Abstract
Objectives:
Biomarkers are needed to identify patients likely to respond to neoadjuvant immunotherapy (NIT) prior to receiving definitive treatment.
Materials and Methods:
We hypothesized that expression of tumor cell HLA class I would correlate with pathologic response (PR) following NIT for primary untreated head and neck cancer. Multispectral immunofluorescence of pre- and post-treatment biopsy specimens from a neoadjuvant study of bintrafusp alfa, a dual TGF-β and PD-L1 inhibitor, was performed.
Results:
Discordant expression of tumor cell HLA class I and PD-L1 measured by multispectral immunofluorescence was observed with most positive tumor cells expressing HLA class I or PD-L1 but not both. Spatial analysis revealed colocalization between tumor parenchyma T cells and HLA class I positive tumors cells, but no clear colocalization between T cells and PD-L1 positive tumor cells. Greater pre-treatment tumor cell HLA class I expression, but not PD-L1 expression or tumor T cell infiltration, correlated with the development of a PR. Additionally, increased tumor cell HLA class I expression after NIT compared to before NIT correlated with development of a PR, whereas inconsistent changes in PD-L1 and T cell infiltration were observed after treatment in all patients.
Conclusions:
These data provide the rationale for the study of tumor cell HLA class I expression in larger prospective studies powered to determine the performance of biomarkers of PR in newly diagnosed HNSCC patients receiving NIT.
Keywords: newly diagnosed, head and neck cancer, HPV negative, neoadjuvant immunotherapy, HLA class I, PD-L1 biomarker, pathologic response
Introduction
Biomarkers can identify patients most likely to respond and clinically benefit from investigational treatments while avoiding risk to those unlikely to respond1. Neoadjuvant immune checkpoint blockade (ICB) targeting the programmed death receptor (PD-1) or ligand (PD-L1) pathway leads to pathologic responses in a subset of patients with newly diagnosed head and neck cancer (HNSCC) not associated with human papillomavirus (HPV) and improves recurrence free survival (RFS)2–8. Minimizing risk is of utmost importance in this patient population that has yet to receive potentially curative standard-of-care treatment. Identifying biomarkers of response to neoadjuvant ICB from initial completed studies may inform which potential biomarkers to include in larger, future prospective studies.
Expression of PD-L1 on tumor and immune cells via immunohistochemistry has been extensively studied as a biomarker of response to PD-pathway ICB in patients with relapsed cancer1. Measurable expression of PD-L1 in tumors positively correlates with objective response to PD-pathway ICB in patients with recurrent or metastatic head and neck cancer not associated with HPV9. Short-term response in neoadjuvant immunotherapy clinical trials is measured by pathologic response PR in the surgical specimen10,11. Contrary to correlations observed between PD-L1 expression and objective response by Response Evaluation Criteria in Solid Tumors (RECIST) in patients with relapsed HNSCC, expression of PD-L1 within tumors does not clearly correlate with pathologic response in numerous neoadjuvant immunotherapy studies3,4,6–8,12.
The type II interferon (IFN)-inducible expression of PD-L1 serves as the biologic basis for its role as a biomarker of response to ICB. Within the tumor microenvironment (TME), tumor antigen specific T cells produce IFN that increases IFN-inducible expression of PD-L1 on tumors cells in proximity through a process termed adaptive immune resistance13. PD-L1 expression on tumor cells in the TME implies current or prior local immune activation. Yet, expression of PD-L1 on tumor and immune cells within the TME is governed by a diverse array of tumor cell intrinsic and extrinsic factors and may be uncoupled from IFN14. We hypothesized that greater expression of another IFN-inducible protein critical for activation of anti-tumor T cell immunity, human leukocyte antigen (HLA) class I, would positively correlate with the development of a pathologic tumor response following neoadjuvant immunotherapy. To explore this hypothesis, we performed multispectral immunofluorescence on tumor biopsies before and after neoadjuvant immunotherapy with dual TGF-β neutralization and PD-L1 blockade in a dataset of newly diagnosed HPV-negative HNSCC where the degree of pathologic response was already established8.
