BRIEF CoMMUNICATIoNS - Nature

b r i e f c o m m u n i c at i o n s

© 2012 Nature America, Inc. All rights reserved.

Identification of a mutation in the extracellular domain of the Epidermal Growth Factor Receptor conferring cetuximab resistance in colorectal cancer Clara Montagut1,2,9, Alba Dalmases1,2,9, Beatriz Bellosillo2,3, Marta Crespo4, Silvia Pairet2,3, Mar Iglesias3,5, Marta Salido3, Manuel Gallen1,2, Scot Marsters6, Siao Ping Tsai6, André Minoche7, Somasekar Seshagiri6, Sergi Serrano2,3,5, Heinz Himmelbauer7, Joaquim Bellmunt1,2,8, Ana Rovira1,2, Jeff Settleman6,9, Francesc Bosch4,9 & Joan Albanell1,2,5,9 Antibodies against epidermal growth factor receptor (EGFR)—cetuximab and panitumumab—are widely used to treat colorectal cancer. Unfortunately, patients eventually develop resistance to these agents. We describe an acquired EGFR ectodomain mutation (S492R) that prevents cetuximab binding and confers resistance to cetuximab. Cells with this mutation, however, retain binding to and are growth inhibited by panitumumab. Two of ten subjects studied here with disease progression after cetuximab treatment acquired this mutation. A subject with cetuximab resistance harboring the S492R mutation responded to treatment with panitumumab. The development of antibodies to EGFR—cetuximab and p anitumumab—was a milestone in metastatic colorectal cancer (mCRC) treatment. Both of the EGFR antibodies prolong survival in subjects with mCRC and are a standard component of therapy in these individuals. KRAS mutation status predicts an individual’s innate resistance to these antibodies, and, because of this, individuals with KRASmutant mCRC are excluded from treatment with these antibodies. However, subjects with mCRC who respond to antibodies to EGFR ultimately acquire resistance to these agents1–6. To identify the mechanisms of acquired cetuximab resistance, we established cetuximab-resistant cells from the highly sensitive human mCRC cell line DiFi (which is wild-type for KRAS, BRAF and PI3K and has an amplification of EGFR) (Fig. 1a and Supplementary Methods). We continuously treated these cells with cetuximab. Five months after beginning treatment, we isolated DiFi-derived cetuximab-resistant clones (DCR) (DCR 7, DCR 9 and DCR 10) (Fig. 1a), which were morphologycally similar to their parental cells (Supplementary Fig. 1). Treatment with cetuximab did not affect the proliferation, apoptosis

(Supplementary Fig. 2) or EGFR signaling to extracellularrelated kinase (ERK) and protein kinase B (AKT) (Fig. 1b) in the DCR clones. A phospho-receptor tyrosine kinase (RTK) array did not reveal any additional activated RTKs in the DCR clones relative to their parental cells (Supplementary Fig. 3); in addition, we did not detect mutations in KRAS, BRAF or PIK3CA or loss of PTEN expression in the DCR clones (data not shown). Notably, parental DiFi cells and DCR clones were equally sensitive to the EGFR kinase inhibitor gefitinib (Fig. 1c,d), suggesting that the DCR clones remained dependent on EGFR for their growth and survival. Viability and EGFR signaling in the DCR clones were effectively decreased by treatment with the antibody panitumumab (Fig. 1e,f and Supplementary Fig. 4). Because treatment with panitumumab blocked EGFR activation by EGF in both cetuximab-sensitive (parental) and cetuximab-resistant (DCR) cells and treatment with cetuximab had this effect only in the parental but not in the DCR cells, we hypothesized that a change in the EGFR cetuximab-binding epitope might have occurred in the DCR clones (Fig. 1g). DNA sequencing of the EGFR coding region (NM_005228.1) revealed a C→A substitution at nucleotide 1,476 in all three DCR clones that was not present in the parental DiFi cells (Fig. 2a). The parental and DCR cells also harbored previously described single-nucleotide polymorphisms (the EGFR mutations C474T and C2709T) and EGFR gene amplification, confirming that the DCR cells had not arisen from a contaminating subpopulation. C1476A causes a substitution of serine to arginine at amino acid 492, which is within the EGFR ectodomain, and a bulky side chain at this position could interfere with cetuximab binding (Fig. 2b). We then expressed wild-type and S492R mutant EGFR in NIH 3T3 fibroblasts lacking endogenous EGFR (Fig. 2c). Both cetuximab and panitumumab inhibited wild-type EGFR activation, whereas in cells carrying the S492R mutation, only panitumumab blocked the activation of EGFR (Fig. 2d). Flow cytometry showed that cetuximab and panitumumab efficiently bound cells expressing wild-type EGFR, but that only panitumumab bound the S942R mutant cells (Fig. 2e). In vitro biochemical binding studies confirmed that the S492R mutant is selectively defective in binding cetuximab but not panitumumab (Fig. 2f). We did not detect the S492R mutation in 156 tumors analyzed from chemotherapy-naive and targeted-agent–naive subjects with mCRC. We also examined pre- and post-therapy specimens from ten individuals with mCRC who experienced disease progression after a prior response to cetuximab with chemotherapy (Supplementary Tables 1 and 2). This study was approved by the local Ethics Board (Clinical Research Ethical Committee of the Parc de Salut Mar, CEIC-IMAS2009/3515/). Informed consent was obtained from all subjects. All pre-treatment biopsies were

