Cancer Treatment Reviews xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Cancer Treatment Reviews journal homepage: www.elsevierhealth.com/journals/ctrv
Anti-Tumour Treatment
A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma Tomohiro Funakoshi a,⇑, Chung-Han Lee b,1, James J. Hsieh b,c,d,2 a
Department of Medicine, Beth Israel Medical Center, University Hospital and Manhattan Campus for the Albert Einstein College of Medicine, New York, NY, USA Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York, NY, USA c Human Oncology and Pathogenesis Program, Memorial Sloan-Kettering Cancer Center, New York, NY, USA d Department of Medicine, Weill Medical College of Cornell University, New York, NY, USA b
a r t i c l e
i n f o
Article history: Received 9 October 2013 Received in revised form 23 November 2013 Accepted 25 November 2013 Available online xxxx Keywords: Vascular endothelial growth factor-targeted therapy Systematic review Metastatic renal cell carcinoma Biomarkers
a b s t r a c t Background: Vascular endothelial growth factor (VEGF)-targeted therapy is the currently standard treatment for advanced and metastatic renal cell carcinoma (RCC). Multiple candidate predictive and prognostic biomarkers have been evaluated. We performed a systematic review and graded the available evidence on the biomarkers for VEGF-targeted therapy in RCC. Methods: We conducted an independent review of PubMed and ASCO databases up to August 2013. Studies were included if biomarkers obtained from metastatic clear-cell RCC patients treated with the FDAapproved VEGF-targeted therapy were assessed for their correlation with clinical outcomes. We graded the studies and determined the Level-of-evidence for each biomarker using a previously published framework. Results: A total of 50 articles were selected for this review. Seven studies assessed the predictive value of biomarkers using the archived specimens from randomized controlled trials. Five predictive biomarkers, such as VEGF, interleukin (IL)-6, hepatocyte growth factor (HGF), osteopontin, single nucleotide polymorphisms in IL-8, satisfied Level II evidence. IL-6 is the most corroborated predictive biomarker based on its consistent predictive value in two different trials. The prognostic value of biomarkers was assessed in 48 studies using the archived specimens from clinical trials, prospective and retrospective observational registries. Three biomarkers, including IL-8, HGF and osteopontin, satisfied Level I evidence for PFS. Conclusion: Though several promising predictive biomarkers for VEGF-targeted therapy have been found, none of them has satisfied the determination of Level I evidence. A more focused development of biomarkers with prospective assessment in clinical trials and clear intent of use in clinical practice is needed. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Clear-cell renal cell carcinoma (RCC) is refractory to cytotoxic chemotherapy. Until recently, interleukin 2 (IL-2) and interferon (IFN) were the standard of care for advanced and metastatic clear-cell RCC, even though they only benefit a small portion of patients at the expense of substantial toxicity. The elucidation of the von Hippel-Lindau (VHL) pathway has led to the development of therapy targeted at specific molecular alterations. VHL gene inactivation, which occurs in the majority of sporadic clear cell RCC, results in accumulation of hypoxia-inducible factor (HIF) alpha subunits, which causes downstream upregulation of a number of ⇑ Corresponding author. Address: 1st Ave at 16th Street, New York, NY 10003, USA. Tel.: +1 917 696 9211; fax: +1 212 420 4615. E-mail addresses:
[email protected] (T. Funakoshi),
[email protected] (C.-H. Lee),
[email protected] (J.J. Hsieh). 1 Tel.: +1 734 717 0356. 2 Tel.: +1 646 888 3263; fax: +1 646 888 3266.
pro-angiogenic factors, including the vascular endothelial growth factor (VEGF) [1]. This led to the development of VEGF signaling pathway inhibitors which have been shown to benefit patients in randomized phase III clinical trials. Currently, there are five angiogenesis inhibitors approved by the United States Food and Drug Administration (FDA) for treatment of advanced and metastatic RCC; four VEGFR targeted tyrosine kinase inhibitors (TKIs), including sorafenib, sunitinib, axitinib and pazopanib, and one anti-VEGF monoclonal antibody, bevacizumab [2–6]. Despite the improved outcomes shown in the clinical trials of VEGF-targeted therapy, the length of response and survival benefit of therapy varies considerably between patients. In addition, VEGF signaling pathway inhibitors have been associated with various toxicities including an increased risk of fatal adverse events [7,8]. Therefore, an identification of biomarkers for efficacy is necessary to select suitable patients for this therapeutic approach. Molecules related to the underlying biology of RCC have been investigated as potential biomarkers for prediction of therapeutic benefit. The
0305-7372/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ctrv.2013.11.008
Please cite this article in press as: Funakoshi T et al. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat Rev (2014), http://dx.doi.org/10.1016/j.ctrv.2013.11.008
2
T. Funakoshi et al. / Cancer Treatment Reviews xxx (2014) xxx–xxx
development of biomarkers requires clinical validity, or the ability of the assay to predict the clinical endpoint of interest as well as clinical utility meaning that the biomarker is actionable for informing treatment decisions in a manner that provides improved patient outcome [9]. In 1996, the American Society of Clinical Oncology Tumor Markers Guidelines Committee recommended the five Levels of evidence (LOE) to determine the clinical validity and utility of a biomarker [10]. Simon et al. proposed an updated revision of the LOE scale providing more precise definitions of key elements for biomarker studies that constitute LOE determination [9]. There has been no systematic attempt to review and grade biomarkers for VEGF-targeted therapy using this standardized LOE scale. Therefore, we conducted a systematic review and graded the LOE in studies that assessed the clinical validity and utility of biomarkers for VEGF signaling pathway inhibitors in patients with metastatic RCC. Methods Study selection Studies were selected and systemically reviewed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [11]. Studies were considered eligible if biomarkers obtained from metastatic clear-cell RCC patients treated with the FDA-approved VEGF-targeted therapy were assessed for their correlation with clinical outcomes. Eligible clinical outcomes included a response rate based on the Response Evaluation Criteria in Solid Tumors (RECIST), progression-free survival (PFS), and overall survival (OS) [12]. We only considered genetic, proteomic and cellular biomarkers related to the pathways targeted by the VEGF-targeted agents or the alternative pathways that may mediate resistance and genomics which can modulate drug metabolism and mediate activity. Studies were excluded if they only assessed laboratory-based factors such as electrolytes, red blood cells, white blood cells, platelets, liver enzymes, lactate dehydrogenase, erythrocyte sedimentation rate and C-reactive protein. We included studies with a sample size greater than 10 patients because the ability to evaluate any biomarkers would be negligible in smaller studies.