Results
Discordant expression of HLA class I and PD-L1 on tumor cells
Patients with newly diagnosed HPV-negative HNSCC were treated on a neoadjuvant immunotherapy study, and the degree of pathologic response was determined in the surgical specimens using standardized criteria10,11 (Figure 1a). Pathologic responses ranged from 3% to 70%, with no tumor demonstrating a major pathologic response (≥90% tumor regression) and 5 of 14 tumors (36%) demonstrating a partial response (≥50% tumor regression). Less than 50% tumor regression was observed in the remaining tumors that were considered; these tumors were considered to have no response. Pre- and post-treatment tumor biopsies from all 14 patients were subjected to multispectral immunofluorescence and digital pathologic analysis to determine whole slide protein level expression of HLA class I (clone EMR8-5, recognizes folded and unfolded conformations), PD-L1 (EL13N) and CD3 (sp7). Representative photomicrographs for these stains are shown in Figure 1b.
Figure 1– Multispectral staining for HLA class I, PD-L1 and CD3.
a, schema of the neoadjuvant immunotherapy clinical trial from which these tumor samples were obtained.
b, photomicrographs of representative multispectral and single-color stains. The bottom row displays higher magnification photomicrographs showing membranous tumor cell HLA class I and PD-L1 staining and membranous CD3 staining. Antibody details are listed in Table I. HNSCC, head and neck squamous cell carcinoma; DAPI, 4′,6-diamidino-2-phenylindole; CK, cytokeratin; HLA, human leukocyte antigen; PD-L1, programmed death-ligand 1.
Whole slide tumor cell expression of HLA class I and PD-L1 was quantified from each specimen using digital pathologic analysis. We first determined the positivity and degree of concordant expression of HLA class I and PD-L1 on 718,962 cytokeratin positive tumor cells in the 14 pre-treatment samples (Figure 2a). The fraction of tumor cells in each sample that displayed measurable expression of these markers is shown in Figure 2b, with patients rank-ordered from greatest to least observed pathologic response following neoadjuvant immunotherapy. A high degree of heterogeneity between pre-treatment samples was observed. On average, 16% (range 4–47%) of tumor cells were positive for both HLA class I and PD-L1, 31% (4–59%) were positive for HLA class I alone, 10% (0–36%) were positive for PD-L1 alone, and 41% (22–74%) were negative for both. Of note, the only two patients to have developed locoregional recurrence within one year of treatment, patients 1 and 4, displayed a reduced percentage of HLA class I positive cells (P=0.02, non-parametric Mann-Whitney test) compared to the other 12 patients. A high degree of discordant tumor cell expression of HLA class I and PD-L1 was observed, with 315,867 of 463,008 (68.2%) of positive tumor cells expressing either HLA class I or PD-L1 but not both.
Figure 2– Discordant tumor cell HLA class I and PD-L1 expression.
a, schema illustrating the digital pathology approach utilized to quantify whole slide, tumor cell-specific expression of HLA class I and PD-L1 including tumor annotation and measurement of single-cell fluorescence intensity.
b, bar plots (top section) display the fraction of HLA class I and/or PD-L1 positive tumor cells on pre-treatment tumor biopsies (n=14) as determined by whole slide digital pathology analysis. Patient samples are rank ordered from left to right by pathologic response following neoadjuvant immunotherapy. The magnitude of primary tumor pathologic response is quantified in a waterfall plot (bottom section), with tumor displaying partial response (≥50% tumor regression) or no response (<50% tumor regression) indicated by color.
HLA, human leukocyte antigen; PD-L1, programmed death-ligand 1.
T cells spatially co-localize with tumor cell HLA class I expression
Expression of HLA class I and PD-L1 on carcinoma cells can be induced by interferon produced by T cells. Given the observed discordance in tumor cell expression of HLA class I and PD-L1 in the pre-treatment tumors, we studied the spatial relationship between expression of these markers on tumor cells and CD3 positive T cells with the hypothesis that CD3 positive cells would predominantly co-localize with tumor cell HLA class I or PD-L1 but not both (Figure 3a). To normalize for any differences in the quantity of CD3 positive cells infiltrating the tumor parenchyma between samples, we considered the fraction of all CD3 positive cells within 20 μm of either HLA class I or PD-L1 positive or negative tumor cells. The distance of 20 μm (or approximately 4–5 cell layers thickness) was chosen based data demonstrating that tumor cells closer in proximity to IFN-producing T cells are more likely to respond to IFN that tumor cells further away15. Proximity analysis revealed a greater fraction of CD3 positive cells within 20 μm of HLA class I positive tumor cells (mean, 92%, range 78–98%) compared to the fraction within 20 μm of HLA class I negative tumor cells (mean 30%, range 11–55%) in all 14 tumors (Figure 3b). No significant difference in the fraction of CD3 positive cells within 20 μm of PD-L1 positive (mean 57%, range 20–70%) or negative (mean 59%, range 42–89%) tumor cells was observed. Similarly, whereas the mean distance between all CD3 positive cells and HLA class I positive cells (mean 10 μm, range 8–17 μm) was less than the mean distance between CD3 positive cells and HLA class I negative tumor cells (mean 37 μm, range 12–52 μm), no significant difference in the mean distance between CD3 positive cells and PD-L1 positive (mean 35 μm, range 12–56 μm) or negative (mean 28 μm, range 13–38 μm) tumor cells was observed (Figure 3c). Together, these findings indicate that CD3 positive cells spatially co-localize with HLA-class I positive tumor cells to a greater degree than PD-L1 positive tumor cells in this cohort of newly diagnosed carcinomas.
