Bevacizumab is one of the most widely used antiangiogenic drugs in oncology, but the overall beneficial effects of this VEGF-A targeting agent are relatively modest, in part due to the lack of a biomarker to select patients most likely to respond favorably. Several molecular aberrations in cancer influence angiogenesis, including mutations in the tumor suppressor gene TP53, which occur frequently in many human malignancies. In this study, we present a multiple regression analysis of transcriptomic data in 123 patients with non–small cell lung cancer (NSCLC) showing that TP53 mutations are associated with higher VEGF-A expression (P = 0.006). This association was interesting given a recent retrospective study showing longer progression-free survival in patients with diverse tumors who receive bevacizumab, if tumors harbor mutant TP53 instead of wild-type TP53. Thus, our current findings linking TP53 mutation with VEGF-A upregulation offered a mechanistic explanation for why patients exhibit improved outcomes after bevacizumab treatment when their tumors harbor mutant TP53 versus wild-type TP53. Overall, this work warrants further evaluation of TP53 as a ready biomarker to predict bevacizumab response in NSCLC and possibly other tumor types. Cancer Res; 75(7); 1187–90. ©2015 AACR.

Non–small cell lung cancer (NSCLC) is a leading cause of cancer-related deaths (1) and represents a heterogeneous group of neoplasms, mostly squamous cell and adenocarcinoma. A diagnosis of NSCLC carries a grim prognosis; 5-year survival is less than 15% (2). Bevacizumab is an antibody that targets vascular endothelial growth factor-A (VEGF-A). Bevacizumab combined with carboplatin and paclitaxel has been approved for the initial treatment of unresectable NSCLC, based on a 2-month increase in survival (12.3 vs. 10.3 months) when compared with the chemotherapy alone (3). Bevacizumab is also used in the treatment of renal and colon cancer and glioblastoma multiforme, with similar modest benefits; its approval for breast cancer was recently revoked by the FDA. The relatively small impact of bevacizumab on outcome may be due to the fact that a subset of patients respond, while others derive no salutary effects or, conceivably, might even be harmed by bevacizumab. Yet, no biomarker for patient selection has been identified. This is especially important because bevacizumab can have serious toxicity, including hypertension and bowel perforation; it is also extremely expensive, costing about 40,000 to 100,000 dollars per year (4).

Of interest in this regard, we recently reported, in a retrospective study of patients with diverse cancers, that use of bevacizumab-containing regimens predicted for longer progression-free survival (PFS) in TP53-mutant tumors (multivariate analysis; P < 0.001; PFS = 11 vs. 5.0 months; mut vs. wt TP53; ref. 5). The mechanism by which this correlation might occur remains unclear. However, several studies suggest a role for TP53 in angiogenesis (6, 7). Importantly in this regard, though early data failed to find a clear association between VEGF expression and outcome after bevacizumab administration, recent data using improved technology suggest that circulating levels of the short isoform of VEGF-A is a strong biomarker candidate for predicting benefit (8).

Herein, we report that, in a transcriptomic evaluation of NSCLC (2), multiple regression analysis showed that VEGF-A expression correlated independently with TP53 mutational status. These data link TP53 mutations directly with VEGF-A, the primary target of bevacizumab, and suggest that TP53 status merits further exploration as a biomarker for bevacizumab response in NSCLC as well as in additional neoplasms.

Patients and tissue samples

Snap-frozen tumor and adjacent normal lung tissue samples from a cohort of 123 patients who underwent complete surgical resection at the Institut Mutualiste Montsouris were used. All tissues were banked after written informed patient consent, and the study was approved by the Ethics Committee of Institut Gustave Roussy.

Gene expression assay and analysis

RNA was extracted with TRIzol, quantified and qualified with a Nanodrop. Gene expression was performed with 244K Human exon array from Agilent (custom design with the content of the 44K Human genome plus 195,000 probes, one for each exon as defined in the refGene list of UCSC build hg18; http://genome.ucsc.edu/). The differential gene expression in tumor versus matched normal lung tissues was calculated in each patient and used in our analysis.

