Characterization of the Novel and Speciﬁc PI3Ka Inhibitor ... · Stephane Ferretti1, Youzhen...

Small Molecule Therapeutics

Characterization of the Novel and Specific PI3Ka InhibitorNVP-BYL719 and Development of the Patient StratificationStrategy for Clinical Trials

Christine Fritsch1, Alan Huang5, Christian Chatenay-Rivauday1, Christian Schnell1, Anupama Reddy5,Manway Liu5, Audrey Kauffmann1, Daniel Guthy1, Dirk Erdmann1, Alain De Pover1, Pascal Furet2, Hui Gao1,Stephane Ferretti1, Youzhen Wang7, Joerg Trappe3, Saskia M. Brachmann1, Sauveur-Michel Maira1,Christopher Wilson6, Markus Boehm4, Carlos Garcia-Echeverria1, Patrick Chene1, Marion Wiesmann1,Robert Cozens1, Joseph Lehar5, Robert Schlegel5, Giorgio Caravatti2, Francesco Hofmann1, andWilliam R. Sellers1

AbstractSomatic PIK3CA mutations are frequently found in solid tumors, raising the hypothesis that selective

inhibition of PI3Kamay have robust efficacy in PIK3CA-mutant cancers while sparing patients the side-effects

associated with broader inhibition of the class I phosphoinositide 3-kinase (PI3K) family. Here, we report the

biologic properties of the 2-aminothiazole derivative NVP-BYL719, a selective inhibitor of PI3Ka and its most

common oncogenic mutant forms. The compound selectivity combined with excellent drug-like properties

translates to dose- and time-dependent inhibition of PI3Ka signaling in vivo, resulting in robust therapeutic

efficacy and tolerability in PIK3CA-dependent tumors. Novel targeted therapeutics such as NVP-BYL719,

designed to modulate aberrant functions elicited by cancer-specific genetic alterations uponwhich the disease

depends, require well-defined patient stratification strategies in order to maximize their therapeutic impact

and benefit for the patients. Here, we also describe the application of the Cancer Cell Line Encyclopedia as a

preclinical platform to refine the patient stratification strategy for NVP-BYL719 and found that PIK3CA

mutation was the foremost positive predictor of sensitivity while revealing additional positive and negative

associations such as PIK3CA amplification and PTENmutation, respectively. These patient selection determi-

nantsarebeingassayed in theongoingNVP-BYL719clinical trials.MolCancerTher; 13(5); 1117–29.�2014AACR.

IntroductionPhosphoinositide-3 kinases (PI3K) are widely exp-

ressed lipid kinases that function as key signal transdu-

cers downstream of cell-surface receptors and key regu-lators of cell metabolism and survival (1, 2).

The PI3K lipid kinase family comprises 8 enzymesorganized into 3 classes (I–III) based on sequence homol-ogy. Class I PI3Ks are further divided into 3 class IAenzyme isoforms, PI3Ka, PI3Kb, and PI3Kd, whereasPI3Kg constitutes class IB. The PI3Ka and PI3Kb isoformsare ubiquitously expressed and several studies suggestthat PI3Ka is thepredominant catalytic isoformengaged inglucose homeostasis regulation (3, 4) and in vasculogen-esis (5), whereas PI3Kb plays a secondary role in insulinsignaling. In contrast, the PI3Kd and PI3Kg isoforms areprimarily expressed in leukocytes and play importantroles in immune responses and inflammation (6).

The PI3K/mTOR pathway is a central oncogenic path-way deregulated in cancer. Aberrant induction of PI3Kpathwayactivity is linked toupstreamgenetic alterations inreceptor tyrosine kinases (RTK), including amplification ofERBB2, through loss-of-function mutations in the tumorsuppressor genes PTEN and TSC1/2, as well as amplifica-tion and mutations in PIK3CA, the gene encoding PI3Ka,among others (7). Specifically, somatic PIK3CA missensemutations were found in a number of common solidtumors, including �25% of breast cancer (8, 9), making

Authors' Affiliations: 1Novartis Institutes for BioMedical Research, DiseaseArea Oncology; 2Novartis Institutes for BioMedical Research, Global Discov-ery Chemistry; 3Novartis Institutes for BioMedical Research, Center forProteomic Chemistry; 4Novartis Pharma AG, Oncology Translational Medi-cine, Basel, Switzerland; 5Novartis Pharma AG,Oncology Translational Med-icine; 6Novartis Institutes for BioMedical Research, Developmental andMolecular Pathways, Cambridge, Massachusetts; and 7Novartis Institutesfor BioMedical Research, Developmental andMolecular Pathways, Shangai,China

Note: Supplementary data for this article are available at Molecular CancerTherapeutics Online (http://mct.aacrjournals.org/).

A.Huang, C.Chatenay-Rivauday, andC. Schnell equally contributed to thiswork.

Current address for C. Garcia-Echeverria: Oncology Drug Discovery andPreclinical Research, Sanofi-Aventis, 94403 Vitry-sur-Seine, France.

Corresponding Author: Christine Fritsch, Novartis Institutes for BioMed-ical Research, Novartis Pharma AG, Disease Area Oncology, CH-4002Basel, Switzerland. Phone: 41616967697; Fax: 41616966381; E-mail:[email protected]

doi: 10.1158/1535-7163.MCT-13-0865

�2014 American Association for Cancer Research.

MolecularCancer

Therapeutics

www.aacrjournals.org 1117

on February 10, 2021. © 2014 American Association for Cancer Research. mct.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2014; DOI: 10.1158/1535-7163.MCT-13-0865

http://mct.aacrjournals.org/

PIK3CA one of the most commonly mutated oncogenes inhuman cancers. Although PIK3CA mutations can bedetected across the entire coding sequence of the gene,80% of the mutations are found in 3 major hotspot clustersin the helical (E542K, E545K) and kinase domains(H1047R).Eachof thesemutations leads togain-of-functionactivation of PI3Ka manifested by increased lipid kinaseactivity, growth-factor independent activation of Akt sig-naling, cellular transformation, and the generation oftumors in a diverse array of preclinical models (10–14).

The strong evidence underscoring the oncogenic natureand the high frequency of PIK3CA mutations raise thehypothesis that selective inhibition of PI3Ka may haverobust antitumor efficacy in the PIK3CA-mutant cancerpopulation. In addition, a selective agent may offer theopportunity to spare patients the side-effects associatedwith broader inhibition of the class I PI3K family. Basedonthis notion, we sought to develop a PI3Ka-isoform selec-tive inhibitor to allow the testing of this therapeutichypothesis in man.

Here we report the biologic properties of the 2-ami-nothiazole derivative NVP-BYL719. The compound is aselective PI3Ka inhibitor equipotent against the wild typeand the most common somatic mutations of PI3Ka. NVP-BYL719 has excellent drug-like properties and in vivoadministration of NVP-BYL719 results in significantdose-dependent antitumor efficacy in mice bearingPIK3CA-dependent tumor xenograft models as well as animproved safetyprofilewith respect to glucosemetabolismwhen compared with pan-PI3K inhibition. To refine thedesign of clinical trials and identify potential predictivebiomarkers forpatient selection,NVP-BYL719wasprofiledacross a large panel of cancer cell lines referred as theCancerCell LineEncyclopedia (CCLE; ref. 15). The analysisof the underlying genetic aberrations drivingNVP-BYL719sensitivity revealed PIK3CAmutation as the foremost pos-itive predictor of sensitivity as well as additional positiveand negative associations such as PIK3CA amplificationand PTENmutation, respectively. NVP-BYL719 predictivefeatures were found to be relevant for predicting responsein vivo in diverse cancer cell line-derived and patient-derived xenograft models tested in mice. Using a novelanalytical approach to define the selectivity index of smallmolecule inhibitors across theCCLE,NVP-BYL719 showedmarkedly selective efficacy in PIK3CAmutants when com-pared with wild-type cell lines and when compared withpan-PI3K inhibitors. These data provided the rationale forconducting the initial clinical testing ofNVP-BYL719 exclu-sively in preselected patients with advanced solid malig-nancies carrying PIK3CA gene alterations (16).

