Circulating Tumor DNA Reveals Genetics, Clonal Evolution, and Residual Disease in Classical Hodgkin Lymphoma

From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

Plenary Paper

LYMPHOID NEOPLASIA Circulating tumor DNA reveals genetics, clonal evolution, and residual disease in classical Hodgkin lymphoma

Valeria Spina,1,* Alessio Bruscaggin,1,* Annarosa Cuccaro,2 Maurizio Martini,3 Martina Di Trani,4 Gabriela Forestieri,1 Martina Manzoni,5 Adalgisa Condoluci,1,6 Alberto Arribas,1 Lodovico Terzi-Di-Bergamo,1 Silvia Laura Locatelli,4 Elisa Cupelli,2 Luca Ceriani,6 Alden A. Moccia,6 Anastasios Stathis,6 Luca Nassi,7 Clara Deambrogi,7 Fary Diop,7 Francesca Guidetti,1 Alessandra Cocomazzi,3 Salvatore Annunziata,8 Vittoria Ruﬁni,8 Alessandro Giordano,8 Antonino Neri,5,9 Renzo Boldorini,10 Bernhard Gerber,6 Francesco Bertoni,1,6 Michele Ghielmini,6 Georg Stussi,¨ 6 Armando Santoro,4,11 Franco Cavalli,1,6 Emanuele Zucca,6 Luigi Maria Larocca,3 Gianluca Gaidano,7 Stefan Hohaus,2,† Carmelo Carlo-Stella,4,11,† and Davide Rossi1,6,†

1Institute of Oncology Research, Bellinzona, Switzerland; 2Institute of Hematology and 3Division of Pathology, Policlinico Gemelli Foundation, Catholic University of the Sacred Heart, Rome, Italy; 4Humanitas Cancer Center, Humanitas Clinical and Research Center, Milan, Italy; 5Department of Oncology and Hemato-oncology, University of Milan, Milan, Italy; 6Oncology Institute of Southern Switzerland, Bellinzona, Switzerland; 7Division of Hematology, Department of Translational Medicine, University of Eastern Piedmont, Novara, Italy; 8Institute of Nuclear Medicine, Policlinico Gemelli Foundation, Catholic University of the Sacred Heart, Rome, Italy; 9Hematology Unit, Foundation Ca’ Granda IRCCS, Ospedale Maggiore Policlinico, Milan, Italy; 10Division of Pathology, Department of Health Science, University of Eastern Piedmont, Novara, Italy; and 11Department of Biomedical Sciences, Humanitas University, Milan, Italy

KEY POINTS The rarity of neoplastic cells in the biopsy imposes major technical hurdles that have so far limited genomic studies in classical Hodgkin lymphoma (cHL). By using a highly sensitive l ctDNA is as an easily accessible source of and robust deep next-generation sequencing approach for circulating tumor DNA tumor DNA for cHL (ctDNA), we aimed to identify the genetics of cHL in different clinical phases, as well as its genotyping. modifications on treatment. The analysis was based on specimens collected from 80 newly diagnosed and 32 refractory patients with cHL, including longitudinal samples collected l ctDNA is a radiation- free tool to track under ABVD (adriamycin, bleomycin, vinblastine, dacarbazine) chemotherapy and longi- residual disease in cHL. tudinal samples from relapsing patients treated with chemotherapy and immunotherapy. ctDNA mirrored Hodgkin and Reed-Sternberg cell genetics, thus establishing ctDNA as an easily accessible source of tumor DNA for cHL genotyping. By identifying STAT6 as the most frequently mutated gene in ∼40% of cases, we refined the current knowledge of cHL genetics. Longitudinal ctDNA profiling identified treatment-dependent patterns of clonal evolution in patients relapsing after chemotherapy and patients maintained in partial remission under immunotherapy. By measuring ctDNA changes during therapy, we propose ctDNA as a radiation-free tool to track residual disease that may integrate positron emission tomography imaging for the early identification of chemorefractory patients with cHL. Collectively, our results provide the proof of concept that ctDNA may serve as a novel precision medicine biomarker in cHL. (Blood. 2018;131(22):2413-2425)

Introduction Interim positron emission tomography/computed tomography (PET/CT) is widely used to predict cHL outcome before com- An unprecedented body of genetic knowledge has been translated pletion of chemotherapy and to inform early treatment in- into biomarkers to refine diagnosis, prognostication, and treatment tensification or de-escalation. However, interim PET/CT results of non-Hodgkin lymphomas (NHLs).1 On the contrary, the genetics are inconsistent with the final outcome in ;20% to 30% of of classical Hodgkin lymphoma (cHL) is less well understood.2 patients, who are thus still exposed to over- or undertreat- Rarity of neoplastic Hodgkin and Reed-Sternberg (HRS) cells in ment.4 By identifying residual disease beyond the sensitivity of the biopsies and their routine formalin fixation impose major techni- imaging and by specifically tracking tumor fingerprints, mo- cal hurdles that have limited the assessments of cHL mutations lecular methods hold the potential of complementing PET/CT in different clinical phases and under different treatments. in assessing tumor response.5 However, molecular minimal residual disease has been so far inapplicable in lymphomas that An unmet medical need in cHL is the early and accurate iden- lack a leukemic component, such as cHL. tification of chemorefractory patients, as they are candidates for treatment intensification to maximize the chances of cure, as well as the early and accurate identification of good-risk patients, as Plasma is a source of circulating tumor DNA (ctDNA) for they are candidates for treatment de-escalation to avoid long- genotyping purposes,6 and studies using copy number ab- term complications of chemoradiotherapy.3 normalities,7 immunoglobulin gene rearrangement,8 or single

A B 80% (12/15) ITPKB CATALYTIC domain 80% 53% 1 946 60% (8/15) 27% 20% 40% (4/15) 13% TNFAIP3 OTU ZF ZF ZF ZF ZF ZF ZF (3/15) 7% (2/15) 20% (1/15) 1 790

% of mutated cases 0% ID3 BTK IRF8 IRF4 B2M ATM MYC TP53 TET2 PIM1 STAT BCL6 CIITA SPEN BTG1 CD58 XPO1 ITPKB STAT6 STAT_int STAT_bindSH2 STAT6_C STAT6 STAT3 TRAF3 PCBP1

CXCR4 alpha GNA13 PRDM1 KMT2D NFKBIE CCND3 ARID1A TNFAIP3 NOTCH1

HIST1H1E 1 847 TNFRSF14 0 missense nonsense 5 splicing 10 missense nonsense frameshift frameshift 15 splicing start loss 3’-UTR No. of mutations 20 C D 100% 20 87.5% 15 80% 10 5 60% 13 84 12

No. of mutations 0 40%

cHL2 UPN8 cHL4cHL6 cHL8cHL5cHL7 cHL12cHL57*UPN10cHL19 cHL11cHL15 cHL14 cHL13 20%

E Biopsy conﬁrmed mutations 0%

100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 91% 75% 66.6% 66.6% 50% 40% Mutation identified both in gDNA and in ctDNA Mutation identified in ctDNA 50% Mutation identified in gDNA

0% identiﬁed in ctDNA cHL2 cHL4cHL8 cHL6 cHL5cHL7 UPN8 cHL13cHL11cHL19 cHL15 cHL14cHL12

