TUTae
`Haemce~hilia
`
`THE INTERNATIONAL WEEKLY JOURNAL OF SCIENCE
`
`Antitumourimmunity
`enhancedbyinhibiting
`PD-L1/PD-1 andidentifying
`AUEICCLmobi 1arXapbs
`
`eviaes
`
`$10.00US $12.99CAN
`
`74 DO NOT REMOVE FROM
`CURRENTPERIODICALS
`ROOM
`
`at
`
`MICROSCOPY
`
`ACCEPT YOUR
`OWN PAPER
`Howsomescientists are
`dupingthe system
`PAGE 480
`
`THE CASE FOR
`VOYOUGeGeel.
`Atomicresolutionis there
`forthe taking
`Haety
`
`MNseeaT
`Meeal
`
`Aad
`
`ENERGY
`
`SUOEMs
`27 November 2014
`
`‘NIGHT-TIME’
`COOLING BY DAY
`Genome & Co. v. Univ. of Chicago
`hued PGR2019-00002
`UNIV. CHICAGO EX. 2056
`
`

`

`PaEAVAREAWAY MICROBIOLOGY
`
`NAWTEIIS wicrosioiocy
`
`DEC 08 2014
`
`sgRNA
`
`UNIVERSITY LIBRARY
`
`
`
`
`ININS
`
`
`
`Genomic DNA
`
`DM
`
`Hi Donor DNA
`
`FREE POSTER
`
`CRISPR-Cas: extraordinary editing
`
`The CRISPR-Cas system evolved to provide adaptive immunity in bacteria and archaea against
`invading genetic elements, such as viruses and plasmids. The mechanism of interference is based on
`CRISPR RNAs, whichare derived from foreign DNA and mediate the detection of invading nucleic acids,
`followed by degradation bythe Casproteins. This poster presents the molecular mechanism of CRISPR-
`Cas interference and highlights the use of specifically engineered guide RNAs and the Cas9 nuclease for
`the editing of virtually any desired genome, in addition to other applications such as imagingin live cells
`and the regulation of gene expression.
`
`
`
`nature
`REVIEWS
`
`
` <== ORIEN
`=:
`
`‘5 Downloadtheposterat:
`www.nature.com/nrmicro/posters/crispr
`
`Produced with support from
`
`y ORIGENF
`
`Your Gene Company
`
`|
`
`nature publishing group &
`
`

`

`CONTENTS
`
`27 November 2014 / Vol 515 / Issue No 7528
`
`NEWS IN FOCUS
`
`COMMENT
`
`THIS WEEK
`
`EDITORIALS
`465 CLIMATE
`Agree to agree
`Other nations mustfollow the lead of
`the United States and China
`
`
`
`§ WORLD VIEW
`_
`467 Openaccess
`is tiring out peer
`reviewers
`Martijn Arns
`Scholarly system must
`adapt to workload
`
`
`SCIENTIFIC LITERATURE
`
`RESEARCH HIGHLIGHTS
`468 SELECTIONS FROM THE
`
`Termite queen rules roost / North China
`earthquakehistory / Old papers back in
`vogue / Tagged fish predated / Remote
`control of epilepsy / Greenland onice
`
`SEVEN DAYS
`470 THE NEWS IN BRIEF
`Oil pipeline rejected / Charity sells
`royalty rights / Holt to head AAAS /
`LHC data go open access / Wave-power
`company enters administration / ‘Polar
`code’ to regulate ships
`
`CAREERS
`
`597 DIVERSITY
`Structural approach
`Materials scientists are trying to
`broaden the range of people
`the field attracts
`
`
`
`47os
`
`Data-managementissues delay
`observatory project
`476 NEUROSCIENCE
`FDA considers how to regulate
`brain-linked devices
`
`47 ~
`
`47 o
`
`47 o
`
`MEDICINE
`Clampdownonclinical-trial reporting
`PUBLISHING
`Open-data dictum notalways followed
`ECOLOGY
`Galapagosgift-shop closure
`shortchanges conservation
`
`FEATURES
`
`
`
`PUBLISHING
`
`Tne weer=
`review scam.
