CLINICAL CANCER RESEARCH | PRECISION MEDICINE AND IMAGING Pretreatment Tissue TCR Repertoire Evenness Is Associated with Complete Pathologic Response in Patients with NSCLC Receiving Neoadjuvant Chemoimmunotherapy Marta Casarrubios1, Alberto Cruz-Bermu dez1, Ernest Nadal2, Amelia Insa3, María del Rosario García Campelo4, Martín Lazaro5, Manuel Do mine6, Margarita Majem7, 8 9 Delvys Rodríguez-Abreu , Alex Martínez-Martí , Javier de Castro-Carpen ~o10, Manuel Cobo11, Guillermo Lo pez-Vivanco , Edel Del Barco , Reyes Bernabe 12 13  Caro , Nuria Vin 14 ~olas15, Isidoro Barneto Aranda , Santiago Viteri , Bartomeu Massuti , Miguel Barquín1, Raquel Laza-Briviesca1, 16 17 18 Belen Sierra-Rodero1, Edwin R. Parra19, Beatriz Sanchez-Espiridion19, Pedro Rocha19, Humam Kadara19, Ignacio I. Wistuba19, Atocha Romero1, Virginia Calvo1, and Mariano Provencio1 Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 ABSTRACT ◥ Purpose: Characterization of the T-cell receptor (TCR) reper- space of CPR patients, achieving an AUC ROC to identify CPR toire may be a promising source for predictive biomarkers of patients of 0.967 (95% confidence interval, 0.897–1.000; P ¼ 0.001), pathologic response to immunotherapy in locally advanced non– and improving the results of PD-L1 tumor proportion score (TPS; small cell lung cancer (NSCLC). AUC ¼ 0.767; P ¼ 0.026) or tumor mutational burden (TMB; AUC Experimental Design: In this study, next-generation TCR ¼ 0.550; P ¼ 0.687). Furthermore, tumors with high pretreatment sequencing was performed in peripheral blood and tissue sam- top 1% clonal space showed similar immune cell populations but a ples of 40 patients with NSCLC, before and after neoadjuvant higher immune reactive gene expression profile. Finally, the selec- chemoimmunotherapy (NADIM clinical trial, NCT03081689), tive expansion of pretreatment tissue top 1% clones in peripheral considering their complete pathologic response (CPR) or non- blood of CPR patients suggests also a peripheral immunosurveil- CPR. Beyond TCR metrics, tissue clones were ranked by their lance, which could explain the high survival rate of these patients. frequency and spatiotemporal evolution of top 1% clones was Conclusions: We have identified two parameters derived from determined. TCR repertoire analysis that could outperform PD-L1 TPS Results: We have found a positive association between an uneven and TMB as predictive biomarkers of CPR after neoadjuvant TCR repertoire in tissue samples at diagnosis and CPR at surgery. chemoimmunotherapy, and unraveled possible mechanisms of Moreover, TCR most frequently ranked clones (top 1%) present in CPR involving enhanced tumor immunogenicity and peripheral diagnostic biopsies occupied greater frequency in the total clonal immunosurveillance. disease that includes patients with potentially resectable tumors that Introduction could theoretically be cured. Lung cancer is one of the main leading causes of death by cancer Recently, in NADIM trial (NCT03081689; ref. 1), we have shown worldwide. Approximately 85% of cases are non–small cell lung that patients treated with neoadjuvant chemoimmunotherapy cancer (NSCLC), of which one third are diagnosed at locally achieved a progression-free survival (PFS) and overall survival (OS) advanced or stage III disease. Stage III NSCLC is a heterogeneous rates at 2 years of 77% and 90%, respectively, whereas 63% of patients 1 Servicio de Oncología Medica, Instituto de Investigacio n Sanitaria Puerta de Note: Supplementary data for this article are available at Clinical Cancer Hierro-Segovia de Arana (IDIPHISA), Hospital Universitario Puerta de Hierro- Research Online (http://clincancerres.aacrjournals.org/). Majadahonda, Madrid, Spain. 2Institut Catala d’Oncologia, L’Hospitalet de Llo- n INCLIVA, Hos- dez contributed equally to this article. M. Casarrubios and A. Cruz-Bermu bregat, L’Hospitalet De Llobregat, Barcelona, Spain. 3Fundacio pital Clínico Universitario de Valencia, Valencia, Spain. 4Hospital Universitario A dez, Servicio de Oncología Me dica, Corresponding Authors: Alberto Cruz-Bermu Corun ~a, A Corun ~a, Spain. 5Hospital Universitario de Vigo, Pontevedra, Spain. n, Sanitaria Puerta de Hierro, Hospital Universitario Instituto de Investigacio 6 Hospital Universitario Fundacio n Jimenez Díaz, Madrid, Spain. 7Hospital de la Puerta de Hierro-Majadahonda, Madrid 28222, Spain. E-mail: Santa Creu i Sant Pau, Barcelona, Spain. 8Hospital Insular de Gran Canaria, Las alberto.cruz.bermudez@gmail.com; and Mariano Provencio, Palmas, Spain. 9Hospital Universitario e Instituto de Oncología Vall d’Hebron mprovenciop@gmail.com. (VHIO), Barcelona, Spain. 10Hospital Universitario La Paz, Madrid, Spain. 11Hos- pital Universitario Regional de Malaga, Malaga, Spain. 12Hospital Universitario Clin Cancer Res 2021;27:5878–90 Cruces, Barakaldo, Spain. 13Hospital Universitario de Salamanca, Salamanca, Spain. 14Hospital Universitario Virgen del Rocio, Sevilla, Spain. 15Hospital Clínic, doi: 10.1158/1078-0432.CCR-21-1200 Barcelona, Spain. 16Hospital Universitario Reina Sofia, Co rdoba, Spain. 17Instituto Oncolo gico Dr. Rosell. Hospital Universitario Quiron Dexeus, Grupo QuironSalud, This open access article is distributed under Creative Commons Attribution- Barcelona, Spain. 18Hospital General de Alicante, Alicante, Spain. 19Department NonCommercial-NoDerivatives License 4.0 International (CC BY-NC-ND). of Translational Molecular Pathology, The University of Texas MD Anderson Cancer Center, Houston, Texas. 2021 The Authors; Published by the American Association for Cancer Research AACRJournals.org | 5878 TCR Repertoire Is Associated with CPR to Chemoimmunotherapy with a proimmflamatory state in the tumor microenvironment of CPR Translational Relevance patients. The advance of chemoimmunotherapy in locally advanced stages of lung cancer has positioned complete pathologic responses (CPR) as a new relevant clinical entity, with implica- Materials and Methods tions for both the differential biology behind these responses Study design and sample collection and their possible use as an endpoint for assessing therapy All the studies presented here regarding TCR repertoire analysis are efficacy. In this exploratory analysis of the NADIM trial, two exploratory in nature and hypothesis generating and therefore will biomarkers [top 1% clonal space and T-cell receptor (TCR) require validation in larger cohorts. Forty-six patients with resectable evenness], associated with T-cell repertoire imbalance, outper- stage IIIA from NADIM clinical trial (NCT03081689) were treated formed the established biomarkers PD-L1 and TMB, regarding with three cycles of nivolumab plus chemotherapy prior to surgery. All CPR prediction after chemoimmunotherapy. In addition, some patients with enough material were included in the study. Patients mechanistic insights are revealed, which imply a higher immu- were classified in two categories according to their tumor pathologic nogenicity of tumors with high top 1% clonal space, as well as response: CPR patients (i.e., 0% of viable tumor cells in tumor bed or the presence of a distinctive peripheral immunosurveillance of any lymph node analyzed) and non-CPR patients (i.e., patients with pretreatment tissue top 1% clones, in patients achieving CPR. any percentage of viable tumor cells in resection specimens). Informed Finally, although these findings have potential clinical impact Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 consent for the collection of research samples and study protocol were and are hypothesis generating, they are exploratory and need to approved by the clinical research ethics committee of Hospital Puerta be confirmed in additional, larger cohorts. de Hierro in accordance with the International Conference on Harmo- nization Guidelines on Good Clinical Practice and the Declaration of Helsinki. Written informed consent to participate in the study was obtained from all patients. achieve a complete pathologic response (CPR). In addition, the PFS Up to 125 samples of peripheral blood (n ¼ 65) and tissue (n ¼ 60) rate in patients with CPR was higher than in non-CPR patients. from these 46 patients were prospectively collected and sequenced to Although these are encouraging results, with most patients maintain- determine their TCR repertoire. TCR sequencing was carried out in ing no evidence of disease status after surgery, some of them do not peripheral blood mononuclear cell (PBMC) samples collected before achieve complete responses and relapse or eventually die. Similar (n ¼ 30) and after (n ¼ 35) neoadjuvant treatment and in tissue results from the phase III Checkmate 816 trial reinforce the superiority samples obtained at diagnosis (n ¼ 22) and at surgery (n ¼ 38). RNA of chemoimmunotherapy compared with chemotherapy alone in the was used instead of DNA to decrease material input requirements and rate of CPR (2). Thus, the identification of biomarkers of CPR