Immunologic Consequences of Sequencing Cancer Radiotherapy and Surgery.

SPECIAL SERIES: MATHEMATICAL ONCOLOGY o iginal epo Immunologic Consequences o Sequencing Cance Radio he apy and Su ge y Juan Ca los L´ opez Al onso, PhD 1 ; Jan Poleszczuk, PhD 2 ; Rachel Walke , PhD 3 ; Sungjune Kim, MD, PhD 3 ; Sha i Pilon-Thomas, PhD 3 ; Jose J. Conejo-Ga cia, MD, PhD 3 ; Ha em Soliman, MD 3 ; B ian Cze niecki, MD 3 ; Louis B. Ha ison, MD 3 ; and Heiko Ende ling, PhD 3 abs ac PURPOSE Ea ly-s age cance s a e ou inely ea ed wi h su ge y ollowed by adio he apy (SR). Radio he apy be o e su ge y (RS) has been widely igno ed o some cance s. We e alua e o e all su i al (OS) and disease- ee su i al (DFS) wi h SR and RS o di e en cance ypes and simula e he plausibili y o RS- and SR-induced an i umo immuni y con ibu ing o ou comes. MATERIALS AND METHODS We analyzed a SEER da a se o ea ly-s age cance s ea ed wi h SR o RS. OS and DFS we e calcula ed o cance s wi h su ficien numbe s o s a is ical powe (cance s o lung and b onchus, esophagus, ec um, ce ix u e i, co pus u e i, and b eas ). We simula ed he immunologic consequences o SR, RS, and adio he apy alone in a ma hema ical model o umo -immune in e ac ions. RESULTS RS imp o ed OS o cance s wi h low 20-yea su i al a es (lung: haza d a io [HR], 0.88; P=.046) and imp o ed DFS o cance s wi h highe su i al (b eas : HR = 0.64; P,.001). Fo ec al cance , wi h in e media e 20-yea su i al, RS imp o ed bo h OS (HR = 0.89; P= .006) and DFS (HR = 0.86; P=.04). Model simula ions sugges ed ha RS could inc ease OS by elimina ing cance o a b oade ange o model pa ame e s and adio he apy-induced an i umo immuni y compa ed wi h SR o selec ed pa ame e combina ions. This could c ea e an immune memo y ha may explain inc eased DFS a e RS o ce ain cance s. CONCLUSION S udy esul s sugges plausibili y ha adia ion o he bulk o he umo could induce a mo e obus immune esponse and be e ha ness he syne gy o adio he apy and an i umo immuni y han pos su gical adia ion o he umo bed. This explo a o y s udy p o ides mo i a ion o p ospec i e e alua ion o immune ac i a ion o RS e sus SR in con olled clinical s udies. Clin Cance In o m. © 2019 by Ame ican Socie y o Clinical Oncology INTRODUCTION Su ge y ollowed by adio he apy (SR) imp o es loco egional con ol in nume ous locally ad anced cance s, 1-3 making adio he apy a i al componen o in en - o-cu e cance ea men . 4 Radio he apy ol- lowed by su ge y (RS) e sus su ge y alone has demons a ed imp o ed ou comes in bladde , ce - ical, esophageal, ec al, and esec able panc ea ic cance s. Clinical ials compa ing SR and RS ou - comes a e sca ce and ha e been limi ed o a ew cance ypes. One success ul example is ec al ca cinoma, 5,6 whe e imp o ed local con ol a e RS e sus SR (wi h o wi hou concu en chemo he apy) was demons a ed and esul ed in a pa adigm shi om SR o RS. Howe e , hese ials ha e ailed o demons a e supe io o e all su i al (OS) a e RS. Simila ly, in a p ospec i e ial o muscle-in asi e bladde cance , 7 no significan di e ence in 3-yea s a is ics be ween RS and SR was ound. A sligh inc ease in OS o pa ien s unde going RS compa ed wi h SR was obse ed in so issue sa coma o he limbs. 8 To da e, he ew published s udies compa ing adio- he apy and su ge y sequencing seem inconclusi e. A SEER analysis showed ha RS can be associa ed wi h a significan ly highe isk o dea h in Siewe ype II gas oesophageal junc ion cance when compa ed wi h SR. 9 Howe e , o locally ad anced esophageal cance , and squamous cell ca cinoma in pa icula , supe io i y o RS o e SR wi h concu en chemo- he apy was demons a ed. 10 Re ospec i e analyses ha e also shown supe io OS a e RS in s age III non–small-lung cance 11 and in cT2 noninflamma o y b eas cance wi h concu en chemo he apy. 12 The e a e also some indica ions ha RS can e- duce he isk o local ecu ence in localized so issue sa coma. 13 I is inc easingly app ecia ed ha adio he apy in- c eases he mu a ional bu den and induces cell s ess ASSOCIATED CONTENT Appendix Da a Supplemen Au ho a ilia ions and suppo in o ma ion (i applicable) appea a he end o his a icle. Accep ed on No embe 27, 2018 and published a ascopubs.o g/jou nal/ cci on Ap il 9, 2019: DOI h ps://doi.o g/10. 