The Evolution of Angiogenic and Inflamed Tumors: The Renal Cancer Paradigm
2020; Cell Press; Volume: 38; Issue: 6 Linguagem: Inglês
10.1016/j.ccell.2020.10.021
ISSN1878-3686
AutoresJames Brugarolas, Satwik Rajaram, Alana Christie, Payal Kapur,
Tópico(s)Renal and related cancers
ResumoGene expression analyses have identified subtypes of conventional renal cell carcinoma broadly distributed into angiogenic and proliferative/ immunogenic clades. Integration with genomic and functional experiments in animal models yields an evolutionary model. Evolutionary trajectories illustrate remarkable plasticity, particularly for a tumor that typically begins with inactivation of a single gene. Gene expression analyses have identified subtypes of conventional renal cell carcinoma broadly distributed into angiogenic and proliferative/ immunogenic clades. Integration with genomic and functional experiments in animal models yields an evolutionary model. Evolutionary trajectories illustrate remarkable plasticity, particularly for a tumor that typically begins with inactivation of a single gene. Conventional clear cell renal cell carcinoma (ccRCC) accounts for over 70% of all adult renal cancers. Approximately 40% of tumors become metastatic. Treatment largely deploys anti-angiogenic (AA) drugs and immune checkpoint inhibitors (ICIs), and most recently their combination. How these therapies should be optimally administered is unclear as actionable predictive biomarkers are lacking. Motzer et al. report integrated genomic analyses from patients with advanced, treatment-naive RCC (with a clear cell or sarcomatoid component) enrolled in the randomized phase 3 IMmotion151 trial evaluating the combination of atezolizumab (anti-PD-L1) and bevacizumab (anti-VEGF) versus sunitinib (VEGFR2 inhibitor) (Motzer et al., 2020Motzer R.J. Banchereau R. Hamidi H. Powles T. McDermott D. Atkins M.B. Escudier B. Liu L.-F. Leng N. Abbas A. et al.Molecular subsets in renal cancer determine outcome to checkpoint and angiogenesis blockade.Cancer Cell. 2020; 38 (this issue): 803-817Abstract Full Text Full Text PDF PubMed Scopus (42) Google Scholar). Samples from 823/915 patients were evaluated, including ∼200 from metastatic sites (799 ccRCC, including 110 with sarcomatoid differentiation; 24 non-clear cell RCC [nccRCC]/sarcomatoid). The authors perform RNA-seq and targeted mutation analyses (FoundationOne) and integrate the results with treatment outcomes. Unsupervised analyses of the most variably expressed genes identify 7 clusters grouped largely into an angiogenic clade (Clusters 1 and 2; 42% patients) and a proliferative clade (Clusters 4–6; 36%) (Figure 1). Cluster 1 (12%) and Cluster 2 (30%) are characterized by an angiogenic signature (including genes encoding VEGF and VEGFR2) and differ by the presence of stroma (Cluster 1). Both are low in immune-associated genes and are enriched for PBRM1 (∼60%) and KDM5C (∼20%) mutations (Figures 1A and 1B). Cluster 2, in particular, is characterized by fatty acid oxidation. Patients with tumors in these clusters had delayed progression with no difference between the treatment arms, which both contain AA drugs (progression free survival [PFS], 13–15 months). With a similar PBRM1/KDM5C mutation profile, Cluster 3 (19%) has less angiogenesis and high expression of complement cascade and cytochrome P450 proteins (Figure 1C). Clusters 4 (14%), 5 (9%), and 6 (13%), the proliferative clusters, have the highest mutation rates of the cell cycle inhibitors CDKN2A/B (20%–40%), express MYC and E2F target genes, and exhibit an anabolic metabolic profile, which may support proliferative demands (e.g., fatty acid and nucleotide synthesis [pentose phosphate pathway]) (Figure 1). Cluster 4 shows a T-effector gene signature (e.g., JAK/STAT and interferon modules). Cluster 5 (n = 74) has an unusually low rate of VHL mutation (20%) and includes 6/24 nccRCC/sarcomatoid as well as 15 tumors subsequently