Purpose:

Fumarate hydratase–deficient renal cell carcinoma (FH-deficient RCC) is a rare and lethal subtype of kidney cancer. However, the optimal treatments and molecular correlates of benefits for FH-deficient RCC are currently lacking.

Experimental Design:

A total of 91 patients with FH-deficient RCC from 15 medical centers between 2009 and 2022 were enrolled in this study. Genomic and bulk RNA-sequencing (RNA-seq) were performed on 88 and 45 untreated FH-deficient RCCs, respectively. Single-cell RNA-seq was performed to identify biomarkers for treatment response. Main outcomes included disease-free survival (DFS) for localized patients, objective response rate (ORR), progression-free survival (PFS), and overall survival (OS) for patients with metastasis.

Results:

In the localized setting, we found that a cell-cycle progression signature enabled to predict disease progression. In the metastatic setting, first-line immune checkpoint inhibitor plus tyrosine kinase inhibitor (ICI+TKI) combination therapy showed satisfactory safety and was associated with a higher ORR (43.2% vs. 5.6%), apparently superior PFS (median PFS, 17.3 vs. 9.6 months, P = 0.016) and OS (median OS, not reached vs. 25.7 months, P = 0.005) over TKI monotherapy. Bulk and single-cell RNA-seq data revealed an enrichment of memory and effect T cells in responders to ICI plus TKI combination therapy. Furthermore, we identified a signature of memory and effect T cells that was associated with the effectiveness of ICI plus TKI combination therapy.

Conclusions:

ICI plus TKI combination therapy may represent a promising treatment option for metastatic FH-deficient RCC. A memory/active T-cell–derived signature is associated with the efficacy of ICI+TKI but necessitates further validation.

Translational Relevance

Fumarate hydratase–deficient renal cell carcinoma (FH-deficient RCC) is a rare and extremely aggressive form of kidney cancer. The identification of optimal treatments and molecular correlates of benefits remains a major unmet need for this disease. We compared different treatments and found immune checkpoint inhibitor plus tyrosine kinase inhibitor (ICI+TKI) combination therapy with superior efficacy. Using bulk and single-cell RNA-sequencing, we further explored the association between transcriptomic features and clinical outcomes. We revealed a cell-cycle progression signature associated with disease progression in localized setting and constructed a memory/active T-cell–derived signature that correlated with therapeutic response to ICI+TKI. These findings provide novel insights into tumor biology of FH-deficient RCC and facilitate precision treatment.

Fumarate hydratase–deficient renal cell carcinoma (FH-deficient RCC) is a rare subtype of kidney cancer, characterized by FH-inactivating alterations (1). Because of the lack of typical morphologic features, the diagnosis of FH-deficient RCC usually depends on negative FH IHC, positive S-(2-succino)-cysteine (2SC) IHC, and/or pathogenic FH alteration in the tumor (2). Despite its rarity, the clinical behavior of FH-deficient RCC is highly aggressive and the cases frequently present in advanced stages at initial diagnosis (3–5). FH-deficient RCC has a dismal prognosis and over half of the patients with metastatic disease die within 24 months (4–6). Unfortunately, owing to limited knowledge of its molecular basis and little evidence from clinical trials, the optimal management for patients with FH-deficient RCC is unclear at present.

A recent phase II trial reported promising results from bevacizumab and erlotinib combination therapy (Bev+Erlo) in patients with advanced hereditary leiomyomatosis and renal cell carcinoma (HLRCC), and therefore it is recommended as a treatment option by the National Comprehensive Cancer Network guidelines (7, 8). Nonetheless, data from FH-deficient RCC patients showed conflicting results of this combination therapy (3, 5, 9). We previously revealed the immune-hot microenvironment of FH-deficient RCC and reported promising clinical activity of immune checkpoint inhibitor (ICI)–based combination therapy in metastatic FH-deficient RCC (4). However, the efficacy of ICI-based treatment still warrants further validation from larger cohorts. In addition, reliable biomarkers are urgently needed to predict treatment benefits and facilitate decision-making in this setting.

In this study, we performed a multicenter retrospective analysis in patients with FH-deficient RCC to evaluate the efficacy of different treatment strategies, explore molecular correlates, and inform clinical practice.

Study design, population, and human specimens

This study is based on a multicenter database of patients with FH-deficient RCC. All suspicious RCC cases were referred for pathologic consultation at the central institution (West China Hospital of Sichuan University, Chengdu, China). FH-deficient RCC was diagnosed according to IHC evidence of negative FH and/or positive 2SC evaluated by two experienced urological pathologists (N. Chen and L. Nie). Subsequently, FH-deficient RCC was further confirmed by DNA sequencing for FH mutation as previously described (4). A total of 91 patients with FH-deficient RCC were finally identified and enrolled for analysis. Whole-exome sequencing (WES) and bulk RNA-sequencing (RNA-seq) were performed on 76 and 45 patients, respectively. Five fresh tumor specimens from 4 patients with metastatic diseases before first-line ICI plus tyrosine kinase inhibitor (TKI) combination therapy (ICI+TKI) were prospectively collected and underwent single-cell RNA-seq (scRNA-seq). Transcriptomic data from a previously published scRNA analysis of FH-deficient RCC (10) were used for validation.

Study approval

The study protocol was approved by the ethics committee review board of each participating hospital. Written informed consent was obtained from each patient and pursued per Declaration of Helsinki.

Data collection and outcomes

Clinicopathologic data were collected, including age at diagnosis, gender, primary tumor size, surgery type, history of skin and uterine leiomyoma, TNM stage, histopathologic pattern, International Society of Urological Pathology (ISUP) grade, International Metastatic RCC Database Consortium (IMDC) risk score, time of metastasis, metastatic sites, and systemic treatment type. Tumor response was assessed using RECIST 1.1; objective response rate (ORR) was defined as complete response (CR) rate + partial response (PR) rate. Disease-free survival (DFS) was defined from diagnosis to local or regional recurrence or distant metastasis or death. Progression-free survival (PFS) was defined from the start of systemic therapy to disease progression or death. Overall survival (OS) was defined from the start of systemic therapy to death from any cause. The last follow-up was on March 9, 2023.