1200/CCI.18.00075 1 Downloaded om ascopubs.o g by H. Lee Mo i Cance Cen e and Resea ch Ins i u e on Ap il 18, 2019 om 206.081.166.225 Copy igh © 2019 Ame ican Socie y o Clinical Oncology. All igh s ese ed. as well as immunogenic cell dea h, he eby exposing a weal h o p e iously hidden and de no o umo -associa ed an igens, s ess p o eins, and dange -associa ed molecula pa e ns o he immune sys em. 14-18 Tumo -infil a ing lym- phocy e en ichmen a e adio he apy was p e iously assessed in 40 pa ien s wi h ec al cance . The densi ies o CD3 + and CD8 + T-lymphocy es significan ly inc eased om p e adio he apy biopsy specimens o pos adio he apy su gically esec ed specimens. 19 I is concei able ha immune- ela ed benefi s o adia ion o he bulk umo compa ed wi h adia ion o he pos su gical ca i y could be gene al phenomena e en in ea ly-s age disease. He ein, we e alua e OS and disease- ee su i al (DFS) ou come da a o a ious cance s and p esen a quan i a i e amewo k o simula e umo –immune sys em dynamics du ing he di - e en ea men sequences, which u he suppo s he plausibili y o inc eased an i umo immuni y wi h RS com- pa ed wi h SR. MATERIALS AND METHODS SEER Analysis We que ied he SEER da abase (No embe 2016 sub- mission wi h addi ional ea men in o ma ion) o local- ized cance s (no lymph node in ol emen and no me as asis; N0M0) wi h a leas 100 cases o each o RS and SR and a ailable in o ma ion abou po en ially con- ounding co a ia es (Fig 1A). These included cance s o he lung and b onchus, esophagus, ec um, ce ix u e i, co pus u e i, and b eas . We selec ed pa ien s who e- cei ed ei he RS o SR and censo ed he ollow-up pe iod o 20 yea s. DFS was calcula ed as p e iously discussed. 19a Haza d a ios (HRs) o DFS and OS adjus ed o age, sex, yea o diagnosis, his ology, ype o su ge y, umo size, and ea men sequencing (RS SR) we e calcula ed using mul i a iable Cox p opo ional haza d models. 20 Modeling o Tumo –Immune Sys em Dynamics Du ing SR and RS Ma hema ical modeling o umo –immune sys em in- e ac ions has a long his o y. 21,22 The es ablished Kuzne so model cap u es fi s -o de p inciples un- de lying he complex disease dynamics and has been calib a ed agains expe imen al da a. 21 This model elies on he ollowing main assump ions: umo s ollow logis ic g ow h dynamics, whe e ini ially exponen ial g ow h decele a es as he umo app oaches he issue-ca ying capaci y; umo g ow h is modula ed by he cy o oxic ac ion o immunocompe en e ec o T cells as pa o specific adap i e immune esponses; cy o oxic e ec o T cells a e ec ui ed in esponse o umo bu den ol- lowing Michaelis-Men en dynamics; umo -infil a ing e ec o T cells may ge exhaus ed by hei an i umo ac i i y and unde go spon aneous dea h; and inna e immuni y o base immune su eillance is ep esen ed as a baseline p esence o e ec o T cells a any ime, e en in he absence o umo cells. We ex ended he Kuzne so model o explo e he po- en ial benefi s and immunologic consequences o a- dio he apy and su ge y sequencing, on he basis o he ollowing addi ional assump ions: adio he apy kills bo h cance and e ec o T cells 23,24 a di e en a es de e mined by he adia ion dose and cell ype–specific adiosensi i i y by some o ms o cell dea h, including au ophagy, apop osis, nec osis, mi o ic ca as ophe, and senescence; i adia ed umo cells may unde go immunogenic cell dea h ha emi s immunos imula ing signals and chemokines, esul ing in he ec ui - men o inflamma o y immune cells o he umo mi- c oen i onmen , including an igen-p esen ing cells such as dend i ic cells and mac ophages, which in u n ec ui and ac i a e cy o oxic e ec o T cells 15,25-28 ; he s eng h o adio he apy-induced an i umo e- sponse, cha ac e ized by subsequen immune cell infil a ion o he umo , is assumed o depend on he numbe o umo cells killed by adio he apy; and su gical esec ion is simula ed by ins an a- neously dec easing bo h cance and e ec o cell popula ions. We deno ed he numbe o iable cance cells (o clo- nogens) in an a bi a y umo olume as C( ), umo - infil a ing ac i a ed immune e ec o T cells as E( ), and i adia ed cance cells unde going immunogenic cell dea h as D( ). The model was o mula ed as a sys em o o dina y di e en ial equa ions gi en by dC/d = ( umo g ow h) − (dea h o umo cells by e ec o T cells) −( adio he apy) − (su gical esec ion), o dC d  C1−C+D k−apEC −δτiTR ,d,C −δτjTS ,C(1) by dE/d = ( umo bu den–s imula ed e ec