Biopsy conﬁrmed mutaions cHL57*UPN10 Patient *chemorefractory sample

Figure 1. ctDNA mirrors the genetics of cHL. (A) Prevalence of nonsynonymous somatic mutations discovered in ctDNA of 15 cHL cases provided with HRS cells microdissected from the paired biopsy. The graph below shows the number and type of nonsynonymous somatic mutations identified in each gene. (B) The position and type of nonsynonymous somatic mutations identified by ctDNA genotyping of the most frequently mutated genes are reported at the top of the proteins. The position and type of nonsynonymous somatic mutations that have been detected in the tumor gDNA of published cHL series19 (ITPKB and TNFAIP3) or B-cell lymphomas of the COSMIC database (version 81; STAT6) are reported at the bottom of the protein. Shapes indicate the type of the mutations, whereas color codes indicate whether they have been identified in the paired microdissected HRS cells (red) or they lacked in the paired microdissected HRS cells (gray). (C) Number of mutations in a given tumor discovered in plasma ctDNA and/or tumor gDNA. Mutations are color coded if they were identified in both plasma ctDNA and tumor gDNA (red), only in plasma ctDNA (blue), or only in tumor gDNA (gray). (D) Venn diagram summarizing the overall number of mutations discovered in both plasma ctDNA and tumor gDNA (red), only in plasma ctDNA (blue), or only in tumor gDNA (gray). The corresponding overall sensitivity of plasma cfDNA genotyping in discovering biopsy-confirmed mutations is shown. (E) For each patient, the fraction of tumor biopsy-confirmed mutations that were detected in plasma ctDNA is shown. Patients are ordered by decreasing detection rates. The red portion of the bars marks the prevalence of tumor biopsy-confirmed mutations that were detected in plasma ctDNA. The gray portion of the bars marks the prevalence of tumor biopsy-confirmed mutations that were not detected in plasma ctDNA. mutations as tumor marker9 identified ctDNA also in cHL. After Methods having shown that ctDNA mirrors HRS cells’ mutational Patients profile, we expanded the applications of ctDNA in cHL by showing that it can be used to noninvasively detect cHL The study had a retrospective observational design. Confirmed mutations without the need for HRS cell microdissection, diagnosis of cHL and availability of biological samples were the longitudinally track cHL clonal evolution under different sole study inclusion criteria. cHL subtyping was according to treatments, and monitor residual disease during multiagent the World Health Organization Classification of Tumors of chemotherapy with ABVD (adriamycin, bleomycin, vinblas- Hematopoietic and Lymphoid Tissues criteria.10 Both pre- tine, dacarbazine). viously untreated (n 5 80) and relapsed/refractory (n 5 32)

2414 blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 SPINA et al From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

Table 1. Characteristics of the 80 newly diagnosed purposes, tumor gDNA from HRS cells and gDNA from biopsy patients with cHL areas devoid of HRS cells were also analyzed. Paired plasma samples were collected in close temporal proximity of the Characteristic Patients, n (%) tumor tissue biopsy (7-14 days after the biopsy). cfDNA from newly diagnosed diffuse large B-cell lymphoma (DLBCL; n 5 19) Male sex 42 (52.4) and primary mediastinal large B-cell lymphoma (n 5 3) and the Median age, y 44 (17-82) corresponding paired normal gDNA were also analyzed. DLBCLs have been previously published and now reanalyzed with a dif- Age range, y 17-82 ferent target region.11 Patients provided informed consent in 17-24 19 (23.75) accordance with local Institutional Review Board requirements 25-44 22 (27.5) and the Declaration of Helsinki. The study was approved by 45-60 20 (25) the Cantonal Ethical Committee of Ticino (CE 3236, BASEC 2017- 61-82 19 (23.75) 01004).

Histology Nodular sclerosis 64 (80) Microdissection, immunohistochemistry, and in Mixed cellularity 12 (15.0) situ hybridization Lymphocyte rich 3 (3.7) The percentage of tumor cells was visually estimated inde- Unclassifiable 1 (1.2) pendently by 2 pathologists (L.M.L. and M. Martini) on hema- toxylin and eosin-stained slides. For each case, multiple specific EBER tumor areas containing 20% to 40% HRS cells were micro- Positive 15 (18.7) dissected to enrich the proportion of neoplastic cells, and were Negative 38 (47.5) pooled. Marking the tumor enriched-regions within tissue sam- Not assessable 27 (33.7) ples relied on the visual assessment of tissue sections or CD30 Ann Arbor Stage immunohistochemical staining of HRS cells. Areas devoid of I 2 (2.5) HRS cells were also microdissected for comparative purposes. II 26 (32.5) Expression of pSTAT6 and MCH-I was assessed by immuno- III 18 (22.5) histochemistry and scored as previously reported (supplemental 12,13 IV 34 (42.5) Appendix, available on the Blood Web site). In situ hybridization of Epstein-Barr virus–encoded small RNAs (EBERs) on Stage strata* formalin-fixed and paraffin-embedded tissue section was carried Limited 14 (17.5) out as described.14 Advanced 66 (82.5)

GHSG risk group CAPP-seq library preparation, ultra-deep NGS, Favorable 3 (3.8) and variant calling Intermediate 17 (21.3) To avoid preanalytic confounding effects, a standardized ap- Unfavorable 60 (75.0) proach was used to extract cfDNA from plasma (supplemental Appendix). A targeted resequencing gene panel including IPS coding exons and splice sites of 77 genes (target region: 191 271 Low, 0-3 63 (78.7) bp) that are recurrently mutated in mature B-cell tumors has High, $4 17 (21.3) been specifically designed for this project (supplemental Table 1) as follows: initial seed genes were chosen if recurrently Bulky disease 16 (20.0) mutated in more than 5% of mature B-cell tumors, in the case of B symptoms 43 (53.8) well-known cancer genes (even if mutated in 1%-5% of cases), or if they were known to be associated with resistance to che- GHSG, German Hodgkin Study Group Score; IPS, International Prognostic Score. motherapy in mature B-cell tumors; and for genes with a well- *Limited stage: IA, IB, or IIA without bulky disease. Advanced stage: III, IV, or I or II with bulky defined hotspot, only coding exons plus splice sites that disease or IIB. included more than 95% of mutations were covered. Tumor and germline gDNA from tissues (median 5 450 ng) were sheared patients with cHL were included. All relapsed/refractory cHL through sonication before library construction to obtain 200-bp failed both first-line therapy as well as autologous transplant fragments. For cfDNA, which is a naturally fragmented DNA, a salvage. The following biological material was analyzed: cell- median amount of 79.63 ng was used for library construction free DNA (cfDNA) isolated from plasma collected at diagnosis without additional fragmentation. The next-generation sequencing before treatment start (n 5 80 patients), during ABVD courses (NGS) libraries were constructed using the KAPA Library Prepa- (onday1ofcourse2atthetimeofinterimreassessment,and ration Kit (Kapa Biosystems), and hybrid selection was performed at the end of treatment; n 5 24 patients), at refractory pro- withthecustomSeqCapEZChoiceLibrary (Roche NimbleGen). In gression (n 5 32 patients), before and after failing auto- the CAncer Personalized Profiling by deep Sequencing (CAPP-seq) transplant salvage (n 5 6), before and after failing brentuximab of cfDNA, the manufacturer’s protocols were modified as pre- vedotin (n 5 5 patients), and before and during therapy with viously reported.15 Multiplexed libraries were sequenced using nivolumab (n 5 5 patients); and normal germline genomic DNA 150- or 300-bp paired-end runs on NextSeq or MiSeq sequencers (gDNA) from peripheral blood granulocytes collected at the (Illumina). A robust and previously validated bioinformatics pipeline same time of the paired plasma sample. For comparative was used for variant calling (supplemental Appendix).11

cHL GENOTYPING ON THE LIQUID BIOPSY blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 2415 From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

missense nonsense frameshift inframe splicing start loss 3’-UTR stop loss 40 30 20 % of mutated cases 10 0% 10% 20% 30% 40%

No. of mutations No. 0 STAT6 37.5% (30/80) TNFAIP3 35% (28/80) ITPKB 27.5% (22/80) GNA13 18.7% (15/80) B2M 16.2% (13/80) ATM 15% (12/80) SPEN 12.5% (10/80) KMT2D XPO1 11.2% (9/80) TP53 ARID1A HIST1H1E 10% (8/80) BTG1 CD36 P2RY8 7.5% (6/80) FBXW7 NFKBIE PIK3CA CIITA 6.2% (5/80) IKBKB BIRC3 CREBBP FBXO11 IRF8 5% (4/80) PCBP1 TBL1XR1 TNFRSF14 RIPK1 BTK CD58 ZMYM3 3.7% (3/80) TRAF3 HIST1H1C MYC BCL2 BCOR CXCR4 IRF4 MAP2K1 MAP3K14 NOTCH1 PIM1 S1PR2 WHSC1 IRAK1 CCND3 CD79B 2.5% (2/80) EZH2 HIST1H3B ID3 KLHL6 NOTCH2 PLCG2 POT1 SIN3A TET2 EGR2 PRDM1 BCL6 BRAF CCND1 EP300 FGFR2 1.2% (1/80) MEF2B STAT3 KLF2 CARD11 CCND2 CD70 CD79A FOXO1 KRAS 0% (0/80) MYD88 NRAS RRAGC TCF3 TRAF2

Figure 2. The mutational proﬁle of newly diagnosed cHL. The heat map shows individual nonsynonymous somatic mutations detected in ctDNA of newly diagnosed cHL (n 5 80). Each row represents a gene, and each column represents a primary tumor. The heat map was manually clustered to emphasize mutational co-occurrence. Mutations are color-coded in red. The number and type of nonsynonymous somatic mutations in any given patients is plotted above the heat map. The horizontal bar graph shows the gene mutation frequency.