IHC and multiple immunofluorescence

IHC was performed using an automatic staining platform, Ventana NexES (Roche). Commercially available primary anti–FH (sc-100743, Santa Cruz Biotechnology), 2SC (crb2005017e, Discovery Antibodies), CD8 (clone C8/144B, Dako), and PD-L1 (740–4859, Roche) were used for IHC. Positive PD-L1 expression was scored using the tumor proportion score. CD8+ T-cell density was defined as the number of CD8+ T cells per mm2. Multiple immunofluorescence staining was performed on formalin-fixed paraffin-embedded (FFPE) sections using the Opal multiplex IHC system (PerkinElmer; NEL800001KT) according to the manufacturer's instructions. Primary antibodies used for immunofluorescence staining were CD8 (66868–1-Ig, Proteintech), CD4 (ab133616, Abcam), and IL7R (ab180521, Abcam). Slides of each tumor were imaged using the Olympus IX83 confocal microscope by scanning 10 random fields on each sample at ×40 magnifications and analyzed with CellProfiler 2.2.0 (11) to detect the total number of nuclei, CD8+, CD4+, IL7R+CD8+, and IL7R+CD4+ cells.

DNA extraction and genomic profiling

High-quality genomic DNA from FFPE sections was extracted using the GeneRead DNA FFPE Kit (180134, Qiagen) following the protocol from the manufacturer. Germline DNA was extracted from white blood cells using the Blood Genomic DNA Mini Kit (CW2087, Cwbiotech). WES was performed on 76 patients. Exome capture was performed using the Agilent SureSelect Human All ExonV5 kit (Technologies) according to the manufacturer's instructions. Targeted and Sanger sequencing were performed on 12 patients as previously described (4).

Somatic mutation analysis

The GATK MuTect2 pipeline was run for paired tumor–normal somatic mutation calling with gnomAD database and a panel of normal made from all normal samples to filter common germline mutations and recurrent technical artifacts. The resulting VCFs were filtered by Mutect2 FilterMutectCalls module, variants outside of the capture kit were removed, and FilterByOrientationBias module was used to filter out false-positive calls from OxoG and FFPE. Somatic single-nucleotide variations (SNV) and small insertions and deletions (INDELS) were detected and further filtered according to the flowing criteria: read depth ≥10 in both tumor and normal samples; mapping quality ≥40 and base quality ≥20; variants allele frequency (VAF) ≥ 5% and supporting reads ≥5 in tumor; VAF in tumor ≥5 times that of the matched normal VAF (12–15). Variants were annotated with Oncotator v1.9.9.0 (RRID:SCR_005183). The tumor mutational burden was determined by analyzing somatic mutations, including coding base substitution and INDELs per mega-base (Mb) for cases with WES data.

Germline mutation analysis

Germline SNVs and INDELs were called by GATK HaplotypeCaller (RRID:SCR_001876). The dbSNP build 151 was used when running the pipeline. Variant quality score recalibration was performed for the germline VCF. The vcfs were annotated by InterVar (16) v2.0.2 to classify variants based on a five-tiered categorization system: pathogenic, likely pathogenic, uncertain significance, likely benign, and benign. Variants were selected whether the InterVar or ClinVar (RRID:SCR_006169) annotation matched “Likely_pathogenic” or “Pathogenic.” The possible pathogenic variants in normal samples with read depth ≥10, genotype quality ≥60, and supporting allele reads ≥2 and VAF ≥0.1 were finally kept.

Multiplex ligation-dependent probe amplification

The multiplex ligation-dependent probe amplification (MLPA) assay was used to detect FH large-scale deletions using the SALSA MLPA Probemix P198-A4 FH Kit (MRC-Holland) as previously described (17).

Bulk RNA-seq analysis

Total RNA was isolated from FFPE samples using the Qiagen RNeasy FFPE Kit (73504, Qiagen) according to the manufacturer's instructions. Gene expression profiles were generated for 45 patients using RNA-seq as previously described (18). Differentially expressed genes (DEG) were determined using the R package “limma” with a cutoff P value of <0.05. Gene set enrichment analysis (GSEA, RRID:SCR_003199) was conducted using the GSEA software version 4.3.2. The initial classification of bulk RNA-seq samples into responder and non-responder clinical outcome groups set the foundation for subsequent analyses. Following normalization, genes were ranked on the basis of their differential expression using the signal-to-noise ratio. The Enrichment Score for each gene set was then calculated to highlight overrepresentation at the top or bottom of the ranked gene list. Furthermore, the computation of the normalized enrichment score, accounting for differences in gene set sizes, enhanced results' interpretability. To ensure the reliability of our findings, permutation tests were conducted 20,000 times to identify critical signatures with statistical significance. Gene sets with |NES| > 1, a P value of < 0.05, and an FDR < 0.05 were identified as profoundly enriched (19, 20). A previously established cell-cycle progression (CCP) signature, three immune-related signatures of clear cell RCC, and hallmark gene sets from the MSigDB database were used (21–24). CIBERSORTx (https://cibersortx.stanford.edu/) was applied to deconvolute the cell composition of our bulk RNA-seq datasets (25). This approach was informed by our prior identification of refined cell subtypes in scRNA-seq data.

Single-cell capture, library preparation, and sequencing

Single-cell suspensions were obtained using either mechanical or enzymatic dissociation methods. The viability and single-cell status were determined via AOPI staining and phase contrast microscopy; only samples with viability greater than 80% were used for subsequent steps. The cells were then suspended at approximately 300 to 600 cells/mL depending on samples and subjected to cDNA library generation using the single cell 3′ Library and Gel Bead Kit V3.1 (10x Genomics, 1000121), Chromium Single Cell G Chip Kit (10x Genomics, 1000120), and Chromium single cell controller (10x Genomics). After quality assessment of the resulting libraries, they were sequenced on an Illumina Novaseq 6000 to a read depth of at least 100,000 reads per cell.