Gene mutations analysis

DNA was extracted with the QIAamp DNA Mini Kit (Qiagen). This was followed by PCR amplification of target exons. Sequence analysis and alignment were performed with SeqScape software (Applied Biosystems). All detected mutations were confirmed in at least one independent PCR reaction. In all 123 samples, full coding sequences of exons, including oncogenic mutational hotspots, were analyzed corresponding to TP53 (NM_000546.4) exons 5–8; KRAS (NM_004448.2) exons 2 and 3; EGFR (NM_005228.3) exons 18–21; PIK3CA (NM_006218.2) exons 10 and 21; BRAF (NM_004333.4) exon 15; and ERBB2 (NM_004448.2) exons 18, 20–24.

Statistical analysis

Associations between TP53 mutational status and RNA expression levels were assessed with a Mann–Whitney test in a univariable analysis. Multiple regression models were fit to assess the best predictors for VEGF-A expression and the association between TP53 status and other variables. Assumptions of multiple regression have been checked graphically. A subanalysis was performed, segregating by histology. P values less than 0.05 were considered statistically significant. All statistical analyses were conducted using SPSS software (v.22.0).

Our study population (N = 123) was composed of 57 patients with adenocarcinoma of the lung (46%), 50 patients with squamous cell carcinoma (41%), 13 patients with large-cell carcinoma (11%), and three were unclassified cases (3%). Testing for aberrations in the TP53, KRAS, EGFR, BRAF, PIK3CA, and ERBB2 genes revealed the following rates of abnormalities: TP53 (N = 31, 25%; 24.6% of adenocarcinoma vs. 28.6% of squamous cell; P = 0.665; Supplementary Results); KRAS (N = 20, 18%); EGFR (N = 13, 12%); PIK3CA (N = 2); and BRAF and ERBB2 (N = 1 each; Supplementary Table S1).

In univariable analysis, transcriptomic data (84 gene products; Supplementary Table S2) showed differential expression as follows: median fold changes of tumor versus normal tissues of VEGF-A (3.9 vs. 3.0; P = 0.015), mTOR (1.8 vs 1.4; P = 0.013), BAX (1.8 vs 1.6; P = 0.033), APAF1 (0.7 vs 0.6; P = 0.028), and AREG (0.1 vs. 0.2; P = 0.015; mut vs. wt TP53). In multiple regression analysis, TP53 mutations correlated independently with higher VEGF-A (P = 0.006) and BAX (P = 0.032) expression (Table 1).

Table 1.

TP53 association with biologic parameters (multiple regression model)

ParametersMut TP53 (N = 31), median (CI 95%)Wt TP53 (N = 91), median (CI 95%)Coefficient95% CIP
VEGF-A 3.9 (3.3–5.6) 3.0 (2.2–3.4) 0.032 0.01–0.05 0.006 
BAX 1.8 (1.7–2.0) 1.6 (1.5–1.7) 0.167 0.02–0.32 0.032 
mTOR 1.8 (0.6–2.0) 1.4 (0.8–1.5) 0.074 −0.09 – 0.24 0.381 
APAF1 0.7 (0.6–1.1) 0.6 (0.5–0.7) 0.177 −0.05 – 0.40 0.124 
AREG 0.1 (0.0–0.2) 0.2 (0.1–0.4) 0.000 −0.002 – 0.001 0.608 
 N (%) N (%)    
Histology   0.143 −0.04 – 0.32 0.115 
Adenocarcinoma 14 (24.6) 43 (75.4)    
Squamous cell 14 (28.6) 35 (71.4)    
KRAS   0.183 −0.04 – 0.41 0.108 
 Mut KRAS 7 (35) 13 (65)    
 Wt KRAS 21 (23) 70 (77)    
EGFR   −0.063 −0.31 – 0.18 0.613 
 Mut EGFR 3 (23) 10 (77)    
 Wt EGFR 26 (26) 73 (74)    
ParametersMut TP53 (N = 31), median (CI 95%)Wt TP53 (N = 91), median (CI 95%)Coefficient95% CIP
VEGF-A 3.9 (3.3–5.6) 3.0 (2.2–3.4) 0.032 0.01–0.05 0.006 
BAX 1.8 (1.7–2.0) 1.6 (1.5–1.7) 0.167 0.02–0.32 0.032 
mTOR 1.8 (0.6–2.0) 1.4 (0.8–1.5) 0.074 −0.09 – 0.24 0.381 
APAF1 0.7 (0.6–1.1) 0.6 (0.5–0.7) 0.177 −0.05 – 0.40 0.124 
AREG 0.1 (0.0–0.2) 0.2 (0.1–0.4) 0.000 −0.002 – 0.001 0.608 
 N (%) N (%)    
Histology   0.143 −0.04 – 0.32 0.115 
Adenocarcinoma 14 (24.6) 43 (75.4)    
Squamous cell 14 (28.6) 35 (71.4)    
KRAS   0.183 −0.04 – 0.41 0.108 
 Mut KRAS 7 (35) 13 (65)    
 Wt KRAS 21 (23) 70 (77)    
EGFR   −0.063 −0.31 – 0.18 0.613 
 Mut EGFR 3 (23) 10 (77)    
 Wt EGFR 26 (26) 73 (74)    