1Department

of Medical Oncology, Hospital del Mar, Barcelona, Spain. 2Cancer Research Program, Institut Municipal d’Investigació Médica (IMIM, Hospital del Mar Research Institute), Barcelona, Spain. 3Department of Pathology, Hospital del Mar, Barcelona, Spain. 4Department of Hematology, Vall d’Hebron University Hospital, Barcelona, Spain. 5Departament de Medicina, Universitat Autonoma de Barcelona, Bellaterra, Spain. 6Discovery Oncology, Genentech, Inc, San Francisco, California, USA. 7Center for Genomic Regulation, Barcelona, Spain. 8Medical School, Universitat Pompeu Fabra (UPF), Barcelona, Spain. 9These authors contributed equally to this work. Correspondence should be addressed to C.M. ([email protected]) or J.A. ([email protected]). Received 15 August 2011; accepted 15 November 2011; published online 22 January 2012; corrected after print 6 July 2012; doi:10.1038/nm.2609

nature medicine VOLUME 18 | NUMBER 2 | FEBRUARY 2012

221

B r i e f c o m m u n i c at i o n s a

b

Cell viability (% of control)

Figure 1 Cetuximab-resistant cells are 120 DiFi DCR7 DCR9 DCR10 sensitive to the EGFR tyrosine kinase –1 Ctx (10 µg ml ) – + – + – + – + inhibitor gefitinib and the monoclonal antibody 100 P-ERK1/2 to EGFR panitumumab. (a) Cells from the DiFi 80 ERK1/2 DiFi human colorectal cancer cell line were made 60 DCR7 P-AKT resistant to cetuximab (Ctx) by continuous DCR9 40 DCR10 AKT exposure to 1 µg ml−1 cetuximab for 5 months. 20 Cell viability was measured after treating DiFi parental cells and DCR clones with increasing 0 DiFi DCR7 DCR9 DCR10 0 0.001 0.01 0.1 1 5 10 20 50 concentrations of cetuximab for 72 h. Gefitinib (100 nM) – + – + – + – + Ctx (µg ml–1) (b) Sensitivity to cetuximab is correlated with P-ERK1/2 the effective inhibition of EGFR downstream DiFi 120 ERK1/2 DCR7 effectors (AKT and ERK). Cell lysates from 100 DCR9 P-AKT DCR10 DiFi cells and DCRs were collected after 2 h of 80 AKT treatment with cetuximab. An immunoblotting 60 analysis was performed using antibodies to the 40 indicated proteins. P-ERK1/2, phosphorylated DiFi DCR7 DCR9 DCR10 ERK1 or ERK2; ERK1/2, ERK1 or ERK2; –1 20 Pnm (10 µg ml ) – + – + – + – + P-AKT, phosphorylated AKT. (c) Parental and 0 P-ERK1/2 resistant cells are equally sensitive to gefitinib. 0 0.01 0.1 1.0 10.0 ERK1/2 Parental DiFi and DCR cells were treated Gefitinib (µM) P-AKT with increasing concentrations of gefitinib, 120 DiFi AKT DCR7 and viable cells were measured after 72 h. 100 DCR9 (d) Sensitivity to gefitinib correlated with DCR10 80 the effective inhibition of EGFR downstream Ctx Pnm Ctx Pnm 60 EGF – effectors (AKT and ERK). Cells were cultured + – + – + – + – + – + –1 (10 ng ml ) for 2 h with gefitinib, and cell lysates were 40 P-EGFR subjected to western blot analysis using 20 antibodies to the indicated proteins. EGFR 0 (e) Parental and DCR cells are equally sensitive 0 0.01 0.1 1.0 10.0 20.0 DiFi DCR10 –1 to panitumumab (Pnm). Parental DiFi and Pnm (µg ml ) DCR cells were treated with increasing concentrations of panitumumab, and viable cells were measured after 72 h. (f) Sensitivity to panitumumab is correlated with the effective inhibition of EGFR downtream effectors (AKT and ERK). Cell lysates from DiFi cells and DCRs were collected after treatment for 2 h with panitumumab. An immunoblotting analysis was performed using antibodies to the indicated proteins. (g) Comparison of the abilities of cetuximab and panitumumab to disrupt the phosphorylation of EGFR after stimulation with EGF in DCR cells. DiFi and DCR cells were treated with cetuximab or panitumumab for 2 h and stimulated with EGF for 15 min. Immunoblotting was performed using antibodies to the indicated proteins. P-EGFR, phosphorylated EGFR.