Single-cell RNA-seq data analysis

The raw sequence data were processed using the cellranger-5.0.1 pipeline (available at https://support.10xgenomics.com/). Briefly, the Illumina base call files for all libraries were demultiplexed and converted into FASTQ files using bcl2fastq (RRID:SCR_015058). The FASTQ files were then aligned to the human reference genome GRCh38 using STAR (RRID:SCR_004463). The aligned reads were further analyzed to trace back to individual cells, and the cell-by-gene digital counts matrices were generated on the basis of the number of UMIs (unique molecular indices) detected in each cell.

The count matrices were processed using the Scanpy (26) package (version 1.9.1). This included excluding genes detected in fewer than three cells and cells that did not meet certain criteria, such as having fewer than 500 genes, more than 5,000 genes, a maximum of 50,000 UMIs, or a maximum mitochondrial content of 20%. Doublets identified by Scrublet (27) were also removed. The resulting counts were then normalized to the total number per cell and log-transformed. Next, Scanpy was used to identify the top 3,000 highly variable genes from the log-transformed combined matrix. This approach is commonly used in gene expression analysis to identify genes that are most informative for distinguishing between different cell populations. Principal component analysis was then performed on the basis of these genes. To reduce variations introduced by inherent differences between individuals, the BBKNN (batch balanced k nearest neighbors; ref. 28) algorithm was used, which improved the clustering of major cell types. Batch-corrected principal components were then used to identify neighbors for Leiden (29) clustering and generate a uniform manifold approximation and projection (UMAP) visualization. The identities of cell clusters were manually determined using either known marker genes or cluster-specific marker genes. The analysis of cell subpopulations followed a similar approach as mentioned above. To quantify the cell type enrichment across tissues, we calculated the ratio of observed-to-expected cell numbers (Ro/e) in each cluster. The expected cell numbers for each combination of cell clusters and tissues are obtained from the χ2 test. One cluster was identified as being enriched in a specific tissue if Ro/e > 1 (30).

FH-deficient RCC immune signature construction

Proportions of T-cell subclusters in responders (CR/PR) and non-responders (stable disease/progressive disease) of ICI+TKI therapy were calculated using scRNA-seq data. DEGs of T-cell subclusters with a higher proportion in responders were extracted, among which we selected genes annotated for immune-related functionality and had a positive correlation with favorable PFS in the ICI+TKI treatment group. Finally, we defined a 6-gene FH-deficient RCC immune signature, consisting of FOXP1, IRF1, IL7R, GPR183, REL, and SELL. Validation of this signature was performed using bulk RNA-seq data from 26 patients treated with ICI+TKI in this study. The signature score for each sample was calculated as the average of the standardized values [log2(FPKM+1)] for the set of genes within the signature.

Statistical analysis

Continuous variables were described using median and range, and categorical variables were summarized by their percentages. Survival outcomes were estimated by the Kaplan–Meier method and compared using the log-rank test. Hazard ratios (HR) and corresponding 95% confidence intervals (CI) were estimated using a Cox proportional hazard model. Univariate and multivariate Cox regression analyses were used to identify independent predictors for survival. ORRs between different treatment groups were compared using the χ2 test. For differential expression analysis between groups of early/late progression and responders/non-responders, a two-sided Wilcoxon rank-sum test was used. For comparisons of signature scores between groups of early/late progression and responders/non-responders, a two-sided Wilcoxon rank-sum test with Benjamini–Hochberg FDR correction was used. For comparisons of the average percentages of IL7R+CD8+ out of CD8+ cells, and IL7R+CD4+ out of CD4+ cells, in samples between groups of responders and non-responders, a two-sided Wilcoxon rank-sum test was used. All analyses were performed using R software v.4.1.0 (R Foundation for Statistical Computing) and SPSS v.26.0 (SPSS Inc., RRID:SCR_002865). A two-sided P value of <0.05 was considered statistically significant. Descriptions of statistical tests performed for each individual analysis are provided in figure legends.

Data availability

The data generated in this study are publicly available in the National Genomics Data Center Genome Sequence Archive database (https://ngdc.cncb.ac.cn/gsa-human/) under the accession numbers: HRA006641 and HRA006802. Any additional data are available upon request from the corresponding author.

Patient and genomic characteristics

A total of 91 patients with FH-deficient RCC were identified from 15 centers from January 2009 to June 2022. The baseline characteristics of the patients at initial diagnosis are summarized in Table 1. The median age was 38 years (range, 13–71 years) and the male-to-female ratio was 2:1. Forty-two patients (42/91, 46%) had regional lymph node metastasis and 43 patients (43/91, 47%) presented distant metastasis. Cutaneous leiomyoma was observed in 3 patients (2/3 germline FH positive) and uterine leiomyoma was observed in 18 of 31 (58%) female patients. Collectively, FH-deficient RCC exhibited a wide range of morphological patterns and 86% (78/91) or tumors had an ISUP grade≥3. FH was negative in 76/91 (83.5%) cases and partially positive in 15 (16.5%) case; 2SC was diffusely positive in 91/91 (100%) cases. Among 32 cases with only somatic FH mutations, all were 2SC positive and 28/31 (90.3%) were FH negative.

Genomic sequencing was performed in 88 patients (Fig. 1A and B). The median tumor mutation burden (TMB) was 1.1 (range, 0.03–10.5). A total of 113 FH genomic alterations were identified, including 61/113 (54%) germline and 52/113 (46%) somatic alterations. Thirteen (13/88, 15%) patients showed both germline and somatic FH alterations. The most common type of FH alteration was missense mutation (51/113, 45%), followed by frameshift mutation (28/113, 25%), large deletion (12/113, 11%), nonsense mutation (10/113, 9%), splice mutation (9/113, 8%), and copy-number loss (3/113, 3%). Apart from FH, other frequent somatic mutations included TTN (16/88, 18%), NF2 (12/88, 14%), FAT1 (10/88, 11%), and KMT2D (8/88, 9%). The TTN gene encodes a large abundant protein (>30,000 amino acids), which poses a high risk of residue alterations due to random DNA repair error. Most missense mutations on this gene are likely to be “passenger” mutations (31).