to maximize valid samples for sequencing. Details of all patients, samples, chemoimmunotherapy induction is therefore a priority in the future and techniques used are summarized in Supplementary Table S1. In scenario of resectable NSCLC. addition, a summary table with the number of paired samples analyzed In the context of immunotherapy, both the determination of PD-L1 for each technique and timepoint is shown in Supplementary Table S2. and tumor mutational burden (TMB) have been carried out to predict All molecular techniques were carried out in a blind fashion, only the which patients are more likely to respond to treatment (3, 4). However, pre- or posttreatment sample origin was known to the researchers the results are not consistent (5–7) and their value in the context of conducting the experiments. neoadjuvant chemoimmunotherapy is limited (1, 2). Thus, it is necessary to explore other biomarkers capable of predicting which RNA extraction patients will benefit most from this treatment (8, 9). PBMCs were isolated from blood samples by gradient density The mechanism by which the adaptive immune system responds to centrifugation using Lymphoprep (Alere Technologies) and cryopre- immunotherapy and recognizes tumor antigens relies mainly on the served until use. RNA from cryopreserved PBMCs was extracted using highly polymorphic T-cell receptors (TCR) present in an individual. Maxwell RSC simply RNA Cells Kit (Promega) as per manufacturer’s That is why the characterization of the TCR repertoire in terms of instructions. RNA from formalin-fixed paraffin-embedded (FFPE) clones, diversity, and antigen specificity by sequencing the CDR3 samples of biopsy at diagnosis and surgically resected specimens were hypervariable region, seems to be a promising approach (10, 11). extracted with the truXTRAC FFPE DNA Kit (Covaris). RNA quan- Other authors have previously reported that parameters extracted tification was carried out using the Qubit RNA BR Assay Kit (Thermo from the TCR repertoire were associated with response to immuno- Fisher Scientific, Catalog No. Q10210) on Qubit apparatus. therapy in different patients with cancer (12–15) and in particularly, NSCLC (16–18). However, it has not been demonstrated its predictive TCR library preparation value in pathologic response determination in neoadjuvant immuno- RNA extracted from PBMCs and FFPE samples was used to prepare therapy for patients with NSCLC (19). The mechanisms by which this the libraries for TCR sequencing. cDNA was obtained from RNA using response to immunotherapy occurs by lymphocytes has not been the SuperScript IV VILO Master Mix (Catalog No. 11756050). RNA elucidated either, although some authors hypothesize that peripheral input for PBMC-derived libraries was 25 and 100 ng for FFPE-derived lymphocytes are probably involved in the reinvigoration of the libraries. TCR libraries from PBMC samples were done using the response (19, 20). For these reasons, we propose the study of TCR Oncomine TCR Beta-LR Assay (Catalog No. A35386). For FFPE- repertoire and highlight its importance in mediating the response to derived RNA samples, TCR libraries were done using the Oncomine immunotherapy. TCR Beta-SR Assay (RNA; Catalog No. A39359). For PBMC-derived For the first time, we describe longitudinally and spatially the TCR libraries, equal volumes from 8 samples at 25 pmol/L were pooled for repertoire in patients with NSCLC receiving neoadjuvant chemoim- sequencing on an Ion 530 chip. For FFPE-derived libraries, equal munotherapy. We also assess its capacity as a possible source for volumes from up to 32 samples at 25 pmol/L were combined for predictive biomarkers of complete pathologic response and its relation sequencing on an Ion 540 chip. AACRJournals.org Clin Cancer Res; 27(21) November 1, 2021 5879 Casarrubios et al. TCR sequencing and data analysis Molecular Diagnostics) following manufacturer’s instructions. Ten Once the libraries were templated, they were sequenced in the Ion out of 11 patients sequenced had valid NGS data for further analysis. GeneStudio S5 Series (Thermo Fisher Scientific) and analysis was done Using the top 1% clonal space cohort median as threshold (≥0.166115), via Ion Reporter version 5.12 (Thermo Fisher Scientific). TCR con- patients with NGS data were divided in two groups (high and low vergence is determined as the aggregate frequency of clones, defined as pretreatment top 1% clonal space). unique TCRB nucleotide sequences, which shared a variable gene and The Bioconductor DESeq2 packages (version4; ref. 23) was used to CDR3 amino acid sequence with at least one other clone. Shannon’s perform the differential analysis expression, screening the differential diversity was calculated using the formula below, in which pi is the expressed genes (DEGs) between high top 1% and low top 1%. To frequency of clonotype i for the sample with N unique clonotypes (21). remove those low expressed genes, we applied an expression filter on X N row count data, selecting genes with, at least, five counts. A hierarchical Shannon Index H ¼  pi log2 ðpi Þ clustering heatmap was performed using pheatmap package in R. i ¼1 A selection of genes with threshold of absolute fold-change ≥ 1 and P < 0.05 were recruited to perform the gene ontology (GO) enrichment Evenness describes how evenly distributed is the TCR repertoire, analysis to identify significant biological processes among those genes. approaching 0 if the repertoire is unbalanced by a reduced number of To analyze the processes involved on each group (high top 1% and low predominant clones and approaching 1 if the repertoire is balanced, top 1%), two groups were established according to their fold-change with similar frequencies of all the clones. It is calculated by dividing the values: those values ≥1 (upregulated on high top 1% group) and values Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 Shannon’s diversity index by log2(N), where N is the total number of ≤1 (upregulated on low top 1%). The significant processes were detected clones in each sample. Evenness high and low categories were ordered according to their P value, selecting the 25 most significant defined using cohort median value (≥0.9077). processes for each group. This analysis was performed with R package Frequency classifications of the clones in the TCR repertoire were bcbioRNASeq (24). done according to their ranked relative abundance in the total T-cell repertoire, named as total clonal space, for each sample. We divided PD-L1 tumor proportion score and TMB assessment them into top 1%, top 1–2%, top 2–5%, and top >5% (19). In this way, As previously reported, the baseline TMB of FFPE tumor diagnostic the top 1% clonal space is defined as the aggregate frequencies of the samples was assessed using an Ion S5 Sequencer (Thermo Fisher top 1% most frequent clones. Top 1% high and low categories were Scientific) with the Oncomine Tumor Mutation Load Assay (Thermo defined using cohort median value (≥0.166115). Fisher Scientific) according to the manufacturer’s instructions. The When looking at the dynamics, we refer to contracted or expanded commercially available PD-L1 IHC assay PD-L1 IHC 22C3 pharmDx clones as those top 1% clones which their clonal frequency in periph- (Dako) was used to assess PD-L1 tumor proportion score (TPS) in eral blood decreases or increases after treatment. FFPE tumor diagnostic samples (1). Jaccard’s index was calculated as: Statistical analysis shared clones ðA þ BÞ Graphs were drawn using GraphPad PRISM v.6. and SPSS Statistics Ji ¼ v25 was used to analyze the data and determine the statistical signif- clones in A þ clones in B  share clones ðA þ BÞ icance, considered by an overall P value <0.05. Statistical significance was adjusted for multiple testing by Bonferroni correction when Tumor microenvironment lymphocytes and gene expression appropriate and was indicated in figure caption. Two-tailed, nonpara- analysis metric Mann–Whitney U test was performed to compare CPR versus Tumor immune cells were identified from FFPE diagnostic biopsies non-CPR groups. Wilcoxon matched-pairs signed rank test was of 11 patients through multiplex immunofluorescence and quantified adopted to make comparisons between pre- and post-neoadjuvant considering tumor or stroma localization as described previous- treatment in paired samples. Spearman rank nonparametric test was ly (1, 22). The immunofluorescence markers used were grouped used for variable correlations. High and low pretreatment top 1% into two 6-antibody panels and allowed the quantification of: clonal space patient groups were defined using the pretreatment TCR total malignant cells or reactive epithelial cells [CKþ (AE1/AE3þ)], cohort median (n ¼ 22). Contingency table analysis when comparing malignant cells or reactive epithelial cells PD-L1þ (CKþPD-L1þ), top 1% high and low patients with clinicopathologic features was done total T cells (CD3þ), cytotoxic T cells (CD3þCD8þ), T cells using Fisher exact test or Chi-square test. The ROC curve analysis was antigen-experienced (CD3þPD-1þ), cytotoxic T cells antigen- done to determine the association of the variables studied with therapy experienced (CD3þCD8þPD-1þ), T cells PD-L1þ (CD3þPD- pathologic response. Regarding gene expression analysis, a multiple L1þ), cytotoxic T cells PD-L1þ (CD3þCD8þPD-L1þ), T cells testing correction was performed to reduce the number of false antigen-experienced PD-L1þ (CD3þPD-1þPD-L1þ), cytotoxic positives, using Benjamini–Hochberg procedure and an adjusted T cells antigen-experienced PD-L1þ (CD3þCD8þPD-1þPD-L1þ), P < 0.05 was considered statistically significant. total macrophages (CD68þ), macrophages PD-L1þ (CD68þPD- L1þ), cytotoxic T cells activated (CD3þCD8þGranzyme Bþ), Availability of data memory T cells (CD3þCD45ROþ), effector/memory cytotoxic T The datasets supporting this study are available from the corre- cells (CD3þCD8þCD45ROþ), regulatory T cells [(CD3þFoxP3þ)- sponding authors upon request. Raw sequencing data are not freely (CD3þCD8þFoxP3þ)], and memory/regulatory T cells available due to lack of specific authorization, regarding potential (CD3þCD45ROþ FoxP3þ). Spearman nonparametric test was relevant germline information release, present in the original consent used for correlations between different cell subpopulations and form signed by the patients during trial enrollment. However, raw data pretreatment top 1% clonal space. are available upon request and complete TCR clone lists identified in A single section of FFPE tissue was used to analyze gene expression pretreatment or posttreatment samples can be found in Supplemen- through the HTG EdgeSeq Precision Immuno-Oncology Panel (HTG tary Files S1 and S2, respectively. 5880 Clin Cancer Res; 27(21) November 1, 2021 CLINICAL CANCER RESEARCH TCR Repertoire Is Associated with CPR to Chemoimmunotherapy Results neoadjuvant treatment, an increase in evenness (P ¼ 0.028), and a Pretreatment tissue TCR evenness is associated with pathologic decrease in number of clones (P ¼ 0.037), between pre- and post- response to neoadjuvant chemoimmunotherapy treatment samples seems to be observed in patients who had CPR, that Forty-six patients with resectable stage IIIA were treated with three were not considered statistically significant after Bonferroni’s correc- cycles of nivolumab plus chemotherapy prior to surgery; 41 of them tion (Fig. 1B). underwent surgical resection. The clinicopathologic characteristics of Extensive analysis to define which ranges of clones were responsible the patients analyzed are shown in Table 1. PBMC samples collected for CPR and non-CPR TCR evenness differences was done. The top 1% before (pre-) and after (post-) neoadjuvant treatment and tumor most common clones occupied higher clonal space, with respect to the samples obtained at diagnosis (pre-) and at surgery (post-) were total repertoire, in CPR patients than in non-CPR patients (P ¼ subjected to TCR sequencing (Supplementary Table S1). From 77 0.00015; Fig. 1C). Opposite results were found for clones below the tissue samples collected, we obtained valid TCR data for 60 (78%); top 5% most common clones (P ¼ 0.001; Fig. 1C). However, these three samples did not have sufficient tissue for RNA extraction and 14 differences were lost in surgery samples, presenting CPR and non-CPR did not yield enough RNA quantity for further analysis, however, all patients similar clonal space values for all clonal ranges. Evaluating samples with enough RNA quantity generated valid NGS data. On the treatment effect, although a decrease in the clonal space occupied by contrary, from 65 blood samples collected, we were able to obtain valid the top 1% seems to occur in posttreatment compared with pretreat- TCR data in all cases (100%). ment samples, this was not statistically significant (P ¼ 0.043; Fig. 1C). Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 As a first approach, TCR repertoire-derived metrics were analyzed. Regarding blood, there were no differences between CPR and non- Pretreatment tissue TCR evenness and top 1% outperform TMB CPR patients in both pre- or posttreatment PBMC samples in terms of and PD-L1 identifiying patients who achieve CPR clones sequenced, convergence, evenness, and diversity (Fig. 1A). The ability of these two potential tissue biomarkers to potentially Also, there were no differences between pre- and posttreatment PBMC predict the pathologic response of these patients was analyzed. The samples in all patients nor stratifying by pathological response for AUC ROC for TCR evenness to distinguish between CPR and non- these parameters (Fig. 1A). CPR was 0.844 [95% confidence interval (CI), 0.667–1.000; P ¼ 0.011]. However, when comparing tissue samples at diagnosis, we found An evenness value lower than 0.863 showed 50% sensitivity and 100% that CPR patients had significantly lower TCR evenness in the TCR specificity identifying patients that will achieve CPR, and therefore, repertoire than non-CPR patients (P ¼ 0.010), reflecting a skewed will be potentially free of disease at the time of surgery (Fig. 2A). repertoire in complete pathologic responders (Fig. 1B). No differences Determination of top 1% clonal space categorizes CPR and non-CPR were found in the other TCR metrics. In contrast, there were no patients with more accuracy than evenness, showing an AUC of 0.967 differences in evenness in surgery samples. Focusing on the effect of (95% CI, 0.897–1.000; P ¼ 0.001; Fig. 2A). Top 1% clonal space higher Table 1. Clinicopathologic characteristics of patients and their association with pretreatment tissue top 1% clonal space. Complete Pretreatment Top 1% clonal Top 1% P value (top 1% Clinicopathologic molecular TCR analysis space high clonal space clonal space characteristics cohort (n ¼ 44) cohort (n ¼ 22) (n ¼ 11) low (n ¼ 11) high vs. low) Age (years) 0.428a Median (IQR) 62 (58.5–70) 65 (59–71) 68 (59–71) 64 (56–71) Sex 0.586b Male 32 (72.73) 18 (81.81) 8 (72.73) 10 (90.91) Female 12 (27.27) 4 (18.18) 3 (27.27) 1 (9.09) Smoking status 1.000b Former 23 (52.27) 10 (45.45) 5 (45.45) 5 (45.45) Smoker 21 (47.73) 12 (54.55) 6 (54.55) 6 (54.55) Histology 0.475c Squamous 15 (34.09) 10 (45.45) 4 (36.36) 6 (54.55) Adenocarcinoma 25 (56.82) 11 (50.00) 6 (54.55) 5 (45.45) NOS 4 (9.09) 1 (4.55) 1 (9.09) 0 (0) Nodal stage 0.338b N0 9 (20.45) 6 (27.27) 4 (36.36)) 2 (18.18) N1 3 (6.82) 0 (0) 0 (0) 0 (0) N2 32 (72.73) 16 (72.73) 7 (63.64) 9 (81.82) Path response 0.006c CPR 25 (56.82) 10 (45.45) 9 (81.82) 1 (9.09) MPR 8 (18.18) 3 (13.64) 0 (0) 3 (27.27) IPR 7 (15.91) 6 (27.27) 1 (9.09) 5 (45.45) Non-resected 4 (9.09) 3 (13.64) 1 (9.09) 2 (18.18) Note: The data shown correspond to number and (%) or median and (IQR). Complete molecular cohort consisted in all patients from who at least one molecular analysis was done (TMB, PD-L1, TILs, HTG, or TCR determination). Pretreatment TCR cohort was composed of all patients from whom TCR metrics were determined in pretreatment tissue sample. Patients with TCR determination in pretreatment tissue were classified as top 1% high or low according to cohort median (0.166115). Abbreviations: IPR, minor pathologic response; IQR, interquartile range; MPR, major pathologic response; NOS, not otherwise specified. a P value was calculated using the Mann–Whitney U test. b P value was calculated using Fisher exact test. c P value was calculated using Chi-square test. AACRJournals.org Clin Cancer Res; 27(21) November 1, 2021 5881 Casarrubios et al. A PBMC samples B Tissue samples Pretreatment Posttreatment Pretreatment Posttreatment Pretreatment Posttreatment Pretreatment Posttreatment 60,000 1.0 90,000 1.0 0.010 60,000 30,000 15,000 Evenness Evenness 40,000 0.8 0.9 Clones Clones 10,000 20,000 0.6 0.8 5,000 0 0.4 0 0.7 CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR Pretreatment Posttreatment Pretreatment Posttreatment Pretreatment Posttreatment Pretreatment Posttreatment 0.10 Shannon's diversity 20 0.3 20 Shannon's diversity 0.2 0.1 0.08 Convergence Convergence 15 15 0.06 0.04 10 10 0.04 0.02 5 5 0.02 Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 0.00 0 0.00 0 CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR C TOP 1% TOP 1-2% TOP 2-5% TOP >5% pre-T post-T pre-T post-T pre-T post-T pre-T post-T 1.0 < 0.001 0.001 CPR Non-CPR 0.8 Clonal space in tissue CPR 0.6 MPR IPR 0.4 0.2 0.0 Figure 1. Pretreatment tissue TCR evenness is associated with pathologic response to neoadjuvant chemoimmunotherapy. A, Differences in metrics derived from TCR repertoire analysis (such as number of clones, evenness, convergence, and Shannon’s diversity index) in peripheral blood samples between CPR and non-CPR patients at pre- and post-neoadjuvant treatment. Pretreatment (pre-T) CPR patients, n ¼ 15; pre-T non-CPR patients, n ¼ 12; posttreatment (post-T) CPR patients, n ¼ 19; post-T non-CPR patients, n ¼ 