Materials and MethodsChemical entities

NVP-BYL719 and NVP-BKM120 were synthesized byGlobal Discovery Chemistry Department (NIBR, Novar-tis, Basel, Switzerland). For in vitro studies, 10 mmol/Lstock solutions were prepared in 100% dimethyl sulfox-

ide. For in vivo experiments,NVP-BYL719was formulatedfor oral administration in solution by solving the com-pound in N-methyl pyrrolidone, polyethylene glycol 300,solutol HS15, and water (10%:30%:20%:40%, v/v) or insuspension in 1% (w/v) carboxymethylcellulose (CMC)þ0.5% (w/v) Tween 80 similar to NVP-BKM120.

AntibodiesAntibodies used for Western blot analysis, reverse

phase protein array (RPPA) and in cell Western assayswere p-Akt (S473) (#9271), total Akt (#9272), p-RPS6(#2211), total RPS6 (#2317), p-p53 (S15) (#9284), and p-ATM (S1981) (#4526) fromCell Signaling and anti-tubulinb from Thermo Fisher.

Cell lines and in vitro compound profilingMechanistic models. To evaluate the isoform-specific

potency of NVP-BYL719 in a cell-based system, an N-terminally myristoylated form of each PI3K class IAisoform was expressed in Rat1 fibroblasts as described inref. 17. The retroviral expression plasmid pBabePurocontaining human p110a, p110b, and p110d with an N-terminal myristoylation (myr) signal followed by an HA-tag were generated. Successfully infected Rat1 cells wereselected in medium containing 4 mg/mL of puromycin,expanded and characterized for expression of the p110isoforms (in 2006). Transgenic expression of the myris-toylated protein was confirmed by increased levels ofphosphorylated Akt. The TSC1�/�-null MEFs mechanis-tic model for mTORC1 constitutive activation has beenobtained from Dr. D. Kwiatkowski in 2007 (Brigham andWomen’s Hospital, Boston, MA).

Human tumor cell lines. The A549 (American TypeCulture Collection, ATCC) and the U2OS (ATCC) wereauthenticated by single-nucleotide polymorphism (SNP)analysis in 2006. The biochemical kinase assays andRPPAassays were conducted as previously described (17, 18).

High-throughput pharmacologic cell line profilingCell lines obtained from ATCC, DSMZ, and HSSRB are

cultured in RPMI or Dulbecco’s modified Eagle mediumplus 10% FBS (Invitrogen) at 37�C5%CO2 using automatedprocessing. Cell line identities were confirmed using a 48variant SNP panel comparing the previous cell line tests asmentioned in ref. 19. A detailed description of the cell linesandof thehigh-throughputcellviabilityassayscanbe foundin ref. 15, see also http://www.broadinstitute.org/ccle.

In vivo studies in mice and ratsAll animal studies were conducted in accordance with

protocols approvedby theNovartis Institutes for BioMed-ical Research Animal Care and Use Committee.

Cell lines–derived tumor models. All in life experi-mentation and efficacy studies were conducted asdescribed previously (17). Tumor xenografts were grownsubcutaneously or orthotopically in nude mice (Harlan,Germany) or nude Rowett rats (Hsd: RH-Fox1rnu, Har-lan, TheNetherlands) by injection of 3� 106 to 1� 107 cells

Fritsch et al.

Mol Cancer Ther; 13(5) May 2014 Molecular Cancer Therapeutics1118




or implantation of tumor fragments of approximately 50mg. Tumor-bearing animalsmicewere treatedwith eithervehicle control, NVP-BYL719, or NVP-BKM120 (p.o.,every day) at the doses indicated.Patient-derived tumor models. Patient-derived xeno-

graft (PDX) models were established by implanting surgi-cal tumor tissues from treatment-na€�ve cancer patients intonude mice. All samples were anonymized and obtainedwith informed consent and under the approval of theinstitutional review boards of the tissue providers andNovartis. All PDX models were histologically character-ized and external diagnosis was independently confirmedby in-house pathologists and were genetically profiledusing various technology platforms after serial passagesin mice. PIK3CA mutation was determined by both RNAand DNA deep sequencing technologies and PIK3CAamplification was determined by SNP array 6.0. For effi-cacy studies, tumor-bearing animals were enrolled whensubcutaneously implanted tumors reached about 200mm3

and treated with NVP-BYL719 at 50 mg/kg daily. Theresponse is reportedaspercentage change in tumorvolumeat last day of treatment relative to day 0 (start of treatment).

Automated calls of NVP-BYL719 sensitivityStarting from the vector of responses Amax or EC50, we

considered the shape of the rank-ordered plot of responsevalues (for Amax, log-transformed EC50) in order to assigncell lines into responder, intermediate, and nonresponderclasses using a combination of EC50 and Amax cutoffsdefined with this method.

Genetic and genomic characterization of cell linesAdetailed description of DNA copy number andmuta-

tion data can be found in ref. 15, see also http://www.broadinstitute.org/ccle. In addition, only dominant neg-ative or functional mutations identified through manualliterature curation were considered in the analyses. Wildtypes were those with no identified sequence variants.

NVP-BYL719 compound differential selectivityanalysisTo assess NVP-BYL719 selectivity on PIK3CAmutants,

an internal panel of drug compounds was profiled across

both the PIK3CAmutant andwild-type cell lines. For eachcompound, the response profiles (Amax values) of the celllines were converted to a selectivity score by multiplyingthe z-score � j Amax j to the compound were z-trans-formed. The z-scores of the mutant responses were thencompared against those of the wild-type lines by takingthe difference between the 2 and then back-converting to aP value. The P values across the different compoundswere then corrected formultiple hypotheses testing usingthe Benjamini–Hochberg false discovery rate (FDR)meth-od (20).

ResultsNVP-BYL719 potently and selectively inhibits PI3Kain vitro

PI3Ka, b, d, and g enzymes share significant amino acidresidue homology with particularly high conservation inthe catalytic kinase domain. The 2-aminothiazole scaffoldwas selected as a starting point for the development ofpotent and selective PI3K inhibitors based on its bindingmode, indicating the potential to use substituents at theamino group to develop interactions with nonconservedamino acids at the ATP pocket entrance (21). Consequent-ly, systematic modification of key moieties and optimi-zation of the drug-like properties led to the identificationof NVP-BYL719 (18).

As previously described in ref. 18, in biochemical assaysNVP-BYL719 inhibits wild-type PI3Ka (IC50¼ 4.6 nmol/L)more potently than the PI3Kd (IC50 ¼ 290 nmol/L) andPI3Kg (IC50¼ 250 nmol/L) isoforms and shows significant-ly reduced activity against PI3Kb (IC50 ¼ 1,156 nmol/L).Here, in addition, we show that NVP-BYL719 potentlyinhibits the 2 most common PIK3CA somatic mutations(H1047R, E545K; IC50�4 nmol/L). The compound alsolacked activity against the class III family member Vps34and the related class IV PIKKprotein kinasesmTOR,DNA-PK, and ATR and was significantly less potent against thedistinct lipid kinase PIK4b (Table 1).

The kinase selectivity profile of NVP-BYL719 was fur-ther examined in in vitro kinase assay panels. Among allthe kinases tested (excluding class I PI3K and PI4Kb) theirrespective IC50 or Kd values were at least 50-fold higher

Table 1. Effects of NVP-BYL719 against PI3K lipid or protein kinases

Enzyme PI3KaE545Kmutant

H1047Rmutant PI3Kb PI3Kd PI3Kg PI4Kb Vps34 mTOR DNA-PK ATR

Methods KG KG KG KG Adapta Adapta KG KG TR-FRET Caliper a-screen

BYL719IC50 nmol/L

4.6 � 0.4 4.0 � 0.6 4.8 � 0.4 1,156 � 77 290 � 180 250 � 140 581 � 42 >9,100 >9,100 >9,100 >15,000

NOTE: In vitro assays were conducted with the indicated recombinant PI3K lipid or protein kinases in the presence of increasingconcentrations of NVP-BYL719 as described in ref. 17. The concentration producing 50% inhibition of the enzymatic activity (IC50) innmol/L as determined in multiple experiments (n ¼ 2–20) is shown as an average � SD.Abbreviation: KG, KinaseGlo.