`
`Evidenceofself-review exposes
`weaknesses in system PAGE 480
`
`484 NUCLEAR POWER
`Desperately seeking plutonium
`Dwindling stockpile forces US rethink
`
`
`
`
`
`
`465 INFECTIOUS DISEASE
`Ebola opportunity
`Slowdown should be put to good use
`466 SPACE
`Moononastick
`473 EMISSIONS
`BOOKS & ARTS
`Crowdfunded mission worthatry
`US-China deal renews hopesfor talks
`MARINE SCIENCE
`
`
`
`487 MICROSCOPY
`Hasten high
`resolution
`Stephen J
`Pennycook &
`Sergei V Kalinin
`Precision
`microscopes are
`needed to analyse
`atomseffectively
`
`
`
`
`MILITARY SCIENCE
`
`Spoils for
`science
`
`Ann Finkbeiner reviews two books on
`the legacies of the cold war PAGE 489
`
`490 ORNITHOLOGY
`Fowl domination
`Ewen Callaway
`491 CLIMATE SCIENCE
`A climate trance
`Richard Van Noorden
`
`
`
`CORRESPONDENCE
`492 Ebola virus control / Brain trainingin its
`infancy / China's progress in research /
`Philae’s marathon achievement
`
`_
`[Py
`
`RebeccaBirch
`
`FUTURES
`600 Ice and
`white roses
`
`27 NOVEMBER 2014 | VOL 515 | NATURE | 459
`
`

`

`CONTENTS
`
`27 November 2014 / Vol 515 / Issue No 7528
`
`NEW ONLINE
`493 Papers published thisweekat nature.com
`494 CLIMATESCIENCE
`NEWS & VIEWS
`
`EI Nifio’s variable history
`Evolution of the El Nifico-Southern
`Oscillation over the past 21,000 years
`JOSSDUIS FBIM Er LETTER Rome
`495 MAMMALIAN EVOLUTION
`A beastofthe southern wild
`Acomplete skull of a Madagascan
`
`ibctaaa ae
`496 CANCER
`Antitumour immunity gets a boost
`Responsesto immune-checkpoint
`blockade in multiple cancers
`Jedd D Wolchok & Timothy A Chan
`SER LETTEar rebase
`498 ASTRONOMY
`Cosmictriangles and
`black-hole masses
`Estimating geometric distances to
`active galactic nuclei and quasars
`Martin Elvis SEE LETTER P.528
`
`499 DEVELOPMENTALBIOLOGY
`Polarize to elongate
`Toll receptors direct cell rearrangement
`UCT naseSesame
`HNC HTORESS SFE ARNOESz
`501 DIET
`Food choices for health and planet
`Projecteddiets linked to greenhouse-
`gas emissions and poorhealth
`Elke Stehfest SEE ARTICLE P.518
`502 IMMUNOLOGY
`Toleranceliesin the timing
`Lck-bound co-receptorrecruitmentis
`rate-limiting for negative selection
`Nicholas RJ Gascoigne
`de del
`aiindioarti
`clowground
`blodiversity
`and ecosystem functioning
`SEESteneree
`ARTICLES
`512 EVOLUTION First cranial remains of a
`gondwanatherian mammalreveal
`remarkable mosaicism
`DW Krause et al. SEE N&V P.495
`
`
`
`
`|
`|
`|
`
`RESEARCH
`
`Pr naale eee
`|
`72ATr
`
`
`
`
`uml
`
`i
`een is
`
`avers
`ATE
`
`518 NUTRITION Globaldiets link
`environmentalsustainability
`and human health
`Ena he Ciases NANGes
`523. DEVELOPMENTALBIOLOGY A positional
`Toll receptor code directs convergent
`extension in Drosophila
`ACParé etal. SEE N&vP.499
`
`531
`
`LETTERS
`528 ASTRONOMYA dust-parallax distanceof
`19 megaparsecsto the supermassive
`blackhole inNGC 4151 __
`SF Honig, D Watson, M Kishimoto
`&J Hjorth SEE N&V P.498
`SPACE PHYSICS An impenetrable barrier
`to ultrarelativisticelectrons in the
`VanAllen radiation belts
`DN Baker et al.
`535 MATERIALS SCIENCE Metallization of
`vanadium dioxide driven by
`large phonon conrehy.
`JD Budajet al.
`540 OPTICS AND PHOTONICS Passive
`radiative cooling below ambientair
`
`
`
`ALLISONBRUCE
`
`perennedetgt
`E Rephaeli & S Fan
`andfabricationofmemory
`545 MOLECULAR ELECTRONICS Design
`
`devices based on nanoscale
`polyoxometalate clusters
`C Buscheetal.
`550 PALAEOCLIMATE Evolution and forcing
`mechanisms ofEl Nifio over
`the past 21,000 years
`Z Liu etal. SEE N&V P.494
`
`554 BIOPHYSICS Multiplex single-molecule
`interaction profilingof DNA-barcoded
`proteins
`| Guetal.
`558 CANCER MPDL3280A(anti-PD-L1)
`treatmentleadsto clinical
`activity in metastatic bladder cancer
`T Powles et al. SEE N&VP. 496
`563 CANCER Predictive correlatesof
`responseto the anti-PD-L1 antibody
`MPDL3280Ain cancerpatients
`RSHerbstetal. SEE N&V P.496
`F
`568 CANCER PD-1 blockade induces
`responsesbyinhibiting
`adaptive immuneresistance
`PC Tumehetal. SEE N&V P.496
`am ©
`;
`vais.Cg eS
`sqoctioinctry sndesinv sequencing
`M Yadav etal. SEE N&V P.496
`‘
`:
`377 CANCER Checkpoint blockadecancer
`immunotherapytargets tumour-
`specific mutant antigens
`NOC aL eee aee
`582 NEUROSCIENCE Histone H2A.Z subunit
`exchangecontrols consolidation of
`recent and remote memory
`|B Zovkic et al.