found to be MiT family translocation RCC (tRCC) (and might include other tumors histologically resembling ccRCC). Both Cluster 4 and Cluster 5 patients had improved PFS on the ICI-containing arm with a significant contribution from tRCC in Cluster 5. Cluster 6 (n = 106) includes 16/24 nccRCC/sarcomatoid tumors and has prominent stroma and myeloid signatures. Finally, Cluster 7 (3%) is characterized by increased expression of small nucleolar RNAs (snoRNAs), which guide chemical RNA modifications, especially SNORDs. This enigmatic cluster has the lowest mutation burden, but is preferentially responsive to the ICI-containing regimen. Admittedly, however, conclusions about Cluster 7 are limited by patient numbers (n = 28). This study follows on previous biomarker analyses of the IMmotion150 phase 2 trial, which largely focused on three gene expression signatures (angiogenesis, T-effector, and myeloid) (McDermott et al., 2018McDermott D.F. Huseni M.A. Atkins M.B. Motzer R.J. Rini B.I. Escudier B. Fong L. Joseph R.W. Pal S.K. Reeves J.A. 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: 749-757Crossref PubMed Scopus (466) Google Scholar), and the results are remarkably concordant across studies. Overall, the distribution into two large clades, one characterized by angiogenesis and fatty acid metabolism and the other by cell proliferation, is reminiscent of the original ccA and ccB classification, which was shown to be prognostic (Brannon et al., 2010Brannon A.R. Reddy A. Seiler M. Arreola A. Moore D.T. Pruthi R.S. Wallen E.M. Nielsen M.E. Liu H. Nathanson K.L. et al.Molecular stratification of clear cell renal cell carcinoma by consensus clustering reveals distinct subtypes and survival patterns.Genes Cancer. 2010; 1: 152-163Crossref PubMed Scopus (222) Google Scholar). A fundamental question is how does this diversification occur? Particularly for a tumor whose initiation is largely driven by the loss of one gene, VHL, this is remarkable. ccRCC often starts with a large deletion on chromosome 3p encompassing VHL (Turajlic et al., 2018Turajlic S. Xu H. Litchfield K. Rowan A. Chambers T. Lopez J.I. Nicol D. O’Brien T. Larkin J. Horswell S. et al.Tracking cancer evolution reveals constrained routes to metastases: TRACERx Renal.Cell. 2018; 173: 581-594Abstract Full Text Full Text PDF PubMed Scopus (288) Google Scholar), which also removes one copy of PBRM1, BAP1, and SETD2. Mutation of the second VHL allele likely follows, and VHL mutation rates are 70%–90% across ccRCC clusters. VHL is the only gene consistently showing truncational mutation (Turajlic et al., 2018Turajlic S. Xu H. Litchfield K. Rowan A. Chambers T. Lopez J.I. Nicol D. O’Brien T. Larkin J. Horswell S. et al.Tracking cancer evolution reveals constrained routes to metastases: TRACERx Renal.Cell. 2018; 173: 581-594Abstract Full Text Full Text PDF PubMed Scopus (288) Google Scholar), and Vhl loss is necessary for ccRCC tumorigenesis, as shown in mice (Gu et al., 2017Gu Y.F. Cohn S. Christie A. McKenzie T. Wolff N. Do Q.N. Madhuranthakam A.J. Pedrosa I. Wang T. Dey A. et al.Modeling renal cell carcinoma in mice: Bap1 and Pbrm1 inactivation drive tumor grade.Cancer Discov. 2017; 7: 900-917Crossref PubMed Scopus (64) Google Scholar). However, Vhl loss alone is insufficient (Gu et al., 2017Gu Y.F. Cohn S. Christie A. McKenzie T. Wolff N. Do Q.N. Madhuranthakam A.J. Pedrosa I. Wang T. Dey A. et al.Modeling renal cell carcinoma in mice: Bap1 and Pbrm1 inactivation drive tumor grade.Cancer Discov. 2017; 7: 900-917Crossref PubMed Scopus (64) Google Scholar). A mutation in the remaining copy of either Pbrm1 or Bap1 appears necessary (Figure 1C). Notably, mutations in PBRM1 and BAP1 split the evolutionary journey at the outset (Peña-Llopis et al., 2012Peña-Llopis S. Vega-Rubín-de-Celis S. Liao A. Leng N. Pavía-Jiménez A. Wang S. Yamasaki T. Zhrebker L. Sivanand S. Spence P. et al.BAP1 loss defines a new class of renal cell carcinoma.Nat. Genet. 