Understanding Decipher GRID For Prostate Cancer - 070
Image made with ChatGPT
In prostate cancer, clinicians associate “Decipher” with a genomic classifier test that reports a risk score for distant metastasis over five years. They use this score to guide critical treatment decisions.
However, there is a major distinction between the test a patient receives and the platform behind it. The Decipher Prostate Genomic Classifier is the clinical product, but it is derived from the Decipher Genomic Resource Information Database (Decipher GRID).
To understand the future of prostate cancer care, we must understand the relationship between the product and the platform.
The product vs. the factory
The Decipher Prostate Genomic Classifier is a 22-gene signature test. When a patient undergoes a biopsy or surgery, clinicians send a small piece of the tumor tissue to a laboratory.
The lab measures the activity of 22 specific genes associated with aggressive cancer. The test uses a machine-learning algorithm to generate a score between 0 and 1.
The Decipher Prostate score is prognostic. It predicts the five-year probability of metastasis. A low score may allow a patient to confidently choose active surveillance, while a high score suggests the tumor is biologically aggressive and requires intensive treatment.
The Decipher GRID is the vast research engine that enabled this 22-gene test. The most significant difference between the test and the platform is the amount of data captured.
When the lab analyzes a tumor for the 22-gene test, it examines the entire transcriptome (the complete set of RNA transcripts in a cell). This process captures the activity levels of approximately 1.4 million markers per patient.
While the patient receives a report based on just 22 markers, the platform saves the other 1.4 million. This massive, deidentified dataset is the Decipher GRID.
The strategic value of the GRID platform
This platform creates a continuous feedback loop that turns every clinical test into a discovery tool.
A patient receives the 22-gene score for their personal care.
The lab simultaneously captures the 1.4 million-marker profile from that same sample and adds it to the GRID.
With over 200,000 patient profiles now in the database, researchers can mine this data to discover new biomarkers associated with prostate cancer.
This data bank is the fundamental value of the GRID. It allows researchers to validate new biomarkers at a speed and scale that is impossible with traditional research methods.
When scientists have a new hypothesis, they do not need to start a decade-long study from scratch. They can query the 200,000 profiles already stored in the GRID and cross-reference them with real-world outcomes.
Where the data comes from
To build a database this powerful, the GRID draws from three main sources:
The GRID Registry: The backbone of the platform is a formal, prospective observational study titled “Prospective Expression Analysis Using The Decipher Genomics Resource for Intelligent Discovery (GRID)”. When patients receive the standard Decipher test, they may choose to share their deidentified data with this study. The goal is to reach one million patients, which provides the statistical power to understand even the rarest types of prostate cancer.
Academic Partnerships: Veracyte, the owner of Decipher GRID, works with top academic researchers. These scientists use the GRID to answer their own research questions, thereby adding new findings and greater value to the database.
Clinical Trial Integration: The GRID includes data from over 90 clinical trials, including Phase 3 trials such as STAMPEDE, CHAARTED, and TITAN. By revisiting old clinical trials and applying genomic signatures from the GRID, researchers can determine why a drug helped some men but failed others. This kind of analysis could rescue a treatment that had been previously written off as a failure by showing which patient subset benefits from the drug.
Linking genes to real-world outcomes
Genomic data becomes meaningful when it is tied to a clinical context. The true power of the GRID lies in its role as an observational registry that links genomic data to three key areas:
Standard information, such as PSA levels and Gleason scores
What actually happened to the patient? Did they have surgery, radiation, or hormone therapy?
Years later, did the cancer spread, or did the patient survive?
Decipher GRID merges deep genomics with long-term clinical follow-up. These mergers allow a researcher to ask a specific question, such as which genes are active in patients who had surgery but still saw their cancer return. This process can now take days instead of years.
The GRID as a biomarker factory
The original 22-gene Decipher test was just the first product off the assembly line. The platform is now producing a new generation of “predictive” biomarkers. While a prognostic test tells you the risk of what might happen, a predictive test tells you which specific therapy will actually work.
Here are the three most significant examples of this evolution:
1. Molecular subtyping - the Prostate Subtyping Classifier (PSC)
Researchers used the GRID’s computational power to analyze the whole transcriptomes of 32,000 prostate cancer profiles. From this, they developed a novel “Prostate Subtyping Classifier” (PSC). They then validated this PSC model on an additional 68,547 profiles.
This PSC model successfully classifies prostate tumors into four distinct molecular subtypes with unique biological and clinical features:
Luminal Differentiated: Respond well to antiandrogen therapy and are associated with less aggressive tumors and the longest time to metastasis after surgery.
Luminal Proliferating: Has higher expression of cell proliferation genes and was shown in a phase 3 trial to derive survival benefit from docetaxel chemotherapy.
Basal Immune: Possesses significant immune infiltration and was the only subtype in one cohort that derived benefit from radiation therapy after surgery.
Basal Neuroendocrine-like: Characterized by lower androgen receptor activity.
2. Predicting hormone therapy response - the PAM50 signature
Antiandrogen therapy is a cornerstone of prostate cancer treatment, but comes with significant side effects, such as bone loss, hot flashes, and fatigue. Can we identify who will actually benefit from it and who can be spared the toxicity?
