- Email: [email protected]
Lymphovascular Invasion Is Associated With Mutational Burden and PD-L1 in Resected Lung Cancer
Journal Pre-proof Lymphovascular Invasion Is Associated with Mutational Burden and PD-L1 in Resected Lung Cancer Kyle G. Mitchell, MD, Marcelo V. Negrao, MD, Edwin R. Parra, MD PhD, Jun Li, PhD, Jianhua Zhang, PhD, Hitoshi Dejima, MD, Ara A. Vaporciyan, MD, Stephen G. Swisher, MD, Annikka Weissferdt, MD DrMed, Mara B. Antonoff, MD, Tina Cascone, MD PhD, Emily Roarty, PhD, Ignacio I. Wistuba, MD, John V. Heymach, MD PhD, Don L. Gibbons, MD PhD, Jianjun Zhang, MD PhD, Boris Sepesi, MD PII:
S0003-4975(19)31416-X
DOI:
https://doi.org/10.1016/j.athoracsur.2019.08.029
Reference:
ATS 33051
To appear in:
The Annals of Thoracic Surgery
Received Date: 6 February 2019 Revised Date:
15 July 2019
Accepted Date: 8 August 2019
Please cite this article as: Mitchell KG, Negrao MV, Parra ER, Li J, Zhang J, Dejima H, Vaporciyan AA, Swisher SG, Weissferdt A, Antonoff MB, Cascone T, Roarty E, Wistuba II, Heymach JV, Gibbons DL, Zhang J, Sepesi B, Lymphovascular Invasion Is Associated with Mutational Burden and PDL1 in Resected Lung Cancer, The Annals of Thoracic Surgery (2019), doi: https://doi.org/10.1016/ j.athoracsur.2019.08.029. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 by The Society of Thoracic Surgeons
Lymphovascular Invasion Is Associated with Mutational Burden and PD-L1 in Resected Lung Cancer Running Title: NSCLC: Tumor Mutational Burden and PD-L1
Kyle G. Mitchell MD1,*, Marcelo V. Negrao MD2,*, Edwin R. Parra MD PhD3, Jun Li PhD4, Jianhua Zhang PhD4, Hitoshi Dejima MD3, Ara A. Vaporciyan MD1, Stephen G. Swisher MD1, Annikka Weissferdt MD DrMed5, Mara B. Antonoff MD1, Tina Cascone MD PhD2, Emily Roarty PhD2, Ignacio I. Wistuba MD3, John V. Heymach MD PhD2, Don L. Gibbons MD PhD2,6,, Jianjun Zhang MD PhD2,, Boris Sepesi MD1, *Drs Mitchell and Negrao are co-first authors.
1
Department of Thoracic and Cardiovascular Surgery. 2Department of Thoracic/Head and Neck Medical Oncology. 3Department of Translational Molecular Pathology. 4Department of Bioinformatics and Computational Biology. 5Department of Pathology. 6Department of Molecular and Cellular Oncology. All authors: The University of Texas MD Anderson Cancer Center, Houston, TX.
Word Count: 4496/4500
Keywords: Non-small cell lung cancer; lymphovascular invasion; tumor microenvironment; tumor mutational burden; PD-L1
Tables and Figures: 2 Tables, 8 Figures, 3 Supplemental Figures
This manuscript was presented at the 55th Annual Meeting of the Society of Thoracic Surgeons; 2019 January 27-29; San Diego, California.
Corresponding Author: Boris Sepesi, MD, FACS Department of Thoracic and Cardiovascular Surgery The University of Texas MD Anderson Cancer Center 1515 Holcombe Blvd, Houston, TX 77030 [email protected]
1
GLOSSARY OF ABBREVIATIONS CD: cluster of differentiation CI: confidence interval ICON: Immunogenomic Profiling of Non-Small Cell Lung Cancer LVI: lymphovascular invasion MCs: malignant cells mIF: multiplex immunofluorescence Mut/Mb: mutations per megabase NSCLC: non-small cell lung cancer OR: odds ratio PD-1: programmed cell death protein 1 PD-L1 programmed death ligand 1 TMB: tumor mutational burden TME: tumor microenvironment
2
ABSTRACT (233/250) Background: High tumor mutational burden (TMB) and PD-L1 expression are leading biomarkers in metastatic non-small cell lung cancer (NSCLC) and predict favorable response to checkpoint inhibitors. We sought to identify clinicopathologic characteristics associated with elevated TMB and PD-L1 expression among patients who underwent resection for NSCLC. Methods: NSCLC patients undergoing primary resection (2016-2018) were prospectively enrolled in an immunogenomic profiling project (ICON). Multiplex immunofluorescence (mIF) quantified densities (cells/mm2) of CD3+, CD3+CD8+, CD3+CD8+PD-1+, malignant cells (MCs), MCsPD-L1+, CD68+, CD68+PD-L1+, and CD20+ cells. Whole exome sequencing quantified TMB (mutations/megabase). TMB and MCsPD-L1+ were dichotomized according to the median of each. Results: 55 patients completed mIF and WES profiling. 41.8% (23/55) had pathologic stage I disease. Median TMB and MCsPD-L1+ were 3.91 and 0.62 cells/mm2, respectively. TMB was higher among smokers (p=0.001) and tumors with lymphovascular invasion (LVI)(p=0.051). TMB was positively correlated with densities of MCsPD-L1+ (r=0.293,p=0.030), CD68+PD-L1+ (r=0.289, p=0.033), and CD20+ (r=0.310,p=0.043) cells. The density of MCsPD-L1+ was associated with increased CD3+CD8+ (r=0.319,p=0.018) and CD68+PD-L1+ (r=0.371,p=0.005) cells. Patients with PD-L1HighTMBHigh tumors (30.9%, 17/55) had higher intratumoral densities of CD3+, CD3+CD8+, CD68+, CD68+PD-L1+, and CD20+ cells. On multivariable analysis LVI was associated with synchronous elevated TMB and PD-L1 expression (OR 3.53,p=0.039). Conclusions: NSCLC tumors with elevated TMB and PD-L1 expression are associated with LVI and increased intratumoral immune cell infiltration. These findings may potentially improve patient selection for checkpoint inhibitor therapy trials in the adjuvant setting.