Figure 3– CD3 co-localization with HLA class I or PD-L1 positive tumor cells.
a, schema illustrating the digital pathology approach utilized to determine the spatial localization of CD3 positive cells and HLA class I or PD-L1 positive tumor cells including proximity analysis of digital spatial maps. Within the representative proximity analysis image, lines connect HLA positive tumor cells (yellow) with CD3 positive cells (green). Above the dot plots are representative photomicrographs of regions of HLA class I or PD-L1 positive or negative tumor cells and CD3 positive cells.
b, dot plots showing the fraction of all CD3 positive cells within 20 um of HLA class I or PD-L1 positive or negative tumor cells in whole slide pre-treatment tumor biopsies (n=14). The two measurements for each tumor are connected with a line. Significance between fractions of CD3 positive cells was determined with a paired, two-tailed t-test.
c, dot plots showing the mean distance between CD3 positive cells and HLA class I or PD-L1 positive or negative tumor cells in whole slide pre-treatment tumor biopsies (n=14). The two measurements for each tumor are connected with a line. Significance between fractions of CD3 positive cells was determined with a paired, two-tailed t-test.
CK, cytokeratin; HLA, human leukocyte antigen; PD-L1, programmed death-ligand 1.
HLA class I expression correlates with pathologic response following neoadjuvant ICB
Multiple studies of neoadjuvant immunotherapy for newly diagnosed head and neck carcinoma have reported no significant correlation between tumor cell PD-L1 expression and the development of a PR3,5–8. In this cohort of newly diagnosed carcinomas, patients that developed a PR following treatment displayed significantly greater tumor cell HLA class I expression pre-treatment compared to patients that did not develop a PR (Figure 4a). A representative example of this staining in a patient that developed a PR and a patient that did not develop a PR is shown in Figure 4b. Consistent with prior results, tumor cell PD-L1 expression did not correlate with the development of a PR (Figure 4c). Similarly, infiltration of CD3 positive cells into the tumor parenchyma did not correlate with the development of a PR (Figure 4d). Together these data suggest that greater tumor cell HLA class I expression, but not tumor cell PD-L1 or T cell infiltration into tumor parenchyma, predicts the development of a PR following neoadjuvant immunotherapy in this cohort of patients with newly diagnosed carcinomas.
Figure 4– Correlation between baseline HLA class I, PD-L1 and CD3 and pathologic response.
a, dot plot shows the whole slide pre-treatment tumor cell HLA class I H-score in patients that did (n=5) or did not (n=9) develop a PR after neoadjuvant immunotherapy. Significance was determined with an unpaired, two-tailed t-test.
b, representative photomicrographs a pre-treatment tumor from a patient that did (patient 3) and did not (patient 10) develop a PR following neoadjuvant immunotherapy. Tumor cell HLA class I H-scores are inset.
c, dot plot shows the whole slide pre-treatment tumor cell PD-L1 H-score in patients that did (n=5) or did not (n=9) develop a PR after neoadjuvant immunotherapy. Significance was determined with an unpaired, two-tailed t-test.
d, dot plot shows the whole slide pre-treatment density of CD3 positive cells in patients that did (n=5) or did not (n=9) develop a PR after neoadjuvant immunotherapy. Significance was determined with an unpaired, two-tailed t-test.
DAPI, 4′,6-diamidino-2-phenylindole; CK, cytokeratin; HLA, human leukocyte antigen; PD-L1, programmed death-ligand 1; PR, pathologic response.