CCP correlated with disease progression in localized FH-deficient RCC

At initial diagnosis, 48 patients presented with localized diseases. The clinicopathologic features of these patients are summarized in Supplementary Table S1. After a median follow-up of 33.1 months, 35 patients progressed to metastatic disease, and the median DFS was 12.2 months. Among multiple clinicopathologic and genomic characteristics, only the clinical N stage was independently associated with DFS (Fig. 2A). To further identify reliable molecular predictors, transcriptomic data of 31 patients with localized FH-deficient RCC were analyzed. Patients were divided into early and late progression groups based on median DFS. A total of 1,571 DEGs were identified between these two groups of tumors (Fig. 2B). We found that there was a strong association of early progression with the enrichment of cell-cycle signaling pathway (E2F targets, G2–M checkpoint, and mitotic spindle) genes and elevated MYC transcriptional activity (Fig. 2C). Further analyses suggested that a predefined CCP signature (32) showed a clinical significance in predicting disease progression in patients with localized FH-deficient RCC (P = 0.006, Fig. 2D) and this association retained even after adjustment for clinicopathologic and genomic variables (Fig. 2A).

Systemic treatment outcomes of metastatic FH-deficient RCC

At the last follow-up, a total of 78 patients presented with metastatic diseases. The clinicopathologic features of these patients are present in Supplementary Table S2. Among them, 35 (45%) had metachronous metastases. No survival difference was noted between patients with synchronous and metachronous metastases (Supplementary Fig. S1A–S1C). The most common site of metastasis was distant lymph node (61%) and bone (32%). Eighty percent (62/78) of patients were stratified into IMDC intermediate/poor-risk group. Eighty-eight percent (69/78) of patients underwent nephrectomy and 87% (68/78) received systemic therapies. The most common first-line treatments were ICI+TKI and TKI monotherapy (Supplementary Fig. S1D). A total of 68 patients with metastatic FH-deficient RCC were evaluable for the best tumor response by RECIST v1.1 (Table 2). The details of systemic therapies for individual patients on each treatment line are shown in Supplementary Fig. S2A. For the first-line setting, the ORR was 29.4%. Patients treated with ICI+TKI had a higher ORR (19/44, 43.2%) than those receiving TKI monotherapy (1/18, 5.6%, P = 0.004). No response was observed in patients receiving other treatments. One patient was treated with Bev+Erlo as first-line therapy and had a stable disease for 17.6 months. For the second-line setting, ICI+TKI was consistently effective with an ORR of 11.1% (2/18) and disease control rate of 100% (18/18).

Survival outcomes for patients with different systemic treatments are summarized in Supplementary Table S3. For the overall first-line setting, the median PFS was 16.5 (95% CI, 12.6–20.4) months and the median OS was 33.7 (95% CI, 23.8–43.6) months (Supplementary Fig. S2B and S2C). Notably, patients treated with first-line ICI+TKI had significantly longer PFS (median PFS: 19.0 vs. 9.3 months, P = 0.001; Fig. 3A) and OS (median OS: not reached vs. 25.7 months, P = 0.005; Fig. 3B) than those receiving TKI monotherapy. To further evaluate the impact of timing on the efficacy of ICI+TKI, we compared patients receiving non-first-line ICI+TKI with those who never received ICI+TKI during their treatment history. The results showed that even used as a second or subsequent-line treatment, ICI+TKI could uniformly provide survival benefits in patients with metastatic FH-deficient RCC (median OS: not reached vs. 17.6 months, P = 0.007; Fig. 3C).

Adverse events (AE) that occurred during treatment in 10% or more patients receiving first-line ICI+TKI are summarized in Supplementary Table S4. The most common AEs were diarrhea (30/44, 68.2%), palmar–plantar erythrodysesthesia syndrome (30/44, 68.2%), and hypothyroidism (27/44, 61.4%). Most AEs were grade 1/2; AEs of grade ≥3 occurred in 7 (15.9%) patients.

Memory/active T cells correlated with treatment response of ICI plus TKI combination therapy

Despite an OS improvement with ICI+TKI, there remains a subset of patients who exhibit resistance or poor response to this combination therapy. Thus, it is crucial to identify clinical and molecular biomarkers that can determine the response to ICI+TKI combination therapy. To this end, we subsequently explored the association between clinicopathologic characteristics (e.g., PD-L1 expression, CD8+ T-cell density) and treatment outcomes, but found a limited predictive value of these variables. In addition, several genomic features (e.g., TMB, FH mutation type, frequent somatic mutations) and immune-related signatures of clear cell RCC (e.g., JAVELIN 101 and IMmotion 150 signatures; refs. 21, 22) also failed to predict the efficacy of ICI+TKI among patients with metastatic FH-deficient RCC (Supplementary Fig. S3, Supplementary Table S5).

Therefore, we sought to characterize the tumor microenvironment using single-cell RNA-seq (scRNA-seq) to explore biomarkers that could predict response to ICI+TKI combination therapy. Cells isolated from 5 fresh tumor samples, including two renal tumors and three metastases, were obtained from 4 patients (2 responders and 2 non-responders) with metastatic FH-deficient RCC before first-line ICI+TKI. The clinical and molecular features of these four patients are summarized in Supplementary Table S6. In total, 43,443 single cells passed quality control and were annotated. Using the UMAP method and canonical lineage markers (Supplementary Table S7), 8 main clusters and 23 sub-clusters across tissues, spanning lymphoid, myeloid, fibroblast, endothelial, and epithelial cells, were identified (Fig. 4A; Supplementary Fig. S4A).

Given the high frequency of T cells and their established function in tumor immune response, we focused on analyzing T cells in FH-deficient RCC (Supplementary Fig. S4B). We identified seven T-cell subsets, including four CD8+ T and three CD4+ T clusters (Fig. 4BD). One CD8 cluster (CXCL13 CD8) highly expressed exhaustion markers (PDCD1, LAG3, and TIGIT) and CXCL13, consistent with an exhausted phenotype (33). Proliferating CD8 cluster expressed genes associated with cell cycle and proliferation (MKI67, TOP2A, and UBE2C). The third CD8 cluster exhibited an effector memory-like phenotype, characterized by high expression of IL7R, GZMA, and GZMK (34, 35). The activated CD8 T cluster showed increased expression of genes induced after activation (HSPA6, HSPA1A, DNAJB1, and BAG3; ref. 36). Among the CD4+ T cells, one cluster highly expressed CCR7, IL7R, and SELL, consistent with the central memory phenotype. Another CD4 T cluster also expressed high levels of HSPA1A, DNAJB1, and BAG3, indicating an activated phenotype. The Treg (regulatory T cell) cluster was characterized by the expression of FOXP3, IL2RA, and BATF.