13. B, Differences in metrics in tissue samples between CPR and non-CPR patients. Pre-T CPR patients, n ¼ 10; pre-T non-CPR patients, n ¼ 9; post-T CPR patients, n ¼ 24; post-T non-CPR patients, n ¼ 14. Comparisons were done between CPR and non-CPR patients and between pre- and posttreatment timepoints. C, Clonal space occupied by each of the percentage rank (top 1%, top 1–2%, top 2–5%, and top > 5%) of the total repertoire in tissue samples. Comparisons between CPR and non-CPR patients are shown for each rank. Pre-T CPR patients, n ¼ 10; pre-T non-CPR patients, n ¼ 9; post-T CPR patients, n ¼ 24; post-T non-CPR patients, n ¼ 14. Each patient is represented by a black symbol. P < 0.0125 was considered statistically significant after Bonferroni’s correction for multiple tests. Only significant differences after Bonferroni’s correction are shown. than 0.1842 showed 80% sensitivity and 100% specificity identifying specificity to classify patients achieving CPR after neoadjuvant che- patients that will achieve CPR after chemoimmunotherapy. Out of the moimmunotherapy than the commonly used PD-L1 and TMB. 19 total patients from our cohort with valid data, we were able to Regarding survival, although the number of patients is limited and classify correctly 8/10 of those with CPR using this cut-off. No due to the efficacy of the therapy the number of events is small, patients association between pretreatment tissue top 1% clonal space using with high pretreatment tissue top 1% clonal space seem to show a the cohort median as threshold with age, sex, smoking status, histol- better prognosis both in their PFS (P ¼ 0.053) and OS (P ¼ 0.059; ogy, or nodal stage at diagnosis was found (Table 1). Fig. 2B). Thus, in the low top 1% group 5 patients have progressed and We next compared the ability of TCR evenness and top 1% clones 3 of them have died, whereas in the high top 1% group only 1 patient with PD-L1 TPS and TMB to potentially predict CPR patients. The has progressed without any death. These patterns are not observed AUC ROCs for PD-L1 TPS and TMB to distinguish CPR and non-CPR for evenness, PD-L1, or TMB using median values as thresholds patients were 0.767 (95% CI, 0.578–0.955; P ¼ 0.026) and 0.550 (95% (Supplementary Fig. S1A). CI, 0.308–0.792; P ¼ 0.687), respectively (Fig. 2A). These results We also assessed the relationships between TCR evenness, top 1%, indicate that both evenness and top 1% showed better sensitivity and PD-L1 TPS, and TMB analyzing the correlations between them. 5882 Clin Cancer Res; 27(21) November 1, 2021 CLINICAL CANCER RESEARCH TCR Repertoire Is Associated with CPR to Chemoimmunotherapy A 1.0 B PFS OS 100 100 0.8 Sensitivity PFS (%) 0.6 OS (%) 50 P = 0.053 50 P = 0.059 0.4 TOP 1% TOP 1% High TOP 1% High Evenness TOP 1% Low TOP 1% Low 0.2 PD-L1 TMB 0 0 0 5 10 15 20 25 30 Months 0 5 10 15 20 25 30 Months 0.0 0.0 0.2 0.4 0.6 0.8 1.0 TOP 1% High 11(0) 11(0) 11(0) 11(0) 10(0) 6(4) 3(7) High 11(0) 11(0) 11(0) 11(0) 10(1) 6(5) 2(9) 1 – Specificity TOP 1% Low 11(0) 10(0) 9(0) 8(0) 7(0) 3(0) 2(0) Low 11(0) 11(0) 10(0) 10(0) 9(0) 9(0) 2(6) C 0.5 CPR D 0.5 0.6 MPR non-CPR R = 0.382 R = 0.230 0.4 IPR 0.4 Top 1% pre-T Top 1% pre-T Top 1% pre-T non-resected P = 0.087 P = 0.329 0.4 0.3 Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 0.3 0.2 0.2 0.2 R = −0.886 0.1 0.1 P < 0.001 0.0 0.0 0.0 0.75 0.80 0.85 0.90 0.95 1.00 0 20 40 60 80 100 0 5 10 15 20 25 Evenness pre-T PD-L1 TMB Figure 2. Pretreatment tissue TCR evenness and top 1% could be better predictors of pathologic response than TMB and PD-L1. A, ROC curve analysis for TMB (square, n ¼ 23), PD-L1 (asterisk, n ¼ 25), evenness (triangle, n ¼ 19), and top 1% (dot, n ¼ 19) determined in pretreatment tissue samples. B, PFS and OS percent survival stratified by pretreatment top 1% clonal space high and low patients (n ¼ 22). C, Correlation between clonal space occupied by top 1% clones and evenness in pretreatment tumor samples (n ¼ 22; CPR, n ¼ 10; non-CPR, n ¼ 9; nonresected, n ¼ 3). D, Correlation between clonal space occupied by the top 1% clones and PD-L1 in pre-T tissue samples (n ¼ 21; CPR, n ¼ 10; non-CPR, n ¼ 8; nonresected, n ¼ 3) and correlation between frequency of the top 1% clones and TMB in pretreatment tissue samples (n ¼ 20; CPR, n ¼ 9; non-CPR, n ¼ 8; nonresected, n ¼ 3). Each patient is represented by a dark gray (CPR), light gray (non-CPR), or white (nonresected) symbol. The black line indicates the linear regression line, and the dotted lines indicate the upper and lower boundaries of the 95% CI. Evenness and top 1% clonal space were strongly negatively correlated node biopsies, for both evenness or top 1% clonal space (Supplemen- (R ¼ 0.886; P < 0.001; Fig. 2C). However, there was no correlation tary Fig. S2D). Also, the differences seen in the frequency of the top 1% between top 1% clonal space and PD-L1 or TMB (R ¼ 0.382, P ¼ 0.087; were maintained regardless of the origin of the sample. Patients with R ¼ 0.230, P ¼ 0.329; Fig. 2D) nor between evenness and PD-L1 or CPR have a higher frequency of the top 1% than non-CPR patients in TMB (R ¼ 0.359, P ¼ 0.109; R ¼ 0.241, P ¼ 0.306; Supplementary both tumor tissue samples (P ¼ 0.030) and in lymph node samples (P Fig. S1B), strengthening the independent value of these biomarkers. ¼ 0.029; data not shown). Finally, we also checked that these metrics The presence of any mutations of potential clinical relevance, and the clonal reproducibility did not significantly vary between specifically KEAP1 (n ¼ 3), EGFR (n ¼ 1), TP53 (n ¼ 9), KRAS different libraries made from the same sample or within technical (n ¼ 3), and HNF1A (n ¼ 4) in this cohort, was not associated with sequencing replicates from the same library. TCR evenness and top 1% changes in tumor top 1% clonal space levels (Supplementary Fig. S1C; were stable across technical replicates (Supplementary Fig. S3A). Supplementary Tables S1 and S3). Clonal reproducibility was almost perfect between sequencing repli- cates, reaching a Jaccard’s index near 1 for all clonal ranges. However, Tissue TCR evenness and top 1% clonal space are independent of the clonal reproducibility between libraries was only maintained for main technical parameters top 1% and top 1% to 2%, decreasing for top 2% to 5% and being almost To assess the robustness of tissue TCR repertoire evenness and top gone for clones below the 5% most common clones (Supplementary 1% clonal space parameters, we tested whether different common Fig. S3B). technical factors (i.e read depth, identified clones, tissue origin, and library preparation) could influence these putative biomarkers. CPR patients showed a selective expansion of pretreatment No correlation between reported read count and evenness (R ¼ tissue top 1% clones in peripheral blood 0.079; P ¼ 0.748) or tissue top 1% clonal space (R ¼ 0.384; P ¼ In view of the importance of the top 1% frequency-ranked clones 0.104) was observed (Supplementary Fig. S2A). In addition, there present in pretreatment tissue samples, we studied the relevance of were no differences in read counts between samples of CPR and this top 1% in the different compartments, deepening in the study of non-CPR patients at diagnosis (P ¼ 0.182) or surgery (P ¼ 0.777; their individual dynamics in peripheral blood and tissue consider- Supplementary Fig. S2B). Also, no correlation was found between ing whether they were expanded or contracted during neoadjuvant the number of clones detected and evenness (R ¼ 0.149; P ¼ 0.542) treatment. or top 1% clonal space (R ¼ 0.193; P ¼ 0.429; Supplementary As a first approach in blood, we found no statistically significant Fig. S2C). differences between CPR and non-CPR patients in the clonal space Furthermore, these biomarkers were not strongly influenced by occupied by the top 1% clones of diagnostic tissue in pre- or post- biopsy’s origin. No differences were found between tumor or lymph treatment PBMC samples (pretreatment, P ¼ 0.234; posttreatment, AACRJournals.org Clin Cancer Res; 27(21) November 1, 2021 5883 Casarrubios et al. A 1.0 Pretreatment Posttreatment E Pretreatment Posttreatment 1.0 Clonal space occupied Clonal space occupied 0.8 CPR by top 1% clones 0.8 by top 1% clones 0.6 Non-CPR 0.6 <0.001 0.4 0.4 CPR 0.2 MPR 0.2 0.1 IPR 0.1 0.0 0.0 CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR B 0.005 F 100 100 % top 1% clones in blood 0.005 % top 1% clones in tissue 80 80 60 Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 60 40 40 20 20 0 Expanded Contracted 0 Expanded Contracted C G Tissue Tissue Contracted Contracted PBMCs PBMCs Expanded Expanded pretreatment posttreatment pretreatment posttreatment 1.0 1.0 0.8 1.0 0.8 0.002 1.0 0.6 Clonal space 0.6 log10 (fold change) Clonal space 0.5 log10 (fold change) 0.5 0.4 0.4 0.0 0.2 0.0 0.2 -0.5 0.10 0.10 -0.5 0.08 0.08 -1.0 0.06 0.06 0.04 -1.0 0.04 -1.5 CPR Non-CPR CPR Non-CPR CPR Non-CPR CPR Non-CPR 0.02 0.02 0.00 0.00 Expanded Contracted Expanded