We next considered whether transcript counts of PD-L1 or HLA class I, determined from bulk RNA sequencing, positively correlated with protein level quantification as determined by immunohistology. PD-L1 transcript counts positively correlated with tumor cell PD-L1 H-score (Supplemental Figure 1). However, HLA class I transcript counts did not positively correlate with tumor cell HLA class I H-score. This data suggests that transcriptional data may not accurately reflect the degree of protein level HLA class I expression on tumor cells in newly diagnosed carcinomas.
We next explored whether the change in expression of tumor cell HLA class I or PD-L1 or tumor parenchyma CD3 positive cell infiltration from pre- to post-neoadjuvant immunotherapy correlated with the development of a PR. Patients that developed a PR after neoadjuvant immunotherapy consistently displayed an increase in tumor cell HLA class I protein expression after treatment (Figure 5a), whereas patients that did not develop a PR generally had stable or decreased expression after treatment. Representative examples of changes in tumor cell HLA class I expression after neoadjuvant immunotherapy are shown for a patient that did (Figure 5b) and did not (Figure 5c) develop a PR. Change in tumor cell PD-L1 expression or tumor parenchyma CD3 positive cell infiltration was inconsistent and no clear associations were observed in patients that did or did not develop a PR. In conclusion, these data indicate that baseline and change in tumor cell HLA class I expression, but not tumor cell PD-L1 expression or tumor parenchyma CD3 positive cell infiltration, correlates with the development of a PR in this cohort of newly diagnosed carcinoma patients receiving neoadjuvant immunotherapy.
Figure 5– Correlation between change in HLA class I, PD-L1 and CD3 with treatment and pathologic response.
a, dot plot showing the log2 of the fold change (post-treatment/pre-treatment) for whole slide tumor cell HLA class I or PD-L1 expression or CD3 positive cell density in patients that did (n=5) or did not (n=9) develop a PR after neoadjuvant immunotherapy. Values greater than zero (vertical line on plot) indicate an increase in expression after treatment, values less than zero indicate a decrease in expression. Significance for each marker was determined with an unpaired, two-tailed t-test.
b, representative pre-treatment and post-treatment photomicrographs of tumor staining from a patient that developed a PR (patient 13). Tumor cell HLA class I H-scores are inset.
c, representative pre-treatment and post-treatment photomicrographs of tumor staining from a patient that did not develop a PR (patient 9). Tumor cell HLA class I H-scores are inset.
DAPI, 4′,6-diamidino-2-phenylindole; CK, cytokeratin; HLA, human leukocyte antigen; PD-L1, programmed death-ligand 1; PR, pathologic response.
Discussion
Here we demonstrate that pre-treatment expression of tumor cell HLA class I may serve as a biomarker for the development of a PR following neoadjuvant immunotherapy in patients with newly diagnosed HNSCC. An increase in tumor cell HLA-class I expression after treatment also correlated with the development of a PR. Additionally, the two patients in this cohort that recurred within one year of treatment demonstrated lower pre-treatment tumor cell HLA class I expression compared to patients that did not recur8. We also demonstrate a lack of correlation between pre-treatment PD-L1 and the development of a PR in the same clinical samples, a finding consistent with multiple prior neoadjuvant clinical studies in HNSCC3–7. This hypothesis-generating correlation between greater tumor cell HLA expression and the development of a PR following neoadjuvant immunotherapy warrants further study in larger prospective studies. A consistent biomarker of response could identify which patients should be subjected to the additional risk associated with neoadjuvant immunotherapy based upon the likelihood of clinical benefit.
Tumor cell surface HLA class I presents tumor antigens to CD8+ T cells for specific, T cell receptor-mediated activation. It is a critical component of the antigen presentation machinery and must be expressed on the surface of tumor cells for T cell-mediated detection and elimination. HLA class I expression is required for baseline or immunotherapy induced T cell-mediated anti-tumor immunity, and its loss through genetic or epigenetic mechanisms renders tumor cells undetectable. To our knowledge, no other prospective neoadjuvant study has evaluated whether tumor cell HLA class I expression can serve as a biomarker of response. This has, however, been studied in patients receiving immunotherapy for relapsed or metastatic disease. In patients with metastatic melanoma, baseline HLA class I expression predicted response to CTLA-4 ICB but not PD-1 ICB16. In patients with different relapsed cancers, heterozygosity at HLA class I loci (as opposed to homozygosity) predicted response to PD-1 or CTLA-4 ICB, indicating that a greater diversity of HLA class I alleles may allow for a greater diversity of T cell antigens to be presented to T cells17. Our results indicate that bulk HLA class I transcript counts derived from multiple cell types do not positively correlate with tumor cell surface-specific HLA class I protein expression, suggesting these previously published data based on bulk RNA-sequencing data must be interpreted cautiously.
