Molecular analysis of aggressive renal cell carcinoma with unclassified histology reveals distinct subsets

Abstract

Renal cell carcinomas with unclassified histology (uRCC) constitute a significant portion of aggressive non-clear cell renal cell carcinomas that have no standard therapy. The oncogenic drivers in these tumours are unknown. Here we perform a molecular analysis of 62 high-grade primary uRCC, incorporating targeted cancer gene sequencing, RNA sequencing, single-nucleotide polymorphism array, fluorescence in situ hybridization, immunohistochemistry and cell-based assays. We identify recurrent somatic mutations in 29 genes, including NF2 (18%), SETD2 (18%), BAP1 (13%), KMT2C (10%) and MTOR (8%). Integrated analysis reveals a subset of 26% uRCC characterized by NF2 loss, dysregulated Hippo-YAP pathway and worse survival, whereas 21% uRCC with mutations of MTOR, TSC1, TSC2 or PTEN and hyperactive mTORC1 signalling are associated with better clinical outcome. FH deficiency (6%), chromatin/DNA damage regulator mutations (21%) and ALK translocation (2%) distinguish additional cases. Altogether, this study reveals distinct molecular subsets for 76% of our uRCC cohort, which could have diagnostic and therapeutic implications.

Figure 1. Recurrent somatic mutations identified in high-grade uRCC.

(a) Key clinicopathological characteristics of our 62-patient uRCC cohort. Pathological (pT) stage was determined according to the 7th edition of the American Joint Committee staging system for renal cancer. The status of metastasis for individual patients was determined at their last follow-up visits or death. (b) Mutational landscape of recurrent somatic mutations based on IMPACT assays. Mutated genes are listed on the left, and denoted by individual rows. Sixty-two individual patient tumours are presented as columns and labelled at the bottom (T#). Mutation frequency (%) and absolute number of cases with non-silent mutations detected on individual genes are listed on the right. Mutation frequency was calculated as the percentage of individual tumours with mutation(s) in the indicated genes.

Figure 2. Molecular characterizations of the NF2 loss uRCC subset.

(a) Genome-wide frequency plot of DNA copy-number gains (blue) and losses (red) across all chromosomes was determined by OncoScan SNP assay in 15 of the 16 uRCC tumours carrying NF2 mutations and/or 22q loss. The y axis denotes frequencies of alteration in individual chromosomal regions. Copy-neutral loss of heterozygosity (CN-LOH) is shown in Supplementary Fig. 1. (b) Summary of NF2 mutations and frequent (>50%) arm-level copy-number alterations (22q, 1p and 3p) detected by sequencing and SNP array analyses of the NF2 loss subset (n=16). Truncating mutations include nonsense mutations, insertions or deletions that alter the reading frame and splice-site mutations. (c) Representative hemizygous losses of chromosome 22q and the NF2 locus were demonstrated by a custom three-probe FISH assay (red, NF2; orange, 22q11; green, chromosome 10 centromere). Scale bar, 10 μm. (d) Representative immunohistochemical stains of NF2 on NF2 wild-type (WT) and NF2 loss tumours are shown. Scale bars, 50 μm. Semiquantitative IHC scores (0—negative; 1—focal/weak staining; 2—moderate staining; 3—strong and diffuse staining) comparing the NF2 loss subset and the other uRCC tumours are presented as a bar graph. Bars, mean values; error bars, 95% CI. (e) Representative images of NF2 WT and NF2 loss uRCC tumours stained by YAP/TAZ and p-YAP antibodies (left panel) are shown. Scale bars, 100 μm. Immunostaining scores (H-scores) for YAP/TAZ and p-YAP nuclear and cytoplasmic staining were determined and presented as a bar graph on NF2 loss (n=16) or other uRCC (n=43) tumours. H-Scores (H=intensity (0–3) × percentage of positive cells (1–100)). Bars, mean values; error bars, 95% CI. (f) GSEA plot of the ranked list of differentially expression genes in uRCC with NF2 loss and those with WT NF2 generated using a previously established YAP/TAZ-regulated gene set. (g) Immunoblots with the indicated antibodies (left) and a bar graph of cell cycle analysis in ACHN cells with YAP1 or control (GFP) knockdown are shown. Bars, mean values; error bars, s.e.m.; replicates n=3. (h) Representative images (left) and a bar graph of colony formed in the soft agar after plating 10⁵ YAP1 or control (GFP) knockdown cells are shown. Scale bars, 200 μm. Bars, mean values; error bars, s.e.m.; replicates n=3. Statistical significance was determined by Mann–Whitney U-test in d and e, and by Student's t-test in g and h. Statistical significance is indicated as ***P<0.001; **P<0.01; *P<0.05.

Figure 3. uRCC subset with hyperactive mTORC1 signalling.

(a) Schematic overview of indicated mutational and immunohistochemical analyses with annotation of mTORC1 hyperactivation on a subset of uRCC (n=16). (b) Depiction of MTOR missense mutations identified in 5 (8%) uRCC tumours. MTOR L2427R mutation recurred in three individual tumours. (c) Functional analyses of MTOR L2427R and V2475M mutants. 293T cells were transfected with the indicated Flag-tagged MTOR expression constructs in conjunction with HA-S6K. Cellular extracts were collected 48 h later and probed with the indicated antibodies. (d) Representative images of p-4EBP1 immunostaining in L2427R and V2475M MTOR mutant tumours. Scale bars, 100 μm. (e) Immunostaining scores (H-scores) of p-4EBP1 and p-S6 were determined and presented as a bar graph on mTORC1 hypearctivation (n=13), NF2 loss (n=16) or the other uRCC (n=31) tumours. H-Scores (H=intensity (0–3) × percentage of positive cells (1–100)). Bars, mean values; error bars, 95% CI. Statistical significance was determined by Mann–Whitney U-test. Statistical significance is indicated as ****P<0.0001.

Figure 4. Clinical outcomes associated with molecular subsets of uRCC.

(a) Overview of molecular features and clinicopathological characteristics of uRCC subsets identified in our cohort. NF2 loss (NF2 loss, n=16), mTORC1 (mTORC1 hyperactive, n=13), FH (FH deficient, n=4), ALK (ALK translocation, n=1), chromatin DNA damage (mutations in chromatin modulation or DNA damage response genes, n=13) and other (tumours with no identifiable recurrent molecular feature, n=15). Truncating mutations include nonsense mutations, insertions or deletions that alter the reading frame and splice-site mutations. *Indicates a FH missense mutation (G401V), likely a passenger mutation. Percentages on the left indicate frequencies of 4 distinct subsets within the uRCC cohort. Percentages on the right indicate mutation frequencies of corresponding gene(s) within the cohort. (b) Progression-free survival (left) and cancer-specific survival (right) associated with NF2 loss, mTORC1, FH, chromatin DNA damage and other groups are presented and colour-coded as in a. Statistical significance was determined by log-rank test. Statistical significance is indicated as ***P<0.001; **P<0.01; *P<0.05.