Seminar article
Renal carcinoma pharmacogenomics and predictors of response: Steps toward treatment individualization

https://doi.org/10.1016/j.urolonc.2013.09.015Get rights and content

Abstract

Molecular knowledge has deeply affected the treatment and outcome of kidney cancer in recent years, and several therapeutic options have become available. However, there are no validated biomarkers to select the best drug for each patient. Already published studies and ongoing investigations could change this scenario in the near future. Regarding antiangiogenic drugs, several works on single nucleotide polymorphisms have achieved promising results, with some SNPs predicting resistance to sunitinib and pazopanib being validated. If more evidence is gained, it could prompt prospective studies exploring a molecularly driven selection of treatment. Another relevant line of investigation for antiangiogenic drugs is the cytokines and antiangiogenic factors. Different studies have found that cytokines and antiangiogenic factors are able to predict the outcome of patients treated with sunitinib, pazopanib, or sorafenib. Issues regarding the thresholds of normality and the best time for assessment are pending, but the communicated results are encouraging. Less evidence is available for mammalian target of rapamycin inhibitors but recent data support a key role of the phosphoinositide 3-kinase/Akt pathway in clear cell renal cell carcinoma and points toward poor response to angiogenic drugs when the pathway is activated. Whether modern phosphoinositide 3-kinase inhibitors could be the best option for these patients is a question that should be addressed. Additionally, a new class of immunomodulators, like anti–programmed death 1 drugs, has demonstrated to achieve long-lasting stabilizations even in some patients with no radiological response or early progression. This is a singular situation where the identification of reliable predictors of efficacy will be key in the development of these drugs in renal cell carcinoma. Finally, germline mutations of the c-Met gene have been proposed as the first predictor of response to targeted therapies in papillary renal cell carcinoma. As a conclusion, translational research will be a cornerstone to move a next step forward in kidney cancer.

Introduction

Major successes achieved in recent years in clear cell renal cell carcinoma (ccRCC) treatment have been based on the molecular knowledge of the disease. This knowledge has led to the approval of 8 agents in this setting (1 antibody against the vascular endothelial growth factor [VEGF]: bevacizumab; 4 inhibitors of its receptors [VEGFR]: sorafenib, sunitinib, pazopanib, and axitinib; 2 inhibitors of the mammalian target of rapamycin [mTOR]: temsirolimus and everolimus; and 1 recombinant form of an endogenous cytokine: interleukin-2 [IL-2]). However, no advances to match the best therapeutic option to each patient have been made. Thus, after an improvement in outcomes with introduction of the initial targeted agents, outcomes have since plateaued and the huge effort of including thousands of patients in clinical trials assessing different combinations of drugs or sequential strategies of treatment has failed to affect overall survival (OS) of RCC.

A “new generation” of studies focusing on molecularly defined populations of patients could be the way to take advantage of the currently available therapeutic arsenal. This review focuses on the so far published approaches toward the individualization of RCC treatment, and gives some clues of expected results and needs for the future. The next sections summarize potential predictive biomarkers proposed for the different RCC drugs.

Interferon (IFN) and IL-2 were largely used in kidney cancer before the development of targeted therapies [1]. Even today, IFN, combined with bevacizumab, and high-dose IL-2, for selected cases, are accepted options in the first-line setting of ccRCC [2]. Concerning markers of response, in 2005 Atkins et al. [3] communicated the results of a nested case-control study of 66 patients with metastatic ccRCC who had received IL-2 within different clinical trials. They described an association between CAIX expression, assessed by immunohistochemistry, and response to IL-2 (P = 0.04). Unfortunately, a prospective validation study (the SELECT study) failed to confirm those results [4]. Thus, currently there are no predictive markers available for cytokine immunotherapy.

Promising results have been recently communicated with the new immunomodulator, nivolumab (BMS-936558), a monoclonal antibody against programmed death 1 (PD-1) [5]. A response rate of 27% in 33 patients with metastatic RCC was achieved. None of the tumors lacking PD1 expression did respond, suggesting PD1 as a biomarker to identify resistant tumors. Confirmatory studies to determine the activity of the drug in RCC and the reliability of PD1 expression as a predictive marker are ongoing. The identification of accurate biomarkers will be key to further develop this class of drugs, because they may not produce immediate radiological responses, and “pseudoprogressions” (initial tumor growth in patients that finally responded) have been described in some cases [6].