3
Landmark phase III clinical trials which utilized immune checkpoint inhibitors in metastatic nonsmall cell lung cancer (NSCLC) identified increased tumor expression of programmed cell death ligand 1 (PD-L1) and tumor mutational burden (TMB) as the best predictive biomarkers of response to antiprogrammed cell death protein (PD-1) therapy (1-5). Other studies have further shown that tumor expression of PD-L1 and neoantigens are associated with local immune cell recruitment and enhanced tumor-specific immune responses (6, 7). Consequently, current guidelines include PD-L1 expression as a selection criterion for checkpoint blockade therapy in patients with metastatic NSCLC (8). Despite the fact that there are currently numerous ongoing clinical trials exploring checkpoint inhibitors in the neoadjuvant and adjuvant settings, the relationships between TMB, tumor expression of PD-L1, and clinicopathologic features in patients with resectable lung cancer have not been fully characterized (9, 10). Because an enhanced understanding of these relationships could potentially assist in identification of the most appropriate patients for enrollment in these novel trials, we examined TMB, PD-L1 expression, and features of the tumor microenvironment (TME) in a cohort of NSCLC patients enrolled in a prospective immunogenomic profiling study (Immunogenomic Profiling of Non-Small Cell Lung Cancer, ICON). In this study of chemotherapy-naïve resected lung cancers, we hypothesized that cancers characterized by synchronous high TMB and PD-L1 expression would be associated with distinct clinicopathologic features and with enhanced intratumoral immune cell infiltration PATIENTS AND METHODS Study Design and Patient Selection The ICON project was a prospective, multifaceted effort at the University of Texas MD Anderson Cancer Center that prospectively enrolled 150 patients prior to resection of primary lung cancer between April 1st, 2016 and August 31st, 2018. Eligible patients with clinical stage IA-IIIA NSCLC who were medically fit to undergo resection were enrolled. Patients included in the present study were those who had multiplex immunofluorescence and whole exome sequencing data available for analysis. Patients who
4
received neoadjuvant chemotherapy were excluded. Tumors were staged using the 7th edition of the American Joint Committee on Cancer’s schema (11). The study was approved by MD Anderson’s Institutional Review Board, and patients gave informed consent before enrollment. Multiplex Immunofluorescence Multiplex immunofluorescence (mIF) analysis was performed using methods that have been described and validated (12). Briefly, four micrometer-thick formalin fixed, paraffin embedded tumor sections were stained using an automated staining system (BOND-MAX; Leica Microsystems) using antibodies against cytokeratin AE1/AE3 (dilution 1:300, Dako), PD-L1 (clone E1L3N, 1:3000, Cell Signaling Technology), CD3 (1:100, Dako), CD8 (clone C8/144B, 1:300, Thermo Fisher Scientific), CD20 (clone L26, 1:250, Dako), and CD68 (clone PG-M1, 1:450, Dako), and PD-1 (clone EPR4877-2, 1:250, Abcam) and fluorophore staining using the Opal 7 kit (catalogue #NEL797001KT; PerkinElmer)(12). Stained slides were scanned using the Vectra 3.0 imaging system (PerkinElmer) with normal human tonsil tissue as a calibration control (12). After scanning, five fields (each 0.3345 mm2) were selected within the tumor using the phenochart 1.0.4 viewer (PerkinElmer). A trained pathologist (E.R.P.) supervised quantification of immune cell densities using InForm image analysis software (PerkinElmer). Marker colocalization was used to identify populations of T lymphocytes (CD3+), cytotoxic T lymphocytes (CD3+CD8+), antigen-experienced T lymphocytes (CD3+CD8+PD-1+), macrophages (CD68+), macrophages expressing PD-L1 (CD68+PD-L1+), B lymphocytes (CD20+, available for 43/55 cases), malignant cells (MCs, AE1/AE3+), and MCs expressing PD-L1 (MCsPD-L1+, AE1/AE+PD-L1+) in the tumor (epithelial nests) and stromal compartments (Supplemental Figure 1). Densities of each colocalized cell population were quantified as the average number of cells/mm2 (12). Cell densities in the tumor compartment and overall tumor (tumor and stromal compartments combined) were analyzed. Whole Exome Sequencing
5
DNA was extracted from frozen tumor tissue using the QIAamp DNA Mini kit (Qiagen) according to the manufacturer’s instructions. Exome capture was performed on 200ng genomic DNA per sample based on KAPA library prep (Kapa Biosystems) using the Agilent SurSelect Human All Exon V4 kit according to manufacturer’s instructions. Paired-end multiplex sequencing of samples was performed on the HiSeq 2500 sequencing platform (Illumina). Blood was used as a normal control. The BWA aligner (bwa-0.7.5a) mapped the raw reads to the human hg19 reference genome (UCSC Genome Browser: genome.UCSC.edu)(13). Duplicate reads were marked using the Picard (version 1.112, http://broadinstitute.github.io/picard/) “MarkDuplicates” module. The Genome Analysis Toolkit (modules “IndelRealigner” and “BaseRecalibrator”) was applied to perform insertion/deletion (indel) realignment and base quality recalibration. MuTect and Pindel were applied to each tumor and its matching normal tissue sample to detect somatic single nucleotide variants (SNVs) and small indels (14, 15). To ensure specificity, the following criteria were applied to filter the detected somatic SNVs and indels: coverage of ≥20 reads for the tumor and ≥10 for the normal control; total number of reads supporting the variant ≥4; and MuTect LOD score ≥6.3. The allele frequency from the normal sample was less than 0.01%. A population frequency threshold of 1% was used to filter out common variants in the databases of 1000 Genome Project, Exome Aggregation Consortium, and ESP6500. Indels were discarded if repetitive sequences were detected within 25 base pairs in the downstream regions. TMB was quantified as the total number of non-synonymous mutations per megabase (mut/Mb). Outcome Definitions and Statistical Analysis Pairwise correlations between continuous variables were assessed using Spearman’s correlations. Differences in continuous variables between groups were analyzed using Mann-Whitney U or KruskalWallis tests, and differences in categorical variables were analyzed using Pearson’s X2 and Fisher’s exact tests. TMB and MCsPD-L1+ were dichotomized according to the observed median values among the study cohort.
Univariable
logistic
regression
was
used
to
identify
associations
between
clinicopathological features (age, sex, histology, pathologic stage, tumor differentiation, and
6
lymphovascular invasion [LVI]) and synchronous elevated (>median) TMB and MCsPD-L1+. Variables with p<0.25 on univariable analysis and those of clinical interest were included in a multivariable model, with stepwise backwards selection then performed with a final selection criterion of p<0.10. Recurrencefree survival (RFS) was defined as the time from resection until disease recurrence or death; all patients without a RFS event at the end of the study period were censored at the date of last follow-up. The Kaplan-Meier method was used to estimate RFS, and differences between groups were analyzed using the log-rank test. All analyses were performed using SPSS version 24 (IBM, Armonk, NY) and STATA version 14.2 (StataCorp, College Station, TX). For all analyses, statistical significance was defined as two-tailed p<0.05. RESULTS Baseline Clinical and Treatment Characteristics Fifty-five patients met inclusion criteria (Supplemental Figure 2).
The majority had
adenocarcinoma, and 42% (n=23) had pathologic stage I disease (Table 1). One patient was diagnosed with stage IV disease due to intraoperative identification of pericardial deposits. Clinicopathologic and Immune Microenvironment Characteristics Associated with Tumor Expression of PD-L1 The median density of malignant cells expressing PD-L1 (MCsPD-L1+) among the entire study cohort was 0.62 (interquartile range [IQR] 0.00-30.19) cells/mm2. Whereas patients with low (≤median) MCsPD-L1+had nearly absent expression of PD-L1, densities of MCsPD-L1+ among patients with higher PD-L1 expression was much more variable (Figure 1). Densities of MCsPD-L1+ were higher in squamous tumors (median 110.69 [IQR 0.32-449.54] versus non-squamous 0.15 [IQR 0.00-6.82], p=0.003), those with poorly-differentiated (2.50 [0.00-710.96] versus moderately- 1.06 [IQR 0.00-29.04] versus well- 0.00 [IQR 0.00-0.23], p=0.041), those with lymphovascular invasion (LVI) (3.88 [IQR 0.00108.55] versus 0.03 [IQR 0.00-7.31], p=0.084), and those with more advanced pathologic stage (stage I 7
0.00 [IQR 0.00-5.34] versus stage II 2.50 [IQR 0.00-449.54] versus stages III/IV 1.03 [IQR 0.00-147.33], p=0.065), although the latter two associations did not reach statistical significance (Figure 2). No differences in MCsPD-L1+ were observed according to smoking status, sex, or age. The density of MCsPD-L1+ was positively correlated with TMB (r=0.293, p=0.030) and intratumoral CD3+CD8+ (r=0.319, p=0.018) and CD68+PDL1+ (r=0.371, p=0.005) cells (Figure 3). Similar associations were observed between MCsPD-L1+ and overall tumor infiltration by CD3+CD8+ (r=0.269, p=0.047) and CD68+PDL1+ cells (r=0.326, p=0.015). These findings demonstrated that the density of malignant NSCLC cells expressing PD-L1 varied according to tumor histopathologic features and was positively associated with TMB and immune cell infiltration. Clinicopathologic and Immune Microenvironment Characteristics Associated with TMB The median observed TMB in the study cohort was 3.91 (IQR 1.59-8.71) mut/Mb; 12 patients (21.8%) had TMB ≥10 mut/Mb. TMB was higher among smokers (median 5.00 [IQR 2.01-10.03] versus non-smokers 1.45 [IQR 0.84-2.33], p=0.001) and among tumors with LVI (5.02 [IQR 3.32-9.01] versus no LVI 2.40 [IQR 1.36-8.71], p=0.051), although the association with LVI did not reach statistical significance (Figure 4A and 4B). No associations were observed between TMB and age, histology, differentiation, or pathologic stage (Figure 4C and 4D). Tumors with increased (>median, n=27/55) TMB had higher densities of CD3+CD8+, CD20+, and CD68+PD-L1+ cells (Figure 5, Supplemental Figure 3). These analyses showed that NSCLC tumors with high TMB were associated with tobacco exposure, pathologic LVI, and increased immune cell infiltrates. Lymphovascular Invasion Is Associated with Synchronous Elevated TMB and PD-L1 Expression Because high tumor PD-L1 expression and elevated TMB have been previously identified as predictive of response to immune checkpoint inhibitors, we next analyzed clinicopathologic patient and tumor characteristics for associations with synchronous elevated (>median) TMB and MCsPD-L1+ (6, 16). PD-L1HighTMBHigh tumors (n=17/55) were more frequently noted to have LVI than were tumors with low (≤median) TMB and MCsPD-L1+ (PD-L1LowTMBLow; 11/17 [64.7%] versus 5/18 [27.8%],
