-benefit ratio. Beyond discrimination/area under the curve, biomarkers should be evaluated for their calibration, specificity at high sensitivity, and their NPV for men with varied prevalence of GG2-5 PCa before being told to forego prostate biopsy. Clinical, biomarker, and imaging-integrated race stratified risk calculators and statistical modeling can be used to improve screening outcomes in minorities. Guidelines should also involve community providers to better reflect the resource limited settings within the US.