NVP-BYL719 PI3Ka Inhibitor and Predictive Response Modeling

www.aacrjournals.org Mol Cancer Ther; 13(5) May 2014 1119




A

CB

D E

0

1,000

900

800

700

600

500

400

300

200

100

0

1,600

1,400

1,200

1,000

800

600

400

200

0

Pla

sm

a c

on

cen

trati

on

(µ

mo

l/L

)

0

10

20

30

40

50

241612842

0 1 2 3 4 5 6 7 8 9

Tu

mo

r le

vels

(n

mo

l/g

)T

um

or

vo

lum

e (

mm

3)

Tu

mo

r vo

lum

e (

mm

3)

Time after treatment (h)

Time of treatment (days)

0

–1.0 –0.8 –0.6 –0.4

Regression (%)

30% regression

Tumor stasis

T/C (%)

–0.2 0.0 0.2 0.4 0.6 0.8 1.0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Time of treatment (days)

Time after treatment (h) Time after treatment (h)

**

*

12.5 mg/kg

0 2416128421

* *

*

25 mg/kg

0 % A

vera

ge v

eh

icle

co

ntr

ol

(S473P-A

kt)

0

20

40

60

80

100

120

140

2416128421

* * * *

*

50 mg/kg

BYL719 tumor levels (mmol/g)

0.0

04

0.0

06

0.0

20.0

40.0

6

0.2

0.6 2 6 20

40

60

200

0.0

1

0.1 1 10

100

% A

vera

ge v

eh

icle

co

ntr

ol

Fra

cti

on

of

tim

e >

IC

80 p

-Akt

(%)

(S473P-A

kt)

0

10

20

30

40

50

60

70

80

90

100

0

10

20

30

40

50

60

70

80

90

100

IC80 in vivo

IC50 in vivo

Figure 1. PK/PD/efficacy relationship of NVP-BYL719 in PI3Ka-dependent tumor mouse models in vivo. A, female athymic mice bearing subcutaneousxenotransplants of Rat1-myr-p110a tumors were treated with single or repeated doses of 12.5, 25, and 50 mg/kg, p.o. At the indicated time points, thegroups of mice (n ¼ 2–15 per group) were sacrificed and blood and tissues were collected. Each tumor tissue was flash frozen, then pulverized andanalyzed by RPPA to determine S473P-Akt levels and, in parallel, the concentration of NVP-BYL719 was quantified. PK/PD relationship for eachdose level was established by comparing % of inhibition of S473P-Akt levels versus control with NVP-BYL719 concentration at each time point indicated(�, P < 0.05). Results are presented as mean � SEM. B, female athymic mice bearing Rat1-myr-p110a subcutaneous xenografts were treated withvehicle (red dots) or NVP-BYL719 at 12.5 (green triangles) mg/kg/day p.o., 25 (red squares) mg/kg/day p.o., and 50 (orange diamonds) mg/kg/day p.o.,every day (n ¼ 8 per group), respectively. (Continued on following page.)

Fritsch et al.





when compared with PI3Ka (Supplementary Fig. 1, Sup-plementary Tables 1–4).To determine the potency and selectivity of NVP-

BYL719 in cellular assay systems, Rat1 cells transformedusing the activated forms of PI3Ka, PI3Kb, or PI3Kd weretested and RPPAs were used to quantify the phosphory-lation of Akt (S473) as a marker of PI3K pathway activity(17). As described in ref. 18, NVP-BYL719 potently inhib-itedAkt phosphorylation in cells transformedwith PI3Ka(IC50 ¼ 74 � 15 nmol/L) and showed significant reducedinhibitory activity in PI3Kb or PI3Kd isoforms trans-formed cells (�15-fold compared with PI3Ka). Here, wereport NVP-BYL719 full dose–response curves as well asits IC80 values on S473P-Akt in Rat1 cells (SupplementaryFig. S2). In addition, treatment of TSC1-null MEF cellswith NVP-BYL719 was not associated with a reduction inphosphorylation of RPS6 (S235/236) when comparedwith the positive control RAD001 (IC50 value < 0.5nmol/L), suggesting that NVP-BYL719 does not inhibitmTORC1 (Supplementary Fig. S3A and S3B). Similarly,NVP-BYL719 does not seem to interfere with the PIKKsinvolved in DNA-damage repair (ATM and ATR) pro-cesses as determined in ATM- and ATR-dependent assaysystems (Supplementary Fig. S3C and S3D). Together,these data strongly support the notion that NVP-BYL719has the relevant in vitro properties of a selective PI3Kainhibitor.

NVP-BYL719 shows robust PK/PD/Efficacyrelationship in PI3Ka-driven tumorsTo examine NVP-BYL719 ability to inhibit the PI3K/

Akt pathway in a PI3Ka-dependent in vivo model,its pharmacokinetic/pharmacodynamic (PK/PD) rela-tionship was assessed in a Rat1-myr-p110a mechanistictumor-bearingmousemodel. Each female athymicmousereceived single or repeated doses of NVP-BYL719 (12.5,25, or 50 mg/kg, p.o.) and plasma and tumors sampleswere collected for PK and PD analysis at different timepoints. Here NVP-BYL719 treatment was associated withdose and time-dependent inhibition of the PI3K/Aktpathway, which notably paralleled time-dependent drugexposure in tumor and plasma (Fig. 1A).To determine whether dose- and time-dependent path-

way inhibition was linked to antitumor activity, Rat1-myr-p110a tumor-bearing nude mice were treated orallyeverydaywith the compound for up to 8 consecutive days(Fig. 1B). Treatments of 12.5, 25, and 50 mg/kg were welltolerated and resulted in a dose-dependent and statisti-cally significant antitumor effect with a T/C of 14.1% and

regressions of 9.6% and 65.2%, respectively. To assess therelative PI3K selectivity in vivo, we further tested NVP-BYL719 in a corresponding Rat1-myr-p110dmodel. NVP-BYL719,when tested at the optimal dose of 50mg/kgp.o.,every day, showed only amodest antitumor effect (T/C of30%; Fig. 1C).

We next sought to better understand the degree ofPI3Ka inhibition that is required for antitumor efficacy.To this end, we first determined the tumor concentrationsgiving 50% (in vivo IC50) and 80% (in vivo IC80) S473P-Aktinhibition (0.4 and 4 mmol/L, respectively) by measuringthe extent of Akt phosphorylation using RPPA and thespecific tumor drug concentration in matched samplesfrom multiple animals and at multiple time points(Fig. 1D). Interestingly,when corrected for plasmaproteinbinding of NVP-BYL719 in mouse (PPB ¼ 91.2%), thein vivo IC50 (35 nmol/L) and IC80 (352 nmol/L) valuesroughly approximate the in vitro cellular IC50 and IC80 of74 and 301 nmol/L, respectively. We next sought todetermine the relationship between exposure, as mea-sured by time over the in vivo IC80, and antitumor efficacy.Here, we found a nearly linear relationship between theantitumor efficacymagnitude and duration of drug expo-sure over the IC80 (R

2 ¼ 0.80, P < 0.001, n ¼ 11; Fig. 1E).From this relationship it seems that 80% inhibition of Aktphosphorylation for at least 29% of the dosing interval isrequired forNVP-BYL719 to induce tumor stasis, and thatthis level of pathway inhibition must be sustained for atleast 45% of the dosing interval to produce 30% tumorregression in the Rat1-myr-p110a tumor-bearing nudemice. In contrast, in the Rat1-myr-p110d tumor-bearingnude mice NVP-BYL719 exposure levels did not achieve80% inhibition of Akt phosphorylation (in vivo IC80 ¼ 29mmol/L; corrected for NVP-BYL719 plasma protein bind-ing in mouse IC80¼ 2,552 mmol/L) most likely explainingthe modest antitumor effect observed and in line withthe modest activity of the compound on p110d. Toexclude the possibility that our finding could be Rat1mouse tumor models specific, NVP-BYL719 was admin-istered in vivo at different doses to nude mice and nuderats bearing a diverse range of cancer cell lines–derivedtumor xenografts. Here as well, we found a nearly linearrelationship between the antitumor efficacy magnitudeand duration of drug exposure over the IC80 (R

2 ¼ 0.77,P < 0.001, n ¼ 27, Supplementary Fig. S4 and Table S5).These data suggest that sustained inhibition of thePI3K/Akt pathway for a fraction of the dosing intervalis required for NVP-BYL719 to produce a robust anti-tumor effect.