Using VEGF-A as the dependent variable, in a multiple linear regression analysis including TP53, KRAS, and EGFR (all mut vs. wt), histology (adenocarcinoma vs. squamous), and the other genes found differentially expressed (BAX, mTOR APAF1, and AREG) as variables, the only statistically significant independent predictor for VEGF-A expression was TP53 mutational status (P = 0.006; Table 2). When forward and backward regression analyses were performed, both tests confirmed that TP53 mutation was the best independent predictor for higher VEGF-A levels (P = 0.008). Segregating by histology, a multiple regression analysis demonstrated that TP53 mutational status was the only independent factor predicting increased VEGF-A expression in adenocarcinoma (P = 0.007) but not in squamous cell carcinoma (P = 0.599), consistent with the observation that expression levels of VEGF-A were significantly higher in TP53-mutated specimens in adenocarcinoma (median 5.95 vs. 3.0; P = 0.012), but not in squamous cell carcinoma (median 3.9 vs. 3.1; P = 0.636; Supplementary Table S3). This difference between adenocarcinoma and squamous cell carcinoma suggests the complexity of the mechanisms involving TP53 and angiogenesis regulation, and illustrates possible mechanistic differences between tissues.

Table 2.

VEGF-A association with biologic parameters (multiple regression model)

ParametersCoefficient95% CIP
TP53 2.039 0.61 – 3.47 0.006 
 Mut TP53    
 Wt TP53    
KRAS −0.693 −2.5 – 1.11 0.448 
 Mut KRAS    
 Wt KRAS    
EGFR 0.594 −1.4 – 2.6 0.552 
 Mut KRAS    
 Wt KRAS    
Histology −1.091 −2.5 – 0.33 0.131 
Adenocarcinoma    
Squamous cell    
Gene expression    
BAX −0.218 −1.5 – 1.0 0.729 
mTOR 0.322 −1.01 – 1.67 0.635 
APAF1 −0.765 −2.6 – 1.1 0.408 
AREG −0.001 −0.01 – 0.01 0.815 
ParametersCoefficient95% CIP
TP53 2.039 0.61 – 3.47 0.006 
 Mut TP53    
 Wt TP53    
KRAS −0.693 −2.5 – 1.11 0.448 
 Mut KRAS    
 Wt KRAS    
EGFR 0.594 −1.4 – 2.6 0.552 
 Mut KRAS    
 Wt KRAS    
Histology −1.091 −2.5 – 0.33 0.131 
Adenocarcinoma    
Squamous cell    
Gene expression    
BAX −0.218 −1.5 – 1.0 0.729 
mTOR 0.322 −1.01 – 1.67 0.635 
APAF1 −0.765 −2.6 – 1.1 0.408 
AREG −0.001 −0.01 – 0.01 0.815 

