methods to detect cancers early, especially cancers for which no screening tests exist currently (such as cancers of the liver and pancreas), and to eliminate precancerous lesions before they become cancerous while minimizing side effects. New Frontiers in Cancer Screening Researchers are working on novel methods and strategies that may improve detection of early cancers and/or precancerous lesions to reduce the death rate from cancer, while minimizing any potential harm from the procedure. Two areas of research with exciting new developments are the use of artificial intelligence and blood-based tests to detect cancer early. Realizing the Potential of Artificial Intelligence for Early Detection of Cancers There have been unprecedented advances in recent years in the use of artificial intelligence (AI) and machine learning (ML) in medicine, including for cancer detection. One of the several ways researchers are using AI and ML for cancer detection is in analyzing large amounts of imaging data collected for the purposes of cancer screening to recognize patterns that are otherwise time-consuming and difficult to discern by eye even by trained health care professionals (see Sidebar 24, p. 66). A recent study found that AI-assisted colonoscopy detected 21 percent more polyps (clumps of usually benign cells that build up on the lining of the colon and can be precursors to colon cancer) compared to a conventional expert-directed colonoscopy (310). Another study described the development of a deep learning model that can predict future risk of developing lung cancer within the next one to six years following a single LDCT scan (311). It is important to note that applying AI in medicine is an area of active research and not all studies have found AI-assisted improvements in cancer detection (312). Additional research is needed to understand the benefit of AI-driven applications in cancer medicine and whether AI mitigates or worsens health disparities. Some of the AI-driven medical devices and software have proven to be highly accurate in clinical trials when compared to current standard practices. In recent years, the number of FDA-approved AI-enhanced software systems for use in medicine, including early detection, has increased significantly (315). Here, we are using two recent examples—ProstatID and SKOUT—to highlight the progress in this exciting area of cancer research. In July 2022, FDA approved ProstatID, which uses AI to measure the volume of the prostate gland from scans obtained using traditional MRI and detect suspicious cancerous lesions. ProstatID is approved for use in a health care facility or hospital to assist trained radiologists in the detection and characterization of potentially cancerous lesions using MRI data. The approval was, in part, based on two clinical studies showing improved detection of prostate cancer and fewer false positives when radiologists used ProstatID. In September 2022, FDA approved SKOUT, a medical device that uses advanced computer vision technology designed to recognize polyps and suspicious tissue and provide realtime feedback to clinicians during colonoscopy. The approval was based on results from a large clinical study showing that detection of polyps and suspicious lesions was substantially improved when colonoscopy was aided with SKOUT compared to standard colonoscopy without the aid of SKOUT (719 versus 562 detections, respectively). The improved detection was even more pronounced for smaller polyps (44 percent increase when using SKOUT versus 29 percent increase when using standard colonoscopy) (313). Examples of recent FDA approvals discussed here underscore the potential applications of AI in the clinic to aid early detection and diagnostic purposes. Uses of AI in other aspects of cancer care—genomic characterization of tumors, drug discovery, and improved cancer surveillance—are also active areas of research (see Artificial Intelligence, p. 146) (316). Moving Toward Minimally Invasive Cancer Screening Most cancer screening tests in current use are designed to detect only one type of cancer. Some of the tests are invasive medical procedures, with potential health risks (see Sidebar 19, p. 57). Researchers are actively working to develop ways to detect cancer using tests that are less invasive and safer. Liquid biopsy is one such way in which a sample of blood, urine, or other body fluid is used to look for signs of cancer. Thanks to research, the knowledge that precancerous lesions and tumors shed a variety of materials (such as cancer cells and small pieces of DNA, RNA, or proteins) has led to the development of tests that can detect these materials in blood, urine, or other body fluids. Liquid biopsy approaches are already in routine use for making treatment decisions and/or monitoring if cancer has returned in patients who have already received cancer treatment, and can be especially beneficial for pediatric patients with cancer (317,318). Liquid biopsy procedures can detect abnormal cells and/or other materials from tumors that are circulating in the blood. It is critical that liquid biopsy tests are highly specific in detecting cancer-derived changes that are absent in healthy cells before these Screening for Early Detection AACR Cancer Progress Report 2023 65
RkJQdWJsaXNoZXIy NTkzMzk=