Summary
A Nature Medicine study reports that urine cell-free RNA sequencing detected localized bladder cancer with 95% sensitivity at 90% specificity. The researchers also investigated its potential for detecting residual disease and predicting response to bladder treatments.
Researchers report that a urine cell-free RNA assay detected localized bladder cancer with 95% sensitivity at 90% specificity in a study published in Nature Medicine on 2 October 2026. The method, called uRARE-seq, was also used to study molecular evidence of residual disease and signals associated with response to treatment. The study analysed 683 urine samples from 515 people; its main detection and treatment-response analyses focused on bladder cancer.
Contents
- How the urine RNA assay works
- Bladder cancer detection results
- Treatment-response signals and next steps
How the urine RNA assay works
Cell-free RNA consists of RNA fragments present in urine supernatant, rather than RNA extracted from the cells that settle out of a urine sample. These fragments are short, so the researchers adapted a sequencing approach designed to analyse fragmented RNA. uRARE-seq uses random priming to prepare RNA for sequencing and affinity capture to enrich selected gene sequences.
The targeted panel covered 4,782 genes, including 3,834 that were rarely detected in urine from non-cancer controls. The researchers designed the panel to help identify signals from cancer while also measuring genes relevant to genitourinary tissues and immune cells. Because the approach analyses gene expression, it can provide information about biological activity as well as the presence of cancer-associated transcripts. The authors describe it as a tumour-naive method: it does not require a matched tumour sample to classify a urine sample.
Bladder cancer detection results
The detection model was developed using urine from 151 people with bladder cancer and 100 non-cancer controls. At its selected threshold, it achieved 95% sensitivity at 90% specificity. Sensitivity measures the proportion of people with cancer identified by the test; specificity measures the proportion of controls correctly classified as negative. An independent validation cohort included 142 people with bladder cancer and 140 controls, and the researchers reported nearly identical performance using the threshold set in the training cohort.
In a paired comparison involving 93 samples from people with bladder cancer, uRARE-seq detected 89 cases (95.7%), while urine cytology detected 32 (34.4%). This comparison concerns detection among those cancer cases, not the assay’s performance in a general screening population. The researchers also developed models to distinguish lower-grade from higher-grade disease and non-muscle-invasive from muscle-invasive cancer. In independent validation cohorts, those models had areas under the receiver operating characteristic curve of 0.84 and 0.75, respectively.
Treatment-response signals and next steps
The team also studied 36 people with intermediate- or high-risk non-muscle-invasive bladder cancer who underwent surgery followed by Bacillus Calmette–Guérin (BCG) immunotherapy. Urine was collected before surgery, before BCG treatment and at follow-up. In this study, a molecular complete response meant that the detection model no longer found bladder-cancer-associated RNA in urine. Such clearance after surgery or BCG was associated with lower recurrence risk than continued detection.
The RNA profiles also differed between groups. Samples from people with a molecular complete response after BCG were enriched for immune-related signals, including T-cell signatures. Samples from people with detectable disease after BCG showed more proliferation-related signals. Using these patterns, the researchers developed a treatment-response model with an area under the curve of 0.93 in leave-one-out cross-validation. Its scores were associated with recurrence in separate cohorts of 57 people treated with BCG and 38 treated with intravesical chemotherapy.
These findings point to a possible way of using urine RNA not only to detect cancer but also to characterise its activity. However, the treatment groups were not randomised, so the study does not establish that the assay can select which treatment will work better for an individual. The case–control study drew participants from two affiliated, geographically adjacent institutions, and the authors say larger prospective studies are needed to establish clinical utility.