Angiogenesis plays a critical role in the growth and spread of cancer, as the resulting new blood vessels supply the tumor with needed nutrients and oxygen; VEGF is probably the most commonly involved proangiogenic factor (9). Our data complement previous preclinical data in NSCLC correlating either aberrant TP53 expression with higher VEGF level (measured by quantitative reverse transcription PCR; ref. 10), or TP53 gene mutations with a strongly positive VEGF immunoreactivity (11). Further, wt TP53 indirectly represses VEGF promoter activity by inhibiting transcription factors, e.g., SP1 and E2F; there is also a TP53-binding site adjacent to the hypoxia-inducible factor-1α (HIF1α) binding site that resides within the VEGF promoter and is essential for VEGF induction during hypoxia (12). In addition, Narendran and colleagues (7) showed that transfection of stromal cells with mutant p53 increased synthesis of VEGF. These mechanistic data are consistent with our observation of high VEGF-A transcripts in the presence of mutant TP53. BAX transcript levels were also higher in tumors harboring mutant TP53 (Table 1). The BAX product, also known as Bcl-2–like protein 4, promotes apoptosis by binding to and antagonizing the Bcl-2 protein. An association between BAX overexpression and specific TP53 mutations of the loop–sheet–helix in NSCLC has previously been reported, though other types of TP53 mutations correlated with lower levels of BAX expression (13).

Using gene expression profiling techniques, prior studies also demonstrated that adenocarcinoma and squamous cell NSCLCs have different expression portfolios (2, 14), which is in line with our observation that the association between TP53 mutations and increased VEGF-A transcripts appears specific for adenocarcinoma of the lung. It is conceivable that the improved outcome that was previously reported by our group (5) in bevacizumab-treated patients who harbored TP53 mutations versus wild-type TP53 is due to the association between TP53 mutations and higher VEGF-A, the target for bevacizumab. However, this association may not hold true for all tumor types. For instance, benefit from bevacizumab could not be correlated with TP53 status in metastatic colorectal cancer (15, 16). On the other hand, these prior studies in colorectal cancer had significant differences in methodology that might have influenced outcome. Kara and colleagues (15) examined only 34 patients and analyzed p53 expression, not mutational status; Ince and colleagues (16) examined survival, not PFS. In contrast, the multivariate analysis demonstrating that a bevacizumab-containing regimen was an independent factor associated with better outcomes in TP53-mutated patients included a variety of tumors and analyzed PFS (5).

The salutary effects of antiangiogenic therapy can differ dramatically, depending on the cancer type. For the majority of tumors, bevacizumab must be combined with other drugs to show benefit. Few responses are observed with monotherapy (9). One exception is ovarian cancer, where monotherapy with bevacizumab can achieve response rates in the 16% to 21% range even in advanced disease (17). One of the most responsive subsets of ovarian cancer is the high-grade serous histology. Of interest in this regard, TP53 mutations are a hallmark of these tumors, with a frequency exceeding 90% (18). In sharp contrast, prostate carcinomas, with a rather low TP53 mutation frequency (approximately 11%; ref. 19), failed to demonstrate benefit from bevacizumab (20).

Our study demonstrates an independent correlation between TP53 mutations and VEGF-A expression in a comprehensive transcriptomic analysis of 123 patients with NSCLC. One of the distinctive features of this dataset is the investigation of differential gene expression in tumor versus matched normal lung tissues, in each patient. This methodology enabled us to discard the noise from the background variability between patients and to pinpoint expression anomalies most likely related to oncogenesis. This unique study feature enabled the demonstration, for the first time, of an independent association between TP53 mutational status and overexpression of VEGF-A. Further interrogation of the data indicates that this correlation pertains to adenocarcinomas, consistent with bevacizumab's approval for adenocarcinomas of the lung.