2012; 44: 751-759Crossref PubMed Scopus (597) Google Scholar). These mutations tend to be mutually exclusive (Peña-Llopis et al., 2013Peña-Llopis S. Christie A. Xie X.J. Brugarolas J. Cooperation and antagonism among cancer genes: the renal cancer paradigm.Cancer Res. 2013; 73: 4173-4179Crossref PubMed Scopus (59) Google Scholar), and the resultant tumors are phenotypically different (Cancer Genome Atlas Research Network, 2013Cancer Genome Atlas Research NetworkComprehensive molecular characterization of clear cell renal cell carcinoma.Nature. 2013; 499: 43-49Crossref PubMed Scopus (2010) Google Scholar; Peña-Llopis et al., 2012Peña-Llopis S. Vega-Rubín-de-Celis S. Liao A. Leng N. Pavía-Jiménez A. Wang S. Yamasaki T. Zhrebker L. Sivanand S. Spence P. et al.BAP1 loss defines a new class of renal cell carcinoma.Nat. Genet. 2012; 44: 751-759Crossref PubMed Scopus (597) Google Scholar). PBRM1-deficient tumors tend to be well differentiated, of low grade, with iconic clear cells (with an expansive cytoplasm of lipids and glycogen) and a prominent vascular network (Kapur et al., 2020Kapur P. Christie A. Rajaram S. Brugarolas J. What morphology can teach us about renal cell carcinoma clonal evolution.Kidney Cancer J. 2020; 18: 68-75Google Scholar). That these traits are linked to VHL and PBRM1 has been shown in mice, where Vhl/ Pbrm1 inactivation reproduces the phenotype (Gu et al., 2017Gu Y.F. Cohn S. Christie A. McKenzie T. Wolff N. Do Q.N. Madhuranthakam A.J. Pedrosa I. Wang T. Dey A. et al.Modeling renal cell carcinoma in mice: Bap1 and Pbrm1 inactivation drive tumor grade.Cancer Discov. 2017; 7: 900-917Crossref PubMed Scopus (64) Google Scholar). Additional Tsc1 disruption in the mice leads to mTORC1 activation and increases tumor grade (Gu et al., 2017Gu Y.F. Cohn S. Christie A. McKenzie T. Wolff N. Do Q.N. Madhuranthakam A.J. Pedrosa I. Wang T. Dey A. et al.Modeling renal cell carcinoma in mice: Bap1 and Pbrm1 inactivation drive tumor grade.Cancer Discov. 2017; 7: 900-917Crossref PubMed Scopus (64) Google Scholar). Notably, TSC1 mutations are enriched in Cluster 3 (together with TSC2, they make 30%), and consistent with an mTORC1 role in this cluster, PTEN is also frequently inactivated (25%). Thus, Cluster 3 may represent a more evolved subtype (Figure 1C). This may also be the case for Cluster 7, which shares a similarly high PBRM1 mutation frequency to that of Clusters 1–3 (60%) but boasts the highest rates of SETD2 mutation (60%) (Figure 1C). SETD2 loss likely follows PBRM1 loss, and these mutations cooperate (Peña-Llopis et al., 2013Peña-Llopis S. Christie A. Xie X.J. Brugarolas J. Cooperation and antagonism among cancer genes: the renal cancer paradigm.Cancer Res. 2013; 73: 4173-4179Crossref PubMed Scopus (59) Google Scholar). Consistent with these findings, Clusters 3 and 7 are more frequently of high grade compared to Clusters 1 and 2 (Figures 1A and 1C). A separate evolutionary journey likely begins with a BAP1 mutation (Peña-Llopis et al., 2012Peña-Llopis S. Vega-Rubín-de-Celis S. Liao A. Leng N. Pavía-Jiménez A. Wang S. Yamasaki T. Zhrebker L. Sivanand S. Spence P. et al.BAP1 loss defines a new class of renal cell carcinoma.Nat. Genet. 2012; 44: 751-759Crossref PubMed Scopus (597) Google Scholar). Unlike PBRM1-deficient tumors, BAP1-deficient tumors are of high grade and associated with worse outcomes (Cancer Genome Atlas Research Network, 2013Cancer Genome Atlas Research NetworkComprehensive molecular characterization of clear cell renal cell carcinoma.Nature. 2013; 499: 43-49Crossref PubMed Scopus (2010) Google Scholar; Peña-Llopis et al., 2012Peña-Llopis S. Vega-Rubín-de-Celis S. Liao A. Leng N. Pavía-Jiménez A. Wang S. Yamasaki T. Zhrebker L. Sivanand S. Spence P. et al.BAP1 loss defines a new class of renal cell carcinoma.Nat. Genet. 