In a single hospital in Chengdu, researchers asked a quiet question of the blood: whether a handful of genetic signals could reveal which men with prostate cancer that had already spread would soon stop responding to hormone-blocking therapy, and how soon that loss would come. Their answer, a three-gene signature drawn from patients enrolled over one year, suggests that the body may leave early traces of a coming change before any scan shows it. The findings remain a promising sign rather than a settled tool, shaped by a single center and a small group of patients, yet they point toward a futu
Three-Gene Blood Signature Predicts Castration Resistance in Metastatic Prostate Cancer
Patients with newly diagnosed metastatic hormone-sensitive prostate cancer face progression to castration-resistant disease, with high-risk patients progressing significantly faster.
Three genes in the blood, sorted by who would progress first
Mark
What's the headline here? A blood test for predicting who will stop responding to hormone therapy?
Mimi
Roughly, yes. Three genes in peripheral blood, SLC25A39, SLC6A8 and YBX3, sorted newly diagnosed metastatic patients into high and low risk. High-risk patients had a median of 466 days before castration resistance.
Luke
And the low-risk group never reached a median, so we don't have a number for them. Also, that hazard ratio of 17.19 is striking, but it comes from a small cohort.
Mark
How does it compare with what clinicians use now?
Mimi
The authors set it against LATITUDE high-risk criteria. Specificity and accuracy were much higher, for example 94.7% versus 26.3% specificity at 545 days.
Luke
But sensitivity wasn't significantly different. It mostly means fewer false alarms, not more cases caught.
Mark
Is the validation solid?
Luke
Not yet. The validation cohort was 44 patients, and the AUC of 0.83 has a confidence interval running from 0.62 to 1.00. That's wide.
Mimi
And the population is single-center, all East Asian, with no external data set.
Mark
What about the SLC6A8 lab work?
Mimi
An inhibitor, ompenaclid, slowed cancer cells and cut glycolysis markers.
Luke
In one cell line. It's suggestive of a mechanism, not proof that blocking it helps patients. The treatment ideas, like skipping docetaxel, are explicitly hypothetical.
Der Puls
- A small set of three blood genes is being tested as an early warning for when metastatic prostate cancer will stop responding to hormone therapy.
- Patients labeled high risk by the model progressed dramatically faster, with a hazard ratio of 17.19 that sharpens the stakes of every classification.
- The signature outperformed the established LATITUDE high-risk criteria on specificity and accuracy, meaning it correctly spared far more patients from false alarms.
- The validation cohort was small and drawn from one East Asian center, so the tool is pressing toward a multicenter prospective trial before anyone can rely on it.
- Researchers are also probing SLC6A8 directly, showing that blocking it slowed cancer cell growth and disrupted their sugar metabolism in the lab.
In a single hospital in Chengdu, researchers asked a quiet question of the blood: whether a handful of genetic signals could reveal which men with prostate cancer that had already spread would soon stop responding to hormone-blocking therapy, and how soon that loss would come. Their answer, a three-gene signature drawn from patients enrolled over one year, suggests that the body may leave early traces of a coming change before any scan shows it. The findings remain a promising sign rather than a settled tool, shaped by a single center and a small group of patients, yet they point toward a future in which treatment decisions might be guided by what the blood is already whispering.
Between February 2024 and February 2025, a team at Sichuan Cancer Hospital in Chengdu screened 230 men newly diagnosed with metastatic prostate cancer, enrolling 157 of them. Each had disease that had already spread when it was found, and each was receiving hormone-blocking treatment. The researchers wanted to know whether a few genes in a simple blood sample could foretell which of these men would stop responding to that therapy, and how soon.
Starting from single-cell data and their own patient measurements, the team narrowed five candidate genes to three: SLC25A39, SLC6A8, and YBX3. They split the patients into development and validation groups and followed them for a median of 611 days, during which 54 progressed. The model predicted one-year progression with an AUC of 0.89 in development and 0.83 in validation. Men labeled high risk moved into castration resistance far sooner, with a median of 466 days, while the low-risk group had not reached that point at all.
Compared with the LATITUDE trial's high-risk criteria, the signature was markedly more specific and accurate, sparing many patients from being wrongly labeled, while sensitivity was broadly similar. The authors also examined SLC6A8 in tumor data from a large public cohort and in laboratory cells, where blocking it slowed growth and disrupted sugar metabolism.