d

f

e




c

wild type for EGFR, KRAS, BRAF and PIK3CA, except for the biopsy from subject 9, whose tumor harbored a V600E mutation in BRAF. Notably, among the biopsies taken after treatment with cetuximab (post-cetuximab biopsies), we detected the S492R mutation in two subjects (subjects 3 and 9). Whereas subject 9 harbored the same nucleotide change that we saw in the DCRs, the mutation in subject 3 involved a different nucleotide substitution that affected the same codon, an A→C change at nucleotide 1,474, which also yielded the S492R substitution (Supplementary Fig. 5). We detected the S492R mutation in subject 9 by deep sequencing at a frequency of 3%, as well as by quantitative RT-PCR. Deep sequencing of the DNA sample from subject 3 revealed that the mutation was present at a frequency of 25%. Deep sequencing of normal cells from subjects 3 (saliva) and 9 (colon) revealed only the wild-type sequence, suggesting that the S492R mutation was somatic. Deep sequencing of pre-treatment tumor specimens from both subjects confirmed that the mutation was not detectably present before treatment (data not shown). Both subjects harboring the S492R mutation had EGFR gene amplification (in both their pre- and postcetuximab specimens) (Supplementary Fig. 6). The post-cetuximab specimen from subject 3 did not harbor known KRAS, BRAF or PIK3CA mutations. The post-cetuximab tumor sample from subject 9 harbored the V600E BRAF mutation, which was also detected in the precetuximab specimen from the same subject. Subject 9 was exitus at the time of the identification of the S492R mutation in his specimen. In subject 3, detection of the acquired EGFR S492R mutation led us to offer him panitumumab according to the approved monotherapy schedule (6 mg per kg of body weight

222

g

every 2 weeks). After 2 months of treatment, a computed tomographic scan showed a more than 50% reduction in the volume in all liver lesions (Fig. 2g) in this subject, with a marked decline in the carcinoembryonic antigen blood tumor marker, thereby reinforcing the preclinical data. At month 5 of treatment with panitumumab, subject 3 showed disease progression. Here we have identified a clinically relevant point mutation within the EGFR extracellular domain that arises during cetuximab therapy and confers resistance to this agent. A missense mutation of the target of a therapeutic antibody has not been previously reported as cause of resistance to that antibody. Notably, the S492R EGFR mutant cell retains the capacity to bind panitumumab. Cetuximab is a chimeric mouse-human immunoglobulin G1 (IgG1), whereas panitumumab is a fully human IgG2. Clinically, this difference translates into distinct toxicity profiles for each antibody, although both antibodies show similar clinical activity 1–6, and medical oncologists generally consider these two agents to be equivalent therapies. However, the present study reveals an opposite preclinical and clinical response to the two antibodies in the presence of the acquired S492R EGFR mutation. This mutation may provide a molecular explanation for the clinical benefit of panitumumab in a subset of subjects with mCRC who do not respond to treatment with cetuximab7, and these findings have substantial immediate clinical implications for persons with mCRC. The specificity of the S492R mutation is expected to facilitate reliable testing to guide the clinical use of panitumumab after cetuximab failure and justifies prospective independent validation of the S492R EGFR mutation. It will