We observed that most positive tumor cells in most patients express HLA class I or PD-L1 and not both concurrently. The underlying cause of this discordant tumor cell HLA class I and PD-L1 expression is unclear. Our finding of co-localization between T cells and HLA class I positive tumor cells within the tumor parenchyma suggests that HLA expression may be primarily driven by interferon. This is consistent with prior observations that baseline tumor cell HLA class I expression is increased upon exposure to type II interferon12. We are unable to comment on whether observed HLA class I negative tumor cells are unable to express class I or rather are not exposed to interferon from T cells, but mutations that would directly abrogate HLA class I expression were not observed previously in this cohort of tumors8. Our observations also offer little insight into the drivers(s) of PD-L1 expression in this cohort, but oncogenic signaling pathways downstream of driver mutations, microenvironmental signals and numerous post-transcriptional and post-translational regulatory mechanisms may all play roles14,18. Follow-up of the highly correlative findings described in this work will require future mechanistic studies investigating heterogeneity in interferon-response and oncogenic signaling pathways as putative drivers of HLA class I and PD-L1. Existing and emerging technologies such as single tumor cell RNA-sequencing and spatial transcriptomics/proteomics may be used to gain additional insight onto these processes.
Our finding that T cell quantification in the tumor parenchyma did not correlate with PR, but that tumor cell HLA class I expression did correlate with PR, suggests that baseline and treatment induced T cell function and production of interferon may be a more important determinant of response to neoadjuvant immunotherapy compared to the presence of T cells alone. This is consistent with previous reports indicating that the magnitude of antigen-specific T cell responses within tumors, and not the quantity of tumor infiltrating T cells, correlates with development of a PR following neoadjuvant immunotherapy4,8.
The major limitation of this study includes small sample size. The findings observed in this exploratory dataset need to be studied in larger prospective studies appropriately powered to draw definitive conclusions about the performance of a biomarker such as tumor cell HLA class I expression. This opportunity exists with large prospective trials of neoadjuvant immunotherapy underway for newly diagnosed HNSCC19. Another limitation of this study is the lack of measurement of T cell activation markers. Based on this preliminary data, future studies should employ techniques that allow simultaneous direct or indirect detection of T cell activation. This could allow, for example, validation that T cells that co-localize with HLA class I positive tumor cells are producing interferon. Previously published bulk transcriptional profiles form this dataset8 demonstrated no clear association between the baseline or change in IFN gene signature and response that, when combined with data presented in this work, support that spatial IFN production by certain populations of T cells may drive biologically relevant changes in the tumor. An additional limitation of this study is the use of one agent with two mechanisms of action in the neoadjuvant setting. As such, we are unable to attribute the observed associations between tumor cell HLA class I expression and pathologic response to either PD-L1 blockade or TGF-β neutralization alone, although to our knowledge no clear mechanistic link between TGF-β signaling and HLA class I expression on carcinoma cells exists. Limitations of this study notwithstanding, these preliminary data provide the data-driven rationale for the study of tumor cell HLA class I expression as a potential biomarker of response in current and future larger prospective studies of neoadjuvant immunotherapy in patients with newly diagnosed head and neck carcinoma.
Methods
Clinical samples
Clinical and correlative results for NCT04247282 have been previously reported8. Briefly, patients with newly diagnosed T2-T4, N0–3, M0 HNSCC not associated with HPV were treated with neoadjuvant bintrafusp alfa, a bifunctional fusion protein composed of the extracellular domain of the human transforming growth factor β receptor II fused via a flexible linker to the C-terminus of each heavy chain of an IgG1 antibody blocking programmed death ligand 1 (anti-PD-L1), prior to surgical resection. Tumor biopsies were performed before neoadjuvant immunotherapy treatment and at the time of surgery. Pathologic responses were defined as the area percentage of viable tumor within the tumor bed (surface area of residual viable tumor/surface area of total tumor bed × 100), assessed by two independent pathologists. A formal PR was assigned to any surgical specimen demonstrating ≥50% regression of tumor within the tumor bed. All H&E slides from each block of each primary tumor surgical specimen were assessed and a weighted (by surface area) pathologic response mean was determined for each patient, rounded to the nearest one percent.