8
respectively; p=0.028) (Figure 6). All (17/17) were former or current smokers. PD-L1HighTMBHigh tumors were marked by higher intratumoral densities of CD3+, CD3+CD8+, CD20+, and CD68+PD-L1+ cells than were PD-L1LowTMBLow (n=18/55) and other (PD-L1LowTMBHigh or PD-L1HighTMBLow, n=20/55) tumors (Figure 7). Upon multivariable analysis, LVI was identified as the only feature associated with PDL1HighTMBHigh tumors (Table 2). Exploratory survival analysis (median follow-up duration for the entire cohort, 16.1 [IQR 10.1-22.6] months) demonstrated worse RFS among patients with LVIPresentPDL1HighTMBHigh than those with LVIAbsentPD-L1LowTMBLow tumors, although this difference did not reach statistical significance (Figure 8). Together, these results demonstrated that PD-L1HighTMBHigh NSCLCs were characterized by the presence of LVI and increased immune cell infiltration. COMMENT In the present study of chemotherapy-naïve, surgically resected lung cancers we observed lower TMB and tumor expression of PD-L1 compared to previously published values for metastatic NSCLC (2, 17). We also identified differential TMB and expression of PD-L1 that varied according to baseline patient and pathologic tumor characteristics. These findings have potential clinical relevance, as future clinical trials utilizing checkpoint inhibitor immunotherapy could explore the presence of LVI, in addition to TMB and PD-L1, as a patient stratification or selection criterion. As immune checkpoint inhibitors continue to be explored in the setting of early NSCLC, an ability to identify patient subgroups most likely to derive therapeutic benefit from these agents is greatly needed (9, 10). Outside of induction clinical trials, the majority of surgically resectable NSCLC patients undergo resection of the primary tumor and mediastinal lymph nodes first, with further decisions about additional therapy guided by the final pathology (8). In patients with metastatic disease, a TMB greater than 10 mut/Mb has been identified as predictive of response to checkpoint blockade (8). The contrast between the frequency of tumors with TMB in excess of that threshold in the present report (22%) and that published in CheckMate 277 (44%) highlights the need for exploration of additional features that can be used singly or in combination to predict therapeutic benefit in patients with resectable disease (8). We
9
submit that our observation that the presence of LVI in the final pathologic specimen is associated with increased TMB and PD-L1 could potentially serve as a guide for the type of adjuvant therapy in NSCLC patients following resection, as assessment of LVI is routinely performed at the time pathologic evaluation. Tumors that demonstrate LVI have a high propensity for distant recurrence, even if resected in very early stages (18). However, current guidelines for adjuvant chemotherapy are based on the tumor size and nodal status rather than the presence or absence of LVI or high TMB and PD-L1 (8). Additionally, the ANVIL trial (NCT02595944) of adjuvant nivolumab does not further stratify patients who meet the inclusion criteria of absent EGFR mutations or ALK rearrangements. As we try to identify patients at increased risk of recurrence following complete surgical resection of early lung cancers, and as we search for more effective and safe adjuvant therapies, we speculate that patients with high TMB, PDL1, and LVI may derive a therapeutic benefit from adjuvant anti-PD-1/PD-L1 agents, either as monotherapy or in combination with chemotherapy. The identification of higher tumor infiltration by cytotoxic T lymphocytes, B lymphocytes, and macrophages expressing PD-L1 among tumors with higher TMB and expression of PD-L1 further supports this interpretation, since the efficacy of checkpoint blockade is greatest in the setting of concurrent high checkpoint expression and effector/helper lymphocyte infiltration (19, 20). Tumor PD-L1 expression has been identified as being associated with increased densities of immune infiltrates, and checkpoint inhibitor-induced release of PD-1+ T lymphocytes from immune cells expressing PD-L1 has been postulated as another mechanism by which antitumor immunity is enhanced by these agents (20-23). The role of B lymphocytes in the microenvironment has not been fully characterized, but recent work has demonstrated that they play important roles in antigen presentation and elaboration of antitumor antibodies and have protective prognostic effects (24). It seems plausible that the group of patients with tumors characterized by high TMB, elevated PD-L1 expression, the presence of LVI, and higher tumor immune cell infiltration would be that which is best suited for novel adjuvant immunotherapy trials in early NSCLC. On the other hand, for patients with a predicted poor anti-tumor immune response, the
10
preferred strategy may employ a combination of anti-PD-1/PD-L1 agents with CTLA-4 blockade or chemotherapy in an effort to improve lymphocyte recruitment (25). Despite the strength provided by the prospective nature of our study, we must acknowledge its limitations. Because only a subset of the tumors in the ICON project had both mIF and TMB data available for analysis, the present study is limited by its small sample size and lack of a validation set. Although the methodology of analysis of cell densities in this study (i.e., mIF) differs from that which is used to quantify PD-L1 expression clinically (i.e., tumor proportion score), mIF permits precise quantification of cell densities and has been previously validated in comparison to chromogenic immunohistochemistry (12). Moreover, the presence of LVI is not reliably detected on preoperative biopsy, and the applicability of these findings may be limited to the adjuvant setting following complete pathological assessment of the surgical specimen. Additionally, the short follow-up in this cohort and few observed RFS events to date precluded a robust analysis of the relationships between TME features with prognostic outcomes while controlling for relevant clinicopathologic characteristics. Further analyses of the prognostic impact of these features will be performed as follow-up data in this cohort matures. In summary, we report that primary NSCLC tumors with synchronous elevated TMB and PD-L1 expression are associated with the presence of LVI and with enhanced intratumoral immune cell infiltration. Further study is warranted to examine whether these findings can improve patient selection for novel trials of immune checkpoint inhibitor therapy in the adjuvant setting.
11
REFERENCES 1.
Rizvi H, Sanchez-Vega F, La K et al. Molecular determinants of response to anti–programmed
cell death (PD)-1 and anti–programmed death-ligand 1 (PD-L1) blockade in patients with non–small-cell lung cancer profiled with targeted next-generation sequencing. J Clin Oncol 2018;36(7):633-641. 2.
Borghaei H, Paz-Ares L, Horn L et al. Nivolumab versus docetaxel in advanced nonsquamous
non-small-cell lung cancer. N Engl J Med 2015;373(17):1627-1639. 3.
Gandhi L, Rodríguez-Abreu D, Gadgeel S et al. Pembrolizumab plus chemotherapy in metastatic
non–small-cell lung cancer. N Engl J Med 2018;378(22):2078-2092. 4.
Hellmann MD, Nathanson T, Rizvi H et al. Genomic features of response to combination
immunotherapy in patients with advanced non-small-cell lung cancer. Cancer Cell 2018;33(5):843852.e844. 5.
Hellmann MD, Ciuleanu T-E, Pluzanski A et al. Nivolumab plus ipilimumab in lung cancer with
a high tumor mutational burden. N Engl J Med 2018;378(22):2093-2104. 6.
Taube JM, Klein A, Brahmer JR et al. Association of PD-1, PD-1 ligands, and other features of
the tumor immune microenvironment with response to anti–PD-1 therapy. Clin Cancer Res 2014;20(19):5064-5074. 7.
Kiyotani K, Park JH, Inoue H et al. Integrated analysis of somatic mutations and immune
microenvironment in malignant pleural mesothelioma. Oncoimmunology 2017;6(2):e1278330. 8.
National Comprehensive Cancer Network. Non-small cell lung cancer. 2018;Version 3.2018.
9.
Forde PM, Chaft JE, Smith KN et al. Neoadjuvant PD-1 blockade in resectable lung cancer. N
Engl J Med 2018;378(21):1976-1986.
12
10.
Owen D, Chaft JE. Immunotherapy in surgically resectable non-small cell lung cancer. J Thorac
Dis 2018;10(Suppl 3):S404-S411. 11.
Goldstraw P, Crowley J, Chansky K et al. The IASLC Lung Cancer Staging Project: Proposals
for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM classification of malignant tumours. J Thorac Oncol 2007;2(8):706-714. 12.
Parra ER, Uraoka N, Jiang M et al. Validation of multiplex immunofluorescence panels using
multispectral microscopy for immune-profiling of formalin-fixed and paraffin-embedded human tumor tissues. Sci Rep 2017;7(1):13380. 13.