(Continued.) Statistics on D tumor volumes and body weights were performed with a one-way ANOVA, post hoc Dunnett and by paired t test on body weightsmeasured at start andat endof treatment respectively (�,P < 0.05). C, female athymicmice bearingRat1-myr-p110d subcutaneous xenograftswere treatedwithvehicle (black dots) or NVP-BYL719 at 12.5 (white triangles) and 50 (orange squares) mg/kg p.o., every day (n ¼ 15 per group). Statistics on D tumorvolumeswereperformedwith aone-wayANOVA,post hocDunnett (�,P<0.05). NVP-BYL719producedadose-dependent andstatistically significant antitumoreffect with a T/C of 70% and 30%, when administered at 12.5 and 50 mg/kg, respectively. D, relationship between tumor tissue concentration and percentS473P-Akt inhibitionmeasuredconcomitantly in theRat1-myr P110a tumors at different timepoints posttreatmentwithNVP-BYL719 at 6.25up to150mg/kgp.o. every day. E, linear correlation observed between tumor growth inhibition (% T/C) or tumor regression and the fraction of time above the in vivo S473P-AktIC80 in the Rat1-myr-P110a tumors (gray dots) and Rat1-myr-p110d tumors (black dots) following NVP-BYL719 treatment (R2 ¼ 0.80; P < 0.001; n ¼ 11).






NVP-BYL719 shows an improved safety profilecompared with pan-class I inhibition

The expected on target side effects of anti-PI3K therapyare insulin resistance and hyperglycemia. To assesswhether NVP-BYL719 perturbs glucose homeostasis,plasma insulin and glucose blood levels were measuredand compared with plasma drug concentrations inmatched samples from multiple animals and at multipletime points. The data here revealed that insulin plasmalevels increasedproportionallywithNVP-BYL719plasmaconcentrations, whereas blood glucose levels were main-tained close to normal up to 20 mmol/L of NVP-BYL719(Fig. 2A and B). However, above 20 mmol/L, we observeda compound concentration-dependent glucose increasewhich led to hyperglycemia despite insulin plasma levelelevation. Thus, we defined 20 mmol/L as NVP-BYL719-related hyperglycemic threshold in mice.

We next hypothesized that the body weight loss weobserved following compound treatment might correlatewith the severity of hyperglycemia. In keeping with thisnotion, we observed a nearly linear relationship betweenthe bodyweight loss magnitude and duration of exposureaboveNVP-BYL719 hyperglycemia threshold (20 mmol/L;R2¼ 0.90;P < 0.001; n¼ 11; Fig. 2C). From this relationship,it seems that the compound exposure levels should besustained for no more than 36% of the dosing intervalabove the hyperglycemia cutoff to maintain body weightloss below 5% in mice.

To determine the therapeutic index of NVP-BYL719with respect to glucose homeostasis, we next comparedthe dose estimated to produce 30% tumor regression (20mg/kg) based on the duration of exposure above the IC80

threshold for S473P-Akt inhibition with the dose estimat-ed to induce 5%bodyweight loss (65mg/kg) based on theduration of exposure above the hyperglycemia threshold.We estimated a therapeutic index value of 3.25 for NVP-BYL719 (Fig. 2D). A similar analysis was conducted withthe pan-PI3K inhibitor NVP-BKM120 (17), leading to atherapeutic index value of 1.1 (Fig. 2E). Thesedata suggestthat PI3Ka-selective agents such as NVP-BYL719 mayimpact physiologic pathways such as glucosemetabolismdifferentially from a broader inhibition of the class I PI3Kfamily.

PIK3CA mutation is a positive predictor and isranked first mutated feature associated with NVP-BYL719 sensitivity

To take an unbiased approach to the discovery andrefinement of predictive biomarkers, NVP-BYL719 sen-sitivity of 474 cancer cell lines from the CCLE wasexamined using an 8-point dose–response matrix fromwhich the EC50 (inflection point) and Amax (maximumobserved response) were determined. To assign celllines into responsive and refractory classes in an unbi-ased manner, we used a combination of EC50 and Amax

cutoffs (3.04 mmol/L and �30%, respectively) definedwith the automated calls method as described in Mate-rials and Methods.

Based on those 2 criteria (EC50 and Amax), 122 cell lineswere categorized as responder (Supplementary Table S6),118 were categorized as intermediate and 234 were cat-egorized as nonresponder to NVP-BYL719 treatment(Fig. 3A). We next asked whether any commonly foundgenetic or pathway aberrations were associated withNVP-BYL719 in vitro responses. PIK3CA mutation is thefirst featurewe examined based on the underlying genetichypothesis for developing a selective PI3Ka inhibitor(Fig. 3B). Interestingly, we found that cell lines that carryPIK3CA mutation are more likely to be NVP-BYL719responsive, with 22 responder cell lines out of total of34, or 64% response rate, significantly higher than the 100of 440 found in PIK3CAwild-type group, or 22% responserate. Hence, the PIK3CA mutation feature alone shouldallow for a 3-fold improvement in response rate versusPIK3CA wild type and a 2.5-fold over random (as overallpopulation has a 25% response rate). Strikingly, using anunbiased approach,PIK3CAmutationwas found to be themost significant mutation feature that predicts NVP-BYL719 response among the 25 mutation features(restricted to known functional mutations for all genestested) included in the analysis (Fig. 3C). A close-upexamination of the lineage distribution among PIK3CAmutants indicated that the preferred sensitivity to NVP-BYL719 can be observed in almost all lineages, suggestingthat the genetic status should represent a reliable patientenrollment criterion across indications (SupplementaryTable S6).

PIK3CA amplification and PTEN mutation alsomodulate response to NVP-BYL719

The CCLE profiling results confirmed that PIK3CAmutation status affectedNVP-BYL719 sensitivity, but alsosuggested the importance of additional, modulating fac-tors. Among the PIK3CA mutant cell lines, 6 were non-responders to NVP-BYL719, whereas among the PIK3CAwild-type cell lines, 100 were responders. Hence, PIK3CAmutation status enriches for but does not uniquely explainNVP-BYL719 response. We consequently examined thedegree of association between the genetic status of othergenes to NVP-BYL719 response, including PTEN, KRAS,NRAS, and BRAFmutation, aswell as ERBB2 and PIK3CAamplification (Fig. 4A). In the overall cell lines population,similar to PIK3CAmutation (P value of 7.5� 10�7, FDR of1.8 � 10�5), PIK3CA amplification, and NRAS mutationwere identified to be positively associated with NVP-BYL719 response (P value of 0.0017, FDR of 0.109 and Pvalue of 0.011, FDR of 0.147, respectively). ERBB2 ampli-ficationshoweda trend tobeassociatedwithNVP-BYL719response and BRAF mutation was rather associated withnonresponse; however, both features just missed the FDR< 0.25 cutoff. KRAS mutation and PTEN mutation, asindependent genetic features, were close to neutral. Con-sidering that the pivotal feature for patient stratification inthe clinic will be mutant PIK3CA, we performed a moredetailed examination of predictive features across thePIK3CAmutant cell linepopulationversuswild type,with

Fritsch et al.





B

C

A

ED

Hyperglycemia threshold

BYL719 plasma levels (µmol/L)

0.0

03

0.0

06

0.0

30.0

6

0.3

0.6 3 6 30

60

0.0

01

0.0

1

0.1 1 10

100

Pla

sm

a in

su

lin

levels

(n

g/m

L)

0.01

0.03

0.1

0.3

1

3

10

30

100

300

Hyperglycemia threshold

BYL719 plasma levels (µmol/L)

0.0

03

0.0

06

0.0

30.0

6

0.3

0.6 3 6 30

60

0.0

01

0.0

1

0.1 1 10

100

Blo

od

glu

co

se

le

ve

ls (

mm

ol/

L)

4

5

6789

10

20

30

40

Bo

dy

we

igh

t d

ec

rea

se

(%

)

0 5 10 15 20 25 30 35 40 45 50 55 60

Fraction of time > hyperglycemia threshold (%)

1

2

3

4

5

6

7

8

9

0

10

30% tumor regression

5% BW loss

Dose BKM120 (mg/kg p.o.)

0 10 20 30 40 50 60 70 80

Fra

cti

on

of

tim

e >

th

res

ho

lds

(%

)

0

10

20

30

40

50

60

70

80

90

100

30% tumor regression

5% BW loss

Dose BYL719 (mg/kg p.o.)

