Data presented with the Department of Urology at Keio University School of Medicine at EAU 2026 in London, from 380 cancer patients and 204 healthy adults.
TOKYO, April 6, 2026 - Craif Inc., a bio-AI company developing a urine analysis platform for early detection of cancer and other diseases, presented joint research on urological cancer with the Department of Urology at Keio University School of Medicine at EAU 2026, the annual congress of the European Association of Urology, held in London from March 13 to 16, 2026. The presentation covered renal cell carcinoma, prostate cancer and urothelial carcinoma, and was made with Yuto Baba, Project Assistant Professor at the time of the congress, and Nobuyuki Tanaka, Senior Assistant Professor, at Keio.
Machine learning models built from microRNA carried in urinary exosomes separated patients with renal cell carcinoma, prostate cancer and urothelial carcinoma from healthy adults with areas under the curve of 0.92, 0.92 and 0.96. Separate models predicted whether disease recurred over 48 months of follow-up at AUC 0.89 for renal cell carcinoma, 0.78 for prostate cancer and 0.74 for urothelial carcinoma. In urothelial carcinoma, urinary microRNA alone showed some ability to estimate tumor immune status, at AUC 0.65 to 0.75.
The work runs on NANO IP (NANO Intelligence Platform), Craif's analytical platform for detecting biomarkers, including DNA and microRNA, in urine and other body fluids, combined with the company's machine learning models. EAU 2026 is one of the largest and most authoritative international urology congresses.
Status of these data
The findings summarized here were presented at a scientific congress. They have not been published in a peer-reviewed journal. The presentation was titled “Machine learning-based analysis of urinary exosomal microRNA: detection, prognostic assessment, and immune profiling of urological cancers.”
Detection: urine microRNA models separated three cancers from healthy controls
To work toward a non-invasive test for urological cancer, the team focused on microRNA carried in urinary exosomes and ran small RNA sequencing on urine from 204 healthy adults and from 380 patients with urological cancer, comprising renal cell carcinoma, prostate cancer and urothelial carcinoma, in whom the share with stage 2 or earlier disease ranged from 61.5 percent to 81.3 percent across the three cancer types.
Machine learning models built to identify each cancer type reached high accuracy against healthy controls: renal cell carcinoma at AUC 0.92, prostate cancer at AUC 0.92 and urothelial carcinoma at AUC 0.96. The sample is urine, so collection is painless.
Prognosis: recurrence prediction over 48 months
A prognostic model built from urinary microRNA predicted whether disease recurred over a 48-month follow-up period at AUC 0.89 in renal cell carcinoma, 0.78 in prostate cancer and 0.74 in urothelial carcinoma. The model included microRNAs that published work has linked to tumor metastasis, an immunosuppressive environment and treatment response.
Tumor immune status: an exploratory signal in urothelial carcinoma
Tumor immune status is associated with prognosis and with response to immune checkpoint inhibitors. To test whether urine could report on it, the team analyzed 102 urothelial carcinoma cases with immunohistochemistry for CD3, CD8, FOXP3 and PD-L1 in tumor tissue available and examined how those findings related to urinary microRNA. The analysis suggested that urinary microRNA alone may estimate tumor immune status with limited accuracy, with an AUC of 0.65 to 0.75.
Study overview
The collaboration developed a method for detecting urological cancer, specifically renal cell carcinoma, prostate cancer and urothelial carcinoma, at an early stage by analyzing microRNA in exosomes carried in urine. The models showed high accuracy in separating each of the three cancer types from healthy controls, and the results suggest they may also prove useful for prognosis, recurrence prediction and assessment of tumor immune status.
Glossary
- CD3: a protein that marks T cells, a type of immune cell. A high count of CD3-positive cells means many T cells have gathered around the tumor, which suggests how strongly the body is mounting an immune response to the cancer.
- CD8: a marker on the T cells that attack cancer cells directly. The more CD8-bearing cells inside a tumor, the more likely the immune system is fighting the cancer.
- FOXP3: a marker for regulatory T cells, which suppress immune activity. High FOXP3 expression may mean the tumor sits in an environment where it can more easily escape immune attack.
- PD-L1: a protein expressed on cancer cells and others that weakens immune cell activity. It is considered an important marker related to the effectiveness of immune checkpoint inhibitors.
About Craif
Craif is a bio-AI company founded in 2018 that develops non-invasive tests for the early detection of cancer. Its platform combines proprietary biomarker detection from urine and other bodily fluids with AI to assess disease risk. Building on one of the world's largest urine datasets, Craif aims to create a urine analysis platform that detects a wide range of diseases early, with cancer as the first step. The company is based in Tokyo and operates a wholly owned laboratory in San Diego, opened in 2026. Learn more at craif.com.
About EAU 2026
The EAU Annual Congress is one of the largest and most authoritative international meetings in urology, drawing urologists, researchers and specialists from medical device and pharmaceutical companies worldwide. It presents new research and clinical trial results across urological cancer and the wider field, shares evidence that feeds into clinical guidelines and updates to standard of care, and contributes to the advancement of urological medicine and clinical practice worldwide.
- Dates: March 13 to March 16, 2026
- Location: London, United Kingdom
- Official site: https://eaucongress.uroweb.org/