Antiangiogenic drugs have become the cornerstone of kidney cancer treatment. Though they all share the VEGF pathway as their main target, relevant differences in the mechanism of action and the spectrum of tyrosine kinases inhibited could make a difference in some populations of patients.

VHL inactivation plays a key role in ccRCC and provides the rationale for the development of antiangiogenic drugs to treat this tumor [7]. In 2008, a retrospective study assessed the value of such alteration as predictor of response in patients with ccRCC treated with any of the 4 different drugs (sorafenib, sunitinib, bevacizumab, and axitinib) in 2 institutions in the United States as continuation of a prior work by Rini et al. [8], [9]. Up to 123 patients of 183 cases could finally be analyzed for both VHL mutations and gene methylation. VHL mutations were found in 60 (49%) and methylation in 12 tumors (10%). A nonsignificant trend toward a better response rate (41% vs. 31%) was found in tumors with VHL inactivation. Such difference became significant in a post hoc analysis when only loss of function mutations were considered. More recently, Garcia-Donas et al. [10] also studied VHL mutations and loss of expression in a prospective cohort of 101 patients treated with first-line sunitinib (SUT-REN 07 study). Of the 31 patients that could be analyzed, 64% presented VHL inactivation, and no association with outcome was observed. Importantly, it must be noted that the average number of cases with VHL inactivation, around 60% in both communications, is far from the 91% reported by some authors [11]. Thus, a more accurate determination of the VHL status and larger sample sizes would be mandatory to reach a reliable conclusion. VHL status is not clinically useful at present.

Single nucleotide polymorphisms (SNPs) have been a major area of interest for researchers in ccRCC (Table 1). As normal endothelial cells and circulating VEGF are the true target of antiangiogenic drugs, the genetic background of the patient could play a critical role determining not only tolerability but also efficacy.

Focusing on toxicity, van Erp et al. [12] assessed, in 2009, 31 polymorphisms in 219 patients diagnosed with renal or gastrointestinal stromal tumors treated with sunitinib. The strongest associations in multivariable analysis were found for the following: leukopenia, FLT3 rs1933437 (T227M), and CYP1A1 rs1048943 (I462V); mucosal inflammation and CYP1A1 rs1048943; any toxicity>grade 2; and VEGFR2 rs2305948 (V297I). A later publication by the same group suggested a protective role for the FLT3 rs1933437 polymorphism in bone marrow toxicities [13]. In 2011, the SUT-REN 07 study communicated that the SNP CYP3A5*1 (rs776746) was associated with an increased risk of sunitinib dose reductions owing to toxicity (HR = 3.7, 95% CI = 1.7–8.4, P = 0.001) [14]. This association was significant after correction for multiple testing. In the same publication, some other SNPs were also associated with hypertension risk: VEGFR2 rs1870377 (Q472H), ABCB1 rs1128503 (G412G), and VEGFA rs699947 (−2578A>C). More recently Kim et al. [15] evaluated 63 patients with mRCC treated with sunitinib and found an association for the VEGF SNP rs2010963 (−634G>C) and higher risk of hypertension.

Concerning efficacy, van der Veldt et al. studied 136 patients with ccRCC, which included those reported formerly by van Erp et al [16], [13]. The SNPs CYP3A5 rs776746, 2 NR1I3 and ABCB1 haplotypes, and VEGFR2 rs1870377 presented the strongest association with survival. García-Donas et al. in the SUT-REN 07 study, found a highly significant association between 2 VEGFR3 polymorphisms, rs307826 (T494A) and rs307821 (R1324L), and worse response (P = 0.005 and P = 0.015, respectively), and shorter PFS (P = 0.0005 and P = 0.0008, respectively) [14]. This association remained significant after correction for multiple testing. In 2013, Beuselinck et al. communicated the results of an independent retrospective study using normal tissue samples from 88 patients diagnosed with metastatic ccRCC who had received sunitinib as first-line treatment. They assessed 16 SNPs formerly associated with either outcome or toxicity. VEGFR3 rs307826 and rs307821 variant allele showed shorter PFS (P = 0.051 and 0.032, respectively). Significant associations were also found for ABCB1 rs1128503, FGFR2 rs2981582, NR1/2 rs2276707, and NR1/3 4073054 [17]. Interestingly, a negative correlation between VEGFR3 overexpression and VEGFR3 rs307826 (r =−0.38, P = 0.002), which may underlie the mechanism of resistance conferred by this SNP, was reported within the SUT-REN 07 study [10].