Notably, four T-cell clusters (effector memory CD8, activated CD8, central memory CD4, and activated CD4 clusters) were more frequent in responders, whereas the other three T-cell clusters (CXCL13 CD8, proliferating CD8, and Treg) were more frequent in non-responders (Fig. 4E; Supplementary Fig. S4C). We next developed expression signatures of these T-cell clusters and observed significant enrichment of the four memory/active T-cell gene signatures in responders (Fig. 4F). Besides, we deconvoluted the scRNAseq-based cell populations in the bulk RNA-seq data using CIBERSORT and observed memory/activated CD4 and CD8 T cells were enriched in responders (Supplementary Fig. S4D). To further validate our findings, we took advantage of a previously published scRNA-seq analysis of FH-deficient RCC treated with ICI-based therapy by Dong and colleagues (10). Consistent with our results, we observed higher proportions of memory/activated CD4 and CD8 T cells in tumors from responders compared with those from non-responders (Supplementary Fig. S5A–S5E). These results indicated a positive correlation between memory/active T cells and favorable response to ICI+TKI combination therapy in FH-deficient RCC.

Identification of an FH-deficient RCC immune signature

To further identify the expression signature for therapeutic response, we extracted DEGs from memory/active T-cell clusters. After refinement based on their immune-related function and the association between their expression and improved PFS, we identified a 6-gene subset (FOXP1, IRF1, IL7R, GPR183, REL, and SELL), herein referred to as the “FH-deficient RCC immune signature” (Supplementary Fig. S6A and S6B). An elevated expression of CD8+ T-cell, IFNα, IFNγ, and inflammatory response pathways was observed in patients with higher expression of the FH-deficient RCC immune signature (≥the median expression of the signature) compared with those with low expression of the signature (Supplementary Fig. S6C). We subsequently validated this signature using the bulk RNA-seq data and found the responders showed higher expression of the FH-deficient RCC immune signature compared with the non-responders (P = 0.042; Fig. 4G). Moreover, patients with higher expression of the signature had significantly longer PFS (median PFS: not reached vs. 17.3 months, P = 0.045; Fig. 4H) than those with lower expression in the combination treatment group. Multiple immunofluorescence staining further confirmed tumor tissues from responders had a higher percentage of IL7R+CD8+ and IL7R+CD4+ T cells (P = 0.040 and P = 0.032, respectively; Fig. 4I and J). Together, these data suggested that FH-deficient RCC immune signature was a prognosticator for ICI+TKI treatment response in metastatic FH-deficient RCC.

FH-deficient RCC is a rare and highly aggressive RCC subtype with no approved standard therapies. Thus, the exploration of effective treatments remains an unmet need for this lethal disease. To our knowledge, this study represents the largest cohort investigating treatment outcomes and molecular correlates of FH-deficient RCC. We found that the CCP score could be a promising molecular tool to predict rapid progression in localized FH-deficient RCC. We also demonstrated the superior efficacy of ICI+TKI to TKI monotherapy in metastatic FH-deficient RCC. More importantly, on the basis of bulk and single-cell transcriptional analysis, we defined an FH-deficient RCC immune signature to select optimal candidates for ICI+TKI combination therapy among patients with metastatic FH-deficient RCC.

To date, only one phase II clinical trial reported preliminary results of Bev+Erlo in patients harboring metastatic HLRCC, demonstrating an ORR of 72.1% and a median PFS of 21.2 months (7). These findings led to the recommendation by the National Comprehensive Cancer Network guidelines. However, real-world studies have shown conflicting clinical activity of Bev+Erlo, with an ORR ranging from 30% to 50% and a median PFS varying between 5.5 to 13.3 months (3, 5, 9). Furthermore, the identification of effective treatment remains an unmet need for patients with somatic FH alterations. Hence, the exploration of alternative therapeutic strategies for overall metastatic FH-deficient RCC patients is undoubtedly imperative.

We previously demonstrated that FH-deficient RCC is highly immunogenic, characterized by abundant tumor-infiltrating lymphocytes and high expression of checkpoint molecules in tumors (4). Consistently, a retrospective study observed extensive CD8+ T-cell infiltration in FH-deficient RCC (10). Besides, a recent study based on a transgenic mouse model revealed that loss of Fh1 could induce the release of mitochondrial DNA to drive innate immunity (37). Collectively, these data provide the molecular basis for the use of ICI-based treatments in patients with FH-deficient RCC. Our results showed favorable tumor response and survival outcomes with ICI+TKI irrespective of treatment line in patients with metastatic FH-deficient RCC compared with TKI monotherapy. Similarly, despite a relatively smaller sample size, two recent retrospective studies also showed that ICI-based treatment was associated with numerically longer PFS and OS in FH-deficient RCC compared with TKI alone (5, 10). One study further demonstrated that ICI+TKI showed better efficacy in FH-deficient RCC in comparison with Bev+Erlo (5). We also firstly reported the safety profiles of ICI+TKI in FH-deficient RCC, which were similar with those presented in clinical trials involving ccRCC (38, 39). A prospective trial (NCT04387500) has been launched by our research team to further validate these findings and shed more light on the treatment of FH-deficient RCC.

Currently, despite great efforts into the molecular characterization of FH-deficient RCC, no reliable biomarker has been identified to predict patient prognosis and treatment response (4, 5, 10). In this study, we explored the gene expression data of FH-deficient RCC and found that the enrichment of cell cycle pathways and MYC signaling were associated with rapid progression in localized FH-deficient RCC. Similarly, a recent study reported significant upregulation of E2F and MYC target signatures in HLRCC, highlighting the role of MYC activation in the tumorigenesis of HLRCC (40). We also revealed the prognostic value of CCP signature in FH-deficient RCC, which has been previously validated in localized ccRCC after RN (24). High expression of CCP signature reflects active growing of cancer cells and inherent aggressiveness of the tumor, and thus ultimately affects outcomes. Taken together, these findings imply that cell-cycle and MYC-associated transcriptional programs may serve as potential therapeutic targets in FH-deficient RCC.