`587 PLANT SCIENCES Epigenetic
`reprogrammingthatprevents
`transgenerationalinheritance
`of the vernalized state
`P Crevillén et al
`591 VIROLOGY A structure-based mechanism
`for tRNA andretroviral RNA
`remodelling during primer annealing
`SB Miller, F Z Yildiz, J A Lo, B Wang
`& VM D'Souza
`
`27 NOVEMBER 2014 |
`
`VOL 515 | NATURE | 461
`
`

`

`nature|joumals
`
`SUN2
`
`27 November 2014 / Vol 515 / Issue No. 7528
`
`5 nature
`
`nature@nature.com
`
`LONDON, PARIS, MUNICH,
`SHANGHAI, TOKYO, SAN FRANCISCO,
`WASHINGTON DC, NEW YORK,
`BOSTON
`EDITOR-IN-CHIEF; Philip Carnpbell
`NATURE EDITORIAL DIRECTOR: Ritu Dhand
`DIRECTOR, AUTHOR AND REVIEWER SERVICES: Véronique Kiermer
`EXECUTIVE EDITOR: Nick Campbell
`HEADOF EDITORIAL PRODUCTION: Lucy Banham
`CHIEF MAGAZINE EDITOR: Rosie Mestel
`CREATIVE DIRECTOR: Kelly Buckheit Krause
`THIS WEEK/NEWS/FEATURES:Alison Abbott, David Adam, Celeste
`Biever, Declan Butler, Ewen Callaway, Davide Castelvecchi,
`Erika Check Hayden, Matthew Crenson, Daniel Cressey,
`David Cyranoski, Elizabeth Gibney, Heidi Ledford, Corie Lok,
`Brendan Maher, Richard Monastersky, Lauren Morello,
`Anna Nagle, Helen Pearson, Sara Reardon, Quirin Schiermeier,
`Jeff Tollefson, Richard Van Noorden, M Mitchell Waldrop,
`Alexandra Witze COMMENT: Sara Abdulla, Joanne Baker,
`Rosalind Cotter, Barbara Kiser, Lucy Odling-Smee NATURE
`PODCAST/VIDEO: Charlotte Stoddart, Noah Baker, Kerri Smith
`NEWS AND VIEWS: Andrew Mitchinson, Jennifer Gardiner,
`Ana Lopes, Marian Turner RESEARCH: Francesca Cesari,
`Karl Ziemelis, -han Chou, Tanguy Chouard, Rosamund Daw,
`Alex Eccleston, Angela Eggleston, Joshua Finkelstein, Henry Gee,
`Patrick Goymer, Hanna Granroth-Wilding, Noah Gray,
`Claire Hansell, Marie-Thérése Heemels, Magdalena Helmer,
`Andrew Jermy, Nathalie Le Bot, Barbara Marte, Juliane
`Méssinger, Leonie Mueck, Leslie Sage, Sadaf Shadan,
`Mara Silva, Magdalena Skipper, Clare Thomas, John VanDecar,
`Liesbeth Venema, Ursula Weiss, Michael White INSIGHTS/
`REVIEWS/PERSPECTIVES: Ursula Weiss CAREERS: Karen
`Kaplan TECHNOLOGY FEATURES: Vivien Marx FUTURES: Colin
`Sullivan SUPPLEMENTS: Herb Brody, Michelle Grayson,
`Kathryn Miller STATISTICAL ADVISER: Theresa Hyslop
`SUBEDITORS:Colin Sullivan, Mary Abraham, Sarah Archibald,
`Nicola Bailey, Adam Born, Nicolas Fanget, Isobel Flanagan,
`Paul Fletcher, Nisha Gaind, Anne Haggart, Dinah Loon,
`Anna Novitzky, Yukie Ozawa, Katherine Perry, David Price,
`Jenny Rooke, David Winters, Anna York PRODUCTION: Nick
`Kemp, Donald McDonald, Stewart Fraser, Susan Gray, Simon
`Gribbin, Hazel Mayhew,lan Pope, Yvonne Strong, Roberta
`Tarran, Charles Wenz ART AND DESIGN: Wesley Fernandes,
`Michael Ayres, Jennifer Burns, Ffion Cleverley,
`Barbara lzdebska, Paul Jackman, Jasiek Krzysztofiak,
`Annthea Lewis, Chris Maddaloni, Chris Ryan, Nik Spencer,
`Claire Welsh JAPAN EDITION: Eiji Matsuda, Mika Ishida,
`Mitsuyo Utsugi, Sayaka Ayame PRESS OFFICE: Alice Henchley,
`Neda Afsarmanesh,Lisa Boucher, Sarah Lahert,
`Michael Stacey, Rebecca Walton NATURE DIGITAL: Tom Scott,
`Cianna Timbers, Alexander Thurrell, Dipti Shah, Dan Foley,
`Robert Sullivan, Matthew Carey, Luke Stavenhagen,
`Marta Rolak, Glynn Phillips, Marcin Arciuch ADMINISTRATION:
`Caroline McLean, Roseann Campbell, Emily Banham,
`Sophie Beasley, Emma Clifford, Dana Coles, Ella Kahn,
`Melissa Rose, Lauren Sequeira, Kenneth Simpson,
`Kimberly Tolleson, Michelle Viscuso, Andrea Von,
`Rebecca White, Diane Yorke
`
`CHIEF EXECUTIVEOFFICER: Steven Inchcoombe
`PUBLISHER: Sarah Greaves
`PUBLISHER (SUPPLEMENTS & PARTNERSHIPS): Richard Hughes
`PUBLISHING PROJECT MANAGER: Anastasia Panoutsou
`
`USA/CANADA/LATIN AMERICA: marketing@us.nature.com
`EUROPE/REST OF WORLD: marketing@nature.com
`JAPAN/CHINA/KOREA: marketing@natureasia.com
`INDIA: marketing-india@nature.com
`
`AMERICAS:institutions@natureny.com
`ASIA-PACIFIC: institutions@natureasia.com
`AUSTRALIA/NEW ZEALAND: nature@macmillan.com.au
`EUROPE/ROW: institutions@nature.com
`INDIA: npgindia@nature.com
`
`
`IONS (incLupiNG Post AND PACKING)
`SUBSC
`180
`UK/EUROPE/REST OF WORLD
`subscriptions@nature.com
`FULL RATE: UK £1,947; Europe €3,014
`PERSONAL RATES: UK £135; Europe €209; Rest of World £210
`STUDENT RATES: UK £79; Europe €122; Rest of World £125
`Individual rates apply to payments by personal cheque or
`credit card, BACK ISSUES: £20.00.
`USA/CANADA/LATIN AMERICA
`subscriptions@us.nature.com
`INSTITUTIONAL/CORPORATERATE: $3,790
`PERSONAL RATE: $199 STUDENT RATE: $99; POSTDOC RATE: $119
`Printed in USA. Individual rates available only to subscribers
`paying by personal checkor credit card. Orders for
`student/postdoc subscriptions must be accompanied by a
`copy of student ID. Rates apply to USA, Canada, Mexico/Central
`& South America. Add 7% GST tax in Canada (Canadian GST
`number 140911595). BACK ISSUES: US$20.00.