In another communication, Beuselinck et al also found an association between the VEGFR1 rs9582036 and outcome [18]. This result is in agreement with prior findings in patients treated with bevacizumab [19].

Finally, Scartozzi et al. [20] analyzed a cohort of 84 patients with ccRCC treated with first-line sunitinib and identified 3 SNPs associated with outcome (VEGFA rs833061 and rs2010963 and VEGFR3 rs6877011). Interestingly, 2 of them (VEGFA rs833061 and rs2010963) have shown to predict pazopanib outcome [21].

Several studies are currently ongoing to confirm the most important findings. As concordant and mutually validated results are becoming available, it looks like the predictive value of some of the proposed SNPs may be confirmed, reaching the clinical practice in a midterm.

Antiangiogenic drugs are known to produce a disruption of tumor vasculature, which leads to the alteration of circulating endothelial cell (CEC) and circulating endothelial progenitor (CEP) levels in the peripheral blood [22], [23], [24]. Thus, they could potentially be used as pharmacodynamic markers of activity of these drugs.

Vroling et al. [25] prospectively assessed the levels of CEC and hematopoietic progenitor cells (HPC) in a series of 24 patients treated with sunitinib. They found that the level of CEC increased and HPC decreased during the first cycle of treatment, with both recovering to baseline in the 2 weeks off treatment period. Higher CEC levels on day 14 were associated with clinical benefit and better PFS (P = 0.034). Another series of 26 patients with ccRCC also found that the level of CEC increased during the first cycle of sunitinib treatment. Concordantly with the former observation, absence of an increase of the level of CEC was associated with shorter PFS [26]. Finally, Farace et al. [27] studied 46 patients treated with sunitinib and 9 with sorafenib. They observed that at baseline high values of CEPs, but not CECs, were associated with worse outcome. Additionally, patients with increased or stable levels of CEP by day 14 had a better outcome.

Altogether, these data point that the increase in the level of CEC along the first cycle is a pharmacodynamic marker of vasculature damage and efficacy for antiangiogenic drugs in RCC. Additional studies are assessing in different contexts whether CECs could be more accurate predictors of tumor progression than radiological evaluations [28]. However, CECs are still far from a clinical application and establishment of thresholds of normality and uniform definitions of CEC and CEPs should be solved before these determinations could reach daily practice.

In 2009, Perez Garcia et al. [29] prospectively studied 31 patients with ccRCC treated with first-line sunitinib. They explored a panel of 174 cytokines in basal serum in 6 patients with extreme phenotypes (3 long-lasting responders vs. 3 primary refractory patients). They obtained 5 candidates of which high levels of tumor necrosis factor α (TNF-α) and metalloprotease-9 (MMP-9) were suggested as markers of worse outcome when the whole cohort was studied.

In 2012, Porta et al. [30] communicated a prospective study of 85 patients with ccRCC treated with suntinib. Serum basic fibroblast growth factor (bFGF), hepatocyte growth factor, and interleukin (IL)-6 were assessed throughout treatment. Though no significant differences were demonstrated, a trend toward an increase of these cytokines before progression was observed.

DePrimo et al. [31] assessed the plasma levels of 4 soluble proteins from 63 patients with advanced RCC in a phase II study. Soluble (s) VEGFR2 and VEGFR3 levels decreased during cycle 1 and tended to return to baseline after 2 weeks off treatment, indicating a dependency on drug exposure. Significantly larger changes in sVEGFR2 and sVEGFR3 levels were observed in patients with better tumor response. In a different phase II clinical trial including 61 patients with mRCC, again, sVEGFR3 levels decreased with sunitinib treatment, and lower sVEGFR3 baseline levels were associated with longer PFS and response [32].

Despite the number of published studies regarding serum markers, the disparity of determinations, the heterogeneity of results, and the lack of validation precludes clear conclusions or current applicability to clinical practice.

microRNAs are short RNA transcripts that regulate gene expression and could constitute biomarkers of drug response. In this respect, Gamez-Pozo et al. [33] analyzed plasma samples from 44 patients treated with first-line sunitinib and defined a signature of circulating microRNAs that could potentially predict response to therapy. Interestingly, some of these microRNAs were related to angiogenesis and apoptosis. Another study, analyzing microRNA tumor expression in 20 cases, found that microRNA-141 was associated with worse response to sunitinib [34].