2012; 44: 751-759Crossref PubMed Scopus (597) Google Scholar). A deterministic role for BAP1 has been shown in mice, where Bap1 loss induces more proliferative, higher-grade tumors (Gu et al., 2017Gu Y.F. Cohn S. Christie A. McKenzie T. Wolff N. Do Q.N. Madhuranthakam A.J. Pedrosa I. Wang T. Dey A. et al.Modeling renal cell carcinoma in mice: Bap1 and Pbrm1 inactivation drive tumor grade.Cancer Discov. 2017; 7: 900-917Crossref PubMed Scopus (64) Google Scholar). In both humans and mice, these tumors are inflamed (Wang et al., 2018Wang T. Lu R. Kapur P. Jaiswal B.S. Hannan R. Zhang Z. Pedrosa I. Luke J.J. Zhang H. Goldstein L.D. et al.An empirical approach leveraging tumorgrafts to dissect the tumor microenvironment in renal cell carcinoma identifies missing link to prognostic inflammatory factors.Cancer Discov. 2018; 8: 1142-1155Crossref PubMed Scopus (55) Google Scholar). Consistent with a link between inflammation and BAP1, Cluster 4, characterized by a T-effector signature, the highest PD-L1 protein expression rates (80%), and preferential responsiveness to the ICI-containing regimen, has the highest BAP1 mutation rate (40%). These tumors may further evolve into sarcomatoid tumors, which show similarly high BAP1 mutation frequency (40%) (Figure 1C). Inflammation likely transcends the tumor confines manifesting systemically by thrombocytosis and anemia (Wang et al., 2018Wang T. Lu R. Kapur P. Jaiswal B.S. Hannan R. Zhang Z. Pedrosa I. Luke J.J. Zhang H. Goldstein L.D. et al.An empirical approach leveraging tumorgrafts to dissect the tumor microenvironment in renal cell carcinoma identifies missing link to prognostic inflammatory factors.Cancer Discov. 2018; 8: 1142-1155Crossref PubMed Scopus (55) Google Scholar). Thrombocytosis, anemia, and neutrophilia have traditionally been linked with poor-risk disease and worse prognosis. Accordingly, there is an enrichment among the poor-risk disease for proliferative clusters (4, 5, 6). Given the emerging link between inflammation and ICI responsiveness, the extent to which prognostic models developed in the AA era will remain prognostic after ICI is unclear. While the foregoing is an oversimplification, it provides an evolutionary framework to understand ccRCC progression. Paralleling the antithetical relationship between PBRM1 and BAP1 (Peña-Llopis et al., 2013Peña-Llopis S. Christie A. Xie X.J. Brugarolas J. Cooperation and antagonism among cancer genes: the renal cancer paradigm.Cancer Res. 2013; 73: 4173-4179Crossref PubMed Scopus (59) Google Scholar), signatures of angiogenesis and inflammation tend not to overlap (Brannon et al., 2010Brannon A.R. Reddy A. Seiler M. Arreola A. Moore D.T. Pruthi R.S. Wallen E.M. Nielsen M.E. Liu H. Nathanson K.L. et al.Molecular stratification of clear cell renal cell carcinoma by consensus clustering reveals distinct subtypes and survival patterns.Genes Cancer. 2010; 1: 152-163Crossref PubMed Scopus (222) Google Scholar; McDermott et al., 2018McDermott D.F. Huseni M.A. Atkins M.B. Motzer R.J. Rini B.I. Escudier B. Fong L. Joseph R.W. Pal S.K. Reeves J.A. 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: 749-757Crossref PubMed Scopus (466) Google Scholar; Wang et al., 2018Wang T. Lu R. Kapur P. Jaiswal B.S. Hannan R. Zhang Z. Pedrosa I. Luke J.J. Zhang H. Goldstein L.D. et al.An empirical approach leveraging tumorgrafts to dissect the tumor microenvironment in renal cell carcinoma identifies missing link to prognostic inflammatory factors.Cancer Discov. 2018; 8: 1142-1155Crossref PubMed Scopus (55) Google Scholar; Motzer et al., 2020Motzer R.J. Banchereau R. Hamidi H. Powles T. McDermott D. Atkins M.B. Escudier B. Liu L.