Contracted Expanded Contracted Expanded Contracted D PBMCs PBMCs H Tissue Tissue pretreatment posttreatment pretreatment posttreatment 0.04 0.08 Clonal space per clone Clonal space per clone 0.03 0.06 0.04 0.02 0.02 0.01 0.010 0.010 0.008 0.008 0.006 0.006 0.004 0.004 0.002 0.002 0.000 0.000 Expanded Contracted Expanded Contracted Expanded Contracted Expanded Contracted Figure 3. CPR patients showed a selective expansion of tissue top 1% clones in peripheral blood. A, Clonal space occupied by the top 1% tissue pre-T clones in pre- and posttreatment PBMC samples. Comparisons between CPR and non-CPR patients in pre- and posttreatment timepoints are shown (n ¼ 13; CPR, n ¼ 6; non-CPR, n ¼ 7). B, Percentage of top 1% tissue pre-T clones that were peripherally expanded or contracted (known as dynamic clones). Comparisons between CPR and non-CPR patients are shown (n ¼ 13; CPR, n ¼ 6; non-CPR, n ¼ 7). C, Clonal space occupied by dynamic clones in peripheral repertoire: PBMCs pre-T and PBMCs post-T. Expressed as frequency of clonal space occupied by the top 1% tissue pre-T clones and fold change between pre- and posttreatment timepoints. Comparisons between CPR and non-CPR patients are shown (n ¼ 13; CPR, n ¼ 6; non-CPR, n ¼ 7). D, Median contribution of peripherally expanded (n ¼ 13; CPR, n ¼ 6; non-CPR, n ¼ 7) or contracted clones (n ¼ 12; CPR, n ¼ 6; non-CPR, n ¼ 6) in pre- and posttreatment peripheral blood. Comparisons between CPR and non- CPR patients are shown. E, Clonal space occupied by the top 1% tissue pre-T in pre- and posttreatment tissue samples. Comparisons between CPR and non-CPR patients in pre- and posttreatment timepoints are shown (n ¼ 18; CPR, n ¼ 10; non-CPR, n ¼ 8). F, Percentage of top 1% tissue pre-T clones that were intratumorally expanded or contracted. Comparisons between CPR and non-CPR patients are shown (n ¼ 18; CPR, n ¼ 10; non-CPR, n ¼ 8). G, Clonal space occupied by top 1% dynamic clones in tissue: tissue pre-T and tissue post-T. Expressed as frequency of clonal space occupied and fold change between pre- and post-timepoints. Comparisons between CPR and non-CPR patients are shown (n ¼ 18; CPR, n ¼ 10; non-CPR, n ¼ 8). H, Median contribution of intratumorally expanded (n ¼ 16; CPR, n ¼ 10; non-CPR, n ¼ 6) or contracted clones (n ¼ 18; CPR, n ¼ 10; non-CPR, n ¼ 8) in pre- and posttreatment peripheral blood. Comparisons between CPR and non-CPR patients are shown. Each patient is represented by a symbol. P < 0.0125 was considered statistically significant after Bonferroni’s correction for multiple tests. Only significant differences after Bonferroni’s correction are shown. 5884 Clin Cancer Res; 27(21) November 1, 2021 CLINICAL CANCER RESEARCH TCR Repertoire Is Associated with CPR to Chemoimmunotherapy P ¼ 0.059). Similarly, there was also no significant changes after Despite the low number of cases analyzed, the RNA sequencing treatment in the clonal space of this pretreatment tissue top 1% clones (RNA-seq) results showed a differential gene expression profile for in blood for both CPR and non-CPR patients (Fig. 3A). tumors with high or low top 1% clonal space, as shown in the We also evaluated the percentage of peripherally expanded or hierarchical heatmap (Fig. 4B). Nearly 200 genes were differentially contracted pretreatment tissue top 1% clones. Patients achieving CPR expressed in high compared with low top 1% tumors (Fig. 4C). showed statistically significant lower percentage of expanded clones Specifically, 139 genes were upregulated, including IFNg (Log2FC and consequently, higher percentage of contracted clones compared 1.14, adjusted P value ¼ 0.022), IL2 (Log2FC 1.78, adjusted with non-CPR patients (P ¼ 0.005; Fig. 3B). However, there were no P value < 0.001), and IL13 (Log2FC 2.16, adjusted P value differences in the clonal space occupied by expanded or contracted < 0.001). Conversely, 53 genes were downregulated, including pretreatment tissue top 1% clones between pathologic response groups VEGFC (Log2FC 1.39, adjusted P value ¼ 0.008), MAPK1 in pre- or posttreatment PBMC samples nor in their fold changes (Log2FC 1.30, adjusted P value ¼ 0.008), or IGF1R (Log2FC 1.31, associated (Fig. 3C). adjusted P value < 0.001). A complete list of differentially expressed Because the percentage of expanded clones in CPR patients was genes can be found at Supplementary Table S4. lower than in non-CPR, but the clonal space remained similar, we Finally, to identify the main biological processes of differentially decided to evaluate the individual contribution of those clones to the expressed genes in tumors with high pretreatment top 1% clonal clonal space. To do this, we normalized the clonal space by the number space, GO enrichment analysis was carried out (Supplementary Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 of clones that were expanded or contracted, thus comparing the Table S5). Top 25 terms for upregulated genes include processes such average clonal space per clone between CPR and non-CPR patients. as: positive regulation of immune effector process (GO:0002699), The tissue top 1% clones that were expanded in the blood of CPR positive regulation of cell killing (GO:0031343), or regulation patients seems to have larger average size per clone than their non-CPR of receptor signaling pathway via JAK-STAT (GO:0046425). counterparts; both in the pre- (P ¼ 0.022) or post- (P ¼ 0.051) PBMCs’ On the other hand, top 25 terms for downregulated genes include clonal space (Fig. 3D). processes such as: epithelial cell proliferation (GO:0050673), cell- Concerning their implication in tissue, the clonal space occupied by cycle arrest (GO:0007050), positive regulation of angiogenesis the pretreatment top 1% clones in tissue samples was significantly (GO:0045766), negative regulation of extrinsic apoptotic signaling reduced after treatment for all patients (P ¼ 0.005). However, strat- pathway (GO:2001237), or positive regulation of protein serine/thre- ifying by pathological response, only CPR patients showed a decrease onine kinase activity (GO:0071902; Fig. 4D). (CPR patients, P ¼ 0.013; non-CPR patients, P ¼ 0.208; Fig. 3E). In addition, the space occupied by the pretreatment top 1% tissue clones Shared and newly emergent posttreatment top 1% clones are seems to be higher in CPR patients compared with non-CPR patients not informative of patient response after treatment, although no statistically significant differences were To reveal possible response mechanisms, we decided to analyze seen (P ¼ 0.237). also the role of the clones belonging to the top 1% posttreatment When looking at their intratumoral dynamics, we observed that tissue according to whether they belong to the pretreatment top 1%, more than 80% of them were contracted in post-neoadjuvant pretreatment non–top 1%, or were newly emerging clones (NEC; i.e., treatment tissue. Unlike blood, no differences were observed in not detected in pretreatment tissue). tissue top 1% dynamics between CPR and non-CPR patients Posttreatment tissue top 1% clonal space analysis showed no (Fig. 3F). Same as previously shown with PBMC samples, the statistically significant differences between these categories (P ¼ clonal space occupied by expanded clones nor the fold change 0.073), however the median clonal space of non–top 1% pretreatment between pre- and posttreatment tissue, showed differences between clones doubles the clonal space of NEC in posttreatment tissue (P ¼ CPR and non-CPR patients (Fig. 3G). In addition, there was a 0.09; Fig. 5A). Anyhow, no differences between CPR and non-CPR higher clonal space occupied by contracted clones in tissue pre- were observed in the relative space of these clone subgroups (Fig. 5B). treatment in CPR patients compared with non-CPR (P ¼ To determine the specific relevance of the pretreatment top 1% 0.002; Fig. 3G). However, no differences were seen in the average clones, their clonal space was compared with the rest of the clones (i.e., clonal space occupied by each clone in pre- and posttreatment the sum of the non–top 1% and NEC clones), showing the latter tissue, in CPR or non-CPR patients (Fig. 3H). category larger values, occupying near 70% of the posttreatment top 1% clonal space (P ¼ 0.010; Fig. 5C). Tumors with high pretreatment top 1% clonal space showed an Regarding NEC, their clonal space was compared with that occupied immune reactive gene expression profile by pretreatment shared clones. The clonal space of shared clones was To understand what characterizes the tumors with high or low top higher than that of NEC, occupying near 80% of the posttreatment top 1% clonal space, we analyzed the relationship of this biomarker with 1% clonal space (P ¼ 0.020; Fig. 5D). the pretreatment tumor immune infiltrate and the gene expression To analyze in blood the dynamics of the top 1% clones of post- profile. treatment samples, we selected those patients from whom we had In terms of immune cell content, no association was found paired samples of the four compartments (tissue and blood at the pre- between top 1% clonal space and tumor, stroma, or total levels of and posttreatment timepoints, n ¼ 12). Clonal space of tissue shared any immune cell subpopulations analyzed (data not shown). Only a clones were not altered in blood after treatment (Supplementary weak correlation was found between T lymphocytes infiltrating the Fig. S4A), being this behavior similar between CPR and non-CPR tumor (CD3þ cells) and the clonal space of the top 1% (R ¼ 0.609, patients (Supplementary Fig. S4B). Concerning NEC, an increase in P ¼ 0.047) that was not considered statistically significant after their clonal space in blood after treatment was observed (P ¼ 0.018; multiple comparisons correction (Fig. 4A). However, stratifiying Supplementary Fig. S4C). Analyzing it by response, it seems that this between high and low top 1% clonal space tumors, using cohort increase is mostly produced in non-CPR patients, showing a trend for median as cutoff, no differences were found in CD3þ levels higher NEC clonal space in posttreatment blood compared with CPR (Fig. 4A). patients (P ¼ 0.020; Supplementary Fig. S4D). AACRJournals.org Clin Cancer Res; 27(21) November 1, 2021 5885 Casarrubios et al. Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 Figure 4. Immune cells and gene expression analysis of tumors with high or low top 1% clonal space. A, Correlation between CD3þ tumor-infiltrating lymphocytes (cells per mm2) and top 1% clonal space in pretreatment tissue. Comparisons between high top 1% and low top 1% patients are shown. P < 0.001 was considered statistically significant after Bonferroni’s correction for multiple tests. B, Hierarchical clustered heatmap showing the expression patterns of all genes analyzed across tumors with high (pink) and low (cyan) top 1% clonal space. The red boxes indicate the upregulated genes, and the blue boxes indicate downregulated genes. C, Volcano plot showing the log10 of adjusted P value and log2 fold change of all genes studied. Red (upregulated) and blue (downregulated) dots represent genes with log2 fold change >|1| and statistically significant (adjusted P value <0.05). D, Dot plots of top 25 enriched GO pathways for downregulated and upregulated genes in tumor with high top 1% clonal space versus tumors with low top 1% clonal space. 5886 Clin Cancer Res; 27(21) November 1, 2021 CLINICAL CANCER RESEARCH TCR Repertoire Is Associated with CPR to Chemoimmunotherapy A B posttreatment clonal space 1.0 posttreatment clonal space 1.0 CPR 0.8 0.8 Relative top 1% Relative top 1% Non-CPR 0.6 0.6 CPR 0.4 0.4 MPR 0.2 IPR 0.2 0.0 0.0 Shared Shared NEC Shared Shared NEC Top 1% non-Top 1% Top 1% non-Top 1% Pre-T Pre-T clones Pre-T clones Pre-T clones clones C D 0.010 0.020 Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 posttreatment clonal space posttreatment clonal space CPR 1.0 1.0 MPR Non-CPR IPR Relative top 1% 0.8 Relative top 1% 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 Shared Shared Shared NEC Top 1% non-Top 1%+ NEC Top 1% Pre-T clones Pre-T clones Pre-T clones Figure 5. Composition of top 1% clones of surgical specimens. A, Relative top 1% posttreatment clonal space occupied by clones shared between pre- and posttreatment, belonging to the top 1% pre-T tissue or non–top 1%, and by NEC (n ¼ 18; CPR, n ¼ 10; non-CPR, n ¼ 8). Comparisons between shared top 1% pre-T clones, shared non– top 1% pre-T clones, and NEC were done. B, Relative top 1% posttreatment clonal space occupied by clones shared between pre- and posttreatment, belonging to the top 1% pre-T tissue or non–top 1%, and by NEC stratified by response (n ¼ 18; CPR, n ¼ 10; non-CPR, n ¼ 8). Comparisons between CPR and non-CPR patients were done. C, Comparison of the relative top 1% posttreatment clonal space occupied between shared top 1% clones and non–top 1% plus NEC (n ¼ 18). D, Comparison of the relative top 1% posttreatment clonal space occupied between total shared clones and NEC (n ¼ 18). Each patient is represented by a symbol. P < 0.05 was considered statistically significant. Only significant differences are shown. In this way, the study of the pretreatment tissue top 1% clonal Discussion space as potential biomarker is encouraging; it reflects tumor There is a need for predictive biomarkers since neither PD-L1 nor immunogenicity and is strongly associated with tumor response TMB are capable of accurately predicting the response to chemoim- being technically robust (21) and affordable considering turn- munotherapy in patients with advanced NSCLC (5–7), or neoadjuvant around times and costs. Thus, if its value is confirmed in larger chemoimmunotherapy in locally advanced patients (1, 2). In this cohorts with longer follow-ups for survival analysis, we believe that study, we performed TCR sequencing in peripheral blood and tissue the analysis of the top 1% clonal space could be implemented in the samples from patients enrolled in NADIM trial (NCT03081689), at clinic. This could allow the personalization of the follow-up and both pre- and post-neoadjuvant treatment timepoints. With this treatment of patients with low top 1% clonal space, who would approach, we have shown for the first time a relationship between presumably not achieve CPR, as well as would enable studies to an uneven TCR repertoire in tissue samples at diagnosis and the determine the value of surgery in patients that will achieve CPR, complete pathologic response in NSCLC patients treated with that likely would present high top 1% clonal space at diagnosis. chemoimmunotherapy. However, limitations similar to those of TMB, such as standard- We hypothesized that this uneven repertoire in CPR patients could ization between platforms and laboratories, as well as democrati- be caused by a small number of dominating clones in the total zation of NGS access, would have to be solved for an effective repertoire. When looking to the most frequently ranked clones, we application in clinical practice. The role of the most frequent clones saw that the top 1% clones occupied higher clonal space in complete as biomarkers was also assessed in other studies, in which a higher responder patients than in non–complete responders. Altogether, frequency of these top-ranked clones in peripheral blood was we have demonstrated that these two novel biomarkers derived associated with higher PFS rates at 9 months in patients treated from the TCR repertoire analysis could predict, with higher accu- with immunotherapy (12) and that a higher frequency of the top 1% racy than PD-L1 and TMB, patients that will be free of disease at clones present at resection was associated with MPR (19). However, time of surgery due to their association with CPR after neoadjuvant this is the first time that baseline TCR parameters showed an chemoimmunotherapy. association with CPR after chemoimmunotherapy. AACRJournals.org Clin Cancer Res; 27(21) November 1, 2021 5887 Casarrubios et al. We have shown that tumors with higher top 1% clonal space of in-between samples (19), the lack of a control group and a validation presented—despite having similar levels of PD-L1, TMB, or cohort, and the phenotype and antigen specificity of the sequenced TIL populations—a higher pro-inflammatory profile reflected by T cells (31). upregulation of biological processes such as leukocyte-mediated immunity, cell killing, and T-cell activation among others. This Conclusions stronger immunogenic profile could explain the repertoire imbal- In conclusion, baseline tissue TCR evenness and top 1% clonal ance and greater pathologic responses, since in such a permissive space are associated with complete pathologic response to neo- microenvironment the specific activation and clonal expansion adjuvant chemoimmunotherapy. In addition, although we cannot after antigen recognition of some antitumoral T-cell clones rule out the role of new clones and the tissue reinvigoration of would be more easily allowed (25, 26). In turn, it seems that this pretreatment clones, we describe the relevance of the peripheral repertoire imbalance does not depend exclusively on the TMB of selective expansion of tissue top 1% clones to achieve complete these tumors, but existing modifying elements between the muta- pathological response. Future studies are warranted in larger tions and their ability to elicit an immune response [i.e., neoantigen cohorts overcoming our limitations to validate the relevance of the presentation (27, 28), or immunosuppressive microenvironment TCR repertoire analysis. factors (29)], that the study of TCR repertoire would consider, but TMB alone would not. Authors’ Disclosures Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 Concerning the mechanism and involvement of pretreatment tissue A. Cruz-Berm udez reports non-financial support from Thermo Fisher Scientific top 1% clones in the response, there are different