data, treatment (line of treatment), sample size, material, methodology, biomarkers studied, clinical outcomes, statistical methods and cut-off point. Outcome definition Biomarkers can play roles in predictive and prognostic characterizations of a patient’s disease. Predictive biomarkers indicate whether a patient will benefit from a given treatment. Prognostic biomarkers provide information about a patient’s likely clinical outcomes with or without treatment. To establish a predictive biomarker, controlled studies are required, while a prognostic biomarker can be established with single arm studies. We examined the clinical validity and utility of biomarkers on the basis of the criteria originally proposed by Hayes et al. and revised by Simon et al. [9,10]. The key elements of biomarker studies that are used to generate a Levels of evidence (LOE) determination included characteristics of (1) clinical trial design; (2) patients and patient data; (3) specimen collection, processing, and archival; (4) statistical design and analysis; and (5) consistency in validation results. According to the scale, Category A study represents prospective randomized clinical trials designed and powered specifically to address biomarker questions. Category B study represents prospective studies not primarily designed to address biomarker questions, rather archive specimens for retrospective analysis of biomarkers. Category C study represents prospective, observational registry studies. Category D study represents retrospective, observational studies. Level I evidence is defined as at least one study from Category A, or one or more studies from Category B with consistent results. Level II evidence includes at least one study from Category B or two or more studies from Category C with consistent results. Level III evidence includes at least one study from Category C, and Levels IV and V evidence includes studies from Category D. Based on this scale, two investigators (TF and CHL) independently graded the studies. If there were any discrepancies between the reviewers, a third investigator (JJH) reviewed the article and finalized the grading. Results Search results
Search strategy We conducted a review of PubMed from January 1966 to August 2013. The search keywords were: , , and . We also searched abstracts and virtual meeting presentations containing the same search terms from the American Society of Clinical Oncology (ASCO) conferences held up to March 2013 in order to identify relevant studies. An independent search of the Web of Science, Embase and Cochrane electronic databases was also performed to ensure that no additional studies were overlooked. In cases of duplicate publications, only the most complete, recent, and updated report of the study was included. Independent reviewers (TF and CHL) screened reports that included the key terms by their titles and abstracts for relevance. Finally, full texts of the relevant articles were retrieved to assess eligibility. Data extraction Two investigators (TF and CHL) independently performed data extraction. Any discrepancies between reviewers were resolved by consensus. The following information was recorded for each study: first author’s name, year of publication, source of patient
Our search strategy yielded 474 potentially relevant articles in PubMed. 371 articles were excluded during the initial title and abstract screening. The remaining 105 articles were retrieved for full review and 59 articles were excluded. We also included four studies from ASCO presentations. Supplement 1 outlines the selection process and reasons for study exclusion. Tran et al. [47] and Figueroa et al. [61] obtained samples from a phase II and a phase III trial and assessed the biomarkers in each trial independently. We considered their analysis of biomarkers using each trial as independent study. A total of 50 articles (52 studies) were selected for this review. Evaluation of biomarker studies We found no prospective controlled trial designed to address biomarkers (Category A study). There were 21 Category B studies which evaluated candidate biomarkers based on archived specimens from previously conducted clinical trials. Seven studies assessed the predictive value of biomarkers for VEGF-targeted therapy using the archived specimens from randomized trials [25,32,35,47,50,60,62]. The prognostic value of biomarkers for VEGF-targeted therapy was evaluated in 17 Category B studies. 20 studies satisfied Category C criteria. 14 studies were prospective
Please cite this article in press as: Funakoshi T et al. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat Rev (2014), http://dx.doi.org/10.1016/j.ctrv.2013.11.008
T. Funakoshi et al. / Cancer Treatment Reviews xxx (2014) xxx–xxx
observational studies and six studies used existing hospital registries. There were 11 Category D studies which were retrospective observational studies. Because the establishment of a predictive biomarker requires controlled studies, the biomarkers assessed in Category C and D studies are considered prognostic (Table 1). Predictive biomarkers Baseline VEGF-A levels in blood were assessed in five studies [25,33,47,50,62]. In the Phase III trial of sorafenib versus placebo, patients with elevated VEGF-A experienced a greater PFS benefit from sorafenib treatment than those with low VEGF-A [25]. In the randomized phase II trial of sorafenib plus IFN-a versus sorafenib, patients with low VEGF-A obtained greater PFS benefit from sorafenib plus IFN-a than those with high VEGF-A [33]. However, there was no significant difference in hazard ratio (HR) for PFS or OS in intention to treat populations according to baseline VEGF-A levels in the other three studies [47,50,62]. Three studies evaluated baseline interleukin 6 (IL-6) levels in blood [33,47,62]. Tran et al. identified that high IL-6 concentrations were predictive of improved relative PFS benefit from pazopanib compared with placebo [47]. In the phase III trial of IFN-a plus bevacizumab versus IFN-a, a significant 3-way interaction between IL-6, hepatocyte growth factor (HGF) and treatment was observed. Patients with high IL-6 and either high or low HGF levels and patients with low IL-6 and low HGF levels had OS benefit from the addition of bevacizumab. Baseline osteopontin levels in blood were also assessed in the three studies [33,47,62]. Zurita et al. reported patients with low osteopontin obtained a greater PFS benefit from sorafenib plus IFN-a than those with high osteopontin [33]. The predictive value of osteopontin was not significant in the two other studies [47,62]. In addition to using only single biomarkers, clusters of cytokines and angiogenic factors (CAFs) have been studied [33,47]. Zurita et al. found that a high six-marker baseline CAFs signature [osteopontin, VEGF-A, carbonic anhydrase IX (CAIX), collagen IV (ColIV), VEGFR-2 (VEGF receptor 2), and tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)] correlated with a greater PFS benefit from sorafenib alone than from sorafenib plus IFN-a [33]. In the phase III trial of pazopanib versus placebo, high concentrations of six CAFs signature [IL-6, IL-8, HGF, osteopontin, VEGF-A, and tissue inhibitor of metalloproteinase 1 (TIMP-1)] demonstrated no significant effect on HR for PFS or OS in intention to treat populations, but a predictive value for an improved relative OS benefit for pazopanib on the basis of treatment received (including crossover) [47] Wu et al. evaluated a predictive value of polymorphisms in IL-8 in two trials [35,60]. They showed a significant interaction between PFS and IL-8 2767A > T polymorphism in the phase III trial of pazopanib versus placebo [35]. However, the HRs for effects of IL-8 polymorphisms were not significantly different in the phase III trial of pazopanib versus sunitinib [60]. Based on these results, no biomarker has satisfied Level I evidence for predicting PFS or OS benefit from VEGF-targeted therapy. VEGF, IL-6 with or without HGF, osteopontin, six-CAFs signature [osteopontin, VEGF-A, CAIX, ColIV, VEGFR-2, and TRAIL] and IL-8 polymorphisms satisfied the determination of Level II evidence. Table 2 summarizes the predictive biomarkers. For these biomarkers to be used in clinical setting, the results need to be confirmed using specimens from another Category B study based on archived tissue from a different trial that has been designed, conducted, and analyzed in a similar manner. Prognostic biomarkers Blood-based biomarker VEGF levels in blood were assessed in 13 studies. One Category B and C study showed high baseline VEGF-A concentrations
3