For patients treated with ICI+TKI, we found neither conventional clinicopathologic biomarkers nor several immune signatures had a positive correlation with treatment efficacy. To address this, we performed scRNA-seq and observed that T cells were the most abundant components in FH-deficient RCC. Among them, exhausted CD8+ T cells were the dominant populations, indicating an immunosuppressive microenvironment, which is consistent with our previous report (4). However, the proportion and signature of exhausted CD8+ T cells could not predict treatment response in patients receiving the ICI+TKI combination therapy. Our further analysis identified an enrichment of memory and effect T cells, characterized by expression of memory/activation genes (IL7R, SEEL, REL, FOXP1, HSPA1A, and DNAJB1; ref. 41) in responders. Moreover, expression signature generated from these memory/activation T-cell subsets was associated with the efficacy of the ICI+TKI combination therapy in FH-deficient RCC. Similarly, a recent study showed that the antitumor activity of chimeric antigen receptor T cells was enhanced through the IL7 signaling axis, leading to better outcomes in multiple preclinical tumor models (42). Another study from Japan also reported that effective antitumor immunity was governed by cancer antigen-specific memory T cells that highly expressed IL7R and SELL in hepatocellular carcinoma (34). Future studies are essential to confirm the correlation between our newly defined immune signature and the effectiveness of immunotherapy in FH-deficient RCC, as well as its potential relevance in other malignancies.

This study is limited by its inherent retrospective nature. Although the present study has reported the clinical outcomes of FH-deficient RCC with the largest cases to date, the relatively small sample size due to the rarity of FH-deficient RCC may still limit the interpretation of the findings. Our single-cell analysis only included 4 patients and involved no post-treatment sample, leading to difficulty in the exploration of factors associated with treatment resistance. Besides, few samples in the validation cohort, inter-sample heterogeneity, and potential impact of other immune/stromal cell populations for treatment response of immune therapy hinder its wide application. Therefore, we will continue to optimize the FH-deficient RCC immune signature for predicting therapeutic effect of ICI+TKI combination therapy for patients with metastatic FH-deficient RCC.

In conclusion, the current study represents the largest cohort investigating treatment outcomes and molecular correlates of FH-deficient RCC. Our findings showed the predictive value of the high CCP scores for progression in patients with localized diseases. We also reported that ICI plus TKI combination therapy was an effective and safe treatment option for patients with metastatic FH-deficient RCC. Moreover, we generated an FH-deficient RCC-associated immune signature based on memory/active T-cell features. This signature showed an association with the effectiveness of ICI+TKI treatment, offering valuable insights for personalized disease management. To enhance the robustness of our findings, it is imperative to conduct independent and prospective studies for validation.

No disclosures were reported.

J. Chen: Data curation, software, formal analysis, validation, investigation, writing–original draft. X. Hu: Data curation, formal analysis, validation, investigation, writing–original draft. J. Zhao: Formal analysis, investigation, methodology, writing–original draft. X. Yin: Formal analysis, investigation, visualization, methodology, writing–original draft. L. Zheng: Formal analysis, validation, investigation, visualization. J. Guo: Software, formal analysis, visualization. J. Chen: Resources, data curation. Y. Wang: Resources, data curation. X. Sheng: Resources, data curation. H. Dong: Resources, data curation. X. Liu: Resources, data curation. X. Zhang: Resources, software, funding acquisition. J. Liang: Resources, software, funding acquisition. H. Liu: Resources, data curation. J. Yao: Resources, data curation. J. Liu: Resources, data curation. Y. Shen: Resources, data curation. Z. Chen: Resources, data curation. Z. He: Resources, data curation. Y. Wang: Resources, data curation, investigation. N. Chen: Resources, data curation, visualization. L. Nie: Resources, data curation, visualization. M. Zhang: Resources, software, visualization. X. Pan: Resources, data curation, visualization. Y. Chen: Resources, data curation. H. Liu: Resources, data curation. Y. Zhang: Resources, data curation. Y. Tang: Resources, data curation. S. Zhu: Resources, data curation. J. Zhao: Resources, data curation. J. Dai: Resources, data curation. Z. Wang: Resources, data curation. Y. Zeng: Resources, data curation. Z. Wang: Resources, data curation. H. Huang: Data curation, methodology, writing–review and editing. Z. Liu: Conceptualization, resources, data curation, supervision, writing–review and editing. P. Shen: Conceptualization, resources, data curation, supervision, writing–review and editing. H. Zeng: Conceptualization, resources, supervision, funding acquisition, writing–review and editing. G. Sun: Resources, data curation, supervision, funding acquisition, writing–review and editing.

The authors thank all patients involved in this study and their families/caregivers. This work was supported by the National Natural Science Foundation of China (82303659, 82203280, 82172785, 82103097, 81902577, 81974398, 81872107, and 81872108), China Postdoctoral Foundation Project (2020M673239, 2021M692286, 2021M702344, 2021M692281, and 2022M722260), Natural Science Foundation of Sichuan Province (2022NSFSC1526), Research Foundation for the Postdoctoral Program of Sichuan University (2021SCU12014 and 2022SCU12042), Post-Doctor Research Project, West China Hospital, Sichuan University (20HXBH026, 2021HXBH036, and 2021HXBH028), 1.3.5 project for disciplines of excellence, West China Hospital, Sichuan University (ZYJC21020), and Sichuan Province Science and Technology Support Program (2021YFS0119 and 2022YF0305).