Gnirke A, Melnikov A, Maguire J et al. Solution hybrid selection with ultra-long oligonucleotides
for massively parallel targeted sequencing. Nat Biotechnol 2009;27:182. 14.
Cibulskis K, Lawrence MS, Carter SL et al. Sensitive detection of somatic point mutations in
impure and heterogeneous cancer samples. Nat Biotechnol 2013;31:213. 15.
Ye K, Schulz MH, Long Q, Apweiler R, Ning Z. Pindel: A pattern growth approach to detect
break points of large deletions and medium sized insertions from paired-end short reads. Bioinformatics 2009;25(21):2865-2871. 16.
Rizvi NA, Hellmann MD, Snyder A et al. Mutational landscape determines sensitivity to PD-1
blockade in non–small cell lung cancer. Science 2015;348(6230):124-128. 17.
Brahmer J, Reckamp KL, Baas P et al. Nivolumab versus docetaxel in advanced squamous-cell
non–small-cell lung cancer. N Engl J Med 2015;373(2):123-135. 18.
Yanagawa N, Shiono S, Abiko M, Ogata S-y, Sato T, Tamura G. Prognostic impact and initial
recurrence site of lymphovascular and visceral pleural invasion in surgically resected stage I non-smallcell lung carcinoma. Eur J Cardiothorac Surg 2013;44(3):e200-e206. 19.
Teng MWL, Ngiow SF, Ribas A, Smyth MJ. Classifying cancers based on T-cell infiltration and
PD-L1. Cancer Res 2015;75(11):2139-2145.
13
20.
Vilain RE, Menzies AM, Wilmott JS et al. Dynamic changes in PD-L1 expression and immune
infiltrates early during treatment predict response to PD-1 blockade in melanoma. Clin Cancer Res 2017;23(17):5024-5033. 21.
Velcheti V, Schalper KA, Carvajal DE et al. Programmed death ligand-1 expression in non-small
cell lung cancer. Lab Invest 2013;94:107. 22.
Cheng H, Borczuk A, Janakiram M et al. Wide expression and significance of alternative immune
checkpoint molecules, B7x and HHLA2, in PD-L1–negative human lung cancers. Clin Cancer Res 2018;24(8):1954-1964. 23.
Yu H, Chen Z, Ballman KV et al. Correlation of PD-L1 expression with tumor mutation burden
and gene signatures for prognosis in early-stage squamous cell lung carcinoma. J Thorac Oncol 2019;14(1):25-36. 24.
Bruno TC, Ebner PJ, Moore BL et al. Antigen-presenting intratumoral B cells affect CD4+
phenotypes in non–small cell lung cancer patients. Cancer Immunol Res 2017;5(10):898-907. 25.
Provencio-Pulla M, Nadal-Alforja E, Cobo M et al. Neoadjuvant chemo/immunotherapy for the
treatment of stages IIIA resectable non-small cell lung cancer (NSCLC): A phase II multicenter exploratory study—NADIM study. J Clin Oncol 2018;36(15_suppl):8521-8521.
14
Table 1: Baseline clinical, pathologic, and treatment characteristics of the study cohort (n=55). Variable Age, median (IQR) (years) Sex Female Male Smoking Never Former/Current Zubrod 0 1 FEV1 (% Predicted) Histology Adenocarcinoma Squamous cell carcinoma Adenosquamous Adenocarcinoma, sarcomatoid type Mixed pleomorphic carcinoma/adenocarcinoma Differentiation Well Moderate Poor Clinical Stage I II III Extent of Resection Sublobar Lobectomy/Bilobectomy Pneumonectomy Pathologic Tumor Size, median (IQR) (cm) Pathologic Stage I II III IV Lymphovascular Invasion Pathologic Margin R0 R1 Adjuvant Therapy Chemotherapy Radiotherapy FEV1: forced expiratory volume in one second
N (%) or Median (IQR) 67.0 (61.0-73.0) 30 (54.5) 25 (45.5) 8 (14.5) 47 (85.5) 51 (92.7) 4 (7.3) 81.0 (76.0-94.0) 38 (69.1) 14 (25.5) 1 (1.8) 1 (1.8) 1 (1.8) 8 (14.5) 30 (54.5) 17 (30.9) 33 (60.0) 19 (34.5) 3 (5.5) 5 (9.1) 47 (85.5) 3 (5.5) 3.5 (2.8-5.5) 23 (41.8) 19 (34.5) 12 (21.8) 1 (1.8) 24 (43.6) 49 (89.1) 6 (10.9) 27 (49.1) 5 (9.1)
15
Table 2: Univariable and multivariable analyses of factors associated with tumors characterized by synchronous high (>median) tumor mutational burden and density of malignant cells expressing PD-L1 (n=17/55) versus all other tumors (n=38/55). Univariable Variable N (%) OR 95% CI Age (years) 1.02 0.96-1.08 Sex (Male) 25 (45.5) 1.55 0.49-4.88 Histology (Squamous)* 15 (27.3) 2.63 0.76-9.08 Pathologic Stage* I 23 (41.8) Reference 19 (34.5) 2.62 0.68-10.06 II III/IV 13 (23.6) 1.60 0.34-7.46 Differentiation (Poor vs Well/Moderate) 17 (30.9) 1.96 0.59-6.55 Lymphovascular Invasion* 24 (43.6) 3.53 1.06-11.70 Smoking status omitted because all PD-L1High/TMBHigh patients were former or current smokers *Included in multivariable model, with stepwise backwards selection then performed CI: confidence interval; OR: odds ratio
p 0.602 0.457 0.128 0.373 0.161 0.549 0.274 0.039
OR
3.53
Multivariable 95% CI
1.06-11.70
p
0.039
16
FIGURE LEGENDS Figure 1: Tumor expression of PD-L1 after dichotomization according to the observed median density of PD-L1+ malignant cells (n=55). Figure 2: Associations between tumor expression of PD-L1 and clinicopathologic and immune microenvironment characteristics (n=55). Increased densities of PD-L1+ malignant cells were observed in (A) squamous and (B) poorly-differentiated tumors. Trends (0.05< p< 0.10) towards increased densities in MCs+PD-L1+ were noted in tumors with (C) lymphovascular invasion and those with (D) advanced pathologic stage. Figure 3: NSCLCs with high (>median) densities of MCsPD-L1+ had higher TMB (A) and tumor infiltration by cytotoxic lymphocytes (CD3+CD8+, B) and macrophages expressing PD-L1 (CD68+PDL1+, C)(n=55). Figure 4: Tumor mutational burden (TMB) was increased among (A) smokers and (B) in tumors with LVI, but was not associated with (C) tumor differentiation or (D) pathologic stage (n=55). Figure 5: Tumors with high TMB had higher intratumoral densities of cytotoxic T lymphocytes (CD3+CD8+, A), B lymphocytes (CD20+, B), and macrophages expressing PD-L1 (CD68+PD-L1+, C)(n=55). Figure 6: Lymphovascular invasion was more frequently present in tumors with synchronous high TMB and elevated densities of malignant cells expressing PD-L1 (PD-L1HighTMBHigh; n=17/55) than in tumors with concomitant low PD-L1 expression and TMB (PD-L1LowTMBLow; n=18/55). Figure 7: Representative images of tumor samples of (A) a patient with pathologic stage II adenosquamous carcinoma with lymphovascular invasion (LVI; A, left panel) and elevated (>median) TMB and PD-L1, and (B) a patient with stage II squamous cell carcinoma with absent LVI and low (≤median) TMB and PD-L1. Whereas (A, right panel) the first patient had high intratumoral lymphocytic
17
infiltrates, the second (B) had low immune cell infiltration. (C) PD-L1HighTMBHigh tumors (n=17/55) had higher intratumoral CD3+CD8+ (left panel), CD20+ (center panel), and CD68+PD-L1+ (right panel) cell densities than did those with PD-L1LowTMBLow (n=18/55) and all other tumors (n=20/55). Figure 8: Recurrence-free survival according to lymphovascular invasion (LVI), density of tumor cells expressing PD-L1 (MCsPD-L1+), and TMB. The cohort was dichotomized according to observed medians of MCsPD-L1+ and TMB. Supplemental Figure 1: Representative multiplex immunofluorescence images of tumors with cells staining positive for CD3, CD8, PD-1, PD-L1, CD68, DAPI (marker of cell nuclei), and AE1/AE3 (marker of malignant cells). Supplemental Figure 2: CONSORT diagram depicting patient selection for the study. Supplemental Figure 3: Tumors with high TMB had higher overall infiltration (tumor and stromal compartments) of cytotoxic T lymphocytes (CD3+CD8+, A), B lymphocytes (CD20+, B), and macrophages expressing PD-L1 (CD68+PD-L1+, C)(n=55).