Validation of hotspot STAT6 mutation The PET acquisition time was at least 3 minutes per bed position. Hotspot STAT6 mutations (c.1249A.T and c.1255G.A) dis- Images were reconstructed with validated and commercially avail- covered by CAPP-seq in cfDNA were validated by allele-speciﬁc able iterative algorithms, and standardized uptake values were polymerase chain reaction (supplemental Appendix). automatically calculated. A central blind analysis was performed by trained nuclear physicians (L.C., S.A., V.R., and A.G.) without in- 18 Imaging studies formationontheclinicaloutcome.The FDG-PET/CT images PET/CT scans were performed at baseline for initial staging, after obtained at baseline and during treatment were analyzed following a 2 cycles of ABVD and at the end of treatment, or at baseline, after standard protocol on a dedicated workstation (Siemens Syngo 4, 10, 17, and 25 cycles of nivolumab. PET and CT images were MMWP Workstation VE36A; Siemens). The interim and end-of- 18 acquired in the same session. CT scans obtained with a low-dose treatment and FDG-PET/CT images scans were visually assessed 18 protocol were used for attenuation-correction of the PET images. according to the Deauville criteria, with FDG uptake of any residual All patients were fasting for at least 6 hours before the injection of lesion scored according to the 5-point scale, using mediastinal blood 250 to 370 MBq (4.5 MBq/kg) 18FDG (ﬂuorodeoxyglucose). Blood pool and liver uptake as reference settings.16 The LYmphoma Re- glucose measured before injection of the radiotracer was lower sponse to Immunomodulatory therapy Criteria (LYRIC) were ap- than 160 mg/dL in all patients. PET data were acquired in 2- or plied on treatment with nivolumab.17 Diffuse uptake in the spleen 3-dimensional mode from the midthigh toward the base of the or marrow on the postchemotherapy scan that was considered a skull after a standardized uptake time of 60 minutes (65 minutes). result of chemotherapy was not scored as active disease.

2416 blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 SPINA et al From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

A Age ≤ 60 years Age > 60 years EBER+ EBER-

NSCHL MCCHL LRCHL NA

Age EBER status cHL subtype STAT6 TNFAIP3 ITPKB GNA13 B B2M p= 0.0229 p= 0.0422 ATM SPEN KMT2D 28/64 XPO1 50% 26/64 TP53 (43.8%) ARID1A (40.1%) HIST1H1E 40% BTG1 CD36 30% P2RY8 FBXW7 20% 2/16 2/16 NFKBIE (12.5%) (12.5%) PIK3CA CIITA 10% IKBKB Mutation frequency BIRC3 0% CREBBP FBXO11 NSCHL other NSCHL other IRF8 PCBP1 STAT6 TNFAIP3 TBL1XR1 TNFRSF14 RIPK1 C BTK CD58 p= 0.024 p= 0.0466 ZMYM3 TRAF3 28/64 HIST1H1C 50% MYC (43.8%) 26/64 BCL2 (40.1%) BCOR 40% CXCR4 IRF4 30% MAP2K1 MAP3K14 20% NOTCH1 1/12 1/12 PIM1 S1PR2 10% (8.3%) (8.3%)

WHSC1 Mutation frequency IRAK1 CCND3 0% CD79B NSCHL MCCHL EZH2 HIST1H3B STAT6 TNFAIP3 ID3 KLHL6 NOTCH2 PLCG2 POT1 D SIN3A TET2 p= 0.0368 p= 0.0545 EGR2 PRDM1 20/38 BCL6 60% BRAF (52.6%) CCND1 17/38 EP300 (44.7%) FGFR2 MEF2B 40% STAT3 KLF2 3/15 CARD11 (20%) 2/15 CCND2 20% (13.3%) CD70

CD79A Mutation frequency FOXO1 KRAS 0% MYD88 EBER+EBER- EBER+EBER- NRAS RRAGC STAT6 TNFAIP3 TCF3 TRAF2

Mutated sample

Unmutated sample

Figure 3. Distribution of mutations among cHL histologic subtypes. (A) The heat map shows individual nonsynonymous somatic mutations detected in ctDNA of newly diagnosed cHL (n 5 80) clustered according to the histological subtype. Each row represents a gene, and each column represents a primary tumor. The heat map was manually clustered to emphasize mutational co-occurrence. Mutations are color-coded in red. EBER status is shown above the heat map. The bar graph compares the prevalence of the most frequently mutated genes between nodular sclerosis (NSCHL) and all the other cHL subtypes (B), between nodular sclerosis (NSCHL) and mixed cellularity (MCCHL) cHL (C), and between EBER positive and negative cases (D).

cHL GENOTYPING ON THE LIQUID BIOPSY blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 2417 From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

A TNFAIP3 IKBKB NFKBIE RIPK1 TRAF3 IRF4 BIRC3 MAP3K14

B ITPKB GNA13 P2RY8 PIK3CA S1PR2 ID3

C STAT6 STAT3

D KMT2D HIST1H1E ARID1A CREBBP BCOR HIST1H3B TET2 EZH2 WHSC1 EP300

E B2M CIITA TNFRSF14 CD58

F SPEN FBXW7 NOTCH1 NOTCH2

Figure 4. Pathways that are recurrently mutated in newly diagnosed cHL. In the heat map, rows correspond to genes and columns represent individual patients. Color coding is based on gene alteration status (gray, wild type; red, mutated). The heat map was manually clustered to emphasize mutational co-occurrence. On the right side is a schematic representation of the mutated pathways. Mutated genes are marked by an arrow, and the prevalence of mutations is reported. The following pathways are shown: NF-kB (A), PI3K/AKT (B), cytokine signaling (C), epigenetics (D), immune surveillance (E), and NOTCH signaling (F).

2418 blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 SPINA et al From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

D 7/13 5/5 (53.8%) (100%) ABC 100% 80%

60%

40%

Frequency 6/13 (46.2%) 20%

0% B2M B2M WT M H 4/10 8/8 (40%) (100%) EFG 100% 80%

60%

40%

Frequency 6/10 (60%) 20%

0% STAT6 STAT6 WT M

Figure 5. Expression of phosphorylated STAT6 and MHC-I proteins in cHL. Expression of MHC-I (A-C), and pSTAT6 (E-G) was examined by immunohistochemistry in formalin-fixed sections of cHL. (A) HRS cells with absent expression of MHC-I in a exemplificative B2M mutated cHL case. (B) HRS cells with absent expression of MHC-I in a exemplificative B2M wild-type (WT) cHL case, in which alternative genetic mechanisms not explored by our assay, may be responsible for the loss of MHC-I expression. (C) HRS cells with normal expression of MHC-I in a exemplificative B2M WT cHL case. Staining for MCH-I was absent on the HRS cell surface of 11/18 patients with cHL, as clearly shown by the lack of surface staining at the cell boundaries in the cluster of HRS cells, including 5/5 cases harboring B2M mutations and 6/13 cases lacking B2M mutations (D). (E) HRS cells with a nuclear overexpression of p-STAT6 in a cHL case harboring STAT6 mutation. (F) HRS cells with a nuclear overexpression of p-STAT6 in a exemplificative STAT6 WT cHL case, consistent with the plethora of genetic mechanisms affecting the cytokine signaling program in this tumor, some of which are not explored by our assay and may be responsible for the overexpression of p-STAT6. (G) HRS cells with absent nuclear expression of p-STAT6 in a exemplificative STAT6 WT cHL case. Staining for nuclear pSTAT6 was positive in HRS cells of 14/18 patients with cHL, including 8/8 cases harboring STAT6 mutations and 6/10 cases lacking STAT6 mutations (H). Original magnifications 3630.