correlated with shorter OS and PFS, respectively [22,47]. Two Category B studies reported a greater increase in VEGF-A levels after treatment was associated with better clinical outcomes [14,41], while two Category C studies reported the opposite results [18,31]. One Category B study identified that baseline VEGF-C levels were prognostic for better response and PFS [16]. Eight studies evaluated soluble VEGFR (sVEGFR) levels. Patients with low sVEGFR-3 levels had better PFS and response in one Category B study [16]. Two Category B studies reported a greater reduction in sVEGFR-3 levels after treatment was associated with better clinical outcomes [14,37]. One Category B study reported the similar result for sVEGFR-2 [14]. Three other angiogenic factors (IL-8, HGF and epidermal growth factor) and 11 immunomodulatory factors (IL-5, IL-6, osteopontin, transforming growth factor b-1, tumor necrosis factor -a, interferon a receptor-2 [IFNaR-2], stromal cellderived factor 1, macrophage colony-stimulating factor and E-selectin) in blood were reported to be prognostic. Tran et al. showed higher baseline interleukin-8, HGF and osteopontin levels were prognostic for a shorter PFS both in the single arm phase II trial of pazopanib and in the pazopanib arm of the phase III trial. Higher baseline IL-6 levels were prognostic for a shorter PFS in the phase II trial, but only prognostic for OS in the phase III trial. They also reported high concentrations of six CAFs signature [IL-6, IL-8, HGF, osteopontin, VEGF-A, and TIMP-1] correlated with shorter PFS in the phase III trial [47]. Table 3 describes the bloodbased biomarkers with statistical significant results and LOE for each blood-based biomarker is shown in Table 4. Tissue-based biomarker Expression of angiogenesis related proteomic markers in tumor specimens, such as VEGF, VEGFR and platelet-derived growth factor receptor (PDGFR), was assessed in five studies. One Category D study showed higher expression of VEGF-A was associated with better PFS and OS [38]. Overexpression of VEGFR-2 and VEGFR-3 was prognostic for better response and PFS in one Category D study respectively [34,58]. While high PDGFR-b expression was associated with better response in one Category D study, high PDGFR-a expression was related to worse PFS in one Category C study [58,40]. Five studies assessed immunomodulatory factors in tissue. Higher expression of IFNaR-2, C-X-C chemokine receptor type 4 (CXCR-4) and programmed cell death 1 ligand 1 (PD-L1) were associated with worse outcomes in one Category C, D and B study, respectively [30,44,61]. Expression of molecular factors related to the PI3K/AKT, hypoxia, metabolism and cell-cycle signaling pathway have also been studied. For the PI3K/AKT pathway, phosphatase and tensin homolog (PTEN) was overexpressed in patients with longer PFS in one Category B study, but one Category D study reported positive PTEN expression was related to better PFS [21,46]. Jonasch et al. reported higher expression of AKT and phosphorylated AKT (pAKT) was related to better and worse survival, respectively in their Category B study [19]. In addition, one Category B study showed patients with longer OS had lower expression of pAKT and phosphorylated S6 ribosomal protein (pS6RP) than patients with shorter OS [21]. For hypoxia signaling pathway, one Category D study showed higher expression of HIF-1a correlated with poor PFS, while Category D study reported higher expression of HIF-1a was associated with poor response and survival. Increased HIF-2a expression was associated with longer PFS and OS in two Category D studies [43,46]. Though higher CAIX expression was reported to be related to worse response and survival in one Category D study, the other three studies did not find CAIX expression status to be prognostic [23,46,47,58]. For metabolism signaling pathway, Tsavachidou-Fenner et al. reported high protein levels of adenosine monophosphate kinase (AMPK) and phosphorylated AMPK were associated with longer survival in their Category B study [21]. For cell-cycle signaling pathway, low levels of cyclin
Please cite this article in press as: Funakoshi T et al. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat Rev (2014), http://dx.doi.org/10.1016/j.ctrv.2013.11.008
4
First author, year
Source of patient data
Treatment (line of treatment)
Sample size
Material
Methodology
Biomarkers studied
Study category
Rini, 2006 [13]
Hospital registry
SUN, AX, BEV + IFN (2nd)
43
FFPE tumor
Sequencing
VHL gene mutational status
C
Deprimo, 2007 [14] Choueiri, 2008 [15]
Single arm phase II trial Hospital registry
SUN (2nd) SUN, SOR, BEV, AX (1st or 2nd)
63 123
FFPE tumor Plasma FFPE tumor
MS-PCR ELISA Sequencing
VHL gene methylation status VEGF-A, sVEGFR-2, PlGF, sVEGFR-3 VHL gene mutational status
B C
Rini, 2008 [16] van Cruijsen, 2008 [17]
Single arm phase II trial Hospital registry
SUN (2nd or 3rd) SUN (1st or 2nd)
61 26
FFPE tumor Plasma Blood
MS-PCR ELISA Flow cytometry
Kontovinis, 2009 [18]
Prospective observational registry Randomised phase II trial
SUN (1st or 2nd)
39
Serum
ELISA
VHL gene methylation status VEGF-A, VEGF-C, sVEGFR-3, PlGF CD11c(hi)CD19( )CD14( )BDCA-1(+)MDC-1, CD11c(+)CD14( )BDCA-3(+)MDC-2 VEGF-A, sVEGFR-2, PDGF
SOR vs SOR + IFN (1st) SUN (2nd) BEV ± Erlotinib
40 (SOR = 22, SOR + IFN = 18) 21 37
FFPE tumor
IHC
AKT, pAKT, S6RP, pS6RP, P70S6K
B
Serum Frozen tumor obtained after 8weeks of treatment
ELISA RPPA ± immunoblotting, RNA microarray
C B
SUN (2nd)
85
Serum
ELISA
TNF-a, MMP-9, ICAM-1, VEGF-A, SDF-1, BDNF AKT, pAKT(S473),pAKT(T308), AMPK, pAMPK, 4EBP1, p4EBP1, pEGFR, pMAPK, pP70S6K, T-S6, pS6RP(S235), pS6RP(S240), PTEN, SRC, pSRC, pGSK-3, pACC-1, TSC-2, pTSC-2, pIRS-1, MEK-1, CyclinB1, Cyclin-D1, Caveolin, P21, P27, pFOXO3a, PKC-a, pPKC-a, BCL-2, PARP-cleaved, E-cadherin , b-catenin, CASP7-cleaved, CD31 VEGF-A, NGAL
SUN, SOR, BEV, Vatalanib (1st) SUN (2nd)
94
FFPE tumor
IHC
CAIX
C
23
Frozen tumor
qRT-PCR
mRNA of VEGF (isoforms 121, 165, 189), VEGFR-1, VEGFR-2, PDGF-
C
Jonasch, 2009 [19] Perez-Gracia, 2009 [20] Tsavachidou-Fenner, 2010 [21]
Hospital registry Single arm phase II trial
Porta, 2010 [22]
Prospective observational registry Hospital registry
Choueiri, 2010 [23] Paule, 2010 [24]
Prospective observational registry
C
C
a, PDGF-b, PDGFR-a, PDGFR-b, FGF-2, HIF-1a, CXCR-4, uPA, uPAR, PAI-1, LYVE-1 VEGF, VEGFR-1, VEGFR-2, PDGFR-b VEGF-A, sVEGFR-2, CAIX, TIMP-1, Ras-P21 VHL gene mutational status CD3(+)CD4(+)CD25(hi)Foxp3(+)T cells
28
Serum FFPE tumor Blood
IHC ELISA Sequencing Flow cytometry
SUN (1st, 2nd or 3rd) SOR (1st or 2nd)
136
FFPE tumor Blood
IHC Genotyping
46
Blood Blood
qRT-PCR Flow cytometry
Prospective observational registry
SUN (1st or 2nd)
26
Blood
Flow cytometry
CD3(+)Foxp3(+)T cells SNPs in ABCB1, ABCG2, NR1I2, NR1I3, CYP3A5, CYP1A1, CYP1A2, VEGFR-2, VEGFR-3, PDGFR-a, FLT-3 TGF-b1, IL-10 mRNA CD4(+)CD25(hi)T cells, CD4(+)CD25(hi)FoxP3(+)T cells, IL-10producing monocytes, TGF-b-producing monocytes CECs
Furuya, 2011 [30]
Prospective observational registry
IFN ± SOR (1st or 2nd)
66
ELISA qRT-PCR
sVEGFR-2 INFaR-1, INFaR-2 mRNA
Farace, 2011 [31]
Prospective observational registry
SUN, SOR (1st or 2nd)
55
Serum Frozen tumor, serum Frozen tumor Blood
Immunoblotting Flow cytometry
Kusuda, 2011 [32]
Retrospective observational registry
SOR (2nd)
45
Plasma FFPE tumor
ELISA IHC
Zurita, 2011 [33]
Randomised phase II trial
SOR vs SOR + IFN (1st)
69 (SOR = 34, SOR + IFN = 35)
Plasma
ELISA
pAKT(S473), AKT, pS6RP(S235) CD45( )CD31(+)CD146(+)7AAD( )CECs, CD45(dim)CD34(+)VEGFR2(+)7AAD( ) progenitor cells VEGF-A, sVEGFR-2, SDF-1a, sVCAM-1 BCL-2, BCL-XL, BAX, Clusterin, Cyclin-D1, Cyclin-D2, Cyclin-D3, Cyclin-E, AKT, pAKT, P44/42 MAPK, pP44/42 MAPK, STAT3, pSTAT3, VEGFR-1, VEGFR-2, VEGFR-3, PDGFR-a, PDGFR-b sCAIX, ColIV, CTACK, EGF, Eotaxin, E-selectin, BFGF, G-CSF, GM-CSF, GRO-alpha, HGF, IFNa-2, IFNc, IL-1b, IL-1RA, IL-2, IL2-RA, IL-3, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12 P40, IL-12 P70, IL-13, IL-16, IL17, IL-18, IP-10, MCP-1, MCP-3, M-CSF, MIF, MIG, MIP-1a, MIP-1b, b-NGF, OPN, PlGF, PDGF-BB, RANTES, SCF, SCGF-b, SDF-1a, TNF-a, TRAIL, VEGF-A, sVEGFR-2
Peña, 2010 [25]
Phase III trial
SOR vs PL (2nd)
712
Adotevi, 2010 [26]
Prospective observational registry
SUN ± BEV (1st or 2nd)
van der Veldt, 2010 [27]
Retrospective observational registry Hospital registry
Gruenwald, 2010 [29]
Busse, 201128
B C
B C
D C
C
C
C
D
B
T. Funakoshi et al. / Cancer Treatment Reviews xxx (2014) xxx–xxx
Please cite this article in press as: Funakoshi T et al. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat Rev (2014), http://dx.doi.org/10.1016/j.ctrv.2013.11.008
Table 1 Characteristics of the studies included in the systematic review.