18
S0003-4975(19)31416-X
DOI:
https://doi.org/10.1016/j.athoracsur.2019.08.029
Reference:
ATS 33051
To appear in:
The Annals of Thoracic Surgery
Received Date: 6 February 2019 Revised Date:
15 July 2019
Accepted Date: 8 August 2019
Please cite this article as: Mitchell KG, Negrao MV, Parra ER, Li J, Zhang J, Dejima H, Vaporciyan AA, Swisher SG, Weissferdt A, Antonoff MB, Cascone T, Roarty E, Wistuba II, Heymach JV, Gibbons DL, Zhang J, Sepesi B, Lymphovascular Invasion Is Associated with Mutational Burden and PDL1 in Resected Lung Cancer, The Annals of Thoracic Surgery (2019), doi: https://doi.org/10.1016/ j.athoracsur.2019.08.029. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 by The Society of Thoracic Surgeons
Lymphovascular Invasion Is Associated with Mutational Burden and PD-L1 in Resected Lung Cancer Running Title: NSCLC: Tumor Mutational Burden and PD-L1
Kyle G. Mitchell MD1,*, Marcelo V. Negrao MD2,*, Edwin R. Parra MD PhD3, Jun Li PhD4, Jianhua Zhang PhD4, Hitoshi Dejima MD3, Ara A. Vaporciyan MD1, Stephen G. Swisher MD1, Annikka Weissferdt MD DrMed5, Mara B. Antonoff MD1, Tina Cascone MD PhD2, Emily Roarty PhD2, Ignacio I. Wistuba MD3, John V. Heymach MD PhD2, Don L. Gibbons MD PhD2,6,, Jianjun Zhang MD PhD2,, Boris Sepesi MD1, *Drs Mitchell and Negrao are co-first authors.
1
Department of Thoracic and Cardiovascular Surgery. 2Department of Thoracic/Head and Neck Medical Oncology. 3Department of Translational Molecular Pathology. 4Department of Bioinformatics and Computational Biology. 5Department of Pathology. 6Department of Molecular and Cellular Oncology. All authors: The University of Texas MD Anderson Cancer Center, Houston, TX.
Word Count: 4496/4500
Keywords: Non-small cell lung cancer; lymphovascular invasion; tumor microenvironment; tumor mutational burden; PD-L1
Tables and Figures: 2 Tables, 8 Figures, 3 Supplemental Figures
This manuscript was presented at the 55th Annual Meeting of the Society of Thoracic Surgeons; 2019 January 27-29; San Diego, California.
Corresponding Author: Boris Sepesi, MD, FACS Department of Thoracic and Cardiovascular Surgery The University of Texas MD Anderson Cancer Center 1515 Holcombe Blvd, Houston, TX 77030 [email protected]
1
GLOSSARY OF ABBREVIATIONS CD: cluster of differentiation CI: confidence interval ICON: Immunogenomic Profiling of Non-Small Cell Lung Cancer LVI: lymphovascular invasion MCs: malignant cells mIF: multiplex immunofluorescence Mut/Mb: mutations per megabase NSCLC: non-small cell lung cancer OR: odds ratio PD-1: programmed cell death protein 1 PD-L1 programmed death ligand 1 TMB: tumor mutational burden TME: tumor microenvironment
2
ABSTRACT (233/250) Background: High tumor mutational burden (TMB) and PD-L1 expression are leading biomarkers in metastatic non-small cell lung cancer (NSCLC) and predict favorable response to checkpoint inhibitors. We sought to identify clinicopathologic characteristics associated with elevated TMB and PD-L1 expression among patients who underwent resection for NSCLC. Methods: NSCLC patients undergoing primary resection (2016-2018) were prospectively enrolled in an immunogenomic profiling project (ICON). Multiplex immunofluorescence (mIF) quantified densities (cells/mm2) of CD3+, CD3+CD8+, CD3+CD8+PD-1+, malignant cells (MCs), MCsPD-L1+, CD68+, CD68+PD-L1+, and CD20+ cells. Whole exome sequencing quantified TMB (mutations/megabase). TMB and MCsPD-L1+ were dichotomized according to the median of each. Results: 55 patients completed mIF and WES profiling. 41.8% (23/55) had pathologic stage I disease. Median TMB and MCsPD-L1+ were 3.91 and 0.62 cells/mm2, respectively. TMB was higher among smokers (p=0.001) and tumors with lymphovascular invasion (LVI)(p=0.051). TMB was positively correlated with densities of MCsPD-L1+ (r=0.293,p=0.030), CD68+PD-L1+ (r=0.289, p=0.033), and CD20+ (r=0.310,p=0.043) cells. The density of MCsPD-L1+ was associated with increased CD3+CD8+ (r=0.319,p=0.018) and CD68+PD-L1+ (r=0.371,p=0.005) cells. Patients with PD-L1HighTMBHigh tumors (30.9%, 17/55) had higher intratumoral densities of CD3+, CD3+CD8+, CD68+, CD68+PD-L1+, and CD20+ cells. On multivariable analysis LVI was associated with synchronous elevated TMB and PD-L1 expression (OR 3.53,p=0.039). Conclusions: NSCLC tumors with elevated TMB and PD-L1 expression are associated with LVI and increased intratumoral immune cell infiltration. These findings may potentially improve patient selection for checkpoint inhibitor therapy trials in the adjuvant setting.
3
Landmark phase III clinical trials which utilized immune checkpoint inhibitors in metastatic nonsmall cell lung cancer (NSCLC) identified increased tumor expression of programmed cell death ligand 1 (PD-L1) and tumor mutational burden (TMB) as the best predictive biomarkers of response to antiprogrammed cell death protein (PD-1) therapy (1-5). Other studies have further shown that tumor expression of PD-L1 and neoantigens are associated with local immune cell recruitment and enhanced tumor-specific immune responses (6, 7). Consequently, current guidelines include PD-L1 expression as a selection criterion for checkpoint blockade therapy in patients with metastatic NSCLC (8). Despite the fact that there are currently numerous ongoing clinical trials exploring checkpoint inhibitors in the neoadjuvant and adjuvant settings, the relationships between TMB, tumor expression of PD-L1, and clinicopathologic features in patients with resectable lung cancer have not been fully characterized (9, 10). Because an enhanced understanding of these relationships could potentially assist in identification of the most appropriate patients for enrollment in these novel trials, we examined TMB, PD-L1 expression, and features of the tumor microenvironment (TME) in a cohort of NSCLC patients enrolled in a prospective immunogenomic profiling study (Immunogenomic Profiling of Non-Small Cell Lung Cancer, ICON). In this study of chemotherapy-naïve resected lung cancers, we hypothesized that cancers characterized by synchronous high TMB and PD-L1 expression would be associated with distinct clinicopathologic features and with enhanced intratumoral immune cell infiltration PATIENTS AND METHODS Study Design and Patient Selection The ICON project was a prospective, multifaceted effort at the University of Texas MD Anderson Cancer Center that prospectively enrolled 150 patients prior to resection of primary lung cancer between April 1st, 2016 and August 31st, 2018. Eligible patients with clinical stage IA-IIIA NSCLC who were medically fit to undergo resection were enrolled. Patients included in the present study were those who had multiplex immunofluorescence and whole exome sequencing data available for analysis. Patients who
4
received neoadjuvant chemotherapy were excluded. Tumors were staged using the 7th edition of the American Joint Committee on Cancer’s schema (11). The study was approved by MD Anderson’s Institutional Review Board, and patients gave informed consent before enrollment. Multiplex Immunofluorescence Multiplex immunofluorescence (mIF) analysis was performed using methods that have been described and validated (12). Briefly, four micrometer-thick formalin fixed, paraffin embedded tumor sections were stained using an automated staining system (BOND-MAX; Leica Microsystems) using antibodies against cytokeratin AE1/AE3 (dilution 1:300, Dako), PD-L1 (clone E1L3N, 1:3000, Cell Signaling Technology), CD3 (1:100, Dako), CD8 (clone C8/144B, 1:300, Thermo Fisher Scientific), CD20 (clone L26, 1:250, Dako), and CD68 (clone PG-M1, 1:450, Dako), and PD-1 (clone EPR4877-2, 1:250, Abcam) and fluorophore staining using the Opal 7 kit (catalogue #NEL797001KT; PerkinElmer)(12). Stained slides were scanned using the Vectra 3.0 imaging system (PerkinElmer) with normal human tonsil tissue as a calibration control (12). After scanning, five fields (each 0.3345 mm2) were selected within the tumor using the phenochart 1.0.4 viewer (PerkinElmer). A trained pathologist (E.R.P.) supervised quantification of immune cell densities using InForm image analysis software (PerkinElmer). Marker colocalization was used to identify populations of T lymphocytes (CD3+), cytotoxic T lymphocytes (CD3+CD8+), antigen-experienced T lymphocytes (CD3+CD8+PD-1+), macrophages (CD68+), macrophages expressing PD-L1 (CD68+PD-L1+), B lymphocytes (CD20+, available for 43/55 cases), malignant cells (MCs, AE1/AE3+), and MCs expressing PD-L1 (MCsPD-L1+, AE1/AE+PD-L1+) in the tumor (epithelial nests) and stromal compartments (Supplemental Figure 1). Densities of each colocalized cell population were quantified as the average number of cells/mm2 (12). Cell densities in the tumor compartment and overall tumor (tumor and stromal compartments combined) were analyzed. Whole Exome Sequencing