0 10 20 30 40 50 60 70 80

Fra

cti

on

of

tim

e >

th

res

ho

lds

(%

)

0

10

20

30

40

50

60

70

80

90

100

Figure 2. Determination of NVP-BYL719 safety profile compared with pan-class I PI3K inhibitors. A, relationship between plasma insulin levelsand plasma NVP-BYL719 concentrations measured in the same probe following NVP-BYL719 treatment. B, relationship between blood glucose levelsand plasma NVP-BYL719 concentrations measured in the same probe following NVP-BYL719 treatment. The in vivo hyperglycemia threshold forNVP-BYL719 (20 mmol/L) is represented by a dotted line. C, linear correlation observed between the fraction of time over plasma hyperglycemiathreshold (20 mmol/L) between two consecutive dosings and body weight loss in the Rat1-myr-P110a tumors (gray dots) and Rat1-myr-p110d tumors(black dots); R2 = 0.90, P < 0.001, n = 11. D, efficacy curve (gray dots) as determined by the fraction of time above the IC80 threshold for S473P-Akt andtolerability curve (orange dots) as determined by the duration of exposure above NVP-BYL719 hyperglycemia threshold (20 mmol/L) in mice treatedorally every day with increasing doses of NVP-BYL719 up to 75 mg/kg. E, efficacy curve (gray dots) as determined by the fraction of time above theIC80 threshold for S473P-Akt and tolerability curve (orange dots) as determined by the duration of exposure above NVP-BKM120 hyperglycemiathreshold (6 mmol/L) in mice treated orally every day with increasing doses of NVP-BKM120 up to 60 mg/kg.






the aim of further refining the selection and enhancingresponse rate. This analysis indicated thatPTENmutationis associated with nonresponse (P value of 0.022, FDR of0.22) in themutantPIK3CA context only,whereasPIK3CAamplification is positively associated with NVP-BYL719sensitivity in the PIK3CA wild-type population only (Pvalue of 0.0037 andFDRof 0.22). Following these findings,we constructed a hypothesis-based predictor of responseto improve upon the single feature model based onPIK3CA mutation status, by combining PIK3CA amplifi-cation and PTEN mutation as significant predictive fea-tures in the respective settings (Fig. 4B). This combinedgenetic predictor is significantly associated with NVP-BYL719 responders versus nonresponders (P value ¼1.49� 10�7). Cross-validation using bootstrapping showsthat the predictor significantly enriched for responders(positive predictive value¼ 76%).However, this predictorhas a sensitivity of only 21%, which means that it missed�80% of the responders. Future work needs to be focusedon identifying features that predict sensitivity in thisremaining80%of the respondersnot explainedbyPIK3CAmutation or amplification.

Genetic alterations in PIK3CA predict NVP-BYL719in vivo efficacy

Next, NVP-BYL719 was administered in vivo at thedose of 50 mg/kg (every day, p.o.) to mice bearing adiverse range of cancer cell lines–derived tumor xeno-grafts (Fig. 5A and Supplementary Table S5) with dif-ferent genetic backgrounds, including the predictivefeatures of the decision tree described previously. Mostof the tumor models that carried a PIK3CA mutation oramplification responded to NVP-BYL719 (responsedefined as T/C < 20%). In contrast, in most of the tumormodels that carried a PTEN mutation or were PIK3CAwild type, we observed progressive disease. In vivo, thepredictor also significantly enriched for responders(positive predictive value ¼ 89%). These data demon-strate that the NVP-BYL719 predictive features derivedfrom the in vitro profiling and analysis of the CCLEseem relevant for predicting response in vivo (P ¼ 0.01,Fisher test).

Considering that PIK3CA mutation or amplificationmight be the key molecular determinants for NVP-BYL719 patient stratification in the clinic, we next per-formed amolecularly defined prospective trial in PIK3CAmutant and/or amplified PDX models in mice with theaim to test our patient selection strategy in a setting thatbest mimics disease response in patients. Tumor-bearinganimalswere treatedwithNVP-BYL719 at 50mg/kg/dayfor 14 to 16 days. Strikingly, 8 of 9 PDXmodels that carry amutation and/or amplification in PIK3CA responded toNVP-BYL719, leading to a response rate of 88% (Fig. 5Band Supplementary Table S7). The PDX response toNVP-BYL719 can beobserved indifferent lineages (breast, lung,gastric, colorectal cancer), suggesting thatPIK3CA geneticstatus should represent a reliable patient enrollment cri-terion across indications.

*

*

Gene mutation

–L

og

10(P

)

122 118

234

EC50 (mmol/L)

AM

AX

(%

)

C

B

A

Figure 3. PIK3CA mutation is the top positive predictor for NVP-BYL719sensitivity.A,NVP-BYL719sensitivityprofile.ScatterplotshowingAmax (%)by EC50 values expressed in mmol/L of NVP-BYL719 in cell viability assaysassessedon474cancer cell lines.Cell lineswithEC50 lowerorequal to3.04mmol/L and Amax lower or equal to�30%were classified as responders toNVP-BYL719 (orange),whereascell lineswith higher values for bothcutoffswere classified as nonresponders (blue). Cell lines that were neitherresponders nor nonresponders were classified as intermediate (yellow).Stars representPIK3CAmutant cell lines.B,NVP-BYL719sensitivity profileseparated according to PIK3CA mutational status. Responder cell linespopulation is significantly enriched in PIK3CA mutant lines as comparedwith the nonresponder population (P value of 7.5 � 10�7). C, unbiasedenrichment analysis across allmutations that are functionallymutated in>3cell lines.Cell lineswerepartitioned into theNVP-BYL719 responder versusnonresponder group. Subsequently, for each gene in the CCLE that wasdominantnegativeor functionallymutated inat least3cell lines in thisstudy,a two-sidedFisher exact test enrichment analysiswasperformed toassessthe significance of the association between mutation status and NVP-BYL719 response. A total of 25 genes were tested. Significance values(P values) were subsequently corrected for multiple hypotheses testingusing the FDRmethodofBenjamini–Hochberg (20). AnFDRcutoff of<0.25was used to establish significance. Blue bars denote those genes that aremore associated with NVP-BYL719 nonresponders, whereas orange barsdenote those thataremoreassociatedwith responders. *, a0.25FDRcutofffor significance.

Fritsch et al.





PIK3CA mutant cell lines are selectively sensitive toNVP-BYL719The above-mentioned approach was useful in defining

what tumors were responsive to PI3Ka inhibition. Anindependent question is askedwhenone considers,whichtherapeutic modality is most selective and hence likely tohave the best therapeutic index in a specific cancer geno-type.Here, using a novel analytical approach to define theselectivity index of small molecule inhibitors across theCCLE, we compared the selectivity profiles across differ-ent compound treatments (�1,000) encompassing morethan 200mechanisms of actions in PIK3CAmutant versuswild-type cell lines and ranked the compounds based onthe magnitude of their effects in these 2 groups (Fig. 6).NVP-BYL719, together with 3 close analogs, showedmarkedly selective efficacy in PIK3CA mutants when

compared with wild-type cell line populations and whencompared with pan-PI3K inhibitors. Conversely, MEKinhibitors were differentially more selectively effective inPIK3CA wild-type cell lines compared with mutants.

DiscussionThe genes comprising the PI3K pathway are commonly

altered in human cancer and targeting this pathwayrepresents an important area for therapeutic develop-ment. Indeed, many agents targeting diverse nodes in thepathway are currently in clinical trials. However, most ofthese compounds are not selective for PI3Ka and inhibitother PI3K isoforms and/or other downstream nodes.Several reported phase I study results for pan-PI3K inhi-bitors showed that pharmacological inhibition of PI3K inhumans is feasible. Preliminary evidence of antitumor

A

BPIK3CA

PIK3CA PTEN

WT

WT

Norm

Nonresponder Nonresponder

ResponderResponder

Amp

Mut

Mut

*

*

* *

*

–L

og

10(P

)