o T-cell e- c ui men ) −(exhaus ion o e ec o T cells by hei an i- umo ac ion) + (physiologic le el and decay o e ec o T cells) + ( adio he apy-induced immunos imula ion) − ( adio he apy) −(su gical esec ion), o dE d  C g+CE−a1−pEC +hEp−E+qD −δτiTR ,d,E−δτjTS ,E(2) and by dD/d = ( umo cell s e ilized by adio he apy) − (lysis o umo cells s e ilized by adio he apy), o dD d δτiTR ,d,C−nD (3) whe e he ime coo dina e on he sys em a iables was omi ed o no a ional simplici y. δ(τ)is heDi acdel a unc ion. Fo demons a ion pu poses, we chose p e iously epo ed specific model pa ame e alues ha simula e a g owing umo and associa ed e ec o cell dynamic in he L´ opez Al onso e al 2© 2019 by Ame ican Socie y o Clinical Oncology Downloaded om ascopubs.o g by H. Lee Mo i Cance Cen e and Resea ch Ins i u e on Ap il 18, 2019 om 206.081.166.225 Copy igh © 2019 Ame ican Socie y o Clinical Oncology. All igh s ese ed. absence o he apy (Table 1). Fo analysis, we changed model pa ame e s ha may ep esen he pa ien -specific biology o indi idual pa ien s, including di e en human leukocy e an igen haplo ypes o inc eased mu a ional bu den o di e en cance s ha may ansla e in o highe immunogenici y. The cy o oxic e ec o adio he apy on cance cells was simula ed using he linea quad a ic (LQ) model ha is A Reco ds o pa ien s wi h cance o b eas , ce ix u e i, co pus u e i, ec um, esophagus, o lung and b onchus (N = 3,483,681) Excluded in o al (n = 3,254,384) (n = 583,418) (n = 1,851,801) (n = 708,477) (n = 7) (n = 1,028) (n = 28,403) (n = 5,273) (n = 3,706) (n = 72,271) P io cance Regional o dis an disease Did no ecei e RT in adju an o neoadju an se ing Unknown age a diagnosis Unknown ace Incomple e da es Missing da a on second p ima y Unable o calcula e cance - ee su i al Unknown/unde ined alue o ≥ 1 co a ia e in mul i a iable analysis B 5101520 Follow-Up (yea s) Su i al (p obabili y) 0 0.25 0.50 0.75 1.00 B eas Ce ix u e i Co pus u e i Rec um Esophagus 203,151 124,384 59,965 19,284 5,380 2,769 1,461 738 252 81 11,940 5,709 2,652 1,068 355 8,136 4,316 1,767 480 129 795 249 107 31 3 2,506 752 251 61 15 No. a isk: Lung and b onchus Su ge y ollowed by adia ion (SR) B eas Ce ix u e i Co pus u e i Rec um Esophagus Lung and b onchus (n = 221,231) (n = 201,631) (n = 2,444) (n = 11,745) (n = 3,142) (n = 189) (n = 2,080) B eas Co pus u e i Rec um Lung and b onchus Ce ix u e i Esophagus Radia ion ollowed B eas Ce ix u e i Co pus u e i Rec um Esophagus Lung and b onchus (n = 8,066) (n = 1,520) (n = 325) (n = 195) (n = 4,994) (n = 606) (n = 426) by su ge y (RS) 0.56 0.54 0.37 0.30 0.13 0.04 FIG 1. (A) SEER in- clusion/exclusion. (B) Kaplan-Meie 20-yea su i al. Sequencing Cance Radio he apy and Su ge y JCO Clinical Cance In o ma ics 3 Downloaded om ascopubs.o g by H. Lee Mo i Cance Cen e and Resea ch Ins i u e on Ap il 18, 2019 om 206.081.166.225 Copy igh © 2019 Ame ican Socie y o Clinical Oncology. All igh s ese ed. widely used in clinical applica ions 32-34 o app oxima e he dose-dependen su i ing ac ion SF(d) o i adia ed cance cells by SFde−ξ(αd+βd2)(4) whe e d(Gy) is he adia ion dose and α(Gy −1 ) and β(Gy −2 ) a e cell ype–specific adiosensi i i y pa ame e s. Com- pelling e idence demons a es ha hypoxic (poo ly oxy- gena ed) cance cells a e g ow h a es ed and es ima ed o be app oxima ely h ee imes mo e esis an o adia ion han no moxic cycling cells. 30,35 We se ξ=1o ξ= 1/3 o espec i ely scale he adiosensi i i y o p oli e a i e and quiescen cance cells as p e iously demons a ed. 30,36 We se adiosensi i i y pa ame e s α=0.3Gy −1 and β=0.03Gy −2 as con en ionally assumed and es ima ed o a a ie y o umo ypes. 29 The p obabili ies o su i ing 2 Gy adia- ion we e SF (2 Gy) = 0.49 o p oli e a ing cance cells (ξ= 1) and SF (2 Gy) = 0.79 o quiescen cance cells (ξ=1/3). E ec o T-cell adiosensi i i y was es ima ed on he basis o expe imen al da a o adio he apy-induced apop osis in lymphocy es ob ained om blood samples. 