The PAM50 molecular signature, derived from breast cancer research, is a biomarker available on the GRID platform for research. The GRID report classifies tumors into one of three subtypes:
Luminal A
Luminal B
Basal
PAM50 was tested in the prospective, randomized phase 2 BALANCE trial. This trial enrolled men with recurrent prostate cancer after surgery. They were randomly assigned to receive either salvage radiation therapy plus a placebo or salvage radiation therapy plus apalutamide, an androgen receptor inhibitor.
All patients were stratified at the start of the trial by PAM50 subtype, determined using the Decipher platform.
The results showed that the benefit of adding apalutamide was entirely confined to a single genomic subtype:
In men with Luminal B tumors, adding apalutamide was highly effective: 72% remained free of biochemical failure, compared with only 54% in the placebo group.
In men with Non-Luminal B tumors, apalutamide provided no benefit: 70% were free of biochemical failure, compared with 71% in the placebo group.
According to Daniel Spratt, M.D.
“Our findings mark the first time, to my knowledge, that a predictive biomarker has been validated in a prospective, biomarker-driven, randomized trial in non-metastatic prostate cancer. Thus, this is an unprecedented advancement for patients who can be more precisely selected to receive hormone therapy or forego the treatment and the potential side effects.”
For a patient with a Non-Luminal B tumor, this test could save them from months or years of unnecessary and toxic treatment.
3. Guiding radiation therapy dose: - the PORTOS signature
For patients who need radiation therapy after surgery, what is the correct dose? A higher dose might be more effective at killing the cancer, but it also increases the risk of side effects and toxicity. A lower dose is safer, but may be less effective.
Researchers again turned to the GRID database to find a genomic signature that could predict response to radiation. They developed the Post-Operative Radiation Therapy Outcomes Score (PORTOS), a gene-expression signature intended to predict benefit from radiation dose escalation after prostatectomy.
The PORTOS signature was validated by analyzing data from two independent Phase 3 randomized trials: SAKK 09/10 and NRG/RTOG 01-26. These trials had already compared dose-escalated radiation to conventional-dose radiation.
The GRID-derived PORTOS signature successfully predicted which patients would benefit from the higher dose. Patients with higher PORTOS scores had a significantly better response to dose-escalated radiation, whereas those with lower scores did not.
A radiation oncologist could use this signature to personalize the intensity of a patient’s radiation plan, escalating the dose only for those patients whose tumor biology suggests it will provide a significant benefit, while potentially sparing all other patients the unnecessary toxicity.
The future - a prostate cancer genomic dashboard
The ultimate vision for the Decipher GRID, with proper validation and regulatory approval, is to move beyond a single risk score. The goal is to use one tissue sample to generate a comprehensive “Prostate Cancer Genomic Dashboard.”
In the near future, when a patient’s tumor is analyzed, the physician could receive a unified report that answers all the critical questions at once:
What is the risk of metastasis? (22-gene Classifier)
What is the tumor’s biological subtype? (PSC)
Will this patient benefit from hormone therapy? (PAM50)
Will this patient benefit from more intense radiation? (PORTOS)
Which chemotherapy is most likely to work? (PSC)
As the platform continues to grow, it is becoming a robust research foundation for more personalized care. By matching a patient’s specific biology to a massive library of real-world outcomes, the GRID provides a roadmap for discovering treatments tailored to the individual.
Even though these new markers still need to undergo formal clinical trials and obtain regulatory approval, they offer a glimpse of the future.
Your doctor can request your Decipher GRID report
If you have had a Decipher Prostate genomic classifier test, your doctor can request your Decipher GRID report at no additional charge to you or your insurance company. Since the Decipher GRID report is for “Research Use Only,” your doctor may be hesitant to use the report to guide your treatment decisions.
I have examined a friend’s GRID report and was impressed with the additional information it provides. Even though it is not supposed to be used to guide prostate cancer management, I believe the most savvy patients and clinicians would find the reports useful. 😉
Until the next newsletter, stay healthy.
Much love,
Keith
This is an excellent explainer, and I appreciate how you separate the clinical product (a 22-gene prognostic score) from the platform (the transcriptomic “factory” that keeps learning). That distinction is often missed and it’s where the real future of precision oncology lives. From a clinician’s lens, the promise here isn’t “more data,” it’s better decisions: identifying who truly needs treatment escalation versus who can safely de-escalate. That’s especially important in prostate cancer, where the harms of overtreatment (urinary/sexual side effects, fatigue, financial toxicity) can be profound. The shift you highlight, from prognostic to predictive signatures (therapy selection, radiation dosing, etc.) is exactly the direction we need. I also appreciate your implied caution: research-use outputs can be informative, but they still need rigorous prospective validation, transparent performance across diverse populations, and clear guardrails for how clinicians should (and shouldn’t) act on them. Really valuable post! This is the kind of systems-level thinking that can reduce “torture” and improve outcomes at scale.
Less torture, more success. Let's hope the government doesn't defund. I wonder how defunding has already affected knowledge. My prayers and best wishes always through your torturous Journey.