`JAPAN/CHINA/KOREA
`JAPAN: nature.asia/japan-subscribe
`CHINA: nature.asia/china-subscribe
`KOREA: www.natureasia.com/ko-kr/subscribe
`
`NPG LONDON: nature@nature.com
`The Macmillan Building, 4 CrinanStreet, London N1 9XW
`Tel: +44 (0)20 7833 4000
`NPG NEW YORK; nature@natureny.com
`75 Varick St, 9th Floor, New York, NY 10013-1917
`Tel: +1 212 726 9200
`NPG ASIA-PACIFIC: nature@natureasia.com
`Chiyoda Building, 2-37 Ichigayatamachi, Shinjuku-ku,
`Tokyo 162-0843
`Tel: +81 3 3267 8751
`NPG GREATER CHINA: china@nature.com
`42F-10/11, The Center, 989 Changle Lu, Shanghai 200031
`Tel: +86 21 2422 5000
`
`Alist of other NPG offices can be found at
`www.nature.com/npg_/contact/offices.html
`
`Nature® (ISSN 0028-0836) published weekly by -
`UK edition: Nature Publishing Group, a division of Macmillan
`Publishers Ltd, The Macmillan Building, 4 Crinan Street,
`London N1 9XW, UK.
`US edition: Nature America, Inc. trading as Nature Publishing
`Group, 75 Varick St, 9th Fl, New York NY 10013-1917, USA.
`Japan edition: NPG Asia-Pacific, Chiyoda Building,
`2-37 Ichigayatamachi, Shinjuku-ku, Tokyo 162-0843, Japan.
`Registered as a newspaperat the British Post Office.
`Nopart of this publication may be reproduced, stored
`in a retrieval system, or transmitted in any form
`(electronic or otherwise) without prior permission from
`permissions@nature.com. In the US, Nature (ISSN
`0028-0836)is published weekly except the last week in
`December by Nature America, Inc. trading as Nature
`Publishing Group, 75 Varick St, 9th Fl, New York NY
`10013-1917, USA. US Periodicals postage paid at New York
`NY, and additional mailing postoffices.
`US POSTMASTER: send address changes to Nature America,
`Inc. Subscription Dept, 75 Varick St, 9th Fl, New York NY
`10013-1917, USA.
`© 2014 Macmillan Publishers Limited. All rights reserved.
`
`ena
`@ AppStore
`
`nature publishing group &te
`
` DISPLAY ADVERTISING:
`
`US/CANADA: display@us.nature.com
`EUROPE; display@nature.com
`ASIA-PACIFIC: nature@natureasia.com
`SPONSORSHIP: Sponsorship@nature.com
`NATURE JOBS: naturejobs@nature.com
`
`464 | NATURE | VOL 515 | 27 NOVEMBER
`
`2014
`
`

`

`This material may be protected by Copyright law (Title 17 U.S. Code)
`
`VRAEN
`
`doi:10.1038/nature14011
`
`Predictive correlates of response to the anti-PD-Ll
`antibody MPDL3280Ain cancer patients
`RoyS. Herbst’, Jean-Charles Soria”, Marcin Kowanetz’, Gregg D. Fine*, Omid Hamid*, Michael S. Gordon”,Jeffery A. Sosman’,
`
`David F. McDermott’, John D. Powderly®, Scott N. Gettinger’, Holbrook E. K. Kohrt’, Leora Horn!®, Donald P. Lawrence",
`Sandra Rost®, Maya Leabman*, Yuanyuan Xiao*, Ahmad Mokatrin®, Hartmut Koeppen’, PritiS. Hegde’, [ra Mellman’,
`Daniel S. Chen? & F. Stephen Hodi'*
`
`tumour-infiltrating immune cells being more common than PD-L1-
`The developmentof humancanceris a multistep process character-
`positive tumourcells. PD-L1-positive tumour-infiltrating immunecells
`ized by the accumulation of genetic and epigenetic alterations that
`included myeloid cells (macrophages,dendritic cells) and T cells; B cells
`drive or reflect tumourprogression. These changes distinguish can-
`were negative for PD-L1 (Fig. Ic).
`cer cells from their normal counterparts, allowing tumoursto be re-
`cognized as foreign by the immunesystem'*. However, tumoursare
`Wedevelopedahigh-affinity human monoclonal immunoglobulin-G1
`(IgG1) antibodyforclinical use that specifically binds to PD-L1 (MPDL-
`rarely rejected spontaneously,reflecting their ability to maintain an
`immunosuppressive microenvironment’. Programmed death-ligand
`3280A;bindingaffinity Ky (dissociation constant) = 0.4nM)and pre-
`ventsits interaction with PD-1 and B7.1. However, the antibody would
`1 (PD-L1; also called B7-H1 or CD274), whichis expressed on many
`leave intact the interaction ofPD-1 with its alternative ligand PD-L2 (also
`cancer and immunecells, plays an important part in blocking the
`called B7-DC or CD273), which is thoughtto have a key role in main-
`‘cancer immunity cycle’ by binding programmeddeath-1 (PD-1) and
`taining peripheraltolerance, particularly in the lung'*"". MPDL3280A
`B7.1 (CD80), both of which are negative regulators of T-lymphocyte
`wasengineeredwith a crystallizable fragment(Fc) domain modification
`activation. Binding of PD-L1to its receptors suppresses T-cell mi-
`eliminating antibody-dependentcellular cytotoxicity at clinicallyrele-
`gration,proliferation andsecretion of cytotoxic mediators, andre-
`vantdoses, preventing depletion ofactivated Tcells'*'® (see Methods).