37 Apo- p osis is conside ed one o he dominan cell dea h p ocesses in esponse o adio he apy, 38 and co ela ion be ween he in ensi y o apop osis in lymphocy es and adia ion dose has been epo ed. 37,39 F om dose- esponse cu es o e ec o CD8 + T cells in i o a e exposu e o acu e doses o 0 o 8 Gy, 37 we de i ed SF E (2 Gy) = 0.61. Radio he apy was simula ed wi h a o al dose o 50 Gy deli e ed in 25 weekday ac ions o d= 2 Gy pe day. The o al dose o 50 Gy was chosen o demons a ion pu poses in line wi h he s anda d dose o b eas cance , he la ges pa ien coho in he SEER da a. Ex ension o he analysis o o he o al doses was s aigh o wa d and did no al e he esul s o his s udy (Appendix Fig A1). A e logis ic umo g ow h and he LQ model in Equa ion 4, 32,40 he ac ion o cance cells (C) s e ilized by adio he apy wi h dose da ime τ i was de e mined by he loss e m T R (τ i ,d,C)in Equa ion 1 gi en by TRτi,d,C1−SFpdCτi1−Cτi+Dτi k +1−SFqdCτiCτi+Dτi k (5) whe e he e ms C(τi)1−C(τi)+D(τi) kand C(τi)C(τi)+D(τi) ka e he numbe o p oli e a ing and quiescen cance cells a ime τ i such ha hei sum is equal o C(τ i ). The adiobi- ologic e ms 1 −SF p (d) and 1 −SF q (d) ep esen e- spec i ely he ac ion o p oli e a ing and quiescen cance cell s e ilized by adio he apy wi h dose dand de e mined by he LQ model in Equa ion 4. TABLE 1. O e iew o Model Pa ame e s and Values Used in Simula ions Pa ame e Desc ip ion Value Uni Fi s Au ho Tumo g ow h a e (1.0 - 4.0) ×10 −1 Day −1 Poleszczuk 22 kTumo issue ca ying capaci y 5.32 ×10 8 Cells Poleszczuk 22 aT cell–cance cell in e ac ions cons an 0.14 ×10 −6 Day −1 cells −1 Poleszczuk 22 pP obabili y ha du ing T cell–cance cell in e ac ions he la e is killed 9.98 ×10 −1 Dimensionless Poleszczuk 22 Magni ude o immune sys em s imula ion by he p esence o cance cells (1.0 - 3.0) ×10 −1 Day −1 Poleszczuk 22 gImmune s imula ion damping coe ficien 0.16 ×10 6 Cells Poleszczuk 22 hE ec o cells decay a e 5.9 ×10 −1 Day −1 Poleszczuk 22 E* Physiologic le el o e ec o cells 0.3 ×10 6 Cells Poleszczuk 22 αRadia ion sensi i i y pa ame e 0.3 Gy −1 Gue e o 29 βRadia ion sensi i i y pa ame e 0.03 Gy −2 Gue e o 29 ξCance cell adia ion esis ance 1 (p oli e a ing cells) Dimensionless Ende ling 30 1/3 (quiescen cells) qRT-induced immunos imula ion (0.0 −6.5) ×10 −1 Day −1 Assumed o demon- s a ion pu pose nLysis a e o umo cells killed by RT 6.9 ×10 −1 Day −1 E imie 31 Abb e ia ion: RT, adio he apy. L´ opez Al onso e al 4© 2019 by Ame ican Socie y o Clinical Oncology Downloaded om ascopubs.o g by H. Lee Mo i Cance Cen e and Resea ch Ins i u e on Ap il 18, 2019 om 206.081.166.225 Copy igh © 2019 Ame ican Socie y o Clinical Oncology. All igh s ese ed. The loss e m T R (τ i ,d,E) in Equa ion 2 simula es he killing e ec s o adio he apy on e ec o T cells (E) a ime τ i and is gi en by TRτi,d,E1−SFEdEτi(6) Su gical esec ion was simula ed by ins an aneously de- c easing bo h cance and e ec o cell popula ions by 99%. A he ime o su ge y τ j , emo al o cell popula ions was inco po a ed in he model by he ollowing loss e ms TSτj,CλCCτj(7) TSτj,EλEEτj(8) whe e su gical esec ion was simula ed wi h λ C =λ E = 0.99. A e su gical esec ion, he dynamics o emaining cell popula ions, i any, con inued o be go e ned by Equa ions 1 o 3. Model simula ions we e ini ialized wi h 10 6 iable cance cells (o clonogens) and 10 5 e ec o T cells in an a bi a y umo olume. Tumo –immune sys em in e ac ion dynamics we e simula ed o di e en combina ions o umo g ow h a e ( ) and ec ui men o e ec o T cells in esponse o umo bu den ( ). Combina ions o pa ame e s and ha yielded an inc ease o cance cells .10 8 we e conside ed o ea men simula ions. No e ha his numbe ep esen s iable cance cells in a umo ha may ha e conside ably la ge adiog aphic olumes. Tumo con ol (TC) was assumed i he numbe o iable cance cells d opped below C( ) ,1. The ma hema ical model was simula ed using Ma lab (www.ma hwo ks.com). RESULTS OS and DFS o RS and SR Cance si es we e e alua ed o 20-yea OS (Fig 1B)and DFS (Appendix Fig A2). A end o RS benefi e sus SR became appa en when cance s we e so ed by hei 20-yea OS (DFS eg ession coe ficien , −0.69; P,.05; OS eg ession coe ficien , +0.35; P,.05; Fig 2). Fo cance s o he lung and b onchus (n = 2,506), RS yielded significan lyimp o edOS(HR=0.88;P= .046; 20-yea su i al, 6.2% 2.5%; Fig 2B). Al hough an OS benefi was no obse ed in b eas cance (n = 203,151) wi h high baseline 20-yea OS a es, RS yielded a significan im- p o emen in DFS (HR = 0.64; P,.001; Fig 2A). In- e es ingly, pa ien s wi h ec al cance (n = 8,136) wi h an in e media e 20-yea su i al a e o 30% may benefi om RS, wi h bo h inc eased OS (HR = 0.89; P= .006) and DFS (HR = 0.86; P= .04). C ucially, RS was no ound o be significan ly in e io o SR in any o he mul i a iable Cox p opo ional model analyses (Da a Supplemen ). Analysis o p opensi y sco e–ma ched coho s is shown in Appendix Figu e A3. E ec s o T ea men O de Negligible Wi hou Simula ed Radio he apy-Induced Immuni y The e ec i eness o adio he apy in elimina ing cance may esul om a combina ion o he le hal e ec o adio he apy and s imula ion o an i umo immuni y. We fi s simula ed ea men s wi hou addi ional adio he apy-induced im- muni y (q= 0 day −1 in Equa ion 2). Simula ions o SR and RS yielded simila ou comes, wi h umo con ol (TC) being dependen on umo g ow h a e and p e ea men umo size. In e es ingly, adio he apy alone could no elimina e Lung and b onchus Rec um HR = 0.86 P = .04 B eas HR = 0.64 P < .001 Co pus u e i Esophagus Ce ix u e i 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 0 102030405060 20-Yea OS (%) DFS HR (neoadju an adju an RT) A HRDFS = –0.69* DFS20y + 1.08† 20-Yea OS (%) Lung and b onchus HR = 0.88 P = .046 Rec um HR = 0.89 P = .006 B eas Co pus u e i Esophagus Ce ix u e i 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 0 102030405060 OS HR (neoadju an adju an RT) B HROS = 0.35* OS20y + 0.84† FIG 2. Haza d a ios (HRs; wi h 95% CIs, P alues, and linea leas squa es eg ession lines weigh ed by he in e ses o si e-specific CIs) o (A) disease- ee su i al (DFS) and (B) o e all su i al (OS) a e p eope a i e (neoadju an ) RT (RS) compa ed wi h pos ope a i e (adju an ) RT (SR). HRs we e adjus ed o age, sex, yea o diagnosis, his ology, ype o su ge y, ype o applied adia ion, and umo size. Equa ion desc ibes he end line, and coe ficien s we e es ed o significan di e ence om 0 using es . (*) P,.05. (†)P,.001. RT, adio he apy Sequencing Cance Radio he apy and Su ge y JCO Clinical Cance In o ma ics 5 Downloaded om ascopubs.o g by H. Lee Mo i Cance Cen e and Resea ch Ins i u e on Ap il 18, 2019 om 206.081.166.225 Copy igh © 2019 Ame ican Socie y o Clinical Oncology. All igh s ese ed. umo s wi h as ew as 10 7 p e ea men iable cance cells (Fig 3). RS Imp o es Ou comes Wi h Radio he apy-Induced Immuni y Because adio he apy-induced cy o oxici y alone was un- likely o e adica e all cance cells wi h he chosen adio- sensi i i y pa ame e s, bu adio he apy o en s e ilized la ge umo s wi h p esumably .10 7 iable cance cells, clinically obse ed comple e umo e adica ion may de- pend on he s eng h o adio he apy-induced an i umo immuni y ac i a ion (q, day −1 in Equa ion 2). The e ficacy o adio he apy alone and RS bu no SR inc eased wi h in- c eased adio he apy-induced immuni y (q=0.25 o 0.65 day −1 ). In addi ion, o s ong adio he apy-induced an i umo immuni y, adio he apy alone esul ed in be e ou comes han SR. RS yielded TC o mo e model pa- ame e combina ions han adio he apy alone and SR (Fig 4). Fo SR, inc eased adio he apy-induced immu- nos imula ion had no no able impac on ou comes, be- cause a majo i y o umo cells we e emo ed p ei adia ion, and adio he apy-induced an i umo immuni y was no su ficien ly achie ed. The benefi o adio he apyalone and RS o e SR was independen o he o al numbe o adia ion ac ions (Appendix Fig A1) and insensi i e o changes in o he model pa ame e s (Appendix Figs A4 o A6). RS Induces S onge An i umo Immuni y Than SR A e su ge y, adia ion is deli e ed wi h he in en o e adica e esidual cance cells o mic oscopic umo s in he umo bed beyond he su gical ma gins. Compa ed wi h SR RSRT alone TC by ea men PD 1.0 × 107 1.0 × 108 P e ea men Viable Cance Cells C ( ) A B 108 106 104 102 100 No. o Cells –17 0 34 68 102 Time (days) (RT) 108 106 104 102 100 No. o Cells –20 –10 0 10 20 30 Time (days) (SR) 108 106 104 102 100 No. o Cells –20 –10 0 10 20 30 Time (days) (RS) (day–1) (day–1) (day–1) (day–1) (day–1) (day–1) (day–1) 0.3 0.2 0.1 0.3 0.2 0.1 0.1 0.2 0.3 0.4 0.3 0.2 0.1 0.3 0.2 0.1 TC PD (day–1) (day–1) (day–1) (day–1) (day–1) 0.1 0.2 0.3 0.4 0.3 0.2 0.1 0.3 0.2 0.1 0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4 RTViable cance cells C( ) E ec o cells E( ) FIG 3. (A) Model-p edic ed ea men ou comes o adio he apy (RT) alone, RT a e su gical esec ion (SR), and RT be o e su gical esec ion (RS) wi hou RT-induced immuni y (ie, q= 0.0 day −1 in Equa ion 2) in a coho o i ual pa ien s wi h di e en combina