Despite the low numbers, both results are of interest as recent findings from the Cancer Genome Atlas have shown the value of microRNAs as prognostic factors in RCC [35]. Thus, this will probably become an intensive area of research in the future.

Xu et al. [21] studied 27 polymorphisms in samples from 397 patients with RCC treated with first-line pazopanib within 3 clinical trials. Two polymorphisms in IL-8, rs1126647 (2767A>T), and rs4073 (−251T>A), and 1 in HIF1A, rs11549467 (A588T), showed a significantly shorter median PFS (P = 0.009, P = 0.01, and P = 0.03, respectively). They also found that 5 polymorphisms in HIF1A, NR1│2, and VEGFA were associated with response (P<0.05) (Table 1).

A currently ongoing translational work is assessing the value of several SNPs in patients recruited in the COMPARZ study (a phase III noninferiority trial that compared pazopanib vs. sunitinib as first-line therapy in patients with ccRCC) [36]. An initial communication pointed to a role of IL-8 SNPs as predictors of OS for both drugs (P = 0.01) [37]. Though mature results are not available yet, the question of whether these SNPs could have a class effect against VEGFR inhibitors or just are prognostic rather than predictive markers will probably remain unanswered.

In 2012, Tran et al. [38] assessed cytokines and angiogenic factors (CAF) in patients treated with pazopanib and included in different clinical trials. They demonstrated that higher concentrations of IL-8, osteopontin, hepatocyte growth factor, and TIMP-1 were associated with shorter PFS. Additionally high concentrations of interleukin 6 were predictive of improved benefit from pazopanib.

A possible association between IL-8 polymorphisms and the circulating levels of the protein has not been explored; however, it is interesting that 2 different studies (Xu et al. and Tran et al.) point toward a potential role of IL-8 as a marker of pazopanib resistance. Moreover, as CAFs can be easily determined in peripheral blood, they have the potential to be used not only as predictive markers at baseline, but also as pharmacodynamic markers along treatment. Additional studies in this regard are warranted.

Pena et al. [39] investigated the effect of VHL alteration on sorafenib response in patients treated within the pivotal trial (TARGET). Only 68 of the 903 patients included in the trial could be analyzed. No association with outcome was found; however, the small numbers preclude reaching any conclusion.

Based on the formerly mentioned works about CAIX and response to cytokines, Choueiri et al [40] studied the role of CAIX as a biomarker of response to antiangiogenics in 2 different sets of patients. First, a single institution retrospective study in patients with ccRCC treated with sunitinib, sorafenib, valatinib, or bevacizumab gave a nonsignificant trend toward better response in sorafenib treated patients. Thus, they performed a second study on samples from the TARGET trial [41]. Of the 903 patients originally enrolled, 133 could be analyzed. However, no association with PFS or tumor shrinkage was found.

The PI3/AKT pathway has recently been confirmed as one of the most important altered pathways in ccRCC [35]. Though classically authors had only studied its association with mTOR inhibitors, in 2010, Jonasch et al. [42] analyzed 40 samples out of 80 patients treated with sorafenib in a prospective phase II trial. They found in a multivariable analysis that high levels of phosphorylation at AKT S473 (pAKT) were associated with worse outcome. Though confirmatory works are needed, several compounds targeting this pathway are under development and could reach the ccRCC arena soon.

Two studies have analyzed the relation between SNPs and sorafenib toxicity. First, an analysis of 178 patients with different solid tumors who received sorafenib or bevacizumab or both demonstrated a higher frequency of hypertension and hand-foot syndrome in patients carrying the VEGFR2 rs1870377 variant [43]. Second, a retrospective analysis performed in 120 patients with cancer who had received sorafenib and in 41 patients with total bilirubin levels collected before and following treatment with the drug, suggested that some genotypes in UGT1A1, UGT1A9, and ABCC2, and serum bilirubin concentration increases, correlated with an abnormally high sorafenib area under the curve [44]. Unfortunately, confirmatory studies are lacking.

Zurita and colleagues studied CAFs in patients treated with sorafenib vs. sorafenib plus IFN [45]. They showed that low baseline concentrations of osteopontin or VEGF were associated with shorter PFS times. Additionally, a signature with 6 CAFs (osteopontin, sCA9, VEGF, tumor necrosis factor–related apoptosis-inducing ligand (TRAIL), collagen IV, and soluble VEGFR2) was generated. Those patients with mRCC who were signature positive obtained more PFS benefit from sorafenib only, whereas signature-negative patients benefited more from sorafenib plus IFN. These results are in some points concordant with those communicated with pazopanib, confirming the interest of CAFs in ccRCC. However, which are the key proteins to be assessed and the optimal cutoff and the moment they should be determined (baseline or along treatment) are issues that need to be addressed.