Xu, 2011 [35] Kim, 2012 [36] Tomita, 2011 [37] Barrios, 2011 [38] Garcia-Donas, 2011 [39] Siraj, 2012 [40] Mancuso, 2012 [41]
de Martino, 2012 [42] Saez, 2012 [43]
D’ Alterio, 2012 [44]
Retrospective observational registry Single arm phase II and phase III trials Retrospective observational registry Single arm phase II trial Single arm phase II trial Prospective observational registry Retrospective observational registry Single arm phase II trial
SUN (1st)
40
FFPE tumor
IHC
PAZ (1st or 2nd)
397
Blood
Genotyping
SUN (1st, 2nd or 3rd) AX (2nd) SUN (1st) SUN (1st)
63
Genotyping
64 116 89
Blood, FFPE, frozen tumor Plasma Plasma Blood, saliva
SUN (1st)
50
FFPE tumor
IHC
sVEGFR-1, sVEGFR-2, sVEGFR-3, VEGF-A, sKIT VEGF-A, sVEGFR-2 SNPs in VEGFR-2, VEGFR-3, PDGFR-a, VEGF-A, IL-8, CYP3A4, CYP3A5, ABCB1, ABCG2 FSHR
SOR (2nd or 3rd)
19
Prospective observational registry Retrospective observational registry
SUN (1st)
22
FFPE tumor FFPE tumor Plasma Blood Serum
Sequencing IHC ELISA Flow cytometry ELISA
B-RAF, PDGFR-a, C-KIT gene mutational status B-RAF VEGF-A CD133(+)CD34(+)CD45( )EPCs, CD146(+)CD31(+)CD45( )CECs 20S proteasome
C
SUN, SOR, PAZ (1st or 2nd)
80
FFPE tumor
IHC
HIF-1a, HIF-2a
D
FFPE tumor FFPE tumor Blood Tumor
Sequencing MS-PCR Genotyping IHC
VHL gene mutational status VHL gene methylation status SNPs in VEGF-A, VEGFR-2 CXCR-4
D
Blood
Genotyping
SNPs in VEGF-A, VEGF-B, VEGF-C, VEGF-D, VEGFR-1, VEGFR-2, PlGF,HIF-1a, HIF-2a, FIH-1, VHL, EGLN2, EGLN1, EGLN3
B
FFPE tumor
IHC
HIF-1a, CAIX, PTEN, P21
D
215 343 (PAZ = 225, PL = 118) 133 (SOR = 66, PL = 67) 75
Plasma Plasma
ELISA, ELISA
IL-6, IL-8, VEGF-A, HGF, TIMP-1, E-selectin, OPN, sCAIX, ColIV IL-6, IL-8, VEGF-A, HGF, TIMP-1, E-selectin, OPN
B B
FFPE tumor
IHC
CAIX
B
Serum
ELISA
BFGF, HGF, IL-6
C
384 (BV + IFN = 192, PL + IFN = 192) 70
Plasma
ELISA
VEGF-A
B
Serum
ELISA
sCAIX
B
38
Blood
RNA microarray
miRNA
C
SOR (1st or 2nd)
18
Plasma
qRT-PCR
cfDNA
C
SUN (1st or 2nd)
88
Frozen tumor
Genotyping
D
SUN (1st)
91
Frozen tumor
Genotyping
SNPs in VEGFR-2, VEGFR-3, PDGFR-a, FGFR-2, IL-8, HIF-1a, ABCB1, CYP3A5, NR1I2, NR1I3 SNPs in VEGFR-1
SUN (1st)
84
Frozen tumor
Genotyping
SNPs in VEGF-A, VEGF-C, VEGFR-1, VEGFR-2, VEGFR-3
D
SOR, SUN, Everolimus, Temsirolimus (1st or 2nd)
58
Blood
Flow cytometry
CD4(+)IFNc(+)IL-4( )cells(Th1), CD4(+)IFNc( )IL-4(+)cells(Th2), CD4(+)CD25(+)FoxP3(+)cells (Treg), CD83(+)cells
C
Retrospective observational registry Phase III trial
SUN (1st)
62
BEV + IFN vs PL + IFN (1st)
Muriel Lópe, 2012 [46]
Retrospective observational registry
Tran-1, 2012 [47] Tran-2, 2012 [47]
Single arm phase II trial Phase III trial
Choueiri, 2012 [48]
Phase III trial
SUN, SOR, BEV, Temsirolimus (1st or 2nd) PAZ (1st or 2nd) PAZ vs PL (1st or 2nd) SOR vs PL (2nd)
110 (BV + IFN = 59, PL+IFN = 51) 135
Porta, 2012 [49]
SUN (1st or 2nd)
Hegde, 2013 [50]
Prospective observational registry Phase III trial
Gigante, 2012 [51]
Randomised phase II trial
Gámez-Pozo, 2012 [52]
Prospective observational registry Prospective observational registry Retrospective observational registry Retrospective observational registry Retrospective observational registry Prospective observational registry
BEV + Temsirolimus vs BEV + IFN vs SUN (1st) SUN (1st)
Lambrechts, 2012 [45]
Feng, 2013 [53] Beuselinck, 2013 [54] Beuselinck, 2013 [55] Scartozzi , 2013 [56] Kobayashi, 2013 [57]
BEV + IFN vs PL + IFN (1st)
ELISA ELISA Genotyping
BCL-2, BCL-XL, BAX, pAKT, pP44/42 MAPK, pSTAT3, VEGFR-1, VEGFR-2, PDGFR-a, PDGFR-b SNPs in VEGFR-2, VEGFR-3, PDGFR-a, IL-8, FGF-2, FGFR-2, HIF-1a, ABCB1, ABCG2, CYP3A4, CYP3A5, NR1I2 SNPs in VEGF-A, VEGFR-2
D B D B B C D B
T. Funakoshi et al. / Cancer Treatment Reviews xxx (2014) xxx–xxx
D
(continued on next page) 5
Please cite this article in press as: Funakoshi T et al. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat Rev (2014), http://dx.doi.org/10.1016/j.ctrv.2013.11.008
Terakawa, 2011 [34]
6
First author, year
Source of patient data
Treatment (line of treatment)
Sample size
Material
Methodology
Biomarkers studied
Study category
Garcia-Donas, 2013 [58]
Prospective observational registry
SUN (1st)
71
FFPE tumor
IHC
C
FFPE tumor FFPE tumor FFPE tumor FFPE tumor FFPE tumor FFPE tumor Blood
Sequencing qRT-PCR IHC RNA microarray Sequencing MS-PCR Genotyping
CAIX, HIF-1a, HIF-2a, VEGF-A, VEGFR-1, VEGFR-2, VEGFR-3, PDGFR-b, P-glycoprotein VHL gene mutational status EGLN3 mRNA HIF-1a, HIF-2a HIF-1a VHL gene mutational status VHL gene methylation status SNPs in IL-8