Statistical analysis cHL, including longitudinal samples obtained under ABVD Progression-free survival was measured from date of ABVD chemotherapy and from relapsing patients treated with che- treatment start to date of progression (event), death from any cause motherapy and immunotherapy. We retrospectively profiled a (event), or last follow-up (censoring). Overall survival was measured total of 349 blood and tissue samples fromdateofABVDstarttodateofdeath(event)orlastfollow-up (censoring). Molecular studies were blinded to the study endpoints. To provide the proof that ctDNA informs on cHL genetics, it was Survival analysis was performed by Kaplan-Meier method. Un- initially genotyped using CAPP-seq, a sensitive (;1023)andvali- supervised hierarchical clustering (Euclidean distance, complete dated targeted ultra-deep-NGS approach for ctDNA,11,15 in a method) according to the mutational status was performed using discovery set of 15 patients (mean coverage: 49413; supplemental the extended package in the R environment (https://cran.r-project. Figure 1A) provided with paired tumor gDNA isolated from HRS org/package5dendextend; R Studio console; RStudio, Boston, cells. Paired normal gDNA was also analyzed to confirm the somatic MA). Categorical variables were compared by x-square test and origin of mutations (mean coverage: 46263; supplemental Fisher’s exact test. Continuous variables were compared by Mann- Figure 1B). High concordance (R2 5 0.978) of variant calling Whitney U test. All statistical tests were 2-sided. Statistical sig- from independent duplicate experiments supported the robust- fi fi , ni cance was de ned as P .05. Correction for multiple ness of CAPP-seq of ctDNA, including detection of low-abundance comparisons was performed by using the false-discovery-rate mutations (supplemental Figure 1D), thus excluding their origin method. The analysis was performed with SPSS v.22 and with R from a batch-specific experimental noise. statistical package 3.3.2 (http://www.r-project.org). Nonsynonymous somatic mutations were discovered in ctDNA of all 15 patients with cHL, including variants of the TNFAIP3, Results ITPKB, GNA13, and B2M genes, which were previously reported Circulating tumor DNA mirrors the genetics of in studies of purified HRS cells (Figure 1A).18,19 The molecular HRS cells spectrum of mutations discovered in ctDNA also reflected that The study was based on specimens collected from 80 newly of published cHL variants (Figure 1A-B), thus supporting their diagnosed patients and 32 patients with relapsed-refractory tumor origin.18,19

cHL GENOTYPING ON THE LIQUID BIOPSY blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 2419 From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

A B Newly diagnosed cHL Refractory cHL (n=80) (n=32) 40% 30% 10%20% 0% 0% 10% 20% 30% 40% 30 STAT6 9 28 TNFAIP3 8 22 ITPKB 6 13 B2M 4 27 33 29 12 ATM 4 9 TP53 1 8 HIST1H1E 1 8 BTG1 0 6 FBXW7 1 6 P2RY8 1 4 PCBP1 1 4 TBL1XR1 1 Tumor speciﬁc at baseline 4 TNFRSF14 0 4 FBXO11 0 Tumor speciﬁc at relapse 4 CREBBP 0 Shared 3 HIST1H1C 1 3 MYC 0 TRAF3 3 0 UPN ZMYM3 C 3 0 6 3 CD58 0 3 RIPK1 0 2 EGR2 0 2 BCL2 0 2 NOTCH2 0 cHL 2 KLHL6 0 72 2 HIST1H3B 0 IRAK1 2 0 cHL 2 S1PR2 0 cHL 55 59 2 MAP3K14 0 cHL 2 MAP2K1 0 52 KLF2 1 0 cHL cHL MEF2B T1 1 0 54 70 1 CCND1 0 1 BCL6 0 0 TCF3 0 cHL 0 NRAS 0 71 0 MYD88 0 0 KRAS 0 cHL 0 CCND2 0 57 cHL 2 PRDM1 1 69 2 SIN3A 1 2 PLCG2 1 2 ID3 1 cHL 2 EZH2 1 51 2 CD79B 1 cHL 2 IRF4 1 68 2 CXCR4 1 T1 T2 1 FGFR2 1 BRAF cHL 1 1 58 Scale by the fraction of mutations 0 TRAF2 1 unique to each timepoint 0 RRAGC 1 0 CARD11 1 0 CD79A 1 D 0 CD70 1 IRF8 cHL 57 4 2 100% 4 BIRC3 2 ARID1A CD58 3 BTK 2 PIK3CA TNFAIP3 2 POT1 2 POT1 BCL2 2 WHSC1 2 80% 2 CCND3 2 PIM1 PIM1 2 2 IRF8 Germline clone 2 NOTCH1 2 TNFAIP3 STAT3 60% 1 2 STAT3 Common ancestral clone 0 FOXO1 2 5 PIK3CA 2 Baseline clone 5 IKBKB 2 40% TET2 NFKBIE 5 NFKBIE 3 CIITA ITPKB Relapse clone 5 CIITA 3 CCND3 NFKBIE CD36 Clonal divergence 6 3 of mutations (%) Fraction STAT6 GNA13 9 XPO1 4 20% 9 KMT2D 4 1 EP300 4 SPEN 10 5 0% 2 BCOR 5 2 TET2 5 8 ARID1A 6 Baseline Relapse 15 GNA13 9

Figure 6. Genotype and clonal evolution of refractory cHL. (A) Comparison between the mutational profile of newly diagnosed cHL (n 5 80; blue bars) vs refractory cHL (n 5 32; red bars). Number and prevalence of mutated cases is given for each gene. (B) Venn diagram showing the number of tumor-specific mutations at baseline (light blue circle), the number of tumor-specific mutations at cHL relapse (magenta circle), and the number of tumor mutations shared between the 2 points (purple circle) in longitudinal

2420 blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 SPINA et al From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