VOLUME 18 | NUMBER 2 | FEBRUARY 2012 nature medicine

B r i e f c o m m u n i c at i o n s

DiFi

b

DCR10

A A T T A T A A G C A A C A G A G G T G A A A A C. A A T T A T A A G M A A C A G A G G T G A A A A C

e EGFR

Counts

a

Ser492 Cetuximab 1476C>A, S492R mutation

EGFR

Cetuximab Panitumumab Ctr hulgG

WT-ECD-Fc /Panitumumab WT-ECD-Fc /Cetuximab S492R-ECD-Fc/Panitumumab S492R-ECD-Fc/Cetuximab

1.5 1.0 0.5 0

0.1 1 10 Ab concentration (ng ml–1)

100

1.0 0.8 0.6 0.4 0.2 0

0

104

S492R EGFR

104

200 160 120 80 40 0 100

S492R EGFR

104

200 160 120 80 40 0 100

WT EGFR 61.81%

101 102 103 PE intensity

9.87%


104

200 160 120 80 40 0 100

EMPTY 8.15% 101 102 103 PE intensity

104

0.01 0.1 1 10 100 1,000 Ab concentration (ng ml–1)

200 160 120 80 40 0 100

WT EGFR 65.83%


63.79%


104

g

S492R-ECD-Fc

1.0

Absorbance 450/620 nm




2.0


Cetuximab Panitumumab Ctr hulgG

WT-ECD-Fc

1.2

2.5

0

–

S492R EGFR Ctx Pnm + – + – +

P-EGFR

EGFR Tubulin

f

WT EGFR Ctx Pnm + – + – +

EGF –1 (10 ng ml ) –

6.68%

200 160 120 80 40 0 100

Pnm (10 µg ml )

Counts

L

EM PT AB Y E_ EG pB FR AB W E_ T EG FR S4 92 pB

O

E_

TR N

AB

O

pB

C

d

EMPTY

–1

R

Wild type

c

Ctx (10 µg ml–1) 200 160 120 80 40 0 100

0.8 0.6 0.4 0.2 0

5 cm

0

0.01 0.1 1 10 100 1,000 Ab concentration (ng ml–1)

Before panitumumab

5 cm After two cycles of panitumumab

Figure 2 Cetuximab-resistant cells harbor a missense mutation (S492R) within the extracellular domain of EGFR. (a) Nucleotide sequence of the EGFR gene in DiFi cells and DCR clones. A heterozygous mutation resulting in the substitution of a serine with an arginine at the position corresponding to amino acid 492 in the extracellular domain of EGFR is observed in DCR clones (arrows). M, mutation. (b) Structural modeling of the interaction between EGFR domain III and cetuximab, confirming the position of EGFR Ser492 at the interface. EGF is shown in purple. (c) Vector (p_BABE) containing the coding sequence of wild-type EGFR (pBABE_EGFRWT) or S492R mutant EGFR (pBABE_EGFRS492R) were retrovirally infected into NIH 3T3 cells to overexpress WT EGFR or S492R EGFR proteins, respectively. NIH 3T3 cells infected with an empty vector (pBABE_EMPTY) were used as control. NIH 3T3 cells were used as negative control (CONTROL). Lysates were analyzed by immunoblotting, which revealed similar expression of EGFR in WT EGFR and S492R EGFR mutant cells. (d) Differing abilities of cetuximab and panitumumab to inhibit the phosphorylation of EGFR after stimulation with EGF in S492R mutant cells. NIH 3T3 cells expressing WT EGFR and S492R mutant EGFR were cultured in the presence of cetuximab or panitumumab for 2 h and stimulated with EGF. Immunoblotting was performed using antibodies to the indicated proteins. (e) Although cetuximab and panitumumab were able to interact with 60% of the cells expressing WT EGFR, only panitumumab was able to bind to cells expressing the S492R EGFR mutation. WT EGFR and S492R EGFR mutant cells were incubated with cetuximab or panitumumab, and antibody binding was analyzed by flow cytometry using a secondary antibody to human IgG conjugated with phycoerythrin (PE). NIH 3T3 cells expressing the empty vector were used as a negative control (EMPTY). The percentage of cells binding to the antibody are shown in the graph. (f) Comparison of the abilities of cetuximab and panitumumab to interact with WT EGFR and the S492R EGFR mutant in vitro, as determined by a direct binding assay (top) or a competitive binding assay (middle and bottom). Ab, antibody; ECD, extracellular domain; Ctr huIgG, control human IgG. Fc, constant fraction. (g) Computed tomographic scan of a target metastatic lesion in the liver (segment VI) from the same subject before beginning panitumumab therapy (left) and after 2 months of treatment with panitumumab (right). The arrows highlight the large volume of the lesion before panitumumab therapy and its decrease after treatment with panitumumab (60% decrease).