ions o p e ea men umo size, umo g ow h a e ( ), and immune ec ui men a e ( ) in esponse o umo bu den. Radia ion is deli e ed o a o al dose o 50 Gy in 25 daily ac ions a 2 Gy pe day, 5 days pe week. Tumo con ol (TC) by ea men (blue) and p og essi e disease (PD; ed) e e o umo e adica ion and escape a e ea men , espec i ely. (B) Time e olu ion o umo and e ec o T cells co esponding o he loca ion ma ked by s a s in panel A. L´ opez Al onso e al 6© 2019 by Ame ican Socie y o Clinical Oncology Downloaded om ascopubs.o g by H. Lee Mo i Cance Cen e and Resea ch Ins i u e on Ap il 18, 2019 om 206.081.166.225 Copy igh © 2019 Ame ican Socie y o Clinical Oncology. All igh s ese ed. SR RSRT alone TC by ea men PD 2.5 × 10–1 4.5 × 10–1 RT-Induced Immunos imula ion q (day –1) A B 108 106 104 102 100 No. o Cells 108 106 104 102 100 No. o Cells 108 106 104 102 100 No. o Cells 108 106 104 102 100 No. o Cells 108 106 104 102 100 No. o Cells 108 106 104 102 100 No. o Cells 108 106 104 102 100 No. o Cells 108 106 104 102 100 No. o Cells 108 106 104 102 100 No. o Cells –27 0 27 54 81 108 Time (days) (RT) –27 0 27 54 81 108 Time (days) (SR) –27 –14 0 14 27 Time (days) (RS) –27 –14 0 14 27 Time (days) (RT) –27 0 27 54 81 108 Time (days) (SR) –27 –14 0 14 27 Time (days) (RS) –17 0 17 34 6851 Time (days) (RT) –170 17345168 Time (days) (SR) –17 0 17 34 Time (days) (RS) (day–1) (day–1) (day–1) (day–1) (day–1) (day–1) (day–1) (day–1) (day–1) (day–1) 0.3 0.2 0.1 0.3 0.2 0.1 0.1 0.2 0.3 0.4 6.5 × 10–1 0.3 0.2 0.1 0.3 0.2 0.1 TC PD (day–1) 0.1 0.2 0.3 0.4 0.3 0.2 0.1 0.3 0.2 0.1 (day–1) 0.1 0.2 0.3 0.4 (day–1) 0.1 0.2 0.3 0.4 (day–1) 0.1 0.2 0.3 0.4 (day–1) 0.1 0.2 0.3 0.4 RT E ec o cells E( ) Viable cance cells C( ) (day–1) 0.3 0.2 0.1 0.1 0.2 0.3 0.4 0.3 0.2 0.1 (day–1) 0.1 0.2 0.3 0.4 0.3 0.2 0.1 (day–1) 0.1 0.2 0.3 0.4 Sequencing Cance Radio he apy and Su ge y JCO Clinical Cance In o ma ics 7 Downloaded om ascopubs.o g by H. Lee Mo i Cance Cen e and Resea ch Ins i u e on Ap il 18, 2019 om 206.081.166.225 Copy igh © 2019 Ame ican Socie y o Clinical Oncology. All igh s ese ed. RS, SR implied a significan ly smalle numbe o cance cells o adio he apy-induced immune ac i a ion. The in eg al o he e m qD in Equa ion 2 simula ed he numbe o e ec o cells ec ui ed o he umo . Simula ion analysis confi med ha he s eng h o umo -specific immuni y induced by RS was significan ly highe han ha induced by SR (Figs 5A and 5B). This may con ibu e o bo h inc eased OS a es and inc eased an i umo immuni y, which may su eil me as a ic deposi s ou side he i adia ion field (abscopal esponse) o inc ease DFS (compa e wi h Fig 2). Radio he apy-Induced Immuni y Sugges s Plausibili y o RS Dose De-escala ion No mal issue oxici y emained a limi ing ac o in adio- he apy, wi h acu e and la e oxici ies la gely being a ib- u able o o al adia ion dose and dose pe ac ion. Model simula ions sugges ed ha o high a es o adio he apy- induced immuni y, RS wi h 2 Gy pe ac ion may be e- duced o as low as hal he numbe o RS ac ions o ou comes compa able o he ull cou se o SR (12- ac ion RS 25- ac ion SR; Fig 5C). Lowe immunogenici y may equi e .13 ac ions bu po en ially p o ides TC p oba- bili y wi h ewe ac ions han SR. Fo high a es o adio he apy-induced immuni y, adio he apy alone may be a po en ial al e na i e o SR, wi h compa able TC a es (Fig 5D). DISCUSSION Accumula ing e idence suppo s he no ion ha adio he apy-induced an i umo immuni y, especially when p omo ed wi h concu en immuno he apy, could esul in long- e m memo y 41,42 and may yield obus sys emic im- mune esponses agains me as a ic ese oi s ou side he i adia ion field, a phenomenon e med he abscopal e ec . 43-45 Radio he apy may no longe be conside ed a pu ely cy o oxic ea men ; a he , i has biologic p op- e ies o ans o m he umo in o an in si u accine. 46 The clinical p omise o he complex bu p omising immune- adia ion syne gy is eflec ed in .100 ac i e clinical ials es ing he combina ion o adio he apy wi h immuno- he apy, mos ly o me as a ic cance . 