To validate our approach and formally confirm the tumor origin subtypes, including mixed cellularity cases, and among EBER- of ctDNA variants, HRS cell-enriched areas (tumor representa- negative cHL compared with EBER-positive cases (Figure 3), tion, 20%-40%) were microdissected from the paired tumor further reinforcing the notion that cHL subtypes have different biopsies and their gDNA analyzed by CAPP-seq blinded to the genetic profiles.18,19 STAT6 mutations also were more frequent mutational profile recovered in the paired ctDNA. To system- among patients aged 60 years or younger, which might reflect atically derive the accuracy of ctDNA genotyping, the results of the enrichment of mixed cellularity and EBER-positive cases plasma ctDNA genotyping and tumor gDNA genotyping (gold among elderly cHL (supplemental Figure 6). standard) were then compared. Mean coverage of the tumor gDNA was 17333 (supplemental Figure 1C). Genotyping of Mutated genes pointed to the molecular deregulation of specific plasma ctDNA identified a total of 106 somatic mutations, programs in cHL. NF-kB was mutated in 46.2% of patients, con- whereas genotyping of the gDNA from HRS cells identified 96 sistent with the strong NF-kB signature of HRS cells (Figure 4A).21,22 somatic mutations (Figure 1C-D). No mutations were detected in At variance with DLBCL, cHL was virtually devoid of mutations in areas of the biopsy devoid of HRS cells, thus validating mutation upstream molecules that converge surface signals to NF-kBand origin from HRS cells. Biopsy-confirmed tumor mutations were that sensitize tumor cells to Bruton tyrosine kinase inhibition detectable in ctDNA samples with an 87.5% (n 5 84/96; 95% (Figure 2).23 PI3K/AKT was affected in 46.2% of patients, consistent confidence interval, 79.2%-92.8%) sensitivity (Figure 1D-E). A with the preclinical evidence that cHL is addicted to this actionable few variants were not identified in tumor gDNA. These variants cellular program (Figure 4B).24,25 Cytokine signaling was mutated in were bona fide neoplastic, according to their molecular profile 37.5% of patients (Figure 4C), although the prevalence of genetic (Figure 1B) and disappearance on disease remission. Their ab- lesions affecting this pathway may be higher and not fully captured sence in microdissected HRS cells was conceivably a result of the byourtargetregionthatdidnotcovertheSOCS1 gene. A number subclonal or anatomical heterogeneity of the tumor, as already of epigenetic genes were cumulatively affected in 35.0% of pa- extensively reported in other ctDNA studies.11,15,20 tients, with the exception of EZH2, whose druggable mutations were rare in cHL (Figures 2 and 4D).26 Genes involved in immune Noninvasive genotyping of newly diagnosed cHL surveillance were mutated in 27.5% of patients (Figure 4E). The After having established that ctDNA mirrors tumor gDNA in cHL, NOTCH pathway was mutated in 20.0% of patients (Figure 4F), in 25,27 we used this source of tumor DNA to noninvasively characterize keeping with the NOTCH signature of HRS cells. the mutational spectrum of 80 newly diagnosed patients whose clinicopathologic characteristics are summarized in Table 1. We pSTAT6 expression in the nucleus of HRS cells from STAT6 mutated detected nonsynonymous somatic mutations in 81.2% of pa- cases provided the proof of principle of the functional consequences tients, with an average of 5 mutations per case. The lack of of STAT6 mutations (Figure 5E-G). Nuclear pSTAT6 expression by correlation between number of mutations and coverage or input HRS cells was broader than STAT6 mutations in our cHL cohort cfDNA for library preparation (supplemental Figure 2A-B) in- (Figure 5H), a fact that may be consistent with the plethora of genetic dicated that the conditions of our CAPP-seq reached a plateau mechanisms affecting the cytokine signaling program in this tumor, in sensitivity that was uniformly maintained across all the study some of which were not explored by our assay (eg: SOCS1 mutations samples. The mean allele frequency of ctDNA mutations was and chromosome 9p amplification. B2M mutated cases always 5.5% (range, 0.29%-74.0%), and the majority (87.3%) had an lacked MHC-I expression in Hodgkin-Reed Sternberg cells, thus allele frequency higher than 1% (supplemental Figure 1C). indirectly validating B2M variants discovered in cfDNA (Figure 5A-C). Pretreatment ctDNA concentration correlated with a stage and MHC-I was absent in a proportion of B2M wild-type patients prognostic group of cHL, thus pointing to ctDNA as a surrogate (Figure 5D). In those cases, alternative genetic mechanisms not marker of tumor load (supplemental Figure 3). explored by our assay, such as B2M gene locus deletion or mutation/ deletion of the HLA genes, may be responsible for the loss of MHC-I Genes recurrently affected by nonsynonymous somatic muta- expression, as previously reported in DLBCL.28 tions in 20% or more of patients included STAT6 (37.5%), TNFAIP3 (35.0%), and ITPKB (27.5%; Figure 2; supplemental We next compared the ctDNA genetic signatures of cHL, DLBCL, Table 2), which often co-occurred in the same patient (Figure 2; and primary mediastinal large B-cell lymphoma by applying a supplemental Figure 4). Hotspot STAT6 mutations (c.1249A.T, probabilistic classifier derived from differentially represented n 5 26; c.1255G.A, n 5 12) discovered in cfDNA were con- mutations. As already shown by gene expression profiling,25 cHL firmed on a different experimental platform by allele-specific and primary mediastinal large B-cell lymphoma clustered together polymerase chain reaction, thus validating CAPP-seq results also genetically, whereas the majority (79.5%) of cHL showed a (supplemental Figure 5). STAT6 and TNFAIP3 mutations were distinct mutational signature compared with DLBCL (supple- enriched in nodular sclerosis cHL compared with other cHL mental Figure 7).

Figure 6 (continued) samples from 13 patients with refractory cHL. (C) Schematic representation of mutational divergence between baseline/relapse pairs of 13 refractory cHL cases. The central node represents baseline (T1), and the distance between baseline and each refractory relapse (edge) is expressed as the fraction of unique mutations to both baseline and relapse points. (D) Phylogenetic tree describing evolutionary distances between sequential tumor pairs in a refractory cHL case. Inferred ancestral clones are also indicated. Evolutionary distance is defined as the fraction of shared mutations between tumor at the baseline and relapse time points (black node), the fraction of mutations specific for the baseline time point (red node), and the fraction of mutations specific for the relapse point (blue node). On the y-axis, 100% represents all mutations identified throughout the longitudinal course of the patient. The time between baseline and relapse is depicted on the x-axis. The leftmost node (gray circle) indicates the germline cell, and the second node from the left (black circle) indicates the last common inferable ancestral clone. The length of the branch between the germline cell and the common ancestral clone reflects the fraction of shared mutations between baseline and relapse samples. The length of the branches between the common ancestral clone and either baseline or relapse samples reflects the fractions of mutations observed only in baseline or relapse samples, respectively. The colored light blue area represents the clonal divergence between baseline and relapse points. Mutated genes shared between baseline and relapse time points (black), mutated genes specific for baseline (red), and mutated genes specific for relapse time point (blue) are annotated.

cHL GENOTYPING ON THE LIQUID BIOPSY blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 2421 From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

