Renal cell carcinoma (RCC) - the most common form of kidney cancer - was diagnosed in over 400,000 people worldwide in 2018 and caused approximately 175,000 deaths. About 25% of patients already have cancer that has spread to other parts of the body when first diagnosed, making effective treatment selection critically important.
The most common subtype, clear-cell RCC, accounts for about 75% of all kidney cancers. Many of these tumors have a mutation in the VHL gene, which normally helps control blood vessel growth. When VHL is lost, the cancer recruits an abnormal blood supply - a process called angiogenesis - that fuels tumor growth.
Kidney cancer is also notable for having high levels of immune cell infiltration. This means the body's immune system is present in the tumor, creating an opportunity for immunotherapy - treatments that help the immune system fight cancer. However, immune checkpoints like PD-L1 can suppress these immune cells, allowing the cancer to escape.
Current treatments include drugs that block blood vessel growth (anti-angiogenics like sunitinib and bevacizumab) and checkpoint inhibitors like atezolizumab that release the immune system's brakes. Not every patient responds the same way, which is why researchers needed to understand the molecular differences between tumors to better match patients to the right treatment.
This study analyzed tumor samples from IMmotion151, a large global Phase III clinical trial that compared two treatments in patients with previously untreated advanced kidney cancer: the combination of atezolizumab plus bevacizumab versus sunitinib alone. A total of 823 out of 915 patients (90%) had pre-treatment tumor samples available for molecular analysis.
Researchers used RNA sequencing (RNA-seq) to measure the activity of thousands of genes in each tumor sample. This technology reads the genetic instructions that are actively being used in cancer cells, providing a snapshot of the tumor's biological programs at the time of treatment.
To group tumors by their molecular similarities, scientists used a statistical method called non-negative matrix factorization (NMF) - an approach that clusters tumors based on patterns in their gene activity without any prior assumptions about what groups should exist. This unbiased approach identified naturally occurring tumor groups.
Additionally, somatic mutation profiling was performed on 715 patients using the FoundationOne assay, which detects specific DNA changes in cancer genes. By combining gene expression data with mutation data (a technique called multi-omics), the team created one of the most comprehensive molecular portraits of advanced kidney cancer ever assembled in a clinical trial.
The analysis identified seven distinct molecular subgroups of kidney cancer tumors, each with a unique biological identity. Two clusters (Clusters 1 and 2) were primarily defined by high angiogenesis - an active blood vessel growth program driven by the VEGF pathway. These tumors also showed higher expression of signaling pathways like TGF-beta, WNT, and NOTCH.
Three clusters (Clusters 4, 5, and 6) showed a very different profile: low angiogenesis but high cell cycle activity, meaning tumor cells were rapidly dividing. These tumors also showed increased anabolic metabolism - the cellular machinery for building new molecules to fuel rapid growth. Cluster 4 stood out as highly immune-infiltrated, with high T-cell presence and PD-L1 expression, earning the label T-effector/Proliferative.
Cluster 3 was characterized by expression of the complement cascade - an immune pathway previously linked to poor prognosis in kidney cancer - and a distinct metabolic program called omega oxidation. This cluster received the label Complement/Omega-oxidation.
A small Cluster 7, called the snoRNA cluster, showed unusual enrichment in small nucleolar RNA molecules, which have been linked to changes in how genes are regulated and have been associated with cancer development. The findings were validated in an independent dataset from a separate trial (IMmotion150), confirming these seven subtypes are real and reproducible.
The seven molecular subtypes predicted outcomes differently depending on which treatment patients received. Patients in the angiogenic clusters (Clusters 1 and 2) did well on both treatments, likely because both sunitinib and the bevacizumab component of the combination target the blood vessel growth program that defines those tumors.
The most striking finding was in the T-effector/Proliferative cluster (Cluster 4): patients in this group showed dramatically better outcomes with the combination of atezolizumab plus bevacizumab compared to sunitinib alone. The response rate was 52% vs. 19.4%, and the risk of disease progression was nearly halved. This makes biological sense - these tumors have a strong immune presence that immunotherapy can harness.
Similarly, patients in the Proliferative cluster (Cluster 5) - which has low angiogenesis but high cell division - showed improved outcomes with the combination therapy (response rate 26.2% vs. only 3.1% with sunitinib). This suggests that even in tumors with limited blood vessels, checkpoint inhibition can provide meaningful benefit.
Importantly, multivariate analyses confirmed that these molecular subgroup differences in treatment benefit were independent of PD-L1 expression and standard clinical risk scores. This means the molecular subtyping adds real information beyond what doctors currently use to guide treatment decisions.