also be of interest to determine whether this mutation contributes to the acquisition of cetuximab resistance in other tumor types. Note: Supplementary information is available on the Nature Medicine website. Acknowledgments This work was supported by Instituto de Salud Carlos III (ISCIII) -Subdirección General de Evaluación y Fomento de la Investigación (PS09/01491, PI08/0211 and PI09/1285) and Plan Nacional (PN) de Investigación Científica, Desarrollo e Innovación Tecnológica (I+D+I), iniciativa Ingenio 2010, programa Consolider and ISCIII/FEDER (RD06/0020/0109); PN de I+D+I 2008-20011 and DIUE Generalitat de Catalunya (2009 SGR 321) grants. We thank the Fundació Privada Cellex (Barcelona) for a generous donation to the Medical Oncology Service, Hospital del Mar. We thank the Tumor Bank of Pathology of Hospital del Mar (MARBiobanc), which is supported by FEDER (RD09/0076/00036), and Xarxa de Bancs de Tumors de Catalunya, which is sponsored by PDO (XBTC). We thank L. Roth and R. Cook for assistance with the EGFR binding studies, J. Guillory and J. Stinson for DNA sequencing, and J. Yélamos for key discussions. AUTHOR CONTRIBUTIONS C.M. designed the overall project. C.M., A.D., A.R., J.S., F.B. and J.A. designed experiments and analyzed data. A.D. performed cell-line and molecular experiments. B.B. and S.P. performed sequencing. M.C. prepared mutant proteins.

nature medicine VOLUME 18 | NUMBER 2 | FEBRUARY 2012

M.S. performed the fluorescence in situ hybridization analysis. H.H. and A.M. performed the deep sequencing. S.M., S.P.T. and S. Seshagiri carried out the sequencing experiments and recombinant protein assays. M.I. collected and evaluated the tumor samples. S. Serrano supervised the tumor assays. C.M., M.G. and J.B. collected clinical information on the subjects. C.M., J.S., F.B. and J.A. wrote the manuscript. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Published online at http://www.nature.com/naturemedicine/. Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html.

1. 2. 3. 4. 5. 6. 7.

Amado, R.G. et al. J. Clin. Oncol. 26, 1626–1634 (2008). Karapetis, C.S. et al. N. Engl. J. Med. 359, 1757–1765 (2008). Van Cutsem, E. et al. N. Engl. J. Med. 360, 1408–1417 (2009). Bokemeyer, C. et al. J. Clin. Oncol. 27, 663–671 (2009). Douillard, J.Y. et al. J. Clin. Oncol. 28, 4697–4705 (2010). Peeters, M. et al. J. Clin. Oncol. 28, 4706–4713 (2010). Saif, M.W., Kaley, K., Chu, E. & Copur, M.S. Clin. Colorectal Cancer 9, 315–318 (2010).

223

co r r i g e n da

Corrigendum: Identification of a mutation in the extracellular domain of the Epidermal Growth Factor Receptor conferring cetuximab resistance in colorectal cancer Clara Montagut, Alba Dalmases, Beatriz Bellosillo, Marta Crespo, Silvia Pairet, Mar Iglesias, Marta Salido, Manuel Gallen, Scot Marsters, Siao Ping Tsai, André Minoche, Seshagiri Somasekar, Sergi Serrano, Heinz Himmelbauer, Joaquim Bellmunt, Ana Rovira, Jeff Settleman, Francesc Bosch & Joan Albanell Nat. Med. 18, 221–223 (2012); published online 22 January 2012; corrected after print 6 July 2012

npg


In the version of this article initially published, due to an oversight by the authors, the first and last names of one the authors, Somasekar Seshagiri, were incorrectly transposed. The error has been corrected in the HTML and PDF versions of the article.