Noninvasive monitoring of clonal evolution in a threshold proposed and validated in DLBCL30 and confirmed refractory cHL as best cutoff to predict progression in our cohort (Figure 7D), was To identify mutations that are enriched in refractory cHL, 32 associated with complete response and cure (Figure 7A,E). Con- patients who relapsed after salvage autologous stem cell versely, a drop of less than 2-log in ctDNA after 2 ABVD courses was transplant were genotyped on ctDNA, and their aggregated associated with progression and inferior survival (Figure 7A,E; fi mutational profile was compared with that of newly diagnosed supplemental Figure 10). Quanti cation of ctDNA complemented patients. Paired analysis of HRS cells microdissected from bi- interim PET/CT in determining residual disease. Indeed, cured opsies collected along with plasma at the time of relapse con- patients who were inconsistently judged as interim PET/CT-positive firmed that ctDNA also mirrors the tumor genotype at this point had a more than 2-log drop in ctDNA, whereas relapsing patients of the disease (Figure 1E). who were inconsistently judged as interim PET/CT negative had a less than 2-log drop in ctDNA (Figure 7A). The lack of correlation Overall, the mutational profiles of newly diagnosed and refractory cHL between log fold change of maximum standardized uptake value were largely overlapping (Figure 6A), suggesting that gene mutations and log fold change of ctDNA between baseline and interim not covered by our target region or molecular mechanisms not evaluations (supplemental Figure 11) is consistent with the notion fl captured by CAPP-seq might contribute to the chemorefractory that maximum standardized uptake value largely re ects the fl phenotype. The origin of ctDNA mutations from clonal hemato- metabolic activity of the in ammatory component of the mass in fl poiesis selected by chemotherapy rather than cHL was ruled out by cHL, whereas ctDNA re ects the tumor load levels. the observation that they did not occur in paired peripheral blood granulocytes collected at the same time of the plasma sample.29 Immune checkpoint inhibitors can cause false-positive PET/CT results in cHL as a result of tumor flares or pseudoprogression.17 To define cHL clonal evolution patterns, we genotyped longitu- We followed longitudinal plasma samples collected during dinal ctDNA samples (n 5 41 from 13 patients) collected before therapy with nivolumab from 5 patients with refractory cHL. frontline treatment (ie, ABVD), at the time of relapse, and during Among 4 patients stabilized in partial remission, ctDNA load did fl salvage therapies comprising transplantation, brentuximab not change under nivolumab treatment, thus re ecting the vedotin, and nivolumab. The amount of input cfDNA for library persistence of the tumor (supplemental Figure 9). One patient preparation and sequencing coverage were similar across longi- achieved a PET/CT complete remission with clearance of ctDNA. tudinal samples, indicating that differences in the representation of He became ctDNA-positive during a PET/CT, and biopsy con- fi fl mutations was not the result of a difference in the sensitivity of the rmed tumor are (supplemental Figure 1E). CAPP-seq (supplemental Figure 2C-D). Clonal shifts between pretreatment and relapse samples were documented in all cases (Figure 6B-C), thus demonstrating that evolution after therapy is the rule. In patients relapsing under/after chemotherapy or brentuximab Discussion vedotin, diagnosis/relapse tumor pairs branched through the ac- The study establishes ctDNA as source of tumor DNA for cHL fi quisition of phase-specific mutations from an ancestral clone that mutational pro ling. By overcoming the major technical hurdles persisted throughout disease course (Figure 6D; supplemental that have so far limited cHL genotyping, our technical approach Figure 8). Mutations of STAT6, GNA13, ITPKB,andTNFAIP3 were based on ctDNA has allowed large-scale assessment of muta- preferentially inferred in the ancestral clones, indicating that they are tions in different clinical phases of the disease, ranging from an early event in cHL (Figure 6D; supplemental Figure 8). Overall, newly diagnosed to refractory disease, and longitudinally during these data point to ctDNA genotyping as a tool for the noninvasive disease treatment. monitoring of cHL clonal evolution and suggest that in a refractory setting, chemotherapy and chemotoxins only partially reshape the Despite targeted resequencing being underpowered compared subclonal composition, leaving intact the ancestral clones. with exome sequencing in discovering new candidates, this study refines the current knowledge of cHL genetics. First, fi Among patients maintaining a partial response under nivolu- STAT6 is identi ed as the most frequently mutated gene in cHL, mab, ancestral clones were cyclically suppressed and replaced which is in keeping with the known importance of cytokine 2,13,25 by novel clones harboring new mutations (supplemental signaling in the biology of this tumor, but was not reported 19 Figure 9). Such a pattern might be interpreted as a drug-promoted in previous exome sequencing studies of cHL cases. Second, fi aggression against cancer neoantigens stemming from mutations mutation recurrence of previously discovered genes is de ned and the attempt made by the tumor to evade treatment by more precisely. As an example, our approach has narrowed the generating new mutations. estimate of the recurrence of B2M mutations in cHL, which is lower than previously reported,19 and consistent with that observed in another cHL study.31 Third, the notion that different Dynamics of circulating tumor DNA during therapy histologic subtypes of cHL are biologically distinct, as previously complements dynamics of PET/CT in cHL shown at the transcriptome level,25 is extended and reinforced at response assessment the genetic level.19,20 Fourth, a few major pathways emerged as We followed serial plasma samples during therapy in a cohort of recurrently mutated, including NF-kB, PI3K-AKT, cytokine and 24 patients with advanced cHL treated with ABVD. Patients NOTCH signaling, and immune evasion. Of note, these path- achieving complete response and cure had a larger drop in ways have been previously identified by gene expression pro- ctDNA load after 2 ABVD courses compared with relapsing filing and functional genomic studies of cHL,13,21,22,24,25,32 patients (Figure 7A-B), and the magnitude of the drop was indicating that mutations act as red flags highlighting cellular maintained until the end of the therapy (Figure 7C). A drop of programs that are relevant for the biology of the disease and 100-fold or 2-log drop in ctDNA after 2 chemotherapy courses, potential therapeutic targets.

2422 blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 SPINA et al From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

A B D

4.5 10000 cHL 66 UPN6 UPN25 cHL 85 cHL 67 cHL 52 cHL 19 UPN4 cHL 23 cHL 8 UPN3 cHL 63 cHL 78 cHL 64 cHL 65 cHL 13 UPN11 cHL 74 cHL 75 cHL 76 cHL 79 cHL 80 cHL 83 cHL 86 Deauville 1 3333344 12224 4 5 4 121 111113 score PD PD PD PD PD PD CR CR CR CR CR CR CR CR CR CR CR CR CR CR CR CR CR CR Outcome 4.0 cHL subtype 0 1000 NSCHL MCCHL 3.5 -1 LRCHL 100 3.0 -2 10 after 2 ABVD courses 2.5 Change in the hGE of ctDNA -3 Standardized log-rank statistic 1 0 p=0.001 2.0

Log fold change in tumor ctDNA -4

-4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 Cured ND disease Log Progressive C E 100 > -2 log fold reduction 1 80 0

-1 60

-2 40

-3 20 -4 Cumulative probability of PFS (%)

Log fold change in tumor ctDNA ≤ 0 -2 log fold reduction p<0.001 ND 0 50 100 150 200 0 24487296 Days from start of therapy Months iPET positive – Progressive disease 18 15 2 2 0 iPET positive – Cured 6 0 0 0 0 iPET negative – Progressive disease iPET negative – Cured

Figure 7. Change in tumor ctDNA is a prognostic biomarker in cHL treated with chemotherapy. (A) Waterfall plot of the log-fold change in ctDNA load after 2 courses of ABVD in 24 advanced stage cHL cases. On top of the graph, the interim PET/CT response scored according to the Deauville criteria and the final outcome of the patient are indicated. Histological subtype of cHL are shown above the plot. Each column is color coded according to the interim PET/CT results and the final patient outcome. Levels of ctDNA are normalized to baseline levels. The dash line tracks the 22-log threshold (CR, complete remission and cure; iPET, interim PET/CT; ND, not detectable; PD, progressive disease). (B) Box plot showing the fold change of ctDNA after 2 ABVD courses among patients who progressed vs patients who were cured. The band inside the box is the median, the bottom and top of the box are the first and third quartiles, and the ends of the whiskers represent the range. P value by Mann- Whitney U test. (C) Spider plot showing the log-fold change in ctDNA load after 2 courses of ABVD and at the end of therapy. Dots represent individual ctDNA measurements at the specified points. Each line is color coded according to the interim PET/CT results and the final patient outcome. (D) Maximally selected log-rank statistics identifies ;2-log-fold reduction in ctDNA levels after 2 ABVD courses as the best cutoff for progression-free survival (PFS) anticipation. (E) Kaplan-Meier curve of PFS stratified according to whether a 2-log reduction in ctDNA was achieved or not after 2 ABVD courses in 24 advanced stage cHL cases. Among patients achieving more than 2-log reduction after 2 ABVD courses, the sole event registered was a death in remission. Conversely, among patients achieving less than 2-log reduction after 2 ABVD courses, all the events were progressions.

By longitudinally profiling patients with cHL treated with ABVD landscape of mutations, which are the major source of neo- or brentuximab vedotin, we provide the evidence for a reservoir antigens in cancer,34 is a mechanism of tumor escape from ancestral population that, as in other lymphoma models,33 is immunotherapy35,36 and might be at the basis of either the long- resistant to chemotherapy and propagates successive disease lasting partial responses or periodical tumor flares that patients relapse. New mutations appearing in ctDNA mark resistant typically experience under treatment with immune checkpoint clones that are selected during the clonal evolution process blockers.37,38 taking place under the selective pressure of treatment, although they may not be directly causative of relapse. Although interim PET/CT response assessment is a novel approach to refine management strategies before completing treatment in Immunotherapy is conversely capable of eradicating ancestral cHL,39-41 meta-analyses demonstrated a certain degree of inac- clones, and its selective pressure elicits a quick clonal evolution curacy of this application.42,43 To fill this gap, an area of growing that was unexpected in cHL, resulting in the complete reshape interest is pairing interim PET/CT with biomarkers to enhance their of the disease mutation profile in few months. The evolving cumulative predictive value.44,45 Our results provide the proof of

cHL GENOTYPING ON THE LIQUID BIOPSY blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 2423 From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