options that are during the conduct of the study. E. Nadal reports grants, personal fees, and non- compatible with each other. financial support from Bristol-Myers Squibb, Roche, and Pfizer; personal fees and On the one hand, it is possible that these clones were not involved in non-financial support from Merck Sharp & Dohme, Takeda, AstraZeneca, Lilly, and the response. Thus, new clones or clones belonging to the non–top 1% Boehringer Ingelheim; grants and personal fees from Merck Serono; and personal fees category would participate in the response, as would support: the from Amgen and Bayer outside the submitted work. A. Insa reports personal fees from increase of tissue NEC in blood posttreatment, the loss of clonal space Bristol, Boehringer Ingelheim, Roche, MSD-Merck, Amgen, and Takeda outside the submitted work. M.R. García Campelo reports personal fees from BMS during the of top 1% pretreatment clones in posttreatment tissue, and the rise of conduct of the study, as well as personal fees from MSD, Roche, AstraZeneca, Takeda, clonal space of pretreatment non–top 1% clones and NEC in top 1% Pfizer, Janssen, Novartis, and Lilly outside the submitted work. M. Lazaro reports clonal space of posttreatment tissue (which combined account for personal fees from BMS, Roche, and MSD outside the submitted work. M. Domine more clonal space than that of the pretreatment top 1% clones). reports personal fees from AstraZeneca, BMS, Boehringer Ingelheim, MSD, Pfizer, However, NEC in the top 1% posttreatment tissue seem to occupy Roche, and Takeda outside the submitted work. M. Majem reports personal fees from half of the space that pretreatment non–top 1% clones do, indicating BMS, MSD, Roche, Sanofi, AstraZeneca, and Boehringer Ingelheim outside the submitted work. D. Rodríguez-Abreu reports personal fees from Bristol Myers a greater importance of reinvigoration of preexisting tumor clones Squibb, Novartis, MSD, Eli Lilly, Pfizer, Roche, and AstraZeneca outside the sub- below the top 1% than of the emergence of new clones. In any case, the mitted work. A. Martínez-Marti reports personal fees and other support from Bristol fact that no differential behavior is observed between responses, Myers Squibb, F. Hoffmann-La Roche, Merck Sharp & Dohme, Pfizer, Boehringer reinforces that the NEC and non–top 1% clones do not seem to have Ingelheim, MSD Oncology, and AstraZeneca outside the submitted work. J. de a specific role to achieve CPR, however it does not rule out their Castro-Carpe~ no reports grants and personal fees from AstraZeneca, Bristol Myers possible participation in responses. Squibb, Merck Sharp & Dohme, and F. Hoffmann-La Roche, as well as personal fees from Boehringer Ingelheim, Amgen, Bayer, Lilly, Janssen-Cilag, Takeda, Novartis, On the other hand, we have shown how this top 1% remained in and Pfizer outside the submitted work. R. Bernabe Caro reports other support from tissue posttreatment and was also present in the blood. Thus, the Bristol Myers Squibb and AstraZeneca, as well as grants from Roche outside the possible role of these clones in the response may be exerted through submitted work. S. Viteri reports personal fees from BMS, Puma, AstraZeneca, Reddy tissue or peripheral mechanisms. In terms of tissue mechanisms, Pharma, Takeda, and Janssen, as well as personal fees and non-financial support from despite this drop of pretreatment tissue top 1% clones, they still MSD, Merck, Roche, and OSE Immunotheraputics outside the submitted work. account for 10% to 20% of the total posttreatment clonal space B. Massuti reports grants from Spanish Lung Cancer Group during the conduct of the (showing a tendency to be more relevant in CPR patients) and occupy study, as well as advisory board travel expenses from Bristol Myers Squibb. H. Kadara reports grants from Johnson & Johnson outside the submitted work. I.I. Wistuba near 30% of the posttreatment top 1% clonal space. Furthermore, this reports grants and personal fees from Genentech/Roche, Bayer, Bristol Myers Squibb, role may have been underestimated given that, at the time of surgery, AstraZeneca/Medimmune, Pfizer, HTG Molecular, Merck, Guardant Health, and there was a reduction of T cells probably related to tumor clearance (1). Novartis; personal fees from GlaxoSmithKline, Oncocyte, MSD, and Daiichi Sankyo; Thus, we cannot rule out the importance of reinvigoration of top 1% in and grants from Adaptive, 4D, Takeda, Karus, Adaptimmune, EMD Serono, Iovance, tumor elimination (30). Regarding the role of pretreatment tissue top Johnson & Johnson, Amgen, and Akoya outside the submitted work. A. Romero 1% clones in the peripheral response, we have seen that in our patients reports personal fees from Boehringer Ingelheim outside the submitted work. V. Calvo reports personal fees from Roche, BMS, MSD, AstraZeneca, and Boehringer the clones identified in tissue are found in blood and are maintained Ingelheim outside the submitted work. M. Provencio reports grants from BMS during after treatment. Furthermore, CPR patients showed a selective the conduct of the study; M. Provencio also reports grants and non-financial support expansion of pretreatment tissue top 1% clones in peripheral blood from Roche, as well as grants from AstraZeneca, Takeda, and MSD outside the compared with non-CPR patients. This could indicate peripheral submitted work. No disclosures were reported by the other authors. immunosurveillance, responsible for eliminating possible relapses at systemic level, which could explain the high survival rate of Authors’ Contributions patients with CPR (1) and the absence of deaths in the high top 1% M. Casarrubios: Formal analysis, investigation, visualization, methodology, writing– group. In this regard, other authors have previously described the original draft. A. Cruz-Berm udez: Conceptualization, resources, data curation, clonal changes in the peripheral TCR repertoire after immunother- formal analysis, supervision, funding acquisition, investigation, visualization, meth- odology, writing–original draft, project administration, writing–review and editing. apy, and identified clones that expanded or contracted peripherally E. Nadal: Resources, data curation, investigation, writing–review and editing. A. Insa: after treatment (12, 15, 19, 20). Resources, data curation, investigation, writing–review and editing. M.R. García Finally, further studies are needed to overcome the limitations of Campelo: Resources, data curation, investigation, writing–review and editing. our study, including but not limited to: the number of patients, the lack M. Lazaro: Resources, data curation, investigation, writing–review and editing. 5888 Clin Cancer Res; 27(21) November 1, 2021 CLINICAL CANCER RESEARCH TCR Repertoire Is Associated with CPR to Chemoimmunotherapy M. D omine: Resources, data curation, investigation, writing–review and editing. Acknowledgments M. Majem: Resources, data curation, investigation, writing–review and editing. Work in the authors’ laboratories was supported by Bristol-Myers Squibb; D. Rodríguez-Abreu: Resources, data curation, writing–review and editing. “Instituto de Salud Carlos III” (ISCIII) PI19/01652 grant cofunded by European A. Martínez-Marti: Resources, data curation, investigation, writing–review and Regional Development Fund (ERDF); Ministry of Science and Innovation RTC2017– editing. J. de Castro-Carpe~ no: Resources, data curation, investigation, writing– 6502–1 “INmunoSIGHT” and RTC2019–007359–1 “BLI-O”; and European Union’s review and editing. M. Cobo: Resources, data curation, investigation, writing– Horizon 2020 research and innovation programme CLARIFY 875160 grant, to review and editing. G. L opez-Vivanco: Resources, data curation, investigation, M. Provencio. Thermo Fisher Scientific provided reagents for TCR sequencing. writing–review and editing. E. Del Barco: Resources, data curation, investigation, A. Cruz-Berm udez received a Spanish Lung Cancer Group (SLCG) grant and is writing–review and editing. R. Bernabe Caro: Resources, data curation, investigation, supported by a ISCIII-“Sara Borrell” contract CD19/00170. M. Casarrubios was writing–review and editing. N. Vi~ nolas: Resources, data curation, investigation, supported by PEJD-2019-PRE/BMD-17006 contract granted to A. Cruz- writing–review and editing. I. Barneto Aranda: Resources, data curation, investi- Berm udez. M. Barquín was supported by an i-PFIS predoctoral fellowship (grant gation, writing–review and editing. S. Viteri: Resources, data curation, investigation, no. IFI18/00051) from ISCIII granted to M. Provencio. R. Laza-Briviesca project administration, writing–review and editing. B. Massuti: Resources, data was supported by PEJ16/MED/AI-1972 and PEJD-2018-PRE/SAL-8641 from curation, investigation, writing–review and editing. M. Barquín: Data curation, European Social Fund (ESF) and Comunidad de Madrid, both granted to formal analysis, investigation, visualization, methodology, writing–review