47 A obus immune esponse a e ocal i adia ion may equi e a su ficien ly la ge popula ion o cance cells o ac i a e immune cells in numbe s la ge enough o dissemina e sys ema ically and es ablish an immunologic memo y. The he ein p esen ed model simula ions sugges a posi i e co ela ion be ween adio he apy-induced an i umo immuni y and umo size a ime o i adia ion and may explain he poo e local ou comes achie ed wi h SR compa ed wi h RS. A e su gical esec ion wi h clean ma gins, adio he apy will, a bes , ac upon mic oscopic umo clus e s in he umo bed, which may be insu ficien o elici a obus immune esponse. Fo cance s wi h poo OS a es e en when de ec ed ea ly (cance s o lung and b onchus, esophagus, ec um), RS may inc ease OS compa ed wi h SR because o addi ional cy o oxici y esul ing om he inc eased immune ac i a ion. Fo umo s wi h highe OS a es (b eas , ec um), benefi s o RS can be seen in DFS, a guably because o highe an i umo immuni y ha may also ac sys emically agains umo s ou side he i adia ion field and u u e disease. 22 Fo some umo s, no significan ou come benefi s we e ob- se able o ei he sequencing o adio he apy and su ge y. Howe e , ends owa d inc eased OS o cance s wi h low su i al a es ( eg ession coe ficien , −0.65; P,.05) as well as inc eased DFS o cance s wi h high su i al a es ( eg ession coe ficien , 0.35; P,.05) we e no ed. One inhe en limi a ion o SEER da a analysis is composed o he po en ially con ounding ac o s ha a e no eco ded, including pa ien selec ion c i e ia. Pa ien s eligible o RS may be disp opo ionally ea ed in clinical ials and hus could ep esen a subs an ially di e en popula ion han pa ien s unde going SR. Mo eo e , RS may be p e e en- ially pe o med by academic, high- olume cen e s ha o en p o ide be e ou comes in gene al. To add ess he issue o unbalanced a ms, we pe o med mul i a ia e Cox p opo ional haza ds model analysis a e p ep ocessing he da a wi h p opensi y sco e ma ching on he basis o all conside ed co a ia es. 48 These da a a e compa able o he esul s ob ained o he whole coho s (Appendix Fig A3). Di e en ma hema ical models ha e been p oposed o simula e umo –immune sys em in e ac ions 21,22,49-59 and he syne gis ic e ec s o adio he apy wi h he immune sys em and immuno he apy. 22,60-63 To demons a e im- munologic consequences o sequencing adio he apy and su ge y, we ex ended he es ablished Kuzne so model and i s pa ame e iza ion as p e iously discussed. 21,22 Model simula ions e ealed ha obse able clinical esponses may be pa ame e dependen , and as such, he model needs o be igo ously calib a ed o indi idual cance s be o e defini e conclusions can be d awn. Taken oge he , he s a is ical analysis o his o ical ou come da a and he ma hema ical model combine o u he add o he g owing in e es in unde s anding he biologic and immunologic consequences o adio he apy. This s udy FIG 4. (A) Model-p edic ed ea men ou comes o adio he apy (RT) alone, RT a e su gical esec ion (SR), and RT be o e su gical esec ion (RS) o umo s o 10 8 iable cance cells p e ea men and inc easing s eng h o RT-induced immunos imula ion (q, day −1 in Equa ion 2) in a coho o i ual pa ien s wi h di e en combina ions o umo g ow h a e ( ) and immune ec ui men a e ( ) in esponse o umo bu den. Radia ion is deli e ed o a o al dose o 50 Gy in 25 daily ac ions a 2 Gy pe day, 5 days pe week. Tumo con ol (TC) by ea men (blue) and p og essi e disease (PD; ed) e e o umo e adica ion and escape a e ea men , espec i ely. (B) Time e olu ion o umo and e ec o T cells co esponding o he loca ions ma ked by diamonds, iangles, and s a s in panel A. L´ opez Al onso e al 8© 2019 by Ame ican Socie y o Clinical Oncology Downloaded om ascopubs.o g by H. Lee Mo i Cance Cen e and Resea ch Ins i u e on Ap il 18, 2019 om 206.081.166.225 Copy igh © 2019 Ame ican Socie y o Clinical Oncology. All igh s ese ed. p o ides a ionale and mo i a ion o addi ional in- es iga ing and unde s anding he e ec s o adio he apy on he immune sys em o pa ien s wi h cance o guide app op ia e p ospec i e alida ion o adio he apy and su ge y sequencing. This should include immune panels on longi udinal blood d aws du ing adio he apy as well as analysis o immune infil a ion in p ei adia ion biopsies and su gically esec ed issues. AFFILIATIONS 1 B aunschweig In eg a ed Cen e o Sys ems Biology, Helmhol z Cen e o In ec ion Resea ch, B aunschweig, Ge many 2 Nalecz Ins i u e o Biocybe ne ics and Biomedical Enginee ing, Polish Academy o Sciences, Wa saw, Poland 3 In eg a ed Ma hema ical Oncology, H. Lee Mo fi Cance Cen e and Resea ch Ins i u e, Tampa, FL CORRESPONDING AUTHOR