principle that ctDNA can measure residual disease during treat- to sample biobanking, molecular studies, and manuscript revision; ment in cHL. Moreover, our data generated the hypotheses that M. Martini and L.M.L. performed pathological revision, immunohistochemical studies, microdissections, and manuscript revision; L.C., S.A., ctDNA quantification after 2 chemotherapy courses may have V.R., and A.G. contributed to imaging revision and manuscript revision; prognostic implications, and that ctDNA may complement interim A.N., F.B., M.G., G.S., A. Santoro, F.C., E.Z., and G.G., contributed to PET/CT in informing on patients’ outcome. Incorporation of both data interpretation and manuscript revision; and S.H. and C.C.-S. PET/CT and ctDNA monitoring into clinical trials should allow us to provided key scientific insights and reagents and contributed to data precisely define their cumulative sensitivity and specificity in an- interpretation and manuscript revision. ticipating the clinical course of patients with cHL. Conflict-of-interest disclosure: D.R. received unrestricted research grants from Gilead and AbbVie and honoraria from Gilead, AbbVie, and Janssen. A. Santoro received honoraria from Merck Sharp & Dome, Bristol, and Acknowledgments Roche. The remaining authors declare no competing financial interests. This study was supported by Grant No. KFS-3746-08-2015, Swiss Cancer League, Bern, Switzerland; Grant No. 320030_169670/1, Swiss National Correspondence: Davide Rossi, Hematology, Oncology Institute of Southern Science Foundation, Bern, Switzerland; iCARE No. 17860, Associazione Switzerland and Institute of Oncology Research, 6500 Bellinzona, Switzerland; Italiana per la Ricerca sul Cancro and Unione Europea, Milan, Italy; e-mail: [email protected]. Fondazione Fidinam, Lugano, Switzerland; Nelia & Amadeo Barletta Foundation, Lausanne, Switzerland; Fondazione Ticinese per la Ricerca sul Cancro, Bellinzona, Switzerland; Grant No. RF2011-02346986, Min- istry of Health, Rome, Italy; Grant No. 16722 and Grant No. 20575, Italian Association for Cancer Research, Milan, Italy; Fondazione Regionale per Footnotes la Ricerca Biomedica, Milan, Italy; Fondazione Umberto Veronesi, Milan, Submitted 3 November 2017; accepted 11 February 2018. Prepublished Italy; Grant No. 10007, Special Program Molecular Clinical Oncology online as Blood First Edition paper, 15 February 2018; DOI 10.1182/ 5 x 1000, Associazione Italiana per la Ricerca sul Cancro, Milan, Italy; blood-2017-11-812073. Grant No. RF-2011-02349712, Ministero della Salute, Rome, Italy; and Grant No. 2015ZMRFEA_004, MIUR-PRIN, Rome, Italy. *V.S. and A.B. contributed equally to this study. Authorship †S.H., C.C.-S., and D.R. contributed equally to this study. Contribution: D.R. designed the study, interpreted data, and wrote The online version of this article contains a data supplement. the manuscript; V.S. and A.B. performed molecular studies and bioinformatics analysis and contributed to data interpretation and manuscript preparation; A. Cuccaro, A. Cocomazzi, A. Condoluci, S.L.L., E.C., There is a Blood Commentary on this article in this issue. A.A.M., A. Stathis, L.N., C.D., F.D., R.B., and B.G. provided study material and clinical data and contributed to manuscript revision; A.A. The publication costs of this article were defrayed in part by page charge and L.T.-D.-B. contributed to bioinformatics analysis and manuscript payment. Therefore, and solely to indicate this fact, this article is hereby preparation; M.D.T., G.F., M. Manzoni, F.G., and A. Cuccaro contributed marked “advertisement” in accordance with 18 USC section 1734.

REFERENCES 8. Herrera AF, Kim HT, Kong KA, et al. Next- clinical characteristics in Hodgkin lymphoma. generation sequencing-based detection of Clin Cancer Res. 2011;17(9):2885-2892. 1. Swerdlow SH, Campo E, Pileri SA, et al. The circulating tumour DNA After allogeneic 2016 revision of the World Health Organiza- stem cell transplantation for lymphoma. 15. Newman AM, Bratman SV, To J, et al. An tion classification of lymphoid neoplasms. Br J Haematol. 2016;175(5):841-850. ultrasensitive method for quantitating circu- Blood. 2016;127(20):2375-2390. lating tumor DNA with broad patient cover- 2. Mathas S, Hartmann S, Kuppers¨ R. Hodgkin 9. Camus V, Stamatoullas A, Mareschal S, et al. age. Nat Med. 2014;20(5):548-554. lymphoma: pathology and biology. Semin Detection and prognostic value of recurrent 16. Cheson BD, Fisher RI, Barrington SF, et al; Hematol. 2016;53(3):139-147. exportin 1 mutations in tumor and cell-free circulating DNA of patients with classical United Kingdom National Cancer Research 3. Armitage JO. Early-stage Hodgkin’s lym- Hodgkin lymphoma. Haematologica. 2016; Institute. Recommendations for initial evalu- phoma. N Engl J Med. 2010;363(7):653-662. 101(9):1094-1101. ation, staging, and response assessment of Hodgkin and non-Hodgkin lymphoma: the 4. Moghbel MC, Mittra E, Gallamini A, et al. 10. Swerdlow SH, Campo E, Harris NL, et al. WHO Lugano classification. J Clin Oncol. 2014; Response assessment criteria and their ap- Classification of Tumors of Haematopoietic 32(27):3059-3068. plications in lymphoma: part 2. J Nucl Med. and Lymphoid Tissues. Lyon: International 2017;58(1):13-22. Agency for Research on Cancer; 2017. 17. Cheson BD, Ansell S, Schwartz L, et al. Refinement of the Lugano classification lym- 5. Luminari S, Galimberti S, Versari A, et al. 11. Rossi D, Diop F, Spaccarotella E, et al. Diffuse phoma response criteria in the era of immu- Positron emission tomography response large B-cell lymphoma genotyping on the nomodulatory therapy. Blood. 2016;128(21): and minimal residual disease impact on liquid biopsy. Blood. 2017;129(14): 2489-2496. progression-free survival in patients with fol- 1947-1957. licular lymphoma. A subset analysis from the 18. Schmitz R, Hansmann ML, Bohle V, et al. FOLL05 trial of the Fondazione Italiana 12. Roemer MG, Advani RH, Redd RA, et al. TNFAIP3 (A20) is a tumor suppressor gene in Linfomi. Haematologica. 2016;101(2):e66-e68. Classical Hodgkin lymphoma with reduced Hodgkin lymphoma and primary mediastinal b2M/MHC class I expression is associated with B cell lymphoma. J Exp Med. 2009;206(5): 6. Scherer F, Kurtz DM, Diehn M, Alizadeh AA. inferior outcome independent of 9p24.1 sta- 981-989. High-throughput sequencing for noninvasive tus. Cancer Immunol Res. 2016;4(11):910-916. disease detection in hematologic malignan- 19. Reichel J, Chadburn A, Rubinstein PG, et al. cies. Blood. 2017;130(4):440-452. 13. Skinnider BF, Elia AJ, Gascoyne RD, et al. Flow sorting and exome sequencing reveal Signal transducer and activator of transcrip- the oncogenome of primary Hodgkin and 7. Vandenberghe P, Wlodarska I, Tousseyn T, tion 6 is frequently activated in Hodgkin and Reed-Sternberg cells. Blood. 2015;125(7): et al. Non-invasive detection of genomic im- Reed-Sternberg cells of Hodgkin lymphoma. 1061-1072. balances in Hodgkin/Reed-Sternberg cells in Blood. 2002;99(2):618-626. early and advanced stage Hodgkin’s lym- 20. Abbosh C, Birkbak NJ, Wilson GA, et al; phoma by sequencing of circulating cell-free 14. Hohaus S, Santangelo R, Giachelia M, et al. PEACE consortium. Phylogenetic ctDNA DNA: a technical proof-of-principle study. The viral load of Epstein-Barr virus (EBV) DNA analysis depicts early-stage lung cancer evo- Lancet Haematol 2015;2(2):e55-e65. in peripheral blood predicts for biological and lution. Nature. 2017;545(7655):446-451.