`stricts tumourcell killing*’°. The PD-L1-PD-1 axis protects the host
`from overactive T-effectorcells not only in cancer but also during
`Patients were treated with MPDL3280A,andpre-treatmentand on-
`treatment tumourspecimenswerecharacterized fromavailable samples.
`microbial infections'!. Blocking PD-L1 should therefore enhanceanti-
`cancer immunity, butlittle is known aboutpredictive factors ofef-
`A total of277 patients with advanced incurable cancer received MPDL-
`3280Aintravenously every 3 weeks (q3w; Extended Data Fig. 1a, band
`ficacy. This study was designedto evaluate the safety, activity and
`Extended Data Table 1; see Methods). Meansingle-dose MPDL3280A
`biomarkers of PD-L1 inhibition using the engineered humanized
`pharmacokinetics were consistent with a typical IgG] at doses =1 mg kg',
`antibody MPDL3280A. Here we showthat across multiple cancer
`with a mean terminalserumhalf-life of ~3 weeks (Extended DataFig.1c).
`types, responses (as evaluated by Response Evaluation Criteria in Solid
`Overall, treatment was well tolerated up to the maximum administered
`Tumours,version 1.1) were observed in patients with tumoursexpres-
`dose of 20 mg kg| q3w (Table 1).
`sing high levels of PD-L1, especially when PD-L1 was expressed by
`Most adverse events (AEs) did not require medical treatment. The
`tumour-infiltrating immunecells. Furthermore, responses were asso-
`most commontreatment-related AE wasfatigue (Table 1), which often
`ciated with T-helper type 1 (T};1) gene expression, CTLA4 express-
`occurred with low-grade fever during thefirst treatment cycle. Pyrexia
`ion and the absence of fractalkine (CX3CL1) in baseline tumour
`was reported in ~21% ofpatients; itmost commonly occurred during
`specimens. Together, these data suggest that MPDL3280Ais most
`effective in patients in which pre-existing immunity is suppressed by
`cycle 1 and was uncommonduring subsequentcycles (Extended Data
`Fig. 2a). Additionally, an ~2-fold increasein activated proliferating CD8 *
`PD-LI, andis re-invigorated on antibody treatment.
`T cells (CD8* HLA-DR‘Ki-67*) and a trendof increased circulating
`Pre-clinical studies demonstrated that anti-PD-L1 treatment of mice
`interferon(IFN)-¥ were observedbythe endofthefirst cycle (Extended
`bearing implanted syngeneic tumourscouldlead to tumourregression
`and the inductionof protective immune memoryinthe setting ofre-
`Data Fig. 2b, c).
`Treatment-related grade 3-4 AEs were observed in 35 patients (13%)
`challenge with tumourcells (Genentech, unpublished data). However,
`and immune-related grade 3-4 AEs were observed in 3 patients (19%)
`most mouse models constitutively express PD-L1 (ref. 12), whichis not
`(see Methodsfor further information regarding AE grades). No cases
`consistent with human tumours. Additionally, only a few syngeneic mod-
`els (notably the MC38 colon carcinoma model) were responsiveto anti-
`of grades 3-5 pneumonitis were seen.
`PD-L1 as a single agent (Genentech, unpublished data). Therefore, a
`The impact of PD-L1 inhibition on metastatic lesions was evaluated
`detailed analysis of PD-L1 expression in human tumours andits asso-
`per Response Evaluation Criteria in Solid Tumours, version 1.1 (RECIST
`ciation with clinical benefit was required.
`v1.1), In the 175efficacy-evaluable patients (with demographic and base-
`line characteristics similar to thosein all patients), confirmed responses
`PD-L1 in humancancers wasinvestigated using an anti-PD-L1 immu-
`(complete and partial responses) were observed in 32 of 175 (18%), 11
`nohistochemistry (IHC) antibody optimizedforstaining offormalin-fixed
`of53 (21%), 11 of43 (26%), 7 of56 (13%) and 3 of 23 (13%) ofpatients
`paraffin-embeddedtissue samples. Stainingofpre-treatment specimens
`with all tumourtypes, non-small cell lung cancer (NSCLC), melanoma,
`submitted for ourclinical study demonstrated expression across a range
`renalcell carcinoma and other tumours(including colorectal cancer, gas-
`ofcancers (Fig. la), PD-L1 staining was observed on tumour cells, as well
`tric cancer, and head and neck squamouscell carcinoma), respectively.
`as on tumour-infiltrating immunecells (Fig. 1b), with PD-L1-positive
`
`'Yale Comprehensive Cancer Center, Yale School of Medicine, 333 Cedar Street, WWW2?21, New Haven, Connecticut 06520, USA, “Gustave Roussy South-Paris University, 1 14 Rue Edouard Vaillant, 94805
`Villefuij, Cedex, France, *Genentech,Inc, 1 DNA Way, South San Francisco, California 94080, USA. “The Angeles Clinic and Research Institute, 11818 Wilshire Blvd, Los Angeles, California 90025, USA.
`“Pinnacle Oncology Hematology, 9055 E Del Camino Dr 100, Scottsdale, Arizona 85258, USA, °Vanderbilt-Ingram Cancer Center, 2220 Pierce Avenue, Nashville, Tennessee 37212, USA. "BethIsrael
`Deaconess Medical Center, 330 Brookline Avenue, Shapiro 9, Boston, Massachusetts 02215, USA, “Carolina BioOncology Institute, 9801 W. Kincey Ave, Suite 145, Huntersville, North Carolina 28078, USA.