and M. Provencio. We thank the patients, their families, all the participating clinical editing. R. Laza-Briviesca: Data curation, investigation, methodology, writing- teams, and all the Spanish Lung Cancer Group, BMS, and Thermo Fisher Scientific review and editing. B. Sierra-Rodero: Data curation, investigation, methodology, staff, for making this study possible. We also would like to give thanks to Maria del writing–review and editing. E.R. Parra: Investigation, writing–review and editing. Rocio Moreno Villa and Auriole Tamegnon for their technical assistance. B. Sanchez-Espiridion: Investigation, project administration, writing–review and Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 editing. P. Rocha: Investigation, writing–review and editing. H. Kadara: Inves- tigation, writing–review and editing. I.I. Wistuba: Resources, writing–review and The costs of publication of this article were defrayed in part by the payment of page editing. A. Romero: Data curation, investigation, writing–review and editing. charges. This article must therefore be hereby marked advertisement in accordance V. Calvo: Resources, data curation, investigation, writing–review and editing. with 18 U.S.C. Section 1734 solely to indicate this fact. M. Provencio: Conceptualization, resources, data curation, supervision, funding acquisition, investigation, writing–original draft, project administration, writing– Received March 31, 2021; revised June 1, 2021; accepted August 3, 2021; review and editing. published first August 10, 2021. References 1. Provencio M, Nadal E, Insa A, García-Campelo MR, Casal-Rubio J, Domine M, 11. Kidman J, Principe N, Watson M, Lassmann T, Holt RA, Nowak AK, et al. et al. Neoadjuvant chemotherapy and nivolumab in resectable non-small-cell Characteristics of TCR repertoire associated with successful immune checkpoint lung cancer (NADIM): an open-label, multicentre, single-arm, phase 2 trial. therapy responses. Front Immunol 2020;11:1–11. Lancet Oncol 2020;21:1413–22. 12. Poran A, Scherer J, Bushway ME, Besada R, Balogh KN, Wanamaker A, et al. 2. Forde P, Spicer J, Lu S, Provencio M, Mitsudomi T, Awad M, et al. Abstract Combined TCR repertoire profiles and blood cell phenotypes predict melanoma CT003 - Nivolumab (NIVO) þ platinum-doublet chemotherapy (chemo) vs patient response to personalized neoantigen therapy plus anti-PD-1. chemo as neoadjuvant treatment (tx) for resectable (IB-IIIA) non-small cell lung Cell Reports Med 2020;1:100141. cancer (NSCLC) in the phase 3 CheckMate 816 trial. In: Proceedings of 13. Cham J, Zhang L, Kwek S, Paciorek A, He T, Fong G, et al. Combination the Annual Meeting of the American Association for Cancer Research; 2021 immunotherapy induces distinct T-cell repertoire responses when administered Apr 10–15; Philadelphia (PA). to patients with different malignancies. J Immunother Cancer 2020;8:1–10. 3. Rizvi H, Sanchez-Vega F, La K, Chatila W, Jonsson P, Halpenny D, et al. 14. Cha E, Klinger M, Hou Y, Cummings C, Ribas A, Faham M, et al. Improved Molecular determinants of response to anti-programmed cell death (PD)-1 and survival with T cell clonotype stability after anti-CTLA-4 treatment in cancer anti-programmed death-ligand 1 (PD-L1) blockade in patients with non-small- patients. Sci Transl Med 2014;6:238ra70. cell lung cancer profiled with targeted next-generation sequencing. J Clin Oncol 15. Wu TD, Madireddi S, de Almeida PE, Banchereau R, Chen YJ, Chitre AS, et al. 2018;36:633–41. Peripheral T cell expansion predicts tumour infiltration and clinical response. 4. Ready N, Hellmann MD, Awad MM, Otterson GA, Gutierrez M, Gainor JF, et al. Nature 2020;579:274–8. First-line nivolumab plus ipilimumab in advanced non-small-cell lung cancer 16. Reuben A, Zhang J, Chiou S-H, Gittelman RM, Li J, Lee W-C, et al. Compre- (CheckMate 568): Outcomes by programmed death ligand 1 and tumor muta- hensive T cell repertoire characterization of non-small cell lung cancer. tional burden as biomarkers. J Clin Oncol 2019;37:992–1000. Nat Commun 2020;11:603. 5. Borghaei H, Hellmann MD, Paz-Ares LG, Ramalingam SS, Reck M, O’Byrne KJ, 17. Anagnostou V, Forde PM, White JR, Niknafs N, Hruban C, Naidoo J, et al. et al. Nivolumab (Nivo) þ platinum-doublet chemotherapy (Chemo) vs chemo Dynamics of tumor and immune responses during immune checkpoint blockade as first-line (1L) treatment (Tx) for advanced non-small cell lung cancer in non–small cell lung cancer. Cancer Res 2019;79:1214–25. (NSCLC) with <1% tumor PD-L1 expression: Results from CheckMate 227. 18. Han J, Duan J, Bai H, Wang Y, Wan R, Wang X, et al. TCR repertoire diversity of J Clin Oncol 2018;36:9001. peripheral PD-1 þ CD8 þ T cells predicts clinical outcomes after immuno- 6. Borghaei H, Langer CJ, Paz-Ares L, Rodríguez-Abreu D, Halmos B, Garassino MC, therapy in patients with non–small cell lung cancer. Cancer Immunol Res 2020;8: et al. Pembrolizumab plus chemotherapy versus chemotherapy alone in patients 146–54. with advanced non-small cell lung cancer without tumor PD-L1 expression: 19. Zhang J, Ji Z, Caushi JX, El Asmar M, Anagnostou V, Cottrell TR, et al. A pooled analysis of 3 randomized controlled trials. Cancer 2020;126:4867–77. Compartmental analysis of T-cell clonal dynamics as a function of pathologic 7. Garassino M, Rodriguez-Abreu D, Gadgeel S, Esteban E, Felip E, Speranza G, response to neoadjuvant PD-1 blockade in resectable non-small cell lung cancer. et al. OA04.06 evaluation of TMB in KEYNOTE-189: pembrolizumab plus Clin Cancer Res 2020;26:1327–37. chemotherapy vs placebo plus chemotherapy for nonsquamous NSCLC. 20. Vroman H, Balzaretti G, Belderbos RA, Klarenbeek PL, Van Nimwegen M, J Thorac Oncol 2019;14:S216–7. Bezemer K, et al. T cell receptor repertoire characteristics both before and 8. Voong KR, Feliciano J, Becker D, Levy B. Beyond PD-L1 testing-emerging following immunotherapy correlate with clinical response in mesothelioma. biomarkers for immunotherapy in non-small cell lung cancer. Ann Transl Med J Immunother Cancer 2020;8:1–7. 2017;5:376. 21. Looney TJ, Topacio-Hall D, Lowman G, Conroy J, Morrison C, Oh D, et al. TCR 9. Li S, Zhang C, Pang G, Wang P. Emerging blood-based biomarkers for predicting convergence in individuals treated with immune checkpoint inhibition for response to checkpoint immunotherapy in non-small-cell lung cancer. cancer. Front Immunol 2020;10. Front Immunol 2020;11:1–11. 22. Parra ER, Uraoka N, Jiang M, Cook P, Gibbons D, Forget MA, et al. Validation of 10. Robins HS, Campregher PV, Srivastava SK, Wacher A, Turtle CJ, Kahsai O, et al. multiplex immunofluorescence panels using multispectral microscopy for Comprehensive assessment of T-cell receptor b-chain diversity in ab T cells. immune-profiling of formalin-fixed and paraffin-embedded human tumor Blood 2009;114:4099–107. tissues. Sci Rep 2017;7:1–11. AACRJournals.org Clin Cancer Res; 27(21) November 1, 2021 5889 Casarrubios et al. 23. Love MI, Huber W, Anders S. Moderated estimation of fold change and 28. Shim JH, Kim HS, Cha H, Kim S, Kim TM, Anagnostou V, et al. HLA- dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:550. corrected tumor mutation burden and homologous recombination 24. Steinbaugh MJ, Pantano L, Kirchner RD, Barrera V, Chapman BA, Piper ME, deficiency for the prediction of response to PD-(L)1 blockade in et al. BcbioRNAseq: R package for bcbio RNA-seq analysis [version 2; peer advanced non-small-cell lung cancer patients. Ann Oncol 2020;31: review: 1 approved, 1 approved with reservations]. F1000Research. F1000 902–11. Research 2018;7:1976. 29. O’Donnell JS, Teng MWL, Smyth MJ. Cancer immunoediting and 25. Gudmundsdottir H, Wells AD, Turka LA. Dynamics and requirements of T cell resistance to T cell-based immunotherapy. Nat Rev Clin Oncol 2019; clonal expansion in vivo at the single-cell level: effector function is linked to 16:151–67. proliferative capacity. J Immunol 1999;162:5212–23. 30. Joshi K, de Massy MR, Ismail M, Reading JL, Uddin I, Woolston A, et al. Spatial 26. Binnewies M, Roberts EW, Kersten K, Chan V, Fearon DF, Merad M, et al. heterogeneity of the T cell receptor repertoire reflects the mutational landscape in Understanding the tumor immune microenvironment (TIME) for effective lung cancer. Nat Med 2019;25:1549–59. therapy. Nat Med 2018;24:541–50. 31. Danilova L, Anagnostou V, Caushi JX, Sidhom J-W, Guo H, Chan HY, et al. 27. Goodman AM, Castro A, Pyke RM, Okamura R, Kato S, Riviere P, et al. MHC-I The mutation-associated neoantigen functional expansion of specific T cells genotype and tumor mutational burden predict response to immunotherapy. (MANAFEST) assay: a sensitive platform for monitoring antitumor immunity. Genome Med 2020;12:45. Cancer Immunol Res 2018;6:888–99. Downloaded from http://aacrjournals.org/clincancerres/article-pdf/27/21/5878/3197516/5878.pdf by guest on 02 February 2023 5890 Clin Cancer Res; 27(21) November 1, 2021 CLINICAL CANCER RESEARCH