Heiko Ende ling, PhD, Depa men o In eg a ed Ma hema ical Oncology, H. Lee Mo fi Cance Cen e and Resea ch Ins i u e, 12902 Magnolia D i e, SRB 4, Tampa, FL 33612; e-mail: heiko.ende ling@mo fi .o g. EQUAL CONTRIBUTION J.C.L.A. and J.P. con ibu ed equally o his wo k. SUPPORT Suppo ed in pa by Ins i u ional Resea ch G an No. 14-189-19 om he Ame ican Cance Socie y (H.E.), by he Ge man Fede al Minis y o Educa ion and Resea ch unding o he eMED p ojec SYSIMIT (G an No. 01ZX1308D; J.C.L.A.), and by he Helmhol z Associa ion o Ge man Resea ch Cen e s Ini ia i e and Ne wo king Fund o he p ojec on educed complexi y models (G an No. ZT-I-0010; J.C.L.A.). AUTHOR CONTRIBUTIONS Concep ion and design: Juan Ca los L´ opez Al onso, Jan Poleszczuk, Sha i Pilon-Thomas, B ian Cze niecki, Louis B. Ha ison, Heiko Ende ling Financial suppo : Juan Ca los L´ opez Al onso, Heiko Ende ling Collec ion and assembly o da a: Jan Poleszczuk, B ian Cze niecki 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 Pa ame e Space Con olled (%) Pa ame e Space Con olled (%) 1 5 10 15 20 2515 10 No. o T ea men F ac ions No. o T ea men F ac ions 15 20 25 CD SR 2 Gy × 25 ac ions q(day–1) 4.5 × 10–1 2.5 × 10–1 6.5 × 10–1 SR 2 Gy × 25 ac ions TC by ea men (SR) 0.1 0.2 0.3 0.4 0.3 0.2 0.1 0.3 0.2 0.1 ATC by ea men (RS) (day–1) (day–1) (day–1) (day–1) 0.1 0.2 0.3 0.4 3 × 107 1 × 107 3 × 106 1 × 106 No. o E ec o Cells B FIG 5. Compa ison o immune ec ui men and ea men ou comes o umo s o 10 8 iable cance cells p e ea men wi h umo g ow h a e ( ) be ween 0.1 and 0.4 day −1 and immune ec ui men a e ( ) in esponse o umo bu den be ween 0.1 and 0.3 day −1 . (A) Rec ui men o e ec o T cells o he umo bed as a esul o adio he apy (RT) –induced immune esponses by su ge y ollowed by RT (SR). (B) Rec ui men o e ec o T cells o he umo bed as a esul o RT-induced immune esponses by RT ollowed by su ge y (RS). Rec ui men o e ec o T cells a e bo h SR and RS is es ima ed by he in eg al o he e m qD in Equa ion 2, wi h q= 4.5 ×10 −1 day −1 . A ows poin o he pa ame e combina ions ha yield umo con ol (TC) by SR and RS (c. . Fig 4A). (C) P opo ion o pa ame e space con olled o SR (25 daily ac ions a 2 Gy pe day, 5 days pe week; blue line) compa ed wi h RS wi h g adually inc easing numbe o ac iona ions. (D) P opo ion o pa ame e space con olled o SR (25 daily ac ions a 2 Gy pe day, 5 days pe week; blue line) compa ed wi h RT alone wi h inc easing numbe o ac iona ions. Pa ame e s a e 0.1 ≤ ≤0.4 day −1 and 0.1 ≤ ≤0.3 day −1 . Di e en s eng hs o RT-induced immunos imula ion a e colo coded (q, day −1 in Equa ion 2). Sequencing Cance Radio he apy and Su ge y JCO Clinical Cance In o ma ics 9 Downloaded om ascopubs.o g by H. Lee Mo i Cance Cen e and Resea ch Ins i u e on Ap il 18, 2019 om 206.081.166.225 Copy igh © 2019 Ame ican Socie y o Clinical Oncology. All igh s ese ed. SR RSRT alone Su i ing F ac ion o E ec o Cells a 2 Gy (SF 2 ) 0.410.610.81 TC by ea men PD (day –1 ) 0.1 0.2 0.3 0.4 TC PD (day –1 ) 0.3 0.2 0.1 (day –1 ) 0.1 0.2 0.3 0.4 (day –1 ) 0.3 0.2 0.1 (day –1 ) 0.1 0.2 0.3 0.4 (day –1 ) 0.3 0.2 0.1 (day –1 ) 0.1 0.2 0.3 0.4 (day –1 ) 0.3 0.2 0.1 (day –1 ) 0.1 0.2 0.3 0.4 (day –1 ) 0.3 0.2 0.1 (day –1 ) 0.1 0.2 0.3 0.4 (day –1 ) 0.3 0.2 0.1 (day –1 ) 0.1 0.2 0.3 0.4 (day –1 ) 0.3 0.2 0.1 (day –1 ) 0.1 0.2 0.3 0.4 (day –1 ) 0.3 0.2 0.1 (day –1 ) 0.1 0.2 0.3 0.4 (day –1 ) 0.3 0.2 0.1 FIG A6. Model-p edic ed ea men ou comes o adio he apy (RT) alone, RT a e su gical esec ion (SR), and RT be o e su gical esec ion (RS) o umo s o 10 8 iable cance cells p e ea men and inc easing su i ing ac ions o e ec o cells a 2 Gy (SF 2 ) in a coho o i ual pa ien s wi h di e en combina ions o umo g ow h a e ( ) and immune ec ui men a e ( ) in esponse o umo bu den. Radia ion is deli e ed o a o al dose o 50 Gy in 25 daily ac ions a 2 Gy pe day, 5 days pe week. Tumo con ol (TC) by ea men (blue) and p og essi e disease (PD; ed) e e o umo e adica ion and escape a e ea men , espec i ely. Fo all simula ions, we se RT-induced an i umo immuni y o q= 4.5 ×10 −1 day −1 . L´ opez Al onso e al 16 © 2019 by Ame ican Socie y o Clinical Oncology Downloaded om ascopubs.o g by H. Lee Mo i Cance Cen e and Resea ch Ins i u e on Ap il 18, 2019 om 206.081.166.225 Copy igh © 2019 Ame ican Socie y o Clinical Oncology. All igh s ese ed.