`Stanford University, CCSR Bldg Room 1110,Stanford, California 94305, USA. !°Vanderbilt-Ingram Cancer Center, 1301 Medical Center Dr, Suite 1710, Nashville, Tennessee 37212, USA. “Massachusetts
`General Hospital, 55 Fruit Street, YAW 9E, Boston, Massachusetts 02114, USA. '*Dana-Farber/Brigham and Women's Cancer Center, 450 Brookline Avenue, Boston, Massachusetts 02215, USA.
`
`27 NOVEMBER 2014 | VOL 515 | NATURE | 563
`
`

`

`ayaLETTER |
`
`@==POD-L1 prevalence determined with a Genentech/Roche anti-PD-L1 IHC assay
`Figure 1 | Programmed death-ligand 1 (PD-L1) prevalence andexpression,
`
`indication
`a
`Percentage of PD-L1 positive (IC)
`Percentage of PD-L1 positive (TC)
`a, PD-L1 prevalence by immunochistochemistry (IHC) in samples collected for
`NSCLG
`184
`26
`a4
`ROG
`88
`25
`10
`PCD4989g. PD-L1 positivity was defined as =5%of tumour-infiltrating
`Melanoma
`58
`36
`5
`immunecells (ICs) or tumour cells (TCs) staining for PD-L1 by IHC.
`HNSCG
`101
`28
`19
`Gastric cancer
`144
`18
`5
`CRC
`v7
`a5
`1
`b, Representative images of PD-L1 by IHC (brown) in tumours frompatients
`Pancreatic cancer
`Ls)
`12
`4
`with non-small cell lung cancer (NSCLC). The PD-L1-negative imageis at
`20X magnification, other images at 40% magnification. ¢, Co-localization
`of PD-L1 withselected tumour-infiltrating immunecell and tumour cell
`markers by immunofluorescence in NSCLCand melanoma tumours. PD-L1
`staining in red; markers of tumour-infiltrating immunecells and tumour
`cells in green; and DAPI stainingin blue. Areas of overlap are indicated with
`white arrowheads. All four images are at 40 resolution. Markers of tumour-
`infiltrating immunecells: CD163 (macrophages), CD11c (dendritic cells)
`and CD3 (Tcells). Marker of tumourcells: cytokeratin (CK). CRC, colorectal
`cancer; HNSCC, head and neck squamouscell carcinoma; NSCLC, non-small
`cell lung cancer; RCC,renal cell carcinoma.
`
`?
`
`. ‘
`f &
`PD-L1 negatives ©
`
`c
`
`f
`q
`E
`4
`7)
`
`c
`le
`
`tae
`wy
`hom
`Fa
`t
`Peee
`PD-L1 positive (IC)
`= =*
`PD-LVCD1ES
`
`PO-L1/C03
`
`
`PO-LVCK
`PO-LVCONIC
`
`
`of responses observed because somepatients who had a best response
`of progressive disease per RECIST went on to develop durable tumour
`shrinkage orprolongedstable disease (pseudoprogression)'’. The med-
`ian progression-free survivalofall patients was 18 weeks. Wealso per-
`formed anexploratory analysis ofpatients with NSCLC anddetected a
`potential trend offormer/current smokers responding better to MPDL-
`3280A versus never smokers (11 of 26 (42%) versus 1 of 10 (10%), re-
`spectively; P = 0.4229using a Fisher exact test; see the accompanying
`paper(ref. 18) for further discussion).
`There appearsto be anassociation between response andthe expres-
`sion ofPD-L1in pre-treatment samples(Fig. 3 and Extended Data Figs 3
`and 4). The association of response to MPDL3280A treatment and
`tumour-infiltrating immunecell PD-L1 expression reachedstatistical
`significance (NSCLC,P = 0.015 (Fig. 3a and ExtendedDataFig.4a); all
`tumours, P = 0.007(Fig. 3b, cand Extended DataFig. 4b)), while the as-
`sociation with tumourcell PD-L1 expression did not (NSCLC, P = 0.920
`(Extended Data Fig. 4c); all tumours, P = 0.079 (Extended Data Fig. 4d)).
`
`
`
`
`
`
`
`Any grade (n (%))
`
`Grade 3-4 (n (%))
`
` Events (=4% ofpatients) Any grade(n (95)) Grade 3-4 (n (%)) Events (=5%ofpatients)
`
`Four more patients had unconfirmed responses(Table 2, Fig. 2 and Ex-
`tended Data Fig. 3a). Responses could also be rapid and durable (Fig. 2b
`and Extended DataFig. 3a), with shrinking orresolving palpable lesions
`detected within days in some responders andnearlyall responders(es-
`pecially patients with NSCLC) continuing to respond andstaying on
`study. In addition, RECIST maynot accurately describethefull spectrum
`Table 1 | Adverse events
`Treatment-related AEs (n = 277)
`AEsregardless of attribution (n = 277)
`
`108 (39.0)
`263 (94.9)
`Any AE
`194 (70.0)
`35 (12.6)
`Any AE
`9 (3.2)
`100 (36.1)
`Fatigue
`67 (24.2)
`5 (1.8)
`Fatigue
`2 (0.7)
`69 (24.9)