Feature Feature Feature

Non-

responder

Responder

Non-

responder

226 96

Responder 8 26

Measured

Pre

dic

ted

Sensitivity 21

Specificity 96

PPV (%) 76

NPV (%) 70

Fisher exact P value = 1.49 × 10–7

Overall population PIK3CA mutant only PIK3CA WT onlyFigure 4. PTEN mutation andPIK3CA amplification/copynumber modulate response toNVP-BYL719. A, hypothesis-based sensitivity association—Barplot representing the �log10 ofthe P value for the 7 hypothesisfeatures tested. Blue bars denotethose genes that are moreassociated with NVP-BYL719nonresponders, whereas orangebars denote those that are moreassociated with responders.Association to NVP-BYL719sensitivity was tested bycomparing the responder to thenonresponder cell line populationseither using a Fisher test forcategorical variables (mutation) ora Wilcoxon test for continuousvariables (copy number). A featuretype–specific correction of Pvalues was performed to accountfor FDR. *, a 0.25 FDR cutoff forsignificance. B, combined geneticpredictor for patient stratification.The combination of PIK3CA andPTEN genetic information issignificantly associated withresponders versus nonresponders(P value ¼ 1.49 � 10�7) andimproves on predictions ofPIK3CA mutation alone.Enrichment analysis of theassociation of the predictions withNVP-BYL719 responders versusnonresponders was performedusing a two-sided Fisher test.Model performance [sensitivity,specificity, positive predictivevalue (PPV) and negativepredictive value (NPV)] wasestimated using 2,000bootstrapped subsets eachrandomly selecting half thesamples.






activity in patients with solid tumors has been observedand the associated adverse events indicate on-target tox-icity such as hyperglycemia (22, 23). However, as clinicaltrials with PI3K inhibitors have so far been conducted inun-preselected patients, the patient response rate and itsextent seemmore modest in comparison to other targetedagents such as BRAF inhibitors in BRAF-mutated mela-noma (24) or crizotinib in ALK-translocated tumors (25).These data raise 2 related key questions as reviewed inref. 26: could the efficacy observed in the clinic be limited

by the safety profile of pan-class I isoforms inhibition andcould the identification of the patient populations that arelikely to benefit the most from the treatment lead to morefrequent and pronounced antitumor effects in humanssimilar to what has been reported preclinically?

The discovery of somatic PIK3CAmissense mutationsand their frequency in a number of common solidtumors raise the possibility that PI3Ka-selective inhibi-tors might be safer and specifically efficacious inpreselected PIK3CA-mutant patients compared with

B

AA

ve

rag

e a

nti

tum

or

eff

ec

ts

(50 m

g/k

g p

.o q

d)

Response distribution across models

Time after treatment (days)

147014701470147014701470147014701470

Fra

cti

on

al

tum

or

vo

lum

e c

han

ge

(% v

s. tim

e 0

)

–100

–50

0

50

100

150

200

250

300

PDX1 PDX2 PDX3 PDX4 PDX5 PDX6 PDX7 PDX8 PDX9

Non-responder

Responder

Non-

responder

511

Responder 81

Measured

Pre

dic

ted

Sensitivity 62

Specificity 92

PPV (%) 89

NPV (%) 69

Fisher exact P value = 0.01

Figure 5. A, genetic alterations inPIK3CA predict NVP-BYL719 invivo efficacy. Female athymicmicebearing tumor xenografts (n ¼ 6–8mice per group) as indicated weretreated with NVP-BYL719 at 50mg/kg p.o. every day or vehicle.Statistics onD tumor volumeswereperformedwith aone-wayANOVA,post hoc Dunnett. Antitumorresponse is defined as T/C <20%.Blue bars denote those genes thatare more associated with NVP-BYL719 nonresponders, whereasorange bars denote those that aremore associated with responders.B, PDX models carrying a PIK3CAmutation and/or amplificationwereestablished by implanting surgicaltumor tissues from treatment-naïve cancer patients into athymicmice. Tumor-bearing mice wereuntreated (n ¼ 1 animal per PDXmodels, represented by whitedots) or treated with NVP-BYL719at 50 mg/kg, every day, p.o. (n ¼ 1animal per PDX models,represented by black dots) for 14to 16 days. The response ispresented in percentage tumorvolume change relative to day 0.

Fritsch et al.





pan-class I inhibitors. In this study, we reported thatNVP-BYL719 potently inhibits the PI3Ka isoform andits 2 most common oncogenic mutants and is selectiveagainst the other class I PI3K isoforms and a wide rangeof other kinases. The compound activity and selectivityprofile combined with excellent drug-like propertiestranslated in vivo, in robust dose- and time-dependentinhibition of PI3Ka signaling, resulting in good thera-peutic efficacy against PIK3CA-dependent tumors.These results showed that selective inhibition of PI3Kalead to robust efficacy in PIK3CA-dependent tumormodels comparable to pan-class I inhibitors as previ-ously reported (17, 27). One additional pending ques-tion is whether PI3Ka inhibitors require a high and/orcontinuous inhibition of the PI3K/Akt pathway to pro-duce a robust antitumor activity. We showed through adetailed PK/PD/efficacy relationship analysis that thefraction of time above the S473P-Akt IC80 value is a keydeterminant associated with NVP-BYL719 efficacy invivo, suggesting that sustained inhibition of the path-way for a fraction of the time period between 2 conse-cutive dosing is likely needed.As PI3Ka-selective inhibitors have been developed

very recently, there is only minimal information availableabout on the safety when compared with broader PI3Kinhibition. Of special concern with PI3K inhibitors is theinduction of insulin resistance as the PI3K pathway playsa predominant role in glucose homeostasis, and hyper-glycemia has been reported to be one of themost frequentadverse events in clinical trials. Jessen and colleagueshave recently demonstrated that PI3Ka-selective inhibi-tion did not perturb glucose homeostasis in rodents incontrast to pan-class I inhibition (28). In our study, byconducting a more detailed PK/PD/efficacy/tolerabilityrelationship analysis we could provide robust evidencethat NVP-BYL719 has a better therapeutic window com-pared with pan-class I inhibition (>3-fold shift), in linewith our working hypothesis.As mentioned previously, one other major challenge in

the clinical development of PI3K inhibitors is to identifypatient populations who will most likely benefit from the

treatment. Thus, active efforts are currentlymade to betterdefine patient stratification methods to maximize thera-peutic responses to such "personalized" therapies. Somepreclinical studies have already found association withactivating PIK3CA mutations and response to pan-PI3Kinhibitors (17, 29, 30).However, one study conductedwithdifferent PI3K inhibitors in a panel of 39 cell lines from 9distinct lineages did not show enhanced activity inPIK3CA mutant lines (31). Here we report for the firsttime a large-scale analysis of sensitivity to a PI3Ka-selec-tive inhibitor using the recently established CCLE (15).The integration of gene expression, chromosomal copynumber, and massive parallel sequencing data from�1,000 human cancer cell lines with pharmacologic pro-files for anticancer drugs allows the interrogation of thedataset toward the identification of genetic, lineage, andgene-expression–based predictors of drug sensitivity. Inour study, the comprehensive analysis has revealed mul-tiple factors that are capable of predicting NVP-BYL719sensitivity. Hence, we found that PIK3CAmutation is notonly associated with NVP-BYL719 response, but is themost significantmutation feature that could predict NVP-BYL719 sensitivity among 25 mutation features includedin the unbiased analysis of such large scale dataset. Thepreferred sensitivity to NVP-BYL719 among PIK3CAmutants can be observed in almost all lineages, suggestingthat the genetic status should represent a reliable patientenrollment criterion across indications. To confirm thatPI3Ka inhibition is the best way to effectively and selec-tively target the PIK3CA-mutant cancer cell populationcompared with the wild-type, we also compared theresponse profiles of PIK3CAmutant versus wild-type celllines across different compound treatments (�1,000)encompassing more than 200 MoAs and ranked the com-pounds based on the magnitude of their effects in the 2groups. In general, PI3Ka inhibitors were selectivelyeffective in PIK3CA mutant cells when compared withwild type, with NVP-BYL719 being among the mostselective and ranking higher than pan-PI3K inhibition.

In addition to PIK3CA mutation, we also foundPIK3CA amplification to be positively associated with

Se

ns

itiv

ity

in

PIK

3C

A W

T v

s.

mu

tan

t

Se

ns

itiv

ity

in

PIK

3C

A m

uta

nt

vs

. W

T

Sele

ctivity s

core

Compounds

20

10

0

–10

–20

–30

–40

–50

–60

PIK3CA compounds (n = 4)

Pan-PI3K compounds (n = 4)

MEK compounds (n = 7)BYL719 (rank = 3)

Figure 6. Identification of selectivityindex of small molecule inhibitorsfor PIK3CAmutant versus PIK3CAwild-type (WT) cell line populationsacross �1,000 differentcompounds.