Multispectral Immunofluorescence
Tyramine Signal Amplification (TSA) Opal™ technology and fully automated staining systems were used for immunofluorescence staining as described. After individual primary antibody optimization, primary and secondary antibody and opal pairings were optimized for minimum background and desired signal amplification in monoplex immunofluorescence using head and neck carcinoma sections. Supplemental Table I lists primary antibody, secondary antibody, and Opal™ details. Whole slide images were obtained at 40X magnification using 5-color whole slide unmixing filters on a Vectra Polaris (Akoya Biosciences). All paired pre- and post-treatment tumors were stained and scanned concurrently.
Image Analysis
Whole slide analysis of each stained slide was performed with HALO Image Analysis software (v3.3, Indica Labs). Tumor annotations were performed using the Random Forest Tissue Classifier Algorithm. Standard nuclear segmentation was used identify individual cells. Fluorescence intensities of tumor cell HLA class I and PD-L1 for each cell were determined using the HALO Highplex FL Analysis Algorithm. Common fluorescence thresholds used to assign scaled intensity (1+, 2+ or 3+) were used for each patient’s paired pre- and post-treatment tumor. H-score was used to describe protein expression in cells and was defined as: (% of 1+ cells × 1)+(% of 2+ cells × 2)+(% of 3+ cells × 3) for a range of 0–300. Density of T cells was defined as the absolute number of CD3 positive cells per unit area (mm2). Spatial analysis between T cells and HLA class I or PD-L1 positive or negative cytokeratin positive tumor cells was performed using the HALO Proximity Analysis Algorithm on spatial plots generated from T cell and tumor cell object cell X&Y coordinates. The fraction of tumor cells positive or negative for HLA class I or PD-L1 was calculated using the HALO Density Heatmap Algorithm. Raw immunofluorescence quantified data are included in Supplemental Table II.
Bulk RNA sequencing
RNA sequencing was performed and analyzed as previously described8. Original data can be accessed through the database of Genotype and Phenotypes (dbGaP), accession number phs002849.v1.p1.
Statistics
Significant differences between paired sets of data were determined with paired, two-tailed t-tests. Significant differences between unpaired sets of data were determined with unpaired, two-tailed t-tests. A P-value (P) of less than 0.05 was considered significant. All analyses were performed in GraphPad Prism (v9) and some schema images were generated in BioRender (BioRender.com). All analyzed data are included in this manuscript.
Supplementary Material
Supplemental Figure 1 - Concordance between PD-L1 and HLA class I transcript counts and protein level expression
Supplemental Table I. Multispectral immunofluorescence reagents.
Supplemental Table II. Raw Multispectral immunofluorescence data.
Highlights:
Tumor cell HLA class I expression is not associated with tumor cell PD-L1 expression
Tumor cell HLA class I expression after treatment associates with pathologic response
Tumor cell PD-L1 expression does not associate with pathologic response
Study of HLA class I expression as a biomarker of response in neoadjuvant studies is warranted
Acknowledgements:
The authors thank Dr. Nyall London and Dr. Charalampos Floudas for their critical review of this work. Funding for this work was provided by the Center for Cancer Research, National Cancer Institute of the National Institutes of Health.
Footnotes
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Study Approval
Clinical study NCT04247282 was approved by the National Institutes of Health Clinical Center Institutional Review Board. Full informed consent was obtained from all patients participating in NCT04247282.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. 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.