`Decreased appetite
`33 (11.9)
`-
`Nausea
`11 (4,0)
`66 (23.8)
`Nausea
`32 (11.6)
`1 (0.4)
`Dyspnoea
`~
`64 (23.1)
`Pyrexia
`32 (11.6)
`=
`Decreased appetite
`=
`60 (21.7)
`Diarrhoea
`29 (10.5)
`7
`Cough
`_
`60 (21.7)
`Rash
`29 (10.5)
`-
`Diarrhoea
`-
`57 (20.6)
`Pruritus
`23 (8.3)
`—
`Pyrexia
`-
`55 (19.9)
`Arthralgia
`22 (7.9)
`=
`Constipation
`-
`49 (17.7)
`Headache
`21 (7.6)
`1 (0.4)
`Headache
`-
`46 (16.6)
`Chills
`19 (6.9)
`-
`Vomiting
`10 (3.6)
`44 (15.9)
`Influenza-like illness
`16 (5.8)
`1 (0.4)
`Anaemia
`-
`43 (15.5)
`Asthenia
`15 (5.4)
`2 (0.7)
`Insomnia
`4(1.4)
`42 (15.2)
`Dyspnea
`15 (5.4)
`2 (0.7)
`Back pain
`-
`41 (14.8)
`Pain
`15 (5.4)
`1 (0.4)
`Arthralgia
`-
`40 (14.4)
`Myalgia
`13 (4.7)
`-
`Rash
`4(1.4)
`34 (12.3)
`Anaemia
`12 (4.3)
`2 (0.7)
`Asthenia
`-
`33 (11.9)
`Dry skin
`12 (4.3)
`=
`Pruritus
`-
`31(11.2)
`Night sweats
`12 (4.3)
`-
`Chills
`-
`30 (10.8)
`Vomiting
`11 (4.0)
`1 (0.4)
`Upper respiratory tract infection
`-
`20 (7.2)
`Other grade 3-4 AEs, =2 patients
`Anxiety
`-
`20 (7.2)
`ALT increased
`Influenza-like illness
`-
`20 (7.2)
`ASTincreased
`Nasal congestion
`3 (1.1)
`20 (7.2)
`Hypoxia
`Urinary tract infection
`4(1.4)
`19 (6,9)
`Hyperglycaemia
`Dehydration
`7 (2.5)
`19 (6.9)
`Hyponatraemia
`Hyperglycaemia
`~
`19 (6.9)
`Cardiac tamponade
`Myalgia
`~
`19 (6.9)
`Hypophosphataemia
`Night sweats
`-
`19 (6.9)
`Tumour lysis syndrome
`Productive cough
`-
`16 (5.8)
`Dry skin
`-
`14 (5.1)
`Dry mouth
`6 (2.2)
`14 (5,1)
`Hypoxia
` Weight decreased - 14 (5.1)
`
`
`
`AE, adverse event; ALT, alanine aminotransferase; AST, aspartate aminotransferase.
`
`6 (2.2)
`4(14)
`4(14)
`4(1.4)
`4(1.4)
`2 (0.7)
`2 (0.7)
`2 (0.7)
`
`3 (1.1)
`3 (1.1)
`3 (1.1)
`2 (0.7)
`2 (0.7)
`2 (0.7)
`2 (0.7)
`2 (0.7)
`
`$64 | NATURE | VOL 515|27 NOVEMBER 2014
`
`

`

`LETTER
`
`
`
`Individual patients with NSCLC receiving MPDL3280A
`
`50
`
`0
`
`-50
`
`-100
`
`85
`
`a 2 aE
`
`3
`
`by Histology
`Non-squamous
`Squamous
`Non-squamous
`Nen-squamous
`Non-squamous
`Squamous
`Noa-squamous
`Squamous
`Non-squamous
`Nen-squamous
`Non-squamous
`Nen-squamous
`
`e
`
`
`
`(HG (1G)
`IHCO
`IHG 3
`IHC O
`IHC 1
`IHG O
`IHG 2
`IHC 3
`IHC
`IHC 0
`IHC 3
`IHC 1
`
`@ On study, on treatment
`On study, post treatment
`
`@ Treatment discontinuation
`» Ongoing response
`
`First response
`
`First PD
`
`0 3 6
`9 12 15 1821 24 27 3 33 WG 99 42 45 48 51 84 S7 60 63 66 69 72 75 Th Al 4
`
`68 (39)
`
`65(37)
`
`33(19)
`
`42
`
`18 (34)
`
`21 (40)
`
`9(17)
`
`16 (38)
`
`16 (38)
`
`7(q17)
`
`2 (18)
`
`5 (46)
`
`2 (18)
`
`11 (26)
`
`18 (42)
`
`4(9)
`
`9 (27)
`
`11 (33)
`
`4(12)
`
`0
`
`1 (25)
`
`4 (80)
`
`3 (75)
`
`0
`
`0
`
`30 (54)
`
`17(30)
`
`18(32)
`
`28 (57)
`
`14 (29)
`
`18 (37)
`
`45
`
`44
`
`46
`
`Al
`
`51
`
`20
`
`0
`
`48
`
`52
`
`Table 2 | Efficacy of MPDL3280Aacross tumour types
`a
`_ 100
`
`
`Tumour types ORR®* (7 (95))=SD (best PD (best SD =24 24-week
`
`
`ze
`NSCLC
`(95% Ch
`response)
`response)
`weeks
`PFS (9%)
`=
`a Non-squamous.
`
`(n (5)
`(n (%))
`(n (%))
`8
`
`100
`
`50
`
`Time (weeks)
`
`®@ Melanoma
`m= NSCLC
`RCC
`@ Other
`
`
`
`MaximumSLDreductionfrombaseline(%)
`
`Overall (n= 175)
`
`NSCLC (n= 53)
`
`Non-squamous
`(n= 42)
`Squamous (n= 11)
`
`Melanoma(n=43)
`
`Cutaneous (n=33)
`
`Mucosal (n = 5)
`
`Ocular (n= 4)
`
`RCC (n= 56)
`
`Clear cell (n = 49)
`
`36 (21)
`(15-27)
`12 (23)
`(12-35)
`9(21)
`(11-36)
`3 (27)
`(8-61)
`13(380)
`(18-45)
`12 (36)
`(21-55)
`1 (20)
`(1-66)
`0
`(0-53)
`8 (14)
`(6-25)
`7(14)
`(7-27)
`1(14)
`(1-55)
`3 (13)
`(4-32)
`
`2 (29)
`
`3 (43)
`
`0
`
`969)
`
`9 (39)
`
`2 (9)
`
`ce
`
`17
`
`24
`
`
`
`Non-clearcell
`(n=7)
`Other (for example,
`CRC, GC and
`HNSCC; n= 23)¢
`Patients dosed by 1 October 1 2012, with=1mgkg- ‘with a baseline tumour assessment. Data cutoff
`was 30 April 2013.Cl, confidence interval; CRC,colorectal cancer; GC, gastric cancer; HNSCC, head and
`neck squamous cell carcinoma; NSCLC, non-small cell lung cancer; ORR,objective response rate; PD,
`progressive disease; PFS, progression-free survival; RCC,renal cell carcinoma; SD, stable disease.