B
FFPE tumor FFPE tumor
IHC IHC
PD-L1 PD-L1
B B
Plasma
ELISA
Ang-2,BFGF,BMP-9, Endoglin, GRO-a, HGF, ICAM-1, IGFBP-1, IGFBP2, IGFBP-3, IL-6, IL-8, MCP-1, OPN, P-selectin, PAI-1 active, PAI-1 total, PDGF-AA, PDGF-BB, PEDF, PlGF, SDF-1, TGFb-1, TGFb-2, TGFbR-3, TSP-2, VCAM-1, VEGF-A, VEGF-C, VEGF-D, sVEGFR-1, sVEGFR-2
B
Choueiri, 2013 [59]
Single arm phase II trial
PAZ (1st or 2nd)
225
Xu, 2013 [60]*
Phase III trial
PAZ vs SUN (1st)
Figueroa-1, 2013 [61]* Figueroa-2, 2013 [61]*
Single arm phase II trial Phase III trial
Nixon, 2013 [62]*
Phase III trial
PAZ (1st or 2nd) PAZ vs PL (1st or 2nd) BEV + IFN vs IFN (1st)
724 (PAZ = 371, SUN = 353) 46 160 (PAZ = 113, PL = 47) 424 (Training set = 286, Validation Set = 138)
B
Abbreviations: 4EBP1, 4E binding protein 1; ABCB1, ATP binding cassette member B1; ABCG2, ATP binding cassette member G2; AMPK, adenosine monophosphate kinase; Ang-2, angiopoietin-2; AX, axitinib; BAX, BCL2-associated X protein; BCL-2, B-cell lymphoma 2; BCL-XL, BCL-extra large; BDNF, brain-derived neurotrophic factor; BEV, bevacizumab; BFGF, basic fibroblast growth factor; BMP9, bone morphogenetic protein 9; CAIX, Carbonic anhydrase IX; CASP7, caspase 7; CECs, circulating endothelial cells; cfDNA, cell-free DNA; ColIV, collagen IV; CTACK, cutaneous T-cell-attracting chemokine; CXCR-4, C-X-C chemokine receptor type 4; CYP, cytochrome P450; EGF, epidermal growth factor;, EGLN, egl nine homolog; ELISA, enzyme-linked immunosorbent assay, FFPE, Formalin fixed paraffin embedded tissues; FGF-2, fibroblast growth factor 2; FGFR-2, FGF receptor 2; FIH-1, factor inhibiting HIF-1a; FLT3, fms-related tyrosine kinase 3; FOXO3a, phosphorylated forkhead Box O 3a; FSHR, follicle stimulating hormone receptor; G-CSF, granulocyte colony-stimulating factor; GM–CSF, granulocyte–macrophage colony-stimulating factor; GRO-a, growth-regulated alpha protein; GSK-3, glycogen synthase kinase 3; HGF, hepatocyte growth factor; HIF, hypoxia-inducible factor; ICAM-1, intercellular Adhesion Molecule 1; IFN, interferon; IGFBP, insulin-like growth factor binding protein; IHC, immunohistochemistry; IL, interleukin; IL-1RA, IL-1 receptor antagonist protein; INFaR, IFN alpha receptor; IP-10, IFNc-induced protein 10; IRS-1, Insulin-receptor substrate 1; LYVE-1, lymphatic vessel endothelial hyaluronan receptor 1; MCP, monocyte chemotactic protein; M-CSF, macrophage colony-stimulating factor; MDC, myeloid dendritic cell; MIF, macrophage migration inhibitory factor; MIG, monokine induced by IFNc; MIP, macrophage inflammatory protein; miRNAs, microRNAs; MMP-9, matrix metallopeptidase 9; MS-PCR, methylation-specific polymerase chain reaction; NGAL, neutrophil gelatinase-associated lipocalin; NGF, nerve growth factor; NR, not reported; OPN, osteopontin; p4EBP1, phosphorylated 4EBP1; P70S6K, P70 ribosomal S6 kinase; pACC-1, phosphorylated acetyl-CoA carboxylase 1; PAI-1, plasminogen activator inhibitor 1; pAKT, phosphorylated AKT; pAMPK, phosphorylated AMPK; PARP, Poly (ADP-ribose) polymerase; PAZ, pazopanib; PDGF, platelet-derived growth factor; PDGFR, PDGF receptor; D-L1, programmed cell death 1 ligand 1; PEDF, pigment epithelium-derived factor; pEGFR, phosphorylated epidermal growth factor receptor; PL, placebo; PlGF, placental growth factor; pMAPK, phosphorylated mitogen-activated protein kinase; pP70S6K, phosphorylated P70 ribosomal S6 kinase; pPKC-a, phosphorylated PKC-a; pS6RP, phosphorylated S6RP; pSRC, phosphorylated SRC; PTEN, phosphatase and tensin homolog; pTSC-2, phosphorylated tuberous sclerosis 2; qRT-PCR, quantitative real-time PCR; RANTES, regulated on activation, normal T-cell expressed and secreted; RPPA, reverse-phase protein arrays; S6RP, S6 ribosomal protein; sCAIX, soluble CAIX; SCF, stem cell factor; SCGF-b, stem cell growth factor beta; SDF-1, stromal cell-derived factor 1; sKIT, soluble stem cell factor receptor; SNP, single-nucleotide polymorphism; SOR, sorafenib; STAT3, signal transducer and activator of transcription 3; SUN, sunitinib; sVCAM-1, soluble vascular cell adhesion molecule 1; sVEGFR, soluble VEGF receptor; TGFb, transforming growth factor b; TGFbR, TGFb receptor; TIMP-1, tissue inhibitor of metalloproteinase 1; TNF-a, tumor necrosis factor alpha; TRAIL, tumor necrosis factorrelated apoptosis-inducing ligand; TSC-2, tuberous sclerosis 2; uPA, urokinase-type plasminogen activator; uPAR, uPA receptor; VEGF, vascular endothelial growth factor; VHL, von Hippel-Lindau. ⁄ ASCO meeting presentation.