2424 blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 SPINA et al From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

21. Bargou RC, Emmerich F, Krappmann D, et al. elderly individuals with clonal hematopoiesis. refractory classic Hodgkin lymphoma. J Clin Constitutive nuclear factor-kappaB-RelA acti- Nat Genet. 2012;44(11):1179-1181. Oncol. 2017;35(19):2125-2132. vation is required for proliferation and survival of Hodgkin’s disease tumor cells. J Clin Invest. 30. Kurtz MD, Scherer F, Newman AM, et al. 39. Johnson P, Federico M, Kirkwood A, et al. 1997;100(12):2961-2969. Prediction of therapeutic outcomes in DLBCL Adapted treatment guided by interim PET-CT from circulating tumor DNA dynamics. J Clin scan in advanced Hodgkin’s lymphoma. 22. Ranuncolo SM, Pittaluga S, Evbuomwan MO, Oncol. 2016;34(15 suppl):7511. N Engl J Med. 2016;374(25):2419-2429. Jaffe ES, Lewis BA. Hodgkin lymphoma re- quires stabilized NIK and constitutive RelB 31. Tiacci E, Penson A, Schiavoni G, et al. New 40. Press OW, Li H, Schoder¨ H, et al. US in- expression for survival. Blood. 2012;120(18): recurrently mutated genes in classical Hodg- tergroup trial of response-adapted therapy for 3756-3763. kin lymphoma revealed by whole-exome se- stage III to IV Hodgkin lymphoma using early quencing of microdissected tumor cells. interim fluorodeoxyglucose-positron emission 23. Young RM, Shaffer AL III, Phelan JD, Staudt LM. Blood. 2016;128:1088. tomography imaging: Southwest Oncology B-cell receptor signaling in diffuse large B-cell Group S0816. J Clin Oncol. 2016;34(17): lymphoma. Semin Hematol. 2015;52(2):77-85. 32. Roemer MG, Advani RH, Ligon AH, et al. 2020-2027. PD-L1 and PD-L2 genetic alterations define 24. Meadows SA, Vega F, Kashishian A, et al. classical Hodgkin lymphoma and predict 41. Borchmann P, Haverkamp H, Lohri A, et al. d PI3K inhibitor, GS-1101 (CAL-101), attenu- outcome. J Clin Oncol. 2016;34(23): Progression-free survival of early interim PET- ates pathway signaling, induces apoptosis, 2690-2697. positive patients with advanced stage Hodg- and overcomes signals from the microenvi- kin’s lymphoma treated with BEACOPP 33. Okosun J, Bod¨ or¨ C, Wang J, et al. Integrated escalated ronment in cellular models of Hodgkin lym- alone or in combination with rituximab (HD18): genomic analysis identifies recurrent muta- phoma. Blood. 2012;119(8):1897-1900. an open-label, international, randomised phase tions and evolution patterns driving the initi- 3 study by the German Hodgkin Study Group. 25. Tiacci E, Doring¨ C, Brune V, et al. Analyzing ation and progression of follicular lymphoma. Lancet Oncol. 2017;18(4):454-463. primary Hodgkin and Reed-Sternberg cells to Nat Genet. 2014;46(2):176-181. capture the molecular and cellular patho- 42. Terasawa T, Lau J, Bardet S, et al. Fluorine-18- 34. Verdegaal EM, de Miranda NF, Visser M, et al. genesis of classical Hodgkin lymphoma. fluorodeoxyglucose positron emission to- Neoantigen landscape dynamics during hu- Blood. 2012;120(23):4609-4620. mography for interim response assessment of man melanoma-T cell interactions. Nature. advanced-stage Hodgkin’s lymphoma and 26. Knutson SK, Kawano S, Minoshima Y, et al. 2016;536(7614):91-95. Selective inhibition of EZH2 by EPZ-6438 diffuse large B-cell lymphoma: a systematic leads to potent antitumor activity in EZH2- 35. Snyder A, Makarov V, Merghoub T, et al. review. J Clin Oncol. 2009;27(11):1906-1914. mutant non-Hodgkin lymphoma. Mol Cancer Genetic basis for clinical response to CTLA-4 43. Adams HJ, Nievelstein RA, Kwee TC. Ther. 2014;13(4):842-854. blockade in melanoma. N Engl J Med. 2014; 371(23):2189-2199. Prognostic value of interim FDG-PET in 27. Schwarzer R, Dorken¨ B, Jundt F. Notch is an Hodgkin lymphoma: systematic review and essential upstream regulator of NF-kB and is 36. Anagnostou V, Smith KN, Forde PM, et al. meta-analysis. Br J Haematol. 2015;170(3): relevant for survival of Hodgkin and Reed- Evolution of neoantigen landscape during 356-366. Sternberg cells. Leukemia. 2012;26(4): immune checkpoint blockade in non-small cell 806-813. lung cancer. Cancer Discov. 2017;7(3): 44. Agostinelli C, Gallamini A, Stracqualursi L, 264-276. et al. The combined role of biomarkers and 28. Challa-Malladi M, Lieu YK, Califano O, et al. interim PET scan in prediction of treatment Combined genetic inactivation of 37. Ansell SM, Lesokhin AM, Borrello I, et al. PD-1 outcome in classical Hodgkin’s lymphoma: a b2-microglobulin and CD58 reveals frequent blockade with nivolumab in relapsed or re- retrospective, European, multicentre cohort escape from immune recognition in diffuse fractory Hodgkin’s lymphoma. N Engl J Med. study. Lancet Haematol. 2016;3(10): large B cell lymphoma. Cancer Cell. 2011; 2015;372(4):311-319. e467-e479. 20(6):728-740. 38. Chen R, Zinzani PL, Fanale MA, et al; 45. Herrera AF, Armand P. Minimal residual disease 29. Busque L, Patel JP, Figueroa ME, et al. KEYNOTE-087. Phase II study of the efficacy assessment in lymphoma: methods and appli- Recurrent somatic TET2 mutations in normal and safety of pembrolizumab for relapsed/ cations. J Clin Oncol. 2017;35(34):3877-3887.

cHL GENOTYPING ON THE LIQUID BIOPSY blood® 31 MAY 2018 | VOLUME 131, NUMBER 22 2425 From www.bloodjournal.org by Fredrik Schjesvold on August 4, 2018. For personal use only.

2018 131: 2413-2425 doi:10.1182/blood-2017-11-812073 originally published online February 15, 2018

Circulating tumor DNA reveals genetics, clonal evolution, and residual disease in classical Hodgkin lymphoma

Valeria Spina, Alessio Bruscaggin, Annarosa Cuccaro, Maurizio Martini, Martina Di Trani, Gabriela Forestieri, Martina Manzoni, Adalgisa Condoluci, Alberto Arribas, Lodovico Terzi-Di-Bergamo, Silvia Laura Locatelli, Elisa Cupelli, Luca Ceriani, Alden A. Moccia, Anastasios Stathis, Luca Nassi, Clara Deambrogi, Fary Diop, Francesca Guidetti, Alessandra Cocomazzi, Salvatore Annunziata, Vittoria Rufini, Alessandro Giordano, Antonino Neri, Renzo Boldorini, Bernhard Gerber, Francesco Bertoni, Michele Ghielmini, Georg Stüssi, Armando Santoro, Franco Cavalli, Emanuele Zucca, Luigi Maria Larocca, Gianluca Gaidano, Stefan Hohaus, Carmelo Carlo-Stella and Davide Rossi

Updated information and services can be found at: http://www.bloodjournal.org/content/131/22/2413.full.html Articles on similar topics can be found in the following Blood collections Lymphoid Neoplasia (2846 articles) Plenary Papers (517 articles)

Information about reproducing this article in parts or in its entirety may be found online at: http://www.bloodjournal.org/site/misc/rights.xhtml#repub_requests

Information about ordering reprints may be found online at: http://www.bloodjournal.org/site/misc/rights.xhtml#reprints

Information about subscriptions and ASH membership may be found online at: http://www.bloodjournal.org/site/subscriptions/index.xhtml

Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by the American Society of Hematology, 2021 L St, NW, Suite 900, Washington DC 20036. Copyright 2011 by The American Society of Hematology; all rights reserved.