`* Per RECISTv1.1. All responses were confirmed exceptfor in one patient with NSCLC, one patient with
`RCC and two patients with melanoma.
`+ Sarcoma(n= 2), ovarian (n= 1), head and neck (n = 6), breast (n = 3), colorectal (n = 6), pancreatic
`(n= 1), gastric (n = 1), oesophageal (n = 1), uterine (n = 1) and pancreatoduodenal(n = 1).
`
`For example, 83% ofpatients with IHC score 3 (tumour-infiltrating im-
`munecell) NSCLC respondedto treatments with only 17% progressing,
`whereas 43% ofpatients with IHC 2 (tumour-infiltrating immunecell)
`NSCLC were limited to diseasestabilization (Fig. 3a and Extended Data
`Fig. 4a; see Methodsfor the IHC score definitions), Ofthe patients with
`IHC 3 (tumourcell) NSCLC,only 38% (3 of8) responded while 38% (3
`of8) progressed (Extended Data Fig. 4c). There wasalso a trend between
`tumour IHC status and median progression-free survival (Fig. 3c).
`When tumoursamples were examinedfor the expressionof differ-
`ent immuneinhibitory factors (see Methods), the expected correlation
`with lack of response to MPDL3280A was not seen (Extended Data
`Fig, 5a, left panel). Instead there wasa trend towardsincreased response
`in PD-L1-positive patients expressing a secondnegative regulator (Ex-
`tended DataFig. 5a, right panel). High PD-L2 expressiondid not appear
`to be associated with resistance to MPDL3280A,and somepatients whose
`pre-treatmenttumourbiopsies showedthehighestlevels of PD-L2 ex-
`pression experienced strong responses to MPDL3280A(for example,
`maximum sum ofthe longest diameter (SLD)decreases of57%, 41% and
`49%). Finally, the expression ofCTLA4 andfractalkinein pre-treatment
`tumours appearedto correlate strongly with either response (CTLA4)
`orprogression(fractalkine) after MPDL3280A (Extended Data Fig, 5b).
`Wecomparedresults obtained for pre-treatment NSCLC tumours
`with thosefor renal cell carcinoma and melanoma(Extended DataFig.6).
`Althoughthe expression ofPD-L1 in MPDL3280A-responsivepatients
`was a commonfeature, other aspects ofthe immune microenvironment
`appeareddifferent. In melanoma, pre-treatment tumours in respond-
`ing patients demonstrated elevated expressionofIFN-y as well as IFN-
`y-inducible genes (for example, [DO] and CXCL9). Theseassociations
`were weakerin patients with NSCLCorrenal cell carcinoma.
`To characterize the immunological events associated with tumourre-
`sponse orprogression,serial on-treatment tumourbiopsies were per-
`formed in 28 patients (Fig. 4a). After treatment, regressinglesions displayed
`a dense immuneinfiltrate and extensive tumour cell necrosis accom-
`panied by the apparentsterilization ofcancercells in somecases (Extended
`Data Fig. 7a, b). A decrease in tumour SLD appeared to be accompanied
`
`
`
`Individual patients with all tumour types receiving MPDL3280A
`
`Figure 2 | Antitumour activity of MPDL3280A.a,A waterfall plot ofpatients
`with non-small cell lung cancer (NSCLC) measuring the maximumreduction
`frombaseline in the sum of the longest diameter (SLD)fortarget lesions; +20%
`and —30% are marked by dashedlines. b, The time to response (Response
`Evaluation Criteria in Solid Tumours version 1.1) and the duration of study
`treatment for patients with NSCLC. The patient with progressive disease
`(PD) experienced ongoingclinical benefit as judged by the investigator. All but
`one response was confirmed. ¢, A waterfall plot of patients with all tumour
`types measuring the maximum reduction from baseline in the SLDfortarget
`lesions; +20% and —30% are marked by dashed lines. IC, tumour-infiltrating
`immunecells; IHC, immunohistochemistry; RCC,renal cell carcinoma.
`
`byan increase in PD-L] expression on tumour-infiltrating immunecells
`and tumourcells (Fig. 4a). The increase in PD-L1 expressionwith treat-
`ment correlated with changes in tumour IFN-y expression (Pearson cor-
`relation coefficient = 0.70; Extended Data Fig. 5c). In addition, RNA
`isolated from regressing lesions wasanalysed for the presence of tran-
`scripts ofimmunological importance using a Fluidigm-based ‘immuno-
`chip’ (iChip, see Methods), and displayed expression patternsindicative
`ofa generalized activation ofCD8 and Ty1 T-cell responses (Extended
`Data Fig, 7c).
`In contrast, most progressing patients with on-treatment biopsies
`showeda lack ofPD-L1 upregulation by either tumour cells or tumour-
`infiltrating immunecells. These growing tumoursdisplayed oneofthree
`pattern