5
DNA was extracted from frozen tumor tissue using the QIAamp DNA Mini kit (Qiagen) according to the manufacturer’s instructions. Exome capture was performed on 200ng genomic DNA per sample based on KAPA library prep (Kapa Biosystems) using the Agilent SurSelect Human All Exon V4 kit according to manufacturer’s instructions. Paired-end multiplex sequencing of samples was performed on the HiSeq 2500 sequencing platform (Illumina). Blood was used as a normal control. The BWA aligner (bwa-0.7.5a) mapped the raw reads to the human hg19 reference genome (UCSC Genome Browser: genome.UCSC.edu)(13). Duplicate reads were marked using the Picard (version 1.112, http://broadinstitute.github.io/picard/) “MarkDuplicates” module. The Genome Analysis Toolkit (modules “IndelRealigner” and “BaseRecalibrator”) was applied to perform insertion/deletion (indel) realignment and base quality recalibration. MuTect and Pindel were applied to each tumor and its matching normal tissue sample to detect somatic single nucleotide variants (SNVs) and small indels (14, 15). To ensure specificity, the following criteria were applied to filter the detected somatic SNVs and indels: coverage of ≥20 reads for the tumor and ≥10 for the normal control; total number of reads supporting the variant ≥4; and MuTect LOD score ≥6.3. The allele frequency from the normal sample was less than 0.01%. A population frequency threshold of 1% was used to filter out common variants in the databases of 1000 Genome Project, Exome Aggregation Consortium, and ESP6500. Indels were discarded if repetitive sequences were detected within 25 base pairs in the downstream regions. TMB was quantified as the total number of non-synonymous mutations per megabase (mut/Mb). Outcome Definitions and Statistical Analysis Pairwise correlations between continuous variables were assessed using Spearman’s correlations. Differences in continuous variables between groups were analyzed using Mann-Whitney U or KruskalWallis tests, and differences in categorical variables were analyzed using Pearson’s X2 and Fisher’s exact tests. TMB and MCsPD-L1+ were dichotomized according to the observed median values among the study cohort.
Univariable
logistic
regression
was
used
to
identify
associations
between
clinicopathological features (age, sex, histology, pathologic stage, tumor differentiation, and
6
lymphovascular invasion [LVI]) and synchronous elevated (>median) TMB and MCsPD-L1+. Variables with p<0.25 on univariable analysis and those of clinical interest were included in a multivariable model, with stepwise backwards selection then performed with a final selection criterion of p<0.10. Recurrencefree survival (RFS) was defined as the time from resection until disease recurrence or death; all patients without a RFS event at the end of the study period were censored at the date of last follow-up. The Kaplan-Meier method was used to estimate RFS, and differences between groups were analyzed using the log-rank test. All analyses were performed using SPSS version 24 (IBM, Armonk, NY) and STATA version 14.2 (StataCorp, College Station, TX). For all analyses, statistical significance was defined as two-tailed p<0.05. RESULTS Baseline Clinical and Treatment Characteristics Fifty-five patients met inclusion criteria (Supplemental Figure 2).
The majority had
adenocarcinoma, and 42% (n=23) had pathologic stage I disease (Table 1). One patient was diagnosed with stage IV disease due to intraoperative identification of pericardial deposits. Clinicopathologic and Immune Microenvironment Characteristics Associated with Tumor Expression of PD-L1 The median density of malignant cells expressing PD-L1 (MCsPD-L1+) among the entire study cohort was 0.62 (interquartile range [IQR] 0.00-30.19) cells/mm2. Whereas patients with low (≤median) MCsPD-L1+had nearly absent expression of PD-L1, densities of MCsPD-L1+ among patients with higher PD-L1 expression was much more variable (Figure 1). Densities of MCsPD-L1+ were higher in squamous tumors (median 110.69 [IQR 0.32-449.54] versus non-squamous 0.15 [IQR 0.00-6.82], p=0.003), those with poorly-differentiated (2.50 [0.00-710.96] versus moderately- 1.06 [IQR 0.00-29.04] versus well- 0.00 [IQR 0.00-0.23], p=0.041), those with lymphovascular invasion (LVI) (3.88 [IQR 0.00108.55] versus 0.03 [IQR 0.00-7.31], p=0.084), and those with more advanced pathologic stage (stage I 7
0.00 [IQR 0.00-5.34] versus stage II 2.50 [IQR 0.00-449.54] versus stages III/IV 1.03 [IQR 0.00-147.33], p=0.065), although the latter two associations did not reach statistical significance (Figure 2). No differences in MCsPD-L1+ were observed according to smoking status, sex, or age. The density of MCsPD-L1+ was positively correlated with TMB (r=0.293, p=0.030) and intratumoral CD3+CD8+ (r=0.319, p=0.018) and CD68+PDL1+ (r=0.371, p=0.005) cells (Figure 3). Similar associations were observed between MCsPD-L1+ and overall tumor infiltration by CD3+CD8+ (r=0.269, p=0.047) and CD68+PDL1+ cells (r=0.326, p=0.015). These findings demonstrated that the density of malignant NSCLC cells expressing PD-L1 varied according to tumor histopathologic features and was positively associated with TMB and immune cell infiltration. Clinicopathologic and Immune Microenvironment Characteristics Associated with TMB The median observed TMB in the study cohort was 3.91 (IQR 1.59-8.71) mut/Mb; 12 patients (21.8%) had TMB ≥10 mut/Mb. TMB was higher among smokers (median 5.00 [IQR 2.01-10.03] versus non-smokers 1.45 [IQR 0.84-2.33], p=0.001) and among tumors with LVI (5.02 [IQR 3.32-9.01] versus no LVI 2.40 [IQR 1.36-8.71], p=0.051), although the association with LVI did not reach statistical significance (Figure 4A and 4B). No associations were observed between TMB and age, histology, differentiation, or pathologic stage (Figure 4C and 4D). Tumors with increased (>median, n=27/55) TMB had higher densities of CD3+CD8+, CD20+, and CD68+PD-L1+ cells (Figure 5, Supplemental Figure 3). These analyses showed that NSCLC tumors with high TMB were associated with tobacco exposure, pathologic LVI, and increased immune cell infiltrates. Lymphovascular Invasion Is Associated with Synchronous Elevated TMB and PD-L1 Expression Because high tumor PD-L1 expression and elevated TMB have been previously identified as predictive of response to immune checkpoint inhibitors, we next analyzed clinicopathologic patient and tumor characteristics for associations with synchronous elevated (>median) TMB and MCsPD-L1+ (6, 16). PD-L1HighTMBHigh tumors (n=17/55) were more frequently noted to have LVI than were tumors with low (≤median) TMB and MCsPD-L1+ (PD-L1LowTMBLow; 11/17 [64.7%] versus 5/18 [27.8%],
8