Bevacizumab is a monoclonal antibody targeting VEGFA, with no inhibitory activity on VEGFR. Thus, genetic studies regarding bevacizumab have focused on VEGFA gene. Interesting associations have been communicated in breast, colorectal, and ovarian cancer [46], [47], [48], [49]. However, it is unknown whether these associations could be valid for ccRCC.

So far, the only published results regarding bevacizumab and SNPs in renal cancer were communicated in 2012 by Lambrechts et al. [19]. In that work, authors used samples from the AVOREN trial (that compared IFN vs. IFN plus bevacizumab as first-line therapy in RCC) and confirmed an association between VEGFR1 SNPs (rs7993418/rs9554316 and rs9513070) and PFS. As mentioned earlier, a similar association has been described for rs9554316 in patients with RCC who were treated with sunitinib [18].

One study evaluated the activation state of 37 proteins in 42 patients who underwent pretreatment with bevacizumab followed by cytoreductive nephrectomy. Its main conclusions were that higher levels of AMP kinase were associated with longer PFS and higher levels of cyclin B were associated with shorter PFS and OS. Thus, a reduced expression and activation of phosphoinositide 3-kinase (PI3K) pathway components and cell cycle factors was seen in patients with prolonged survival after therapy [50]. These results confirm the importance of this pathway, not only regarding the mTOR inhibitors but also for the antiangiogenic drugs.

Only 1 study has been communicated regarding axitinib and SNPs. It was performed within the pivotal AXIS trial comparing axitinib and sorafenib. It found 3 VEGFA polymorphisms, rs699947, rs833061 and rs1570360, significantly associated with PFS but only in univariate analysis [51]. Though differences in PFS among genotypes were driven more by patients in axitinib than in sorafenib arm (n = 127), it is not possible to determine whether these SNPs are truly predictors of response or just prognostic factors. Additional work, testing SNPs deemed as relevant with other antiangiogenics, would be of major interest and could help to determine if this second generation of VEGFR inhibitors (axitinib or tivozanib) could be better options for patients harboring SNPs of resistance.

Only 2 retrospective studies have been communicated regarding mTOR inhibitors and tumor predictive markers in ccRCC. In 2007, Cho et al. [52] published a retrospective analysis of 20 patients from a phase II clinical trial. They determined the expression of CAIX, phospho-S6 (pS6), and pAkt and assessed VHL mutations. An association between higher expression of pS6 and clinical benefit from temsirolimus (P = 0.02) was found, and high expression of pAKT demonstrated a trend toward a better outcome (P = 0.07).

In 2009, Figlin et al. [54] studied samples from patients included in the temsirolimus pivotal trial [53]. They determined HIF-1 alpha and PTEN expression, but did not find any association with outcome.

Cho's results, together with sorafenib and bevacizumab studies, could suggest that tumors with PI3K activation would do better with PI3K or mTOR inhibitors than with antiangiogenics. However, prospective clinical trials will be needed to test this hypothesis.

Lactate dehydrogenase (LDH) is an enzyme regulated by the TORC1 complex. A retrospective study analyzing baseline LDH in patients treated within the pivotal trial of temsirolimus vs. IFN or the combination of both has recently been communicated [55]. Up to 416 from the 626 patients included in the original clinical trial were analyzed. Not surprisingly, elevated LDH was associated with worse outcome with a hazard ratio for death of 2.81 (95% CI = 2.01–3.94, P<0.001). However, interestingly, patients with high LDH values obtained a significant benefit in terms of OS with temsirolimus when compared with IFN (P<0.002). Unfortunately, LDH did not demonstrate to be a reliable marker in the pivotal trial of everolimus (RECORD-1), another mTOR inhibitor [56]. Thus, the usefulness of LDH as a predictive marker must be further investigated.

Additionally a substudy of patients included in the TORAVA trial (that compared the combination of temsirolimus plus bevacizuamb vs. bevacizumab plus IFN vs. sunitinib in ccRCC) assessed serum levels of CAIX as potential predictive marker [57]. Unfortunately, final results were not significant.