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

1.
Moch
H
,
Amin
MB
,
Berney
DM
,
Compérat
EM
,
Gill
AJ
,
Hartmann
A
, et al
.
The 2022 World Health Organization classification of tumours of the urinary system and male genital organs-part A: renal, penile, and testicular tumours
.
Eur Urol
2022
;
82
:
458
68
.
2.
Muller
M
,
Guillaud-Bataille
M
,
Salleron
J
,
Genestie
C
,
Deveaux
S
,
Slama
A
, et al
.
Pattern multiplicity and fumarate hydratase (FH)/S-(2-succino)-cysteine (2SC) staining but not eosinophilic nucleoli with perinucleolar halos differentiate hereditary leiomyomatosis and renal cell carcinoma-associated renal cell carcinomas from kidney tumors without FH gene alteration
.
Mod Pathol
2018
;
31
:
974
83
.
3.
Carril-Ajuria
L
,
Colomba
E
,
Cerbone
L
,
Romero-Ferreiro
C
,
Crouzet
L
,
Laguerre
B
, et al
.
Response to systemic therapy in fumarate hydratase–deficient renal cell carcinoma
.
Eur J Cancer
2021
;
151
:
106
14
.
4.
Sun
G
,
Zhang
X
,
Liang
J
,
Pan
X
,
Zhu
S
,
Liu
Z
, et al
.
Integrated molecular characterization of fumarate hydratase–deficient renal cell carcinoma
.
Clin Cancer Res
2021
;
27
:
1734
43
.
5.
Xu
Y
,
Kong
W
,
Cao
M
,
Wang
J
,
Wang
Z
,
Zheng
L
, et al
.
Genomic profiling and response to immune checkpoint inhibition plus tyrosine kinase inhibition in FH-deficient renal cell carcinoma
.
Eur Urol
2023
;
83
:
163
72
.
6.
Gleeson
JP
,
Nikolovski
I
,
Dinatale
R
,
Zucker
M
,
Knezevic
A
,
Patil
S
, et al
.
Comprehensive molecular characterization and response to therapy in fumarate hydratase–deficient renal cell carcinoma
.
Clin Cancer Res
2021
;
27
:
2910
9
.
7.
Srinivasan
R
,
Gurram
S
,
Al Harthy
M
,
Singer
EA
,
Sidana
A
,
Shuch
BM
, et al
.
Results from a phase II study of bevacizumab and erlotinib in subjects with advanced hereditary leiomyomatosis and renal cell cancer (HLRCC) or sporadic papillary renal cell cancer
.
J Clin Oncol
2020
;
38
:
5004
.
8.
National Comprehensive Cancer Network
.
NCCN clinical practice guidelines in oncology
.
Kidney cancer (version 3.2024)
.
Available from
: https://www.nccn.org/professionals/physician_gls/pdf/kidney.pdf.
9.
Choi
Y
,
Keam
B
,
Kim
M
,
Yoon
S
,
Kim
D
,
Choi
JG
, et al
.
Bevacizumab plus erlotinib combination therapy for advanced hereditary leiomyomatosis and renal cell carcinoma-associated renal cell carcinoma: a multicenter retrospective analysis in Korean patients
.
Cancer Res Treat
2019
;
51
:
1549
56
.
10.
Dong
P
,
Zhang
X
,
Peng
Y
,
Zhang
Y
,
Liu
R
,
Li
Y
, et al
.
Genomic characteristics and single-cell profiles after immunotherapy in fumarate hydratase–deficient renal cell carcinoma
.
Clin Cancer Res
2022
;
28
:
4807
19
.
11.
Carpenter
AE
,
Jones
TR
,
Lamprecht
MR
,
Clarke
C
,
Kang
IH
,
Friman
O
, et al
.
CellProfiler: image analysis software for identifying and quantifying cell phenotypes
.
Genome Biol
2006
;
7
:
R100
.
12.
Vadgama
N
,
Pittman
A
,
Simpson
M
,
Nirmalananthan
N
,
Murray
R
,
Yoshikawa
T
, et al
.
De novo single-nucleotide and copy number variation in discordant monozygotic twins reveals disease-related genes
.
Eur J Hum Genet
2019
;
27
:
1121
33
.
13.
Weinreb
I
,
Piscuoglio
S
,
Martelotto
LG
,
Waggott
D
,
Ng
CK
,
Perez-Ordonez
B
, et al
.
Hotspot activating PRKD1 somatic mutations in polymorphous low-grade adenocarcinomas of the salivary glands
.
Nat Genet
2014
;
46
:
1166
9
.
14.
Schrijver
W
,
Selenica
P
,
Lee
JY
,
Ng
CKY
,
Burke
KA
,
Piscuoglio
S
, et al
.
Mutation profiling of key cancer genes in primary breast cancers and their distant metastases
.
Cancer Res
2018
;
78
:
3112
21
.
15.
Ellrott
K
,
Bailey
MH
,
Saksena
G
,
Covington
KR
,
Kandoth
C
,
Stewart
C
, et al
.
Scalable open science approach for mutation calling of tumor exomes using multiple genomic pipelines
.
Cell Syst
2018
;
6
:
271
81
.
e7
.
16.
Li
Q
,
Wang
K
.
InterVar: clinical interpretation of genetic variants by the 2015 ACMG-AMP guidelines
.
Am J Hum Genet
2017
;
100
:
267
80
.
17.
Zheng
L
,
Zhang
X
,
Pan
X
,
Huang
Z
,
Zhang
M
,
Xian
J
, et al
.
AKR1B10 is a new sensitive and specific marker for fumarate hydratase–deficient renal cell carcinoma
.
Mod Pathol
2023
;
36
:
100303
.
18.
Sun
G
,
Chen
J
,
Liang
J
,
Yin
X
,
Zhang
M
,
Yao
J
, et al
.
Integrated exome and RNA-sequencing of TFE3-translocation renal cell carcinoma
.
Nat Commun
2021
;
12
:
5262
.
19.
Goeman
JJ
,
Bühlmann
P
.
Analyzing gene expression data in terms of gene sets: methodological issues
.
Bioinformatics
2007
;
23
:
980
7
.
20.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
.
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci USA
2005
;
102
:
15545
50
.