respectively; p=0.028) (Figure 6). All (17/17) were former or current smokers. PD-L1HighTMBHigh tumors were marked by higher intratumoral densities of CD3+, CD3+CD8+, CD20+, and CD68+PD-L1+ cells than were PD-L1LowTMBLow (n=18/55) and other (PD-L1LowTMBHigh or PD-L1HighTMBLow, n=20/55) tumors (Figure 7). Upon multivariable analysis, LVI was identified as the only feature associated with PDL1HighTMBHigh tumors (Table 2). Exploratory survival analysis (median follow-up duration for the entire cohort, 16.1 [IQR 10.1-22.6] months) demonstrated worse RFS among patients with LVIPresentPDL1HighTMBHigh than those with LVIAbsentPD-L1LowTMBLow tumors, although this difference did not reach statistical significance (Figure 8). Together, these results demonstrated that PD-L1HighTMBHigh NSCLCs were characterized by the presence of LVI and increased immune cell infiltration. COMMENT In the present study of chemotherapy-naïve, surgically resected lung cancers we observed lower TMB and tumor expression of PD-L1 compared to previously published values for metastatic NSCLC (2, 17). We also identified differential TMB and expression of PD-L1 that varied according to baseline patient and pathologic tumor characteristics. These findings have potential clinical relevance, as future clinical trials utilizing checkpoint inhibitor immunotherapy could explore the presence of LVI, in addition to TMB and PD-L1, as a patient stratification or selection criterion. As immune checkpoint inhibitors continue to be explored in the setting of early NSCLC, an ability to identify patient subgroups most likely to derive therapeutic benefit from these agents is greatly needed (9, 10). Outside of induction clinical trials, the majority of surgically resectable NSCLC patients undergo resection of the primary tumor and mediastinal lymph nodes first, with further decisions about additional therapy guided by the final pathology (8). In patients with metastatic disease, a TMB greater than 10 mut/Mb has been identified as predictive of response to checkpoint blockade (8). The contrast between the frequency of tumors with TMB in excess of that threshold in the present report (22%) and that published in CheckMate 277 (44%) highlights the need for exploration of additional features that can be used singly or in combination to predict therapeutic benefit in patients with resectable disease (8). We
9
submit that our observation that the presence of LVI in the final pathologic specimen is associated with increased TMB and PD-L1 could potentially serve as a guide for the type of adjuvant therapy in NSCLC patients following resection, as assessment of LVI is routinely performed at the time pathologic evaluation. Tumors that demonstrate LVI have a high propensity for distant recurrence, even if resected in very early stages (18). However, current guidelines for adjuvant chemotherapy are based on the tumor size and nodal status rather than the presence or absence of LVI or high TMB and PD-L1 (8). Additionally, the ANVIL trial (NCT02595944) of adjuvant nivolumab does not further stratify patients who meet the inclusion criteria of absent EGFR mutations or ALK rearrangements. As we try to identify patients at increased risk of recurrence following complete surgical resection of early lung cancers, and as we search for more effective and safe adjuvant therapies, we speculate that patients with high TMB, PDL1, and LVI may derive a therapeutic benefit from adjuvant anti-PD-1/PD-L1 agents, either as monotherapy or in combination with chemotherapy. The identification of higher tumor infiltration by cytotoxic T lymphocytes, B lymphocytes, and macrophages expressing PD-L1 among tumors with higher TMB and expression of PD-L1 further supports this interpretation, since the efficacy of checkpoint blockade is greatest in the setting of concurrent high checkpoint expression and effector/helper lymphocyte infiltration (19, 20). Tumor PD-L1 expression has been identified as being associated with increased densities of immune infiltrates, and checkpoint inhibitor-induced release of PD-1+ T lymphocytes from immune cells expressing PD-L1 has been postulated as another mechanism by which antitumor immunity is enhanced by these agents (20-23). The role of B lymphocytes in the microenvironment has not been fully characterized, but recent work has demonstrated that they play important roles in antigen presentation and elaboration of antitumor antibodies and have protective prognostic effects (24). It seems plausible that the group of patients with tumors characterized by high TMB, elevated PD-L1 expression, the presence of LVI, and higher tumor immune cell infiltration would be that which is best suited for novel adjuvant immunotherapy trials in early NSCLC. On the other hand, for patients with a predicted poor anti-tumor immune response, the
10
preferred strategy may employ a combination of anti-PD-1/PD-L1 agents with CTLA-4 blockade or chemotherapy in an effort to improve lymphocyte recruitment (25). Despite the strength provided by the prospective nature of our study, we must acknowledge its limitations. Because only a subset of the tumors in the ICON project had both mIF and TMB data available for analysis, the present study is limited by its small sample size and lack of a validation set. Although the methodology of analysis of cell densities in this study (i.e., mIF) differs from that which is used to quantify PD-L1 expression clinically (i.e., tumor proportion score), mIF permits precise quantification of cell densities and has been previously validated in comparison to chromogenic immunohistochemistry (12). Moreover, the presence of LVI is not reliably detected on preoperative biopsy, and the applicability of these findings may be limited to the adjuvant setting following complete pathological assessment of the surgical specimen. Additionally, the short follow-up in this cohort and few observed RFS events to date precluded a robust analysis of the relationships between TME features with prognostic outcomes while controlling for relevant clinicopathologic characteristics. Further analyses of the prognostic impact of these features will be performed as follow-up data in this cohort matures. In summary, we report that primary NSCLC tumors with synchronous elevated TMB and PD-L1 expression are associated with the presence of LVI and with enhanced intratumoral immune cell infiltration. Further study is warranted to examine whether these findings can improve patient selection for novel trials of immune checkpoint inhibitor therapy in the adjuvant setting.
11
REFERENCES 1.
Rizvi H, Sanchez-Vega F, La K et al. Molecular determinants of response to anti–programmed
cell death (PD)-1 and anti–programmed death-ligand 1 (PD-L1) blockade in patients with non–small-cell lung cancer profiled with targeted next-generation sequencing. J Clin Oncol 2018;36(7):633-641. 2.
Borghaei H, Paz-Ares L, Horn L et al. Nivolumab versus docetaxel in advanced nonsquamous
non-small-cell lung cancer. N Engl J Med 2015;373(17):1627-1639. 3.
Gandhi L, Rodríguez-Abreu D, Gadgeel S et al. Pembrolizumab plus chemotherapy in metastatic
non–small-cell lung cancer. N Engl J Med 2018;378(22):2078-2092. 4.
Hellmann MD, Nathanson T, Rizvi H et al. Genomic features of response to combination
immunotherapy in patients with advanced non-small-cell lung cancer. Cancer Cell 2018;33(5):843852.e844. 5.
Hellmann MD, Ciuleanu T-E, Pluzanski A et al. Nivolumab plus ipilimumab in lung cancer with
a high tumor mutational burden. N Engl J Med 2018;378(22):2093-2104. 6.
Taube JM, Klein A, Brahmer JR et al. Association of PD-1, PD-1 ligands, and other features of
the tumor immune microenvironment with response to anti–PD-1 therapy. Clin Cancer Res 2014;20(19):5064-5074. 7.
Kiyotani K, Park JH, Inoue H et al. Integrated analysis of somatic mutations and immune
microenvironment in malignant pleural mesothelioma. Oncoimmunology 2017;6(2):e1278330. 8.
National Comprehensive Cancer Network. Non-small cell lung cancer. 2018;Version 3.2018.
9.
Forde PM, Chaft JE, Smith KN et al. Neoadjuvant PD-1 blockade in resectable lung cancer. N
Engl J Med 2018;378(21):1976-1986.
12
10.
Owen D, Chaft JE. Immunotherapy in surgically resectable non-small cell lung cancer. J Thorac
Dis 2018;10(Suppl 3):S404-S411. 11.
Goldstraw P, Crowley J, Chansky K et al. The IASLC Lung Cancer Staging Project: Proposals
for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM classification of malignant tumours. J Thorac Oncol 2007;2(8):706-714. 12.
Parra ER, Uraoka N, Jiang M et al. Validation of multiplex immunofluorescence panels using
multispectral microscopy for immune-profiling of formalin-fixed and paraffin-embedded human tumor tissues. Sci Rep 2017;7(1):13380. 13.
Gnirke A, Melnikov A, Maguire J et al. Solution hybrid selection with ultra-long oligonucleotides
for massively parallel targeted sequencing. Nat Biotechnol 2009;27:182. 14.
Cibulskis K, Lawrence MS, Carter SL et al. Sensitive detection of somatic point mutations in
impure and heterogeneous cancer samples. Nat Biotechnol 2013;31:213. 15.