NVP-BYL719 sensitivity in the overall and interestinglyalso in the PIK3CA WT cell line population. NRASmutation correlated with higher drug sensitivity withinthe entire cell lines tested or within the PIK3CA WT cellline population although here it just missed the FDR <0.25 cutoff. This is to our knowledge, the first report thatrevealed NRAS mutation being associated with PI3Kaselective inhibition.

The identification of the patient population that wouldmost benefit from PI3K inhibitors such as NVP-BYL719not only requires the determination of positive predictorsbut similarly of molecular features that are associatedwith primary resistance. Interestingly, we observed anegative association of PTEN mutation in the PIK3CA-mutated population only. This result seems to contrastwith a previously reported study where cell lines harbor-ing double alterations in PIK3CA and PTEN were signif-icantly more sensitive to the pan-class I PI3K inhibitorGDC0941 than cell lines with no detectable alterations inthe signaling pathway (29). However, it has been dem-onstrated in preclinical models that PTEN-deficienttumors are more dependent on PI3Kb signaling than onPI3Ka (32–34), hence, treatment of dual PIK3CA- andPTEN-mutated cancers might require inhibitors withactivity against both PI3Ka and PI3Kb such as pan-classI PI3K inhibitors. In contrast topreviously reported results(35), concurrent KRASmutation did not confer resistancetoNVP-BYL719 inPIK3CA-mutated cell lines. SomeKRASmutations preferentially signal through the PI3K path-way; however, KRASmutation has also been described asa resistant factor for PI3K inhibitors (12, 36) and thisopposite interplay may be context or lineage specific.

Importantly, the hypothesis-based predictor we havedeveloped by combining PIK3CA mutations with othersignificant features, PIK3CA amplification and PTENmutation showed an improvedPPVof 76%and translatedin vivo across a selection of cancer cell lines aswell as PDXtumor xenograft mouse models. However, this predictor,with a sensitivity of 21%, missed �80% of the respondersthat are not associated with the selected 3 features, indi-cating that other predictive markers of NVP-BYL719response may be very context specific and may requirefurther investigation of selected subgroup of cell lines orindications that may not be represented in sufficientnumbers to allow a statistically significant score in thecurrent dataset.

Based on the results obtained so far demonstrating thatNVP-BYL719 is a selective PI3Ka inhibitor with gooddrug-like and pharmacologic properties and that geneticalterations in PIK3CA is the most significant predictivefeature of selective sensitivity to the compound, NVP-BYL719 has been the first PI3Ka-selective inhibitor to

enter phase I clinical development in preselected adultpatients with advanced solid malignancies carryingPIK3CA gene alterations. Preliminary clinical data avail-able indicate that NVP-BYL719 is well tolerated, withmanageable side effects, a predictable PK profile andsome objective responses and prolonged disease stabi-lization with tumor shrinkage have already beenreported (16, 37). These encouraging first clinical resultssuggest that the validity of the hypothesis, the quality ofthe molecule, and the patient selection criteria identifiedand assessed preclinically are likely to translate inpatient benefit. We therefore aim at integrating currentand to be discovered patient selection criteria in thedesign of the future-clinical trials with the objective ofenhancing response rate and benefit for the patientsbearing tumors with the highest likelihood of beingsensitive to NVP-BYL719.

Disclosure of Potential Conflicts of InterestM. Boehm has ownership interest (including patents) in Novartis

Pharma AG. C. Garcia-Echeverria has ownership interest (includingpatents) in Sanofi. R. Schlegel has ownership interest (includingpatents) in Novartis Pharmaceuticals. No potential conflicts of interestwere disclosed by the other authors.

Authors' ContributionsConception and design: C. Fritsch, A. Huang, C. Chatenay-Rivauday,C. Schnell, Y. Wang, S.M. Brachmann, C. Garcia-Echeverria, G. Caravatti,F. Hofmann, W.R. SellersDevelopment ofmethodology:A.Huang, C. Chatenay-Rivauday,M. Liu,A. Kauffmann, D. Guthy, A. De Pover, S. Ferretti, C. WilsonAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): C. Chatenay-Rivauday, C. Schnell, D. Erdmann,A. De Pover, H. Gao, S. Ferretti, Y. Wang, J. Trappe, S.M. Brachmann,C. WilsonAnalysis and interpretation of data (e.g., statistical analysis, biostatis-tics, computational analysis): C. Fritsch, A. Huang, C. Chatenay-Rivau-day, C. Schnell, A. Reddy, M. Liu, A. Kauffmann, A. De Pover, P. Furet,J. Trappe, S.M. Brachmann, C. Wilson, R. Cozens, J. LeharWriting, review, and/or revision of themanuscript: C. Fritsch, C. Schnell,M. Liu, A. Kauffmann, D. Erdmann, S.M. Brachmann, S.-M. Maira,C. Wilson, M. Boehm, R. Cozens, F. Hofmann, W.R. SellersAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): M. Liu, W.R. SellersStudy supervision: C. Fritsch, A. Huang, C. Chatenay-Rivauday,C. Schnell, H. Gao, Y. Wang, C. Wilson, C. Garcia-Echeverria, P. Chene,M. Wiesmann, R. Cozens, R. Schlegel, F. Hofmann, W.R. Sellers

AcknowledgmentsThe authors thank E. Pfister, A. Gluck-Gad�e, P. Hauser, M. Ristov,

M. Klopfenstein, S. Barb�e, J. Koenig, T. Ferrat, F. Wenger, L. Martinuzzi-Duboc, W. Tinetto, D. Sterker, C. Kowalik, M. Scheibler, and G. Yang forexcellent technical assistance, thank C. Quadt, E. Di Tomaso, and S. Cosulichfor very helpful discussion and also thank M. Schlabach, E. Billy, andT. Schmelzle for their help during the revision of the manuscript.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received October 14, 2013; revised February 7, 2014; accepted February23, 2014; published OnlineFirst March 7, 2014.

References1. Cantley LC. The phosphoinositide 3-kinase pathway. Science

2002;296:1655–7.2. Sansal I, Sellers WR. The biology and clinical relevance of the PTEN

tumor suppressor pathway. J Clin Oncol 2004;22:2954–63.


Fritsch et al.




3. Knight ZA, Gonzales B, Feldman ME, Zunder ER, GoldenbergDD, Williams O, et al. A pharmacological map of the PI3K familydefines a role for p110a in insulin signaling. Cell 2006;125:733–47.

4. Sopasakis VR, Liu P, Suzuki R, Kondo T, Winnay J, Tran TT, et al.Specific roles of the p110a isoform of phosphatidylinositol 3-kinase inhepatic insulin signaling and metabolic regulation. Cell Metab 2010;11:220–30.

5. Graupera M, Guillermet-Guibert J, Foukas LC, Phng LK, Cain RJ,Salpekar A, et al. Angiogenesis selectivity requires the p110a isoformof PI3K to control endothelial cell migration. Nature 2008;453:662–6.

6. Rommel C, Camps M, Ji H. PI3K d and PI3K g: partners in crime ininflammation in rheumatoid arthritis and beyond? Nat Rev Immunol2007;7:191–201.

7. Rodon J, Dienstmann R, Serra V, Tabernero J. Development of PI3Kinhibitors: lessons learned from early clinical trials. Nat Rev Clin Oncol2013;10:143–53.

8. Bachman KE, Argani P, Samuels Y, Silliman N, Ptak J, Szabo S, et al.The PIK3CA gene is mutated with high frequency in human breastcancers. Cancer Biol Ther 2004;3:776–7.

9. WuG,XingM,MamboE,HuangX, LiuX,GuoZ, et al. Somaticmutationand gain of copy number of PIK3CA in human breast cancer. BreastCancer Res 2005;7:R609–16.

10. Ikenoue T, Kanai F, Hikiba Y, Obata T, Tanaka Y, Imamura J, et al.Functional analysis of PIK3CA gene mutations in human colorectalcancer. Cancer Res 2005;65:4562–7.

11. Isakoff SJ, Engelman JA, Irie HY, Luo J, Brachmann SM, Pearline RV,et al. Breast cancer-associated PIK3CA mutations are oncogenic inmammary epithelial cells. Cancer Res 2005;65:10992–1000.

12. Engelman JA, Chen L, Tan X, Crosby K, Guimaraes AR, Upadhyay R,et al. Effective use of PI3K and MEK inhibitors to treat mutant KrasG12D and PIK3CA H1047R murine lung cancers. Nat Med 2008;14:1351–6.