References
- 1.Nishino M, Ramaiya NH, Hatabu H, Hodi FS. Monitoring immune-checkpoint blockade: response evaluation and biomarker development. Nat Rev Clin Oncol. 2017;14(11):655–668. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Uppaluri R, Campbell KM, Egloff AM, et al. Neoadjuvant and Adjuvant Pembrolizumab in Resectable Locally Advanced, Human Papillomavirus-Unrelated Head and Neck Cancer: A Multicenter, Phase II Trial. Clin Cancer Res. 2020;26(19):5140–5152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Schoenfeld JD, Hanna GJ, Jo VY, et al. Neoadjuvant Nivolumab or Nivolumab Plus Ipilimumab in Untreated Oral Cavity Squamous Cell Carcinoma: A Phase 2 Open-Label Randomized Clinical Trial. JAMA Oncol. 2020;6(10):1563–1570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ferrarotto R, Bell D, Rubin ML, et al. Impact of Neoadjuvant Durvalumab with or without Tremelimumab on CD8(+) Tumor Lymphocyte Density, Safety, and Efficacy in Patients with Oropharynx Cancer: CIAO Trial Results. Clin Cancer Res. 2020;26(13):3211–3219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ferris RL, Spanos WC, Leidner R, et al. Neoadjuvant nivolumab for patients with resectable HPV-positive and HPV-negative squamous cell carcinomas of the head and neck in the CheckMate 358 trial. J Immunother Cancer. 2021;9(6). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wise-Draper TM, Gulati S, Palackdharry S, et al. Phase II Clinical Trial of Neoadjuvant and Adjuvant Pembrolizumab in Resectable Local-Regionally Advanced Head and Neck Squamous Cell Carcinoma. Clin Cancer Res. 2022;28(7):1345–1352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Vos JL, Elbers JBW, Krijgsman O, et al. Neoadjuvant immunotherapy with nivolumab and ipilimumab induces major pathological responses in patients with head and neck squamous cell carcinoma. Nat Commun. 2021;12(1):7348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Redman JM, Friedman J, Robbins Y, et al. Enhanced neoepitope-specific immunity following neoadjuvant PD-L1 and TGF-b blockade in HPV-unrelated head and neck cancer. J Clin Invest. 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Burtness B, Harrington KJ, Greil R, et al. Pembrolizumab alone or with chemotherapy versus cetuximab with chemotherapy for recurrent or metastatic squamous cell carcinoma of the head and neck (KEYNOTE-048): a randomised, open-label, phase 3 study. Lancet. 2019;394(10212):1915–1928. [DOI] [PubMed] [Google Scholar]
- 10.Stein JE, Lipson EJ, Cottrell TR, et al. Pan-Tumor Pathologic Scoring of Response to PD-(L)1 Blockade. Clin Cancer Res. 2020;26(3):545–551. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cottrell TR, Thompson ED, Forde PM, et al. Pathologic features of response to neoadjuvant anti-PD-1 in resected non-small-cell lung carcinoma: a proposal for quantitative immune-related pathologic response criteria (irPRC). Ann Oncol. 2018;29(8):1853–1860. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ferris RL, Whiteside TL, Ferrone S. Immune escape associated with functional defects in antigen-processing machinery in head and neck cancer. Clin Cancer Res. 2006;12(13):3890–3895. [DOI] [PubMed] [Google Scholar]
- 13.Tumeh PC, Harview CL, Yearley JH, et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature. 2014;515(7528):568–571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Chen S, Crabill GA, Pritchard TS, et al. Mechanisms regulating PD-L1 expression on tumor and immune cells. J Immunother Cancer. 2019;7(1):305. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hoekstra ME, Bornes L, Dijkgraaf FE, et al. Long-distance modulation of bystander tumor cells by CD8(+) T cell-secreted IFNgamma. Nat Cancer. 2020;1(3):291–301. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Rodig SJ, Gusenleitner D, Jackson DG, et al. MHC proteins confer differential sensitivity to CTLA-4 and PD-1 blockade in untreated metastatic melanoma. Sci Transl Med. 2018;10(450). [DOI] [PubMed] [Google Scholar]
- 17.Chowell D, Morris LGT, Grigg CM, et al. Patient HLA class I genotype influences cancer response to checkpoint blockade immunotherapy. Science. 2018;359(6375):582–587. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Shklovskaya E, Rizos H. Spatial and Temporal Changes in PD-L1 Expression in Cancer: The Role of Genetic Drivers, Tumor Microenvironment and Resistance to Therapy. Int J Mol Sci. 2020;21(19). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Uppaluri R, Lee NY, Westra W, et al. KEYNOTE-689: Phase 3 study of adjuvant and neoadjuvant pembrolizumab combined with standard of care (SOC) in patients with resectable, locally advanced head and neck squamous cell carcinoma. Journal of Clinical Oncology. 2019;37(15_suppl):TPS6090–TPS6090. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental Figure 1 - Concordance between PD-L1 and HLA class I transcript counts and protein level expression
Supplemental Table I. Multispectral immunofluorescence reagents.
Supplemental Table II. Raw Multispectral immunofluorescence data.