The authors suggested the signature might one day guide intensified therapy, earlier radiotherapy, or decisions about docetaxel, and help select aggressive cases for trials. They were careful to call these ideas hypothetical. The study's limits are real: one center, an East Asian population, a small validation group, and short follow-up. A prospective multicenter study across several Chinese hospitals is now being planned.
Between February 1, 2024, and February 1, 2025, researchers at Sichuan Cancer Hospital in Chengdu screened 230 men with newly diagnosed metastatic prostate cancer. They enrolled 157 of them, drew blood, and sequenced the RNA. The question was whether a few genes in that blood could say which patients would stop responding to hormone-blocking treatment, and how soon.
The results were published October 14, 2026, in Frontiers in Oncology. All the patients had de novo metastatic hormone-sensitive prostate cancer, meaning it had already spread when it was found. Each had at least one measurable lesion, an ECOG performance status of 2 or lower, and received androgen deprivation therapy plus an androgen receptor pathway inhibitor, with or without docetaxel. Progression to metastatic castration-resistant prostate cancer was judged by Prostate Cancer Clinical Trials Working Group 3 criteria.
The team began with single-cell transcriptomic data from the GSE288878 data set to find genes tied to castration resistance. Intersecting those with genes that differed in their own cohort left five candidates: SLC25A39, GRB10, SELENBP1, SLC6A8, and YBX3. Elastic net–regularized Cox regression then narrowed these to three: SLC25A39, SLC6A8, and YBX3. The patients were randomly split 7:3 into development and validation groups, and 147 finished follow-up, 103 and 44 respectively. At a median follow-up of 611 days, 54 had progressed.
For predicting one-year castration-resistance-free survival, the model reached an AUC of 0.89 in the development cohort and 0.83 in the validation cohort. C-index values were 0.849 and 0.783, and Brier scores were 0.056 and 0.060. Patients labeled high risk progressed much faster than those labeled low risk (hazard ratio 17.19; P <.0001). Median time free of castration resistance was 466 days in the high-risk group and had not been reached in the low-risk group. In multivariate analysis, higher blood levels of SLC25A39 and SLC6A8 were independent predictors of progression.
The authors also compared the model with the high-risk criteria from the LATITUDE trial. At 365 days, specificity was 82.1% versus 51.3% and accuracy 81.0% versus 52.4%. At 545 days, specificity was 94.7% versus 26.3%, accuracy 81.1% versus 29.3%, and balanced accuracy 75.0% versus 30.6%. Sensitivity did not differ significantly, at 67.2% versus 66.4% at 365 days and 55.3% versus 34.9% at 545 days. Decision curve analysis showed more net benefit than treating everyone or no one at threshold probabilities of roughly 0.2 to 0.6.
The work also looked at SLC6A8 itself. In The Cancer Genome Atlas prostate data (n = 498), high SLC6A8 in tumor tissue independently predicted shorter progression-free survival. In C4-2B castration-resistant cells, the SLC6A8 inhibitor ompenaclid slowed proliferation and promoted apoptosis in a dose-dependent way. It lowered the glycolysis-related genes HK2, PFKP, and ALDH1B1, and reduced lactic acid, pyruvic acid, and fructose-1,6-bisphosphate.
Corresponding author Shengke Yang, of Sichuan Cancer Hospital's urology department, and colleagues proposed uses for the signature. It might guide triplet therapy or earlier metastasis-directed radiotherapy where clinical risk is equivocal, such as oligometastatic bone disease or borderline PSA. It might also support omitting docetaxel in high-volume disease with a favorable profile, and help enrich trials for aggressive phenotypes. They stressed these applications are hypothetical until prospectively validated.
The limitations are substantial: one center, an exclusively East Asian population, no external validation data set, and a small validation cohort with few events and wide confidence intervals. Follow-up supports only 1- and 1.5-year projections, and the mechanistic work used one inhibitor in one cell line. A prospective multicenter validation study across three to five tertiary centers in China is being planned.
Bemerkenswerte Zitate
The signature could guide intensification with triplet therapy or earlier local treatment for patients with equivocal clinical risk, though these applications remain hypothetical pending prospective validation.— Shengke Yang and coinvestigators, Sichuan Cancer Hospital and Institute
Möchten Sie die ganze Geschichte?
Das Original lesen bei CancerNetwork ↗