21.
Motzer
RJ
,
Robbins
PB
,
Powles
T
,
Albiges
L
,
Haanen
JB
,
Larkin
J
, et al
.
Avelumab plus axitinib versus sunitinib in advanced renal cell carcinoma: biomarker analysis of the phase 3 JAVELIN Renal 101 trial
.
Nat Med
2020
;
26
:
1733
41
.
22.
McDermott
DF
,
Huseni
MA
,
Atkins
MB
,
Motzer
RJ
,
Rini
BI
,
Escudier
B
, et al
.
Clinical activity and molecular correlates of response to atezolizumab alone or in combination with bevacizumab versus sunitinib in renal cell carcinoma
.
Nat Med
2018
;
24
:
749
57
.
23.
Liberzon
A
,
Birger
C
,
Thorvaldsdóttir
H
,
Ghandi
M
,
Mesirov
JP
,
Tamayo
P
.
The Molecular Signatures Database (MSigDB) hallmark gene set collection
.
Cell Syst
2015
;
1
:
417
25
.
24.
Morgan
TM
,
Mehra
R
,
Tiemeny
P
,
Wolf
JS
,
Wu
S
,
Sangale
Z
, et al
.
A multigene signature based on cell-cycle proliferation improves prediction of mortality within 5 yr of radical nephrectomy for renal cell carcinoma
.
Eur Urol
2018
;
73
:
763
9
.
25.
Newman
AM
,
Steen
CB
,
Liu
CL
,
Gentles
AJ
,
Chaudhuri
AA
,
Scherer
F
, et al
.
Determining cell type abundance and expression from bulk tissues with digital cytometry
.
Nat Biotechnol
2019
;
37
:
773
82
.
26.
Wolf
FA
,
Angerer
P
,
Theis
FJ
.
SCANPY: large-scale single-cell gene expression data analysis
.
Genome Biol
2018
;
19
:
15
.
27.
Wolock
SL
,
Lopez
R
,
Klein
AM
.
Scrublet: computational identification of cell doublets in single cell transcriptomic data
.
Cell Syst
2019
;
8
:
281
91
.
28.
Polański
K
,
Young
MD
,
Miao
Z
,
Meyer
KB
,
Teichmann
SA
,
Park
JE
.
BBKNN: fast batch alignment of single cell transcriptomes
.
Bioinformatics
2020
;
36
:
964
5
.
29.
Traag
VA
,
Waltman
L
,
van Eck
NJ
.
From Louvain to Leiden: guaranteeing well-connected communities
.
Sci Rep
2019
;
9
:
5233
.
30.
Cheng
S
,
Li
Z
,
Gao
R
,
Xing
B
,
Gao
Y
,
Yang
Y
, et al
.
A pan-cancer single-cell transcriptional atlas of tumor infiltrating myeloid cells
.
Cell
2021
;
184
:
792
809
.
31.
Tan
H
,
Bao
J
,
Zhou
X
.
Genome-wide mutational spectra analysis reveals significant cancer-specific heterogeneity
.
Sci Rep
2015
;
5
:
12566
.
32.
Cuzick
J
,
Swanson
GP
,
Fisher
G
,
Brothman
AR
,
Berney
DM
,
Reid
JE
, et al
.
Prognostic value of an RNA expression signature derived from cell cycle proliferation genes in patients with prostate cancer: a retrospective study
.
Lancet Oncol
2011
;
12
:
245
55
.
33.
Liu
B
,
Zhang
Y
,
Wang
D
,
Hu
X
,
Zhang
Z
.
Single-cell meta-analyses reveal responses of tumor-reactive CXCL13(+) T cells to immune-checkpoint blockade
.
Nat Cancer
2022
;
3
:
1123
36
.
34.
Mizukoshi
E
,
Nakagawa
H
,
Tamai
T
,
Kitahara
M
,
Fushimi
K
,
Nio
K
, et al
.
Peptide vaccine-treated, long-term surviving cancer patients harbor self-renewing tumor-specific CD8(+) T cells
.
Nat Commun
2022
;
13
:
3123
.
35.
Principe
N
,
Kidman
J
,
Goh
S
,
Tilsed
CM
,
Fisher
SA
,
Fear
VS
, et al
.
Tumor infiltrating effector memory antigen-specific CD8(+) T cells predict response to immune checkpoint therapy
.
Front Immunol
2020
;
11
:
584423
.
36.
Krishna
C
,
DiNatale
RG
,
Kuo
F
,
Srivastava
RM
,
Vuong
L
,
Chowell
D
, et al
.
Single-cell sequencing links multiregional immune landscapes and tissue-resident T cells in ccRCC to tumor topology and therapy efficacy
.
Cancer Cell
2021
;
39
:
662
77
.
37.
Zecchini
V
,
Paupe
V
,
Herranz-Montoya
I
,
Janssen
J
,
Wortel
IMN
,
Morris
JL
, et al
.
Fumarate induces vesicular release of mtDNA to drive innate immunity
.
Nature
2023
;
615
:
499
506
.
38.
Motzer
R
,
Alekseev
B
,
Rha
SY
,
Porta
C
,
Eto
M
,
Powles
T
, et al
.
Lenvatinib plus pembrolizumab or everolimus for advanced renal cell carcinoma
.
N Engl J Med
2021
;
384
:
1289
300
.
39.
Rini
BI
,
Plimack
ER
,
Stus
V
,
Gafanov
R
,
Hawkins
R
,
Nosov
D
, et al
.
Pembrolizumab plus axitinib versus sunitinib for advanced renal-cell carcinoma
.
N Engl J Med
2019
;
380
:
1116
27
.
40.
Valcarcel-Jimenez
L
,
Rogerson
C
,
Yong
C
,
Schmidt
C
,
Yang
M
,
Cremades-Rodelgo
M
, et al
.
HIRA loss transforms FH-deficient cells
.
Sci Adv
2022
;
8
:
eabq8297
.
41.
Hurton
LV
,
Singh
H
,
Najjar
AM
,
Switzer
KC
,
Mi
T
,
Maiti
S
, et al
.
Tethered IL15 augments antitumor activity and promotes a stem-cell memory subset in tumor-specific T cells
.
Proc Natl Acad Sci USA
2016
;
113
:
E7788
-e97.
42.
Shum
T
,
Omer
B
,
Tashiro
H
,
Kruse
RL
,
Wagner
DL
,
Parikh
K
, et al
.
Constitutive signaling from an engineered IL7 receptor promotes durable tumor elimination by tumor-redirected T cells
.
Cancer Discov
2017
;
7
:
1238
47
.
This open access article is distributed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license.

Supplementary data