Beyond gene activity patterns, specific DNA mutations were linked to the molecular subtypes. Mutations in the PBRM1 gene were associated with the highly angiogenic clusters and predicted better outcomes with sunitinib. PBRM1 is involved in regulating which genes are turned on or off, and its loss appears to amplify the hypoxia and angiogenesis response that makes tumors dependent on blood vessel growth.
In contrast, alterations in CDKN2A/B - genes that normally act as brakes on cell division - were linked to the proliferative clusters and to worse overall prognosis. However, patients with these alterations showed significantly better outcomes when treated with atezolizumab plus bevacizumab vs. sunitinib, including a response rate of 42% vs. 20%.
TP53 mutations were similarly associated with proliferative, aggressive tumor biology and trended toward better outcomes with the combination therapy. Mutations in ARID1A and KMT2C - genes involved in regulating how DNA is packaged and read - were linked to improved outcomes with the combination therapy, suggesting that tumors with disrupted gene regulation may be particularly sensitive to immunotherapy.
These mutation findings suggest that a targeted genetic test could help guide treatment selection for kidney cancer patients - a step toward truly personalized medicine, where the treatment is matched to the specific molecular weaknesses of an individual patient's tumor.
About 20% of kidney cancer tumors contain sarcomatoid elements - aggressive cells that resemble a different type of cancer (sarcoma) and are associated with rapid spread and poor outcomes. Historically, these tumors responded poorly to standard anti-angiogenic therapies.
This study revealed why: sarcomatoid tumors have lower angiogenesis activity and fewer PBRM1 mutations, explaining their poor response to drugs targeting blood vessel growth. Instead, they show higher expression of cell cycle genes, and critically, much higher levels of PD-L1 (63% vs. 39% in non-sarcomatoid tumors) and more immune cell infiltration.
Molecularly, sarcomatoid tumors cluster with the T-effector/Proliferative, Proliferative, and Stromal/Proliferative groups - the same groups that benefit most from checkpoint inhibitor combination therapy. They also have more frequent alterations in CDKN2A/B (26% vs. 15%) and PTEN, suggesting these gene losses may drive the aggressive dedifferentiation process.
These findings provide the biological explanation for why checkpoint inhibitor-based therapies have shown remarkable efficacy - including complete responses - in patients with sarcomatoid kidney cancer. The molecular profile of these tumors makes them inherently more sensitive to immunotherapy, which is now supported by this mechanistic data.
This study represents the largest integrated molecular analysis of advanced kidney cancer in a randomized clinical trial, and its findings have important practical implications. The identification of seven molecular subtypes that predict different treatment responses provides a roadmap for personalizing therapy - moving beyond one-size-fits-all treatment decisions.
For patients whose tumors fall into angiogenic clusters, current treatments targeting blood vessel growth appear optimal. For those with immune-active or proliferative tumors, checkpoint inhibitor combinations are clearly superior. If molecular subtyping were routinely applied, doctors could match patients to the treatment most likely to benefit them from the start, rather than learning through trial and error.
The findings also point toward future treatment strategies. Patients with CDKN2A/B-altered tumors may benefit from CDK4/6 inhibitors - drugs that target the cell cycle proteins these tumors depend on. Preclinical data suggest that combining CDK4/6 inhibitors with checkpoint immunotherapy might be especially powerful. Similarly, PBRM1-mutant tumors may benefit from newer agents targeting HIF-2A, a key driver of the angiogenic response in these tumors.
The molecular insights from this kidney cancer study are also likely applicable to other cancer types, since the same combination therapies are now being tested in liver cancer, lung cancer, and endometrial cancer. Understanding which molecular features predict response or resistance could accelerate progress across multiple disease areas.
This groundbreaking study shows that kidney cancer is not one disease but seven biologically distinct diseases, each responding differently to available treatments. The molecular subtype of a patient's tumor - determined by analyzing which genes are active and which are mutated - can predict which treatment is most likely to work.
For patients and families, this research represents progress toward a world where treatment decisions for kidney cancer are guided by the specific biology of each person's tumor, not just the cancer's stage or standard risk categories. It supports the idea that asking for molecular testing before starting treatment may help identify the best path forward.
The study also validates that the combination of bevacizumab plus atezolizumab is superior to sunitinib for patients whose tumors have an active immune environment or high cell proliferation - a finding that can already influence treatment decisions for advanced kidney cancer patients today.
Looking ahead, this research lays the foundation for biomarker-driven clinical trials that enroll patients based on their tumor's molecular profile. This approach - testing treatments in the specific patient populations most likely to respond - is the path to faster development of more effective, less toxic therapies for kidney cancer.