A new study suggests that a blood test utilizing artificial intelligence (AI) to identify cancer-related genetic alterations and protein biomarkers may aid in screening women for early indications of ovarian cancer.
According to research conducted by the Johns Hopkins Kimmel Cancer Center in partnerships with several institutions across the United States and Europe, a blood test that employs artificial intelligence (AI) could potentially identify early signs of ovarian cancer in women.
The research, shared on September 30 in the journal Cancer Discovery, published by the American Association for Cancer Research, applied AI analysis to DNA fragments along with two specific protein biomarkers to pinpoint women with ovarian cancer. The biomarkers, known as cancer antigen 125 (CA-125) and human epididymis protein 4 (HE4), have previously been recognized as indicators of ovarian cancer but were inadequate on their own for reliable detection. However, when these biomarkers are combined with AI-based identification of cancer-associated DNA fragment patterns in the bloodstream, the precision of screening increases, allowing for better differentiation between cancerous tumors and benign growths.
Victor E. Velculescu, M.D., Ph.D., the senior author of the study, noted, “By integrating artificial intelligence with cell-free DNA fragment analysis and a couple of protein biomarkers into a straightforward blood test, we enhanced the detection of ovarian cancer, even in earlier stages of the disease. This AI-driven methodology could provide an affordable and accessible solution for widespread ovarian cancer screening.”
Ovarian cancer ranks as the fifth leading cause of cancer-related deaths among women in the United States, with around a 50% five-year survival rate, as reported by the Centers for Disease Control and Prevention (CDC).
Co-first author Jamie Medina, Ph.D., a postdoctoral fellow at the Johns Hopkins Kimmel Cancer Center, stated, “Timely detection of ovarian cancer can save lives, but many women are diagnosed at advanced stages when survival odds significantly drop. The absence of distinct early symptoms or effective biomarkers has complicated early detection efforts.”
Previously, the researchers showcased the AI-powered DELFI (DNA Evaluation of Fragments for early Interception) method, which employs fragmentomics—a novel approach for liquid biopsies that improves the identification of DNA fragments in blood and is effective for lung cancer detection. This technology leverages the difference in DNA organization between healthy and cancer cells; in healthy cells, DNA is neatly arranged, while in cancer cells, it becomes disordered. Upon the death of healthy cells, they leave a predictable set of DNA fragments, whereas cancer cell death results in irregular and chaotic DNA fragments.
The recent study analyzed blood samples from 94 women diagnosed with ovarian cancer, 203 women with benign ovarian tumors, and 182 women without known ovarian growths. Participants were treated at hospitals in the Netherlands and Denmark. The researchers utilized the DELFI-Pro test, combining AI-based analysis of cell-free DNA with tests for CA-125 and HE4, for ovarian cancer screening. This test successfully identified significantly more ovarian cancer cases compared to tests of either protein alone, achieving high accuracy with minimal false positives. Specifically, it identified 72%, 69%, 87%, and 100% of ovarian cancer cases at stages I-IV, respectively. In contrast, CA-125 alone detected 34%, 62%, 63%, and 100% for the same stages.
To validate their findings, the researchers employed the test on a second group of American women, which included 40 patients with ovarian cancer, 50 with benign ovarian growths, and 22 without known lesions. Even in this smaller cohort, the test demonstrated similar success, detecting 73% of all cancers and 81% of high-grade serous ovarian carcinoma— the most aggressive type—with nearly no false positives among cancer-free women. The DELFI-Pro test was also effective in differentiating between benign growths and malignant tumors, an advantage over ultrasound exams.
“Ovarian cancers exhibit a distinct DNA fragmentation pattern absent in benign lesions,” remarked Akshaya Annapragada, co-first author and an M.D./Ph.D. student at the Johns Hopkins University School of Medicine. This differentiation is crucial because the next step for women with ovarian growths detected by ultrasound typically involves exploratory surgery. Utilizing “liquid biopsy” tests could prevent unnecessary surgeries for women with benign growths.
Velculescu and his team aim to further validate the utility of the test through larger randomized clinical trials, but he expressed optimism about the findings: “This study reinforces the advantages of using genome-wide, cell-free DNA fragmentation analysis and artificial intelligence for highly accurate cancer detection. Our results indicate that this combined strategy outperforms existing biomarkers in screening efficacy.”
Co-authors of the study included Sarah Short, Adrianna L. Bartolomucci, Dimitrios Mathios, Shashikant Koul, Noushin Niknafs, Michaël Noë, Zachariah H. Foda, Daniel C. Bruhm, Carolyn Hruban, Nicholas A. Vulpescu, Renu Dua, Jenna V. Canzoniero, Stephen Cristiano, Vilmos Adleff, Lori J. Sokoll, Stephen B. Baylin, Robert B. Scharpf, and Jillian Phallen from Johns Hopkins; Pien Lof, Daan van den Broek, Beatriz Carvalho, Gerrit A. Meijer, and Christine A.R. Lok from The Netherlands Cancer Institute; Euihye Jung, Heather Symecko, Susan M. Domchek, and Ronny Drapkin from the University of Pennsylvania; Michael F. Press from the University of Southern California; Dennis J. Slamon and Gottfried E. Konecny from UCLA; Christina Therkildsen from Hvidovre Hospital in Denmark; and Claus Lindbjerg Andersen from Aarhus University Hospital and Aarhus University in Denmark.
This research received funding from the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, the Gray Foundation, the SU2C in-Time Lung Cancer Interception Dream Team Grant, the Stand Up to Cancer-Dutch Cancer Society International Translational Cancer Research Dream Team Grant; as well as several other foundations and NIH grants.
Medina, Annapragada, Scharpf, Phallen, and Velculescu hold intellectual property rights for patent applications filed by Johns Hopkins University related to cell-free DNA and ovarian cancer. Additionally, some co-authors have involved patent applications that have been licensed to DELFI Diagnostics and other entities for cancer detection, a company that several authors co-founded.