Ye K, Schulz MH, Long Q, Apweiler R, Ning Z. Pindel: A pattern growth approach to detect
break points of large deletions and medium sized insertions from paired-end short reads. Bioinformatics 2009;25(21):2865-2871. 16.
Rizvi NA, Hellmann MD, Snyder A et al. Mutational landscape determines sensitivity to PD-1
blockade in non–small cell lung cancer. Science 2015;348(6230):124-128. 17.
Brahmer J, Reckamp KL, Baas P et al. Nivolumab versus docetaxel in advanced squamous-cell
non–small-cell lung cancer. N Engl J Med 2015;373(2):123-135. 18.
Yanagawa N, Shiono S, Abiko M, Ogata S-y, Sato T, Tamura G. Prognostic impact and initial
recurrence site of lymphovascular and visceral pleural invasion in surgically resected stage I non-smallcell lung carcinoma. Eur J Cardiothorac Surg 2013;44(3):e200-e206. 19.
Teng MWL, Ngiow SF, Ribas A, Smyth MJ. Classifying cancers based on T-cell infiltration and
PD-L1. Cancer Res 2015;75(11):2139-2145.
13
20.
Vilain RE, Menzies AM, Wilmott JS et al. Dynamic changes in PD-L1 expression and immune
infiltrates early during treatment predict response to PD-1 blockade in melanoma. Clin Cancer Res 2017;23(17):5024-5033. 21.
Velcheti V, Schalper KA, Carvajal DE et al. Programmed death ligand-1 expression in non-small
cell lung cancer. Lab Invest 2013;94:107. 22.
Cheng H, Borczuk A, Janakiram M et al. Wide expression and significance of alternative immune
checkpoint molecules, B7x and HHLA2, in PD-L1–negative human lung cancers. Clin Cancer Res 2018;24(8):1954-1964. 23.
Yu H, Chen Z, Ballman KV et al. Correlation of PD-L1 expression with tumor mutation burden
and gene signatures for prognosis in early-stage squamous cell lung carcinoma. J Thorac Oncol 2019;14(1):25-36. 24.
Bruno TC, Ebner PJ, Moore BL et al. Antigen-presenting intratumoral B cells affect CD4+
phenotypes in non–small cell lung cancer patients. Cancer Immunol Res 2017;5(10):898-907. 25.
Provencio-Pulla M, Nadal-Alforja E, Cobo M et al. Neoadjuvant chemo/immunotherapy for the
treatment of stages IIIA resectable non-small cell lung cancer (NSCLC): A phase II multicenter exploratory study—NADIM study. J Clin Oncol 2018;36(15_suppl):8521-8521.
14
Table 1: Baseline clinical, pathologic, and treatment characteristics of the study cohort (n=55). Variable Age, median (IQR) (years) Sex Female Male Smoking Never Former/Current Zubrod 0 1 FEV1 (% Predicted) Histology Adenocarcinoma Squamous cell carcinoma Adenosquamous Adenocarcinoma, sarcomatoid type Mixed pleomorphic carcinoma/adenocarcinoma Differentiation Well Moderate Poor Clinical Stage I II III Extent of Resection Sublobar Lobectomy/Bilobectomy Pneumonectomy Pathologic Tumor Size, median (IQR) (cm) Pathologic Stage I II III IV Lymphovascular Invasion Pathologic Margin R0 R1 Adjuvant Therapy Chemotherapy Radiotherapy FEV1: forced expiratory volume in one second
N (%) or Median (IQR) 67.0 (61.0-73.0) 30 (54.5) 25 (45.5) 8 (14.5) 47 (85.5) 51 (92.7) 4 (7.3) 81.0 (76.0-94.0) 38 (69.1) 14 (25.5) 1 (1.8) 1 (1.8) 1 (1.8) 8 (14.5) 30 (54.5) 17 (30.9) 33 (60.0) 19 (34.5) 3 (5.5) 5 (9.1) 47 (85.5) 3 (5.5) 3.5 (2.8-5.5) 23 (41.8) 19 (34.5) 12 (21.8) 1 (1.8) 24 (43.6) 49 (89.1) 6 (10.9) 27 (49.1) 5 (9.1)
15
Table 2: Univariable and multivariable analyses of factors associated with tumors characterized by synchronous high (>median) tumor mutational burden and density of malignant cells expressing PD-L1 (n=17/55) versus all other tumors (n=38/55). Univariable Variable N (%) OR 95% CI Age (years) 1.02 0.96-1.08 Sex (Male) 25 (45.5) 1.55 0.49-4.88 Histology (Squamous)* 15 (27.3) 2.63 0.76-9.08 Pathologic Stage* I 23 (41.8) Reference 19 (34.5) 2.62 0.68-10.06 II III/IV 13 (23.6) 1.60 0.34-7.46 Differentiation (Poor vs Well/Moderate) 17 (30.9) 1.96 0.59-6.55 Lymphovascular Invasion* 24 (43.6) 3.53 1.06-11.70 Smoking status omitted because all PD-L1High/TMBHigh patients were former or current smokers *Included in multivariable model, with stepwise backwards selection then performed CI: confidence interval; OR: odds ratio
p 0.602 0.457 0.128 0.373 0.161 0.549 0.274 0.039
OR
3.53
Multivariable 95% CI
1.06-11.70
p
0.039
16
FIGURE LEGENDS Figure 1: Tumor expression of PD-L1 after dichotomization according to the observed median density of PD-L1+ malignant cells (n=55). Figure 2: Associations between tumor expression of PD-L1 and clinicopathologic and immune microenvironment characteristics (n=55). Increased densities of PD-L1+ malignant cells were observed in (A) squamous and (B) poorly-differentiated tumors. Trends (0.05< p< 0.10) towards increased densities in MCs+PD-L1+ were noted in tumors with (C) lymphovascular invasion and those with (D) advanced pathologic stage. Figure 3: NSCLCs with high (>median) densities of MCsPD-L1+ had higher TMB (A) and tumor infiltration by cytotoxic lymphocytes (CD3+CD8+, B) and macrophages expressing PD-L1 (CD68+PDL1+, C)(n=55). Figure 4: Tumor mutational burden (TMB) was increased among (A) smokers and (B) in tumors with LVI, but was not associated with (C) tumor differentiation or (D) pathologic stage (n=55). Figure 5: Tumors with high TMB had higher intratumoral densities of cytotoxic T lymphocytes (CD3+CD8+, A), B lymphocytes (CD20+, B), and macrophages expressing PD-L1 (CD68+PD-L1+, C)(n=55). Figure 6: Lymphovascular invasion was more frequently present in tumors with synchronous high TMB and elevated densities of malignant cells expressing PD-L1 (PD-L1HighTMBHigh; n=17/55) than in tumors with concomitant low PD-L1 expression and TMB (PD-L1LowTMBLow; n=18/55). Figure 7: Representative images of tumor samples of (A) a patient with pathologic stage II adenosquamous carcinoma with lymphovascular invasion (LVI; A, left panel) and elevated (>median) TMB and PD-L1, and (B) a patient with stage II squamous cell carcinoma with absent LVI and low (≤median) TMB and PD-L1. Whereas (A, right panel) the first patient had high intratumoral lymphocytic
17
infiltrates, the second (B) had low immune cell infiltration. (C) PD-L1HighTMBHigh tumors (n=17/55) had higher intratumoral CD3+CD8+ (left panel), CD20+ (center panel), and CD68+PD-L1+ (right panel) cell densities than did those with PD-L1LowTMBLow (n=18/55) and all other tumors (n=20/55). Figure 8: Recurrence-free survival according to lymphovascular invasion (LVI), density of tumor cells expressing PD-L1 (MCsPD-L1+), and TMB. The cohort was dichotomized according to observed medians of MCsPD-L1+ and TMB. Supplemental Figure 1: Representative multiplex immunofluorescence images of tumors with cells staining positive for CD3, CD8, PD-1, PD-L1, CD68, DAPI (marker of cell nuclei), and AE1/AE3 (marker of malignant cells). Supplemental Figure 2: CONSORT diagram depicting patient selection for the study. Supplemental Figure 3: Tumors with high TMB had higher overall infiltration (tumor and stromal compartments) of cytotoxic T lymphocytes (CD3+CD8+, A), B lymphocytes (CD20+, B), and macrophages expressing PD-L1 (CD68+PD-L1+, C)(n=55).
18