T. Funakoshi et al. / Cancer Treatment Reviews xxx (2014) xxx–xxx
Please cite this article in press as: Funakoshi T et al. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat Rev (2014), http://dx.doi.org/10.1016/j.ctrv.2013.11.008
Table 1 (continued)
7
T. Funakoshi et al. / Cancer Treatment Reviews xxx (2014) xxx–xxx Table 2 Predictive biomarkers with statistical significant results. First author, year
Treatment (line of treatment)
Study category
Putative predictive biomarkers
Outcome according to biomarker
Statistical methods
Cut-off point
Blood-based protein biomarkers Peña, 2010 [25] SOR vs PL (2nd)
B
VEGF-A
Multivariate Cox regression
75th percentile (254 pg/ml)
Zurita, 2011 [33]
SOR vs SOR + IFN (1st)
B
VEGF-A
Multivariate Cox regression
Median (19.6 ng/ ml)
Tran-2, 2012 [47]
PAZ vs PL (1st or 2nd)
B
IL-6
Cox regression
Median (13.07 pg/ml)
Nixon, 2013 [62]
BEV + IFN vs IFN (1st)
B
IL-6 and HGF
Multivariate Cox regression
Median
Zurita, 2011 [33]
SOR vs SOR + IFN (1st)
B
Osteopontin
Multivariate Cox regression
Median (51.3 ng/ ml)
Zurita, 2011 [33]
SOR vs SOR + IFN (1st)
B
Six factor signature (OPN, sCAIX, VEGF-A, TRAIL, ColIV, sVEGFR2)
For PFS, treatment-by-biomarker status interaction test was significant (P = 0.02 in first half of data and P = 0.023 in second half of data) For PFS, treatment-by-biomarker status interaction test was significant (P = 0.01) For PFS and OS, treatment-bybiomarker status interaction test was significant (P = 0.009 and 0.005) For OS, 3-way interaction between IL6, HGF and treatment was significant(P < 0.0001) For PFS, treatment-by-biomarker status interaction test was significant (P = 0.004) For PFS, treatment-by-biomarker status interaction test was significant (P = 0.0002)
Multivariate Cox regression
CAF index 4
B
IL-8 rs1126647
Multivariate Cox regression
TT vs AA genotypes
Blood-based SNP biomarkers Xu, 2011 [35] PAZ vs PL (1st or 2nd)
For PFS, treatment-by-biomarker status interaction test was significant (P = 0.04)
Table 3 Blood-based prognostic biomarkers with statistical significant results. Treatment (line of treatment)
Study category
Putative prognostic biomarkers
Outcome according to biomarker
Statistical methods
Cut-off point
Angiogenesis Deprimo, 2007 [14]
SUN (2nd)
B
VEGF-A
SUN (1st or 2nd)
C
VEGF-A
Student’s t-test and Wilcoxon rank-sum t test Log-rank test (PFS) and Student’s t-test (response)
not reported
Kontovinis, 2009 [18]
Porta, 2010 [22]
SUN (2nd)
C
VEGF-A
Pts with PR had a greater increase in levels than pts with SD or PD (P < 0.05) Greater increase in levels by the end of cycle 2 was associated with worse PFS (HR = 0.2 [0.059–0.68], P = 0.01). Pts with PD had a greater increase in levels by the end of cycle 2 and higher level at the end of cycle 2 than pts with PR or SD (P = 0.01 and P = 0.033) High baseline level was associated with worse PFS (HR = 2.04 [1.24–3.27], P = 0.005)
Bivariate Cox regression
Farace, 2011 [31]
SUN, SOR (1st or 2nd)
C
VEGF-A
Log-rank test (PFS and OS)
Mancuso, 2012 [41]
SOR (2nd or 3rd)
B
VEGF-A
not reported
not reported
Tran-1, 2012 [47]
B
VEGF-A
Cox regression
not reported
Rini, 2008 [16]
PAZ vs PL (1st or 2nd) SUN (2nd or 3rd)
B
VEGF-C
Wilcoxon rank-sum t test
Median (722.1 pg/ml)
Deprimo, 2007 [14]
SUN (2nd)
B
sVEGFR-2
SUN (2nd)
B
sVEGFR-3
Rini, 2008 [16]
SUN (2nd or 3rd)
B
sVEGFR-3
Student’s t-test and Wilcoxon rank-sum t test Student’s t-test and Wilcoxon rank-sum t test Wilcoxon rank-sum t test
not reported
Deprimo, 2007 [14]
Tomita, 2011 [37]
AX (2nd)
B
sVEGFR-3
Greater increase in levels between day 1 and day 14 was associated with worse OS (P = 0.02) Responders had a greater increase in levels from baseline to the end of the first 3 cycles and higher level at the end of the first 3 cycles than non-responders (P < 0.0001 and P = 0.010) High baseline level was associated with worse OS (P = 0.004) Low baseline level was associated with better PFS (HR = 0.37 [0.16–0.60], P = 0.0006). Pts with PR had a lower baseline level than pts with SD or PD Pts with PR had a greater decrease in levels than pts with SD or PD (P < 0.05) Pts with PR had a greater decrease in levels than pts with SD or PD (P < 0.05). Low baseline level was associated with better PFS (HR = 0.45 [0.21–0.78], P = 0.006). Pts with PR had a lower baseline level than pts with SD or PD Greater reduction in levels from baseline to cycle 2 day 1 was associated with better PR rate (P = 0.045) and PFS (HR = 0.42 [0.21– 0.83], P = 0.010).
Upper value of the normal range (707 pg/ml) 270 pg/ml
First author, year
Fisher’s exact test (response) and log-rank test (PFS)
Mean (6.7 fold increase)
not reported Median (47,000 pg/ ml) Median
(continued on next page)
Please cite this article in press as: Funakoshi T et al. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat Rev (2014), http://dx.doi.org/10.1016/j.ctrv.2013.11.008
8
T. Funakoshi et al. / Cancer Treatment Reviews xxx (2014) xxx–xxx
Table 3 (continued)
⁄
First author, year
Treatment (line of treatment)
Study category
Putative prognostic biomarkers
Outcome according to biomarker
Statistical methods
Cut-off point
Tran-1, 2012 [47]
PAZ (1st or 2nd)
B
IL-8
Cox regression
not reported
Tran-2, 2012 [47]
PAZ (1st or 2nd)
B
IL-8
High baseline level was associated with worse PFS (P = 0.013) High baseline level was associated with worse PFS (P = 0.006) and OS (P = 0.003)
Cox regression
Median (16.42 pg/ml)
Immunomodulator Tran-1, 2012 [47]
PAZ (1st or 2nd)
B
IL-6
Cox regression
not reported
Tran-2, 2012 [47]
PAZ (1st or 2nd)
B
IL-6
Cox regression
Tran-1, 2012 [47]
PAZ (1st or 2nd)
B
HGF
Median (13.07 pg/ml) not reported
Tran-2, 2012 [47]
PAZ (1st or 2nd)
B
HGF
High baseline level was associated with worse PFS (P = 0.021) High baseline level was associated with worse OS (P < 0.0001) High baseline level was associated with worse PFS (P = 0.013) High baseline level was associated with worse PFS (P = 0.01) and OS (P = 0.002)
Tran-1, 2012 [47]
PAZ (1st or 2nd)
B
Osteopontin
Cox regression
Tran-2, 2012 [47]
PAZ (1st or 2nd)
B
Osteopontin
High baseline level was associated with worse PFS (P = 0.041) Higher baseline level was associated with worse PFS (P = 0.0004) and OS (P < 0.0001)
Tran-2, 2012 [47]
PAZ (1st or 2nd)
B
SUN (2nd)
C
Porta, 2010 [22]
SUN (2nd)
C
NGAL
Busse, 2011 [28]
SOR (1st or 2nd)
C
Furuya, 2011 [30]
IFN ± SOR (1st or 2nd)
C
TGFb-1 mRNA INFaR-2
Farace, 2011 [31]