-F. Leng N. Abbas A. et al.Molecular subsets in renal cancer determine outcome to checkpoint and angiogenesis blockade.Cancer Cell. 2020; 38 (this issue): 803-817Abstract Full Text Full Text PDF PubMed Scopus (42) Google Scholar). However, how the mutations and gene expression patterns are linked is not well understood, although PBRM1 loss may amplify the HIF-VEGF angiogenic response downstream of VHL. Representing potential extremes of angiogenic (AA-responsive) and proliferative/inflamed (ICI-responsive) tumors are ccRCCs with pancreatic metastases and those with sarcomatoid differentiation, respectively (Kapur et al., 2020Kapur P. Christie A. Rajaram S. Brugarolas J. What morphology can teach us about renal cell carcinoma clonal evolution.Kidney Cancer J. 2020; 18: 68-75Google Scholar). ccRCC with metastasis to the pancreas (independently of other organs) is associated with improved survival and epitomizes PBRM1-deficient low-grade ccRCC, with preferential AA drug responsiveness (Kapur et al., 2020Kapur P. Christie A. Rajaram S. Brugarolas J. What morphology can teach us about renal cell carcinoma clonal evolution.Kidney Cancer J. 2020; 18: 68-75Google Scholar). In contrast, sarcomatoid ccRCCs are resistant to AA and ICI responsive (Motzer et al., 2020Motzer R.J. Banchereau R. Hamidi H. Powles T. McDermott D. Atkins M.B. Escudier B. Liu L.-F. Leng N. Abbas A. et al.Molecular subsets in renal cancer determine outcome to checkpoint and angiogenesis blockade.Cancer Cell. 2020; 38 (this issue): 803-817Abstract Full Text Full Text PDF PubMed Scopus (42) Google Scholar). At some level, it is puzzling that the identification of angiogenic, AA-responsive, and inflamed, ICI-responsive, subtypes has required gene expression analyses. The notion is, at least to some extent, axiomatic and reinforced by the fact that neither AA nor ICI exert direct effects on tumor cells but rather act through vascular and immune cells, respectively. In addition, angiogenesis and inflammation (as well as the presence of stroma and of proliferating cells), are traits appreciable through phenotypic analyses of histological slides. Thus, histology is well poised to advance the biomarker field, and its complexity is beginning to be dismantled (Kapur et al., 2020Kapur P. Christie A. Rajaram S. Brugarolas J. What morphology can teach us about renal cell carcinoma clonal evolution.Kidney Cancer J. 2020; 18: 68-75Google Scholar). With the advent of digital pathology, features imperceptible to the human eye may become appreciable, new associations may be uncovered, and findings will be both objectivized and made more generalizable. The dichotomy between angiogenic and proliferative/inflamed tumors raises questions about the current therapeutic paradigm combining AA and ICI. Inasmuch as the combination has something to offer to different subtypes, unprecedented response rates are unsurprising. However, this comes at the cost of possibly unnecessary therapy for some patients as well as added toxicity and expense. While the field is moving toward ever-broader approaches including triple-combination therapies, these efforts should be balanced through the identification of predictive biomarkers supporting precision medicine. J.B., A.C., and P.K. are supported by NIH ( P50 CA196516 ) and by the Cancer Research & Prevention Institute of Texas (CPRIT; RP180192 ). J.B. and A.C. are also supported by CPRIT ( RP180191 ). P.K. is supported by NIH ( R01CA244579 , R01CA154475 , and R01DK115986 ), DOD ( W81XWH1910710 ), and CPRIT ( RP200233 ). S.R. is supported by startup funds provided through the Lyda Hill Department of Bioinformatics . J.B. is a consultant for and has received research funding from Arrowhead Pharmaceuticals and is a consultant for Exelixis. J.B. holds US Patent No. 15/761,534. Molecular Subsets in Renal Cancer Determine Outcome to Checkpoint and Angiogenesis BlockadeMotzer et al.Cancer CellNovember 5, 2020In BriefMotzer et al. perform integrative multi-omics analyses of 823 renal cancer tumors from a randomized clinical trial. A robust molecular classification scheme, based on transcriptional and gene alteration profiles and differential clinical outcomes to VEGF blockade alone or in combination with anti-PD-L1, informs personalized treatment strategies and future therapeutic development in RCC. Full-Text PDF Open Archive
Referência(s)