13. Meyer DS, Brinkhaus H, M€uller U, M€uller M, Cardiff RD, Bentires-Alj M.Luminal expression of PIK3CAmutant H1047R in the mammary glandinduces heterogeneous tumors. Cancer Res 2011;71:4344–51.

14. Liu P, Cheng H, Santiago S, Raeder M, Zhang F, Isabella A, et al.Oncogenic PIK3CA-driven mammary tumors frequently recur via PI3Kpathway-dependent and PI3K pathway-independent mechanisms.Nat Med 2011;17:1116–20.

15. Barretina J, Caponigro G, StranskyN, Venkatesan K,Margolin AA, KimS, et al. The Cancer Cell Line Encyclopedia enables predictive model-ling of anticancer drug sensitivity. Nature 2012;483:603–7.

16. JuricD,Rodon J,Gonzalez-AnguloAM,BurrisHA,Bendell J, Berlin JD,et al. BYL719, a next generation PI3Ka-specific inhibitor: preliminarysafety, PK, and efficacy results from the first-in-human study. CancerRes 2012;72:8 (suppl; abstr CT-01).

17. Maira SM, Pecchi S, Huang A, Burger M, Knapp M, Sterker D, et al.Identification and characterization of NVP-BKM120, an orally availablepan-class I PI3-kinase inhibitor. Mol Cancer Ther 2012;11:317–28.

18. Furet P, Guagnano V, Fairhurst RA, Imbach-Weese P, Bruce I, KnappM, et al. Discovery of NVP-BYL719 a potent and selective phospha-tidylinositol-3 kinase a inhibitor selected for clinical evaluation. BioorgMed Chem Lett 2013;23:3741–8.

19. Guagnano V, Kauffmann A, W€ohrle S, Stamm C, Ito M, Barys L, et al.FGFR genetic alterations predict for sensitivity to NVP-BGJ398, aselective pan-FGFR inhibitor. Cancer Discov 2012;2:1118–33.

20. Benjamini Y, Hochberg Y. Controlling the false discovery rate: apractical and powerful approach to multiple testing. J R Stat Soc B1995;57:289–300.

21. Bruce I, Akhlaq M, Bloomfield GC, Budd E, Cox B, Cuenoud B, et al.Development of isoform selective PI3-kinase inhibitors as pharmaco-logical tools for elucidating the PI3K pathway. Bioorg Med Chem Lett2012;22:5445–50.

22. Moreno Garcia V, Baird RD, Shah KJ, Basu B, Tunariu N, Blanco M,et al. A phase I study evaluatingGDC-0941, anoral phosphoinositide-3kinase (PI3K) inhibitor, in patients with advanced solid tumors ormultiple myeloma. J Clin Oncol 2011;29:15 (suppl; abstr 3021).

23. Bendell JC, Rodon J, Burris HA, de Jonge M, Verweij J, Birle D, et al.Phase I, dose-escalation study of BKM120, an oral pan-class I PI3Kinhibitor, in patients with advanced solid tumors. J Clin Oncol 2012;30:282–90.

24. Flaherty KT, Puzanov I, Kim KB, Ribas A, McArthur GA, Sosman JA,et al. Inhibition of mutated, activated BRAF in metastatic melanoma.N Engl J Med 2010;363:809–19.

25. Kwak EL, Bang YJ, Camidge DR, ShawAT, Solomon B,Maki RG, et al.Anaplastic lymphoma kinase inhibition in non-small-cell lung cancer.N Engl J Med 2010;363:1693–703.

26. Brana I, Siu LL. Clinical development of phosphatidylinositol 3-kinaseinhibitors for cancer treatment. BMC Med 2012;10:161.

27. Raynaud FI, Eccles SA, Patel S, Alix S, Box G, Chuckowree I, et al.Biological properties of potent inhibitors of class I phosphatidylinosi-tide 3-kinases: from PI-103 through PI-540, PI-620 to the oral agentGDC-0941. Mol Cancer Ther 2009;8:1725–38.

28. Jessen KA, Kessler L, Kucharski J, Guo X, Staunton J, Elia M, et al.INK1117: a potent and orally efficacious PI3Ka-selective inhibitor forthe treatment of cancer. Cancer Res 2011;71:8 (suppl; abstr 4501).

29. Kwei KA, Baker JB, Pelham RJ. Modulators of sensitivity and resis-tance to inhibition of PI3K identified in a pharmacogenomic screen ofthe NCI-60 human tumor cell line collection. PLoS ONE 2012;7:e46518.

30. O'Brien C, Wallin JJ, Sampath D, GuhaThakurta D, Savage H, Pun-noose EA, et al. Predictive biomarkers of sensitivity to the phospha-tidylinositol 30 kinase inhibitor GDC-0941 in breast cancer preclinicalmodels. Clin Cancer Res 2010;16:3670–83.

31. Dan S, Okamura M, Seki M, Yamazaki K, Sugita H, Okui M, et al.Correlating phosphatidylinositol 3-kinase inhibitor efficacy with sig-naling pathway status: in silico and biological evaluations. Cancer Res2010;70:4982–94.

32. Wee S, Wiederschain D, Maira SM, Loo A, Miller C, deBeaumont R,et al. PTEN-deficient cancers depend on PIK3CB. Proc Natl Acad SciU S A 2008;105:13057–62.

33. Jia S, Liu Z, Zhang S, Liu P, Zhang L, Lee SH, et al. Essential roles of PI(3)K-p110b in cell growth, metabolism and tumorigenesis. Nature2008;454:776–9.

34. Ni J, Liu Q, Xie S, Carlson C, Von T, Vogel K, et al. Functionalcharacterization of an isoform-selective inhibitor of PI3K-p110b as apotential anticancer agent. Cancer Discov 2012;2:425–33.

35. Huang A, Fritsch C, Wilson C, Reddy A, Liu M, Lehar J, et al. Singleagent activity of PIK3CA inhibitor BYL719 in a broad cancer cell linepanel. Cancer Res 2012;72:8 (suppl; abstr 3749)

36. Ihle NT, Lemos R Jr, Wipf P, Yacoub A, Mitchell C, Siwak D, et al.Mutations in the phosphatidylinositol-3-kinase pathway predict forantitumor activity of the inhibitor PX-866 whereas oncogenic Ras is adominant predictor for resistance. Cancer Res 2009;69:143–50.

37. Juric D, Argiles G, Burris HA, Gonzalez-Angulo AM, Saura C, Quadt C,et al. Phase I study of BYL719, an a-specific PI3K inhibitor, in patientswith PIK3CA mutant advanced solid tumors: preliminary efficacy andsafety in patients with PIK3CA mutant ER-positive (ERþ) metastaticbreast cancer (MBC). Cancer Res 2012;72:24 (suppl; abstr P6-10-07).






2014;13:1117-1129. Published OnlineFirst March 7, 2014.Mol Cancer Ther Christine Fritsch, Alan Huang, Christian Chatenay-Rivauday, et al. for Clinical TrialsNVP-BYL719 and Development of the Patient Stratification Strategy

InhibitorαCharacterization of the Novel and Specific PI3K

Updated version

10.1158/1535-7163.MCT-13-0865doi:

Access the most recent version of this article at:

Material

Supplementary

http://mct.aacrjournals.org/content/suppl/2014/03/11/1535-7163.MCT-13-0865.DC1

Access the most recent supplemental material at:

Cited articles

http://mct.aacrjournals.org/content/13/5/1117.full#ref-list-1

This article cites 37 articles, 15 of which you can access for free at:

Citing articles

http://mct.aacrjournals.org/content/13/5/1117.full#related-urls

This article has been cited by 43 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://mct.aacrjournals.org/content/13/5/1117To request permission to re-use all or part of this article, use this link



http://mct.aacrjournals.org/lookup/doi/10.1158/1535-7163.MCT-13-0865

http://mct.aacrjournals.org/content/suppl/2014/03/11/1535-7163.MCT-13-0865.DC1

http://mct.aacrjournals.org/content/13/5/1117.full#ref-list-1

http://mct.aacrjournals.org/content/13/5/1117.full#related-urls

http://mct.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://mct.aacrjournals.org/content/13/5/1117


Characterization of the Novel and Speciﬁc PI3Ka Inhibitor ... · Stephane Ferretti1, Youzhen...

Documents

Transcript of Characterization of the Novel and Speciﬁc PI3Ka Inhibitor ... · Stephane Ferretti1, Youzhen...