Bevacizumab has previously been hailed as the best-selling drug in oncology. However, for most cancers in which it is used, including NSCLC, renal, and colon cancer as well as glioblastoma multiforme, bevacizumab increases survival by only a couple of months. Furthermore, the FDA recently acted to rescind its approval in breast cancer because of the lack of proof a survival advantage, despite previous evidence of some activity in this disease (8). Most likely, in relevant malignancies, a subgroup of patients is responsive to bevacizumab, but a biomarker defining this subset has remained elusive. TP53 mutations are found in diverse cancers, and 25% of our NSCLC patients had a TP53 alteration. Indeed, TP53 is one of the most commonly aberrant genes across tumors, yet there is no approved therapy that targets it. Our data have previously suggested a clinical association between TP53 mutations and better PFS after bevacizumab treatment (5). Our current observations show that TP53 mutations are an independent predictor of high expression of VEGF-A, the primary target of bevacizumab. These observations suggest that upregulation of transcription of the VEGF-A gene (12) may link TP53 status to antiangiogenic therapy outcome. Prospective investigation of TP53 as a biomarker for response to bevacizumab, and possibly other antiangiogenic agents, in NSCLC, as well as other malignancies, is therefore warranted.

R. Kurzrock is founder of RScueRX. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M. Schwaederlé, V. Lazar, P. Validire

Development of methodology: M. Schwaederlé, V. Lazar, P. Validire, L. Lacroix

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V. Lazar, P. Validire, L. Lacroix, J.C. Soria

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Schwaederlé, V. Lazar, L. Lacroix, J.C. Soria, Y. Pawitan, R. Kurzrock

Writing, review, and/or revision of the manuscript: M. Schwaederlé, V. Lazar, J. Hansson, J.C. Soria, R. Kurzrock

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases):P. Validire

Study supervision: M. Schwaederlé, V. Lazar, J. Hansson

Other (coordinator of EU FP6 Integrated Project CHEMORES providing support for the project): J. Hansson

Other (approval and final editing): R. Kurzrock

The authors thank Drs. Sarah Murray and Lisa Madlensky for their help on the classification of TP53 alterations.

This study was financially supported by Joan and Irwin Jacobs Fund, MyAnswerToCancer philanthropic fund, and Chemores (www.chemores.org), an EU FP6 funded program.