Recently, foretinib, a new c-MET inhibitor, has been tested in a phase II trial in papillary renal cell carcinoma [58]. Authors studied potential predictive biomarkers in 67 out of the 74 treated patients. Though alterations affecting the c-MET gene were common in most of the tumors, the presence of germline mutations were associated with 50% of responses compared with 9% in the rest of the series. If confirmed, these mutations could predict foretinib benefit in patients with inherited papillary renal carcinoma with germline c-MET mutations. Unfortunately c-MET mutations are rare in sporadic papillary RCC and no other c-MET pathway biomarkers were significantly associated with outcome in the sporadic papillary RCC cohort in this study.

Section snippets

Summary

Though no kidney cancer biomarker has been implemented in the clinic yet, intensive research is focusing on this issue. SNPs and CAFs are probably the best-studied candidates and some promising results have been concordant between authors. Thus, they are most likely closer than any other biomarker to a clinical application. Regarding tumor alterations, PI3K pathway alteration and the potential therapeutic value of new agents against this pathway are foreseen as one of the major areas of

References (59)

  • National Comprehensive Cancer Network. http://www.nccn.org [accessed...
  • M. Atkins et al.

    Carbonic anhydrase IX expression predicts outcome of interleukin 2 therapy for renal cancer

    Clin Cancer Res

    (2005)
  • D.F. McDermott et al.

    The high-dose aldesleukin (HD IL-2) SELECT trial in patients with metastatic renal cell carcinoma (mRCC)

    J Clin Oncol

    (2010)
  • S.L. Topalian et al.

    Safety, activity, and immune correlates of anti-PD-1 antibody in cancer

    N Engl J Med

    (2012)
  • E.J. Lipson et al.

    Durable cancer regression off-treatment and effective reinduction therapy with an anti-PD-1 antibody

    Clin Cancer Res

    (2013)
  • J.R. Gnarra et al.

    Mutations of the VHL tumour suppressor gene in renal carcinoma

    Nat Genet

    (1994)
  • B.I. Rini et al.

    Clinical response to therapy targeted at vascular endothelial growth factor in metastatic renal cell carcinoma: impact of patient characteristics and Von Hippel-Lindau gene status

    BJU Int

    (2006)
  • T.K. Choueiri et al.

    Von Hippel-Lindau gene status and response to vascular endothelial growth factor targeted therapy for metastatic clear cell renal cell carcinoma

    J Urol

    (2008)
  • M.L. Nickerson et al.

    Improved identification of von Hippel-Lindau gene alterations in clear cell renal tumors

    Clin Cancer Res

    (2008)
  • N.P. van Erp et al.

    Pharmacogenetic pathway analysis for determination of sunitinib-induced toxicity

    J Clin Oncol

    (2009)
  • N.P. van Erp et al.

    Myelosuppression by sunitinib is flt-3 genotype dependent

    Br J Cancer

    (2010)
  • J.J. Kim et al.

    Association of VEGF and VEGFR2 single nucleotide polymorphisms with hypertension and clinical outcome in metastatic clear cell renal cell carcinoma patients treated with sunitinib

    Cancer

    (2012)
  • A.A. van der Veldt et al.

    Genetic polymorphisms associated with a prolonged progression-free survival in patients with metastatic renal cell cancer treated with sunitinib

    Clin Cancer Res

    (2011)
  • B. Beuselinck et al.

    Single-nucleotide polymorphisms associated with outcome in metastatic renal cell carcinoma treated with sunitinib

    Br J Cancer

    (2013)
  • Beuselinck B, Karadimou A, Lambrechts D. et al. VEGFR1 single nucleotide polymorphisms associated with outcome in...
  • M. Scartozzi et al.

    VEGF and VEGFR polymorphisms affect clinical outcome in advanced renal cell carcinoma patients receiving first-line sunitini

    Br J Cancer

    (2013)
  • C.F. Xu et al.

    Pazopanib efficacy in renal cell carcinoma: evidence for predictive genetic markers in angiogenesis-related and exposure-related genes

    J Clin Oncol

    (2011)
  • S.J. Shin et al.

    Circulating vascular endothelial growth factor receptor 2/pAkt-positive cells as a functional pharmacodynamic marker in metastatic colorectal cancers treated with antiangiogenic agent

    Invest New Drugs

    (2013)
  • M. Manzoni et al.

    Circulating endothelial cells and their apoptotic fraction are mutually independent predictive biomarkers in Bevacizumab-based treatment for advanced colorectal cancer

    J Cancer Res Clin Oncol

    (2012)
  • Cited by (0)

    View full text