SUN, SOR (1st or 2nd)
C
SDF-1a
Zurita, 2011 [33]
SOR ± IFN (1st)
B
IL-5
Zurita, 2011 [33]
SOR ± IFN (1st)
B
EGF
Zurita, 2011 [33]
SOR ± IFN (1st)
B
M-CSF
Tran-1, 2012 [47]
PAZ (1st or 2nd)
B
E-selectin
Gigante, 2012 [51]
BEV + Temsirolimus, BEV + IFN, SUN (1st)
C
CAIX
High signature was associated with worse PFS (P = 0.014) and OS (P < 00001) High level was associated with worse OS (P = 0.045). Pts with PD had higher level than pts with PR or SD (P = 0.009) High baseline level was associated with worse PFS (HR = 1.65 [1.04–2.78], P = 0.034) High level was associated with better PFS (P = 0.005) and OS (P = 0.039) In 21 pts treated with IFN + SOR, pts with PR or CR had higher baseline levels than pts with SD or PD (P = 0.0003) Decrease in levels between day 1 and day 14 was associated with worse PFS (P = 0.002) and OS (P = 0.007). High baseline level was associated with better PFS (HR = 0.48, P = 0.015) High baseline level was associated with worse PFS (HR = 2.45, P = 0.003) High baseline level was associated with worse PFS (HR = 1.86, P = 0.04). High baseline level was associated with better PFS (P = 0.002) High baseline level was associated with worse OS (HR = 2.65 [1.19-5.92], P = 0.014)
Cox regression
Perez-Gracia, 2009 [20]
Six CAF signature⁄ TNF-a
Matrix modifier Perez-Gracia, 2009 [20]
SUN (2nd)
C
MMP-9
Tran-2, 2012 [47]
PAZ (1st or 2nd)
B
TIMP-1
Other Gámez-Pozo, 2012 [52]
SUN (1st)
C
Feng, 2013 [53]
SOR (1st or 2nd)
C
Poor response model 3.3 miRNAs cfDNA
Cox regression Cox regression
Cox regression
Median (311.30 pg/ ml) not reported Median (191.63 ng/ ml) Median
Log-rank test (OS) and Student’s t-test or Mann– Whitney U-test (response) Bivariate Cox regression
Median (50 pg/ml)
Multivariate Cox regression Mann–Whitney U-test
not reported not reported
Log-rank test (PFS and OS)
not reported
Multivariate Cox regression Multivariate Cox regression Multivariate Cox regression Cox regression
Median
Cox regression
Median (20.6 pg/ml)
High baseline level was associated with higher risk of PD (OR = 2.84 [1.11–10.44], P = 0.024) and worse TTP (P = 0.042). Pts with PD had higher level than pts with PR or SD (P = 0.027) High baseline level was associated with worse PFS (P = 0.006) and OS (P < 0.0001)
Logistic regression (risk of PD), log-rank test (TTP) and Student’s t-test or Mann–Whitney U-test (response) Cox regression
Median (4062 ng/ml)
Poor response group was associated with worse OS (HR = 0.46 [0.22–0.98], P = 0.044)
Multivariate Cox regression
not reported
Low levels at week 12, 16, 24 were associated with better OS (P = 0.047, 0.018, 0.018). Pts with PR or SD had lower levels measured from 8 to 24 weeks than pts with PD (P < 0.05)
Log-rank test (OS) and Mann-Whitney U-test (response)
Median (4.976 (W12), 5.185 (W16), 5.226 (W24) g/ml)
110 ng/ml
Median Median not reported
not reported
Six CAF signature includes IL-6, IL-8, VEGF, HGF, TIMP-1, and osteopontin. Poor response model 3.3 miRNAs includes miR-192, miR-193a-5p, and miR-501-3p.
B1 were associated with longer survival in one Category B study and high levels of P21 were associated with shorter survival in one Category D study [21,46]. The prognostic value of gene alterations (mutation or hypermethylation) was examined in seven studies. One Category C study showed a loss of function mutation
of VHL was a prognostic factor for improved response [15]. B-RAF mutations (V600E and K601E) were associated with worse PFS in one Category B study [41]. Table 5 describes the tissue-based biomarkers with statistical significant results and the LOE for each tissue-based biomarker is shown in Table 4.
Please cite this article in press as: Funakoshi T et al. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat Rev (2014), http://dx.doi.org/10.1016/j.ctrv.2013.11.008
9
T. Funakoshi et al. / Cancer Treatment Reviews xxx (2014) xxx–xxx Table 4 Level-of-evidence determination for prognostic biomarkers. Response Blood-based biomarkers I II
VEGF-A (baseline), VEGF-A (change), VEGF-C (baseline), sVEGFR-2 (change), sVEGFR-3 (baseline), sVEGFR-3 (change)
III
TNF-a (baseline), INFaR-2 (baseline), MMP-9 (baseline), cfDNA (change)
PFS
OS
IL-8 (baseline), HGF (baseline), Osteopontin (baseline) VEGF-C (baseline), sVEGFR-3 (baseline), sVEGFR-3 (change), IL-6 (baseline), Six factor signature⁄ (baseline), IL-5 (baseline), EGF (baseline), M-CSF (baseline), E-selectin (baseline), TIMP-1 (baseline) VEGF-A (baseline), VEGF-A (change), NGAL (baseline), TGFb-1 mRNA (baseline), SDF-1a (change), MMP-9 (baseline/TTP)
VEGF-A (baseline), IL-8 (baseline), IL-6 (baseline), HGF (baseline), Osteopontin (baseline), Six factor signature⁄ (baseline), TIMP-1 (baseline) VEGF-A (change), TNF-a (baseline), TGFb-1 mRNA (baseline), SDF-1a (change), CAIX (baseline), Poor response model 3.3 miRNAs (baseline), cfDNA (change)
IV Tissue-based biomarkers I II III IV SNP biomarkers I II III IV
Cellular biomarkers I II III
VEGF-A mRNA, INFaR-2, VHL mutational status VEGFR-2, PDGFR-b, FSHR, CXCR-4, PTEN, HIF-1a, HIF-2a, CAIX, P21
VEGF-A (rs833061), HIF-1a (rs11549467), NR1I2 (rs3814055) VEGFR-1 (rs9582036), VEGFR-3 (rs307826)
CD1c(+)CD11c(hi)CD14( )CD19( ) MDC-1 (baseline)
PD-L1, pAKT (S473), AMPK, Cyclin B1, B-RAF mutational status PDGFR-a
pAKT (S473), AKT, pS6RP (S235), AMPK, pAMPK, Cyclin B1 VEGF-A mRNA
VEGF-A, VEGFR-2, VEGFR-3, CXCR-4, HIF-1a, HIF-2a, CAIX, P21
VEGF-A, HIF-1a, HIF-2a, EGLN3 mRNA
VEGFR-1 (rs7993418), VEGFR-1 (rs9513070), IL-8 (rs1126647), HIF-1a (rs11549467) VEGFR-3 (rs307826), VEGFR-3 (rs307821) VEGF-A (rs833061), VEGF-A (rs2010963), VEGFR-1 (rs9582036/TTP), VEGFR-1 (rs9554320/TTP), VEGR-3 (rs6877011), VEGF-A (rs3025039) & VEGFR-2 (rs2305948), FGFR-2 (rs2981582), ABCB1 (rs1128503), CYP3A5 (6986A/G), NR1I2 (rs2276707), NR1I3 (5719C/ T, 7738A/C, 7837T/G), NR1I3 (rs4073054)
IL-8 (rs1126647)
CD1c(+)CD11c(hi)CD14( )CD19( ) MDC-1 (baseline), CD45(dim)CD34(+)VEGFR2(+)7AAD( ) Progenitor cells (baseline & change), Th1/Th2 cell ratio (baseline)
VEGF-A (rs833061), VEGFR-3 (rs307826), VEGFR-3 (rs307821), VEGR-3 (rs6877011), ABCB1 (3435C/T, 1236C/T, 2677G/T), ABCB1 (rs1128503), NR1I3 (rs2307424), NR1I3 (rs4073054)
CD3(+)CD4(+)CD25(hi)Foxp3(+) T cells (change), CD45(dim)CD34(+)VEGFR2(+)7AAD( ) Progenitor cells (baseline)
IV ⁄
Six CAF signature includes IL-6, IL-8, VEGF, HGF, TIMP-1, and osteopontin. Poor response model 3.3 miRNAs includes miR-192, miR-193a-5p, and miR-501-3p.