1.
Siegel
R
,
DeSantis
C
,
Virgo
K
,
Stein
K
,
Mariotto
A
,
Smith
T
, et al
Cancer treatment and survivorship statistics, 2012.
CA Cancer J Clin
2012
;
62
:
220
41
.
2.
Lazar
V
,
Suo
C
,
Orear
C
,
van den Oord
J
,
Balogh
Z
,
Guegan
J
, et al
Integrated molecular portrait of non-small cell lung cancers.
BMC Med Genomics
2013
;
6
:
53
.
3.
Sandler
A
,
Gray
R
,
Perry
MC
,
Brahmer
J
,
Schiller
JH
,
Dowlati
A
, et al
Paclitaxel-carboplatin alone or with bevacizumab for non–small-cell lung cancer.
N Engl J Med
2006
;
355
:
2542
50
.
4.
Avastin (bevacizumab) price [Internet]
.
Available from
: http://www.news-medical.net/health/Avastin-(Bevacizumab)-Price.aspx
5.
Said
R
,
Hong
DS
,
Warneke
CL
,
Lee
JJ
,
Wheler
JJ
,
Janku
F
, et al
P53 Mutations in advanced cancers: clinical characteristics, outcomes, and correlation between progression-free survival and bevacizumab-containing therapy.
Oncotarget
2013
;
4
:
705
14
.
6.
Ravi
R
,
Mookerjee
B
,
Bhujwalla
ZM
,
Sutter
CH
,
Artemov
D
,
Zeng
Q
, et al
Regulation of tumor angiogenesis by p53-induced degradation of hypoxia-inducible factor 1alpha.
Genes Dev
2000
;
14
:
34
44
.
7.
Narendran
A
,
Ganjavi
H
,
Morson
N
,
Connor
A
,
Barlow
JW
,
Keystone
E
, et al
Mutant p53 in bone marrow stromal cells increases VEGF expression and supports leukemia cell growth.
Exp Hematol
2003
;
31
:
693
701
.
8.
Lambrechts
D
,
Lenz
H-J
,
de Haas
S
,
Carmeliet
P
,
Scherer
SJ
. 
Markers of response for the antiangiogenic agent bevacizumab.
J Clin Oncol Off J Am Soc Clin Oncol
2013
;
31
:
1219
30
.
9.
Bellou
S
,
Pentheroudakis
G
,
Murphy
C
,
Fotsis
T
. 
Anti-angiogenesis in cancer therapy: Hercules and hydra.
Cancer Lett
2013
;
338
:
219
28
.
10.
Yuan
A
,
Yu
C-J
,
Luh
K-T
,
Kuo
S-H
,
Lee
Y-C
,
Yang
P-C
. 
Aberrant p53 expression correlates with expression of vascular endothelial growth factor mRNA and interleukin-8 mRNA and neoangiogenesis in non–small-cell lung cancer.
J Clin Oncol Off J Am Soc Clin Oncol
2002
;
20
:
900
10
.
11.
Niklińska
W
,
Burzykowski
T
,
Chyczewski
L
,
Nikliński
J
. 
Expression of vascular endothelial growth factor (VEGF) in non–small cell lung cancer (NSCLC): association with p53 gene mutation and prognosis.
Lung Cancer Amst Neth
2001
;
34
:
S59
64
.
12.
Farhang Ghahremani
M
,
Goossens
S
,
Haigh
JJ
. 
The p53 family and VEGF regulation: “It's complicated.”
Cell Cycle Georget Tex
2013
;
12
:
1331
2
.
13.
Huang
C
,
Kohno
N
,
Inufusa
H
,
Kodama
K
,
Taki
T
,
Miyake
M
. 
Overexpression of bax associated with mutations in the loop-sheet-helix motif of p53.
Am J Pathol
1999
;
155
:
955
65
.
14.
Liu
J
,
Yang
XY
,
Shi
WJ
. 
Identifying differentially expressed genes and pathways in two types of non-small cell lung cancer: adenocarcinoma and squamous cell carcinoma.
Genet Mol Res GMR
2014
;
13
:
95
102
.
15.
Kara
O
,
Duman
BB
,
Kara
B
,
Erdogan
S
,
Parsak
CK
,
Sakman
G
. 
Analysis of PTEN, VEGF, HER2 and P53 status in determining colorectal cancer benefit from bevacizumab therapy.
Asian Pac J Cancer Prev APJCP
2012
;
13
:
6397
401
.
16.
Ince
WL
,
Jubb
AM
,
Holden
SN
,
Holmgren
EB
,
Tobin
P
,
Sridhar
M
, et al
Association of k-ras, b-raf, and p53 status with the treatment effect of bevacizumab.
J Natl Cancer Inst
2005
;
97
:
981
9
.
17.
Sato
S
,
Itamochi
H
. 
Bevacizumab and ovarian cancer.
Curr Opin Obstet Gynecol
2012
;
24
:
8
13
.
18.
Wojnarowicz
PM
,
Oros
KK
,
Quinn
MCJ
,
Arcand
SL
,
Gambaro
K
,
Madore
J
, et al
The genomic landscape of TP53 and p53 annotated high grade ovarian serous carcinomas from a defined founder population associated with patient outcome.
PLoS ONE
2012
;
7
:
e45484
.
19.
cbioportal [Internet]
.
Available from
: http://www.cbioportal.org/public-portal/index.do
20.
Ogita
S
,
Tejwani
S
,
Heilbrun
L
,
Fontana
J
,
Heath
E
,
Freeman
S
, et al
Pilot phase II trial of bevacizumab monotherapy in nonmetastatic castrate-resistant prostate cancer.
ISRN Oncol
2012
;
2012
:
242850