SNP biomarker Single nucleotide polymorphisms (SNPs) have been investigated as potential biomarkers, particularly in genes related to VEGF-targeted therapy pharmacodynamics, VEGF-independent alternative pro-angiogenic pathways or pharmacokinetics. Polymorphisms in genes involved in VEGF and PDGF- dependent angiogenesis such as VEGF, VEGFR, and PDGFR were assessed in nine studies. SNP rs833061 in VEGF-A was associated with response and survival in one Category B and D study, respectively [35,56]. SNP rs2010963 in VEGF-A also correlated with survival in one Category D study [56]. Four VEGFR-1 polymorphisms were associated with response and survival. Garcia-Donas et al. and Beuselinck et al. independently reported that two SNPs in VEGFR-3 (rs307826 and rs307821) correlated with response and survival in their Category C and D study, respectively [39,54]. Another SNP rs6877011 in VEGR-3 was also associated with survival in one Category D study [56]. SNPs in VEGF-independent alternative pro-angiogenic pathways (IL-8 and fibroblast growth factor receptor 2 [FGFR-2]) were evaluated in four studies. Wu et al. consistently reported SNP rs1126647 in IL-8 correlated with survival in their two Category B studies [35,60]. One Category D study showed PFS was associated with SNP in FGFR-2 [54]. Polymorphisms in
genes related to pharmacokinetics (i.e., absorption, such as ABCB1, or metabolism, such as CYP3A5, NR1I2 and NR1I3) were examined in three studies. Two polymorphic variants of ABCB1 were associated with survival in two Category D studies [27,54]. van der Veldt et al. reported PFS was associated with SNP in CYP3A5 in their Category D study [54]. Two NR1I2 polymorphisms were associated with response and survival in one Category B and one Category D study, respectively [35,54]. Three NR1I3 polymorphisms were also related to survival in Category D studies [27,54]. Table 6 describes the SNP biomarkers with statistical significant results and the LOE for each SNP biomarker is shown in Table 4. Cellular biomarker A prognostic value of cellular biomarkers was assessed in five Category C studies. van Cruijsen et al. showed high baseline myeloid dendritic cell levels were prognostic for improved PFS [17]. In study of Farace et al. high baseline circulating progenitor cell values were associated with worse PFS and OS [31]. Adotevi et al. reported a greater decrease of regulatory T cells after treatment correlates with worse OS [26]. Supplement 2 describes the cellular biomarkers with statistical significant results and the LOE for each cellular biomarker is shown in Table 4.
Please cite this article in press as: Funakoshi T et al. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat Rev (2014), http://dx.doi.org/10.1016/j.ctrv.2013.11.008
10
Treatment (line of treatment)
Study category
Putative prognostic biomarkers
Outcome according to biomarker
Statistical methods
Cut-off point
Angiogenesis Garcia-Donas, 2013 [58]
SUN (1st)
D
VEGF-A
High expression was associated with better PFS (HR = 2.56 [1.04–6.29], P = 0.041) and OS (HR = 4.29 [1.43–12.8], P = 0.0092)
Paule, 2010 [24]
SUN (2nd)
C
Terakawa, 2011 [34]
SUN (1st)
D
VEGF-A (isoforms 121, 165) mRNA VEGFR-2
High VEGF(121)/VEGF(165) ratio was associated with worse OS (HR = 5.8 [1.4–24.5], P = 0.02). Pts with RP or SD had higher levels of VEGF(121) and VEGF(165) than pts with PD (P = 0.04 and 0.04) High expression was associated with better response (P = 0.039) and PFS (HR = 2.91 [1.15– 7.41], P = 0.0025)
Univariate log-rank test (PFS) and multivariate Cox regression (OS) Cox regression (OS)
Garcia-Donas, 2013 [58] Kusuda, 2011 [32]
SUN (1st) SOR (2nd)
D C
VEGFR-3 PDGFR-a
High expression was associated with better PFS (HR = 0.403 [0.20–0.82], P = 0.012) High expression was associated with worse PFS (HR = 6.29 [1.18–33.33], P = 0.032)
Garcia-Donas, 2013 [58] Siraj, 2012 [40]
SUN (1st) SUN (1st)
D D
PDGFR-b FSHR
High expression was associated with better response (OR = 0.04 [0.002–0.68], P = 0.026) Pts with PR had higher percentage of FSHR-positive vessels than pts with SD or PD (P < 0.0001)
Multivariate logistic regression not reported
Stained tumour cells >25% Third quartile (1.25) Stained tumour cells >10% Median Stained tumour cells >10% Median not reported
C
INFaR-2
D
CXCR-4
In 21 pts treated with IFN + SOR, pts with SD or PD had higher expression than pts with PR or CR (P = 0.0003) High expression was associated with worse response (P = 0.026) and PFS (HR = 2.04 [1.08– 3.84], P = 0.027)
Mann–Whitney U-test
D’ Alterio, 2012 [44]
IFN ± SOR (1st or 2nd) SUN (1st)
Figueroa-1, 2013 [61]
PAZ (1st or 2nd)
B
PD-L1
High expression was associated with worse PFS (HR = 6.5 [2.6–15.9], P = 0.0005)
Chi-square test (respnose) and multivariate Cox regression (PFS) Log-rank test
BEV ± Erlotinib
B
PTEN
Pts with OS >30 months had higher expression than pts with OS 30 months had lower expression than pts with OS 30 months had lower expression than pts with OS 20% H-Score >3 not reported Expression intensity ± not reported not reported not reported not reported Stained tumour cells >85% Stained tumour cells >10% Stained tumour cells >10% Median
T. Funakoshi et al. / Cancer Treatment Reviews xxx (2014) xxx–xxx
Please cite this article in press as: Funakoshi T et al. A systematic review of predictive and prognostic biomarkers for VEGF-targeted therapy in renal cell carcinoma. Cancer Treat Rev (2014), http://dx.doi.org/10.1016/j.ctrv.2013.11.008
Table 5 Tissue-based prognostic biomarkers with statistical significant results.
T. Funakoshi et al. / Cancer Treatment Reviews xxx (2014) xxx–xxx
11
not reported Mancuso, 2012 [41]
SOR (2nd or 3rd)
B
B-RAF mutational status
V600E and K601E mutations were associated with worse PFS (P < 0.05)
As described in the outcome As described in the outcome Multivariate logistic regression Loss of function mutations was associated with higher CR or PR rate (P = 0.03) VHL mutational status C
In 65 pts treated with sunitinib, high expression was associated with worse response (P = 0.025) and PFS (P < 0.001). For 28 pts treated with sorafenib, high expression was associated with worse PFS (P = 0.01) P21 D SUN, SOR, BEV, Temsirolimus (1st or 2nd)
SUN, SOR, BEV, AX (1st or 2nd) Genetic biomarker Choueiri, 2008 [15]
not reported Youden index 0.15 Univariate Cox regression (PFS and OS) v2 test (response) and log-rank test (PFS) High expression was associated with worse PFS (P = 0.038) and OS (P = 0.028) Cyclin B1 B
pAMPK B
BEV ± Erlotinib
Cell-cycle signaling pathway Tsavachidou-Fenner, 2010 [21] Muriel Lópe, 2012 [46]
not reported not reported Univariate Cox regression (PFS and OS) and Student’s t-test (OS) Student’s t-test High expression was associated with better PFS (P = 0.038) and OS (P = 0.061). Pts with OS >30 months had higher expression than pts with OS 30 months had higher expression than pts with OS