Introduction
Immunotherapy has grown to a rapidly advancing sphere of research on modern strategies for the treatment of cancer (Fridman
2017). A number of patients have already benefited from immune checkpoint blockades, and various drugs are currently under evaluation in clinical trials (Zhao and Subramanian
2018). In oesophageal adenocarcinoma (EAC)—a rapidly increasing cancer entity with a worse prognosis—surgery, chemotherapy and radiation remain the basis of treatment of EAC (Gore
2005; Grierson et al.
2017). The role of evolving immunotherapy has yet to be examined for EAC.
Immune checkpoints are a well-known form of immunomodulation, leading to a down-regulated immune response in the tumour microenvironment. Several of these checkpoints have been detected, such as programmed death cell protein 1 (PD-1), which is common and has occurred during the treatment of melanoma, non-small-cell lung cancer (NSCLC), renal cell carcinoma and urothelial carcinoma (Andrews
2017; Roberts
2017; Ma
2017). In addition to PD-1, lymphocyte activation protein-3 (LAG3) presents a targetable checkpoint, as can contribute to therapy strategies, including treatment of EAC.
LAG3 belongs to the immunoglobulin superfamily (IgSF) and is particularly displayed on several forms of T-lymphocytes (CD4+, CD8+, regulatory T-cells [Treg], tumour infiltrating lymphocytes [TILs]), as well as B-lymphocytes and dendritic cells (He
2016). LAG3 shares approximately 20% identity with the CD4 gene, and it binds to major histocompatibility complex 2 (MHC II) with greater affinity than CD4 (He
2016). The LAG3/MHC II complex on CD4+ cells negatively modulates T-cell activity and enhances antigen self-tolerance when displayed on CD8+ cells. Conversely, LAG3 binding to MHC II on Treg cells advances the suppressive effect on T-lymphocytes, enforcing the negative immune regulation effect of LAG3 (Andrews
2017). Studies have suggested that LAG3 is a negative regulator of T-cell activation and function, since the blockade of LAG3 on human CD4 clones resulted in enhanced proliferation, with an elevated production of IL-2, IL-4, IFN-γ and TNFα (Previte, et al.
2019; Goldberg and Drake
2011).
Analysis of LAG3 overexpression on TILs has revealed evidence for a pathological role, which involves down-regulating the immune response for various cancer entities, such as chronic lymphatic leukaemia, colorectal cancer, ovarian cancer, melanoma and hepatocellular carcinoma, leading to a worse prognosis in LAG3 positive malignancies (Shapiro
2017; Li
2013; Hemon
2011; Huang
2015; Chen and Chen
2014). Interestingly, and in contrast to these findings, recent studies in breast cancer patients showed a favourable outcome in LAG3 positive tumours regarding the overall survival (OS) of the patients, while other studies showed a worse prognosis in breast cancer (Sidaway
2017; Burugu
2017).
The aim of the present study is to assess the expression of LAG3 on TILs at the protein level, as well as the mRNA level, in EAC and correlate the expression profile with clinico-pathological and molecular data and the prognosis of individual patients.
Material and methods
Patients and tumour samples
The formalin-fixed and paraffin embedded samples from 477 patients with EACs, who underwent primary surgical resection or resection after neoadjuvant therapy between 1999–2014 at the Department of General, Visceral and Cancer Surgery, University of Cologne, Germany, were analysed. The standard surgical procedures were laparotomic or laparoscopic gastrolysis and right transthoracic en-bloc oesophagectomy, with intrathoracic oesophagogastrostomy, including two-field lymphadenectomy of mediastinal and abdominal lymph nodes, transhiatal extended distal oesophagectomy with transabdominal intrathoracic or cervical anastomosis as described previously (Holscher
2007). Patients with advanced oesophageal cancer (cT3), or lymph node metastasis in clinical staging, received preoperative chemoradiation (5-Fluouracil, cisplatin, 40 Gy) or chemotherapy. Follow-up data were available for all patients. Patient characteristics are given in Table
1. Depending on the effect of neoadjuvant chemo- or radiochemotherapy, there was a preponderance of minor responders, defined as having a histopathological residual tumour of ≥ 10% (Schneider
2008).
Table 1
Baseline characteristics of the entire patient cohort
Sex | | |
Female | 42 | 10.0 |
Male | 379 | 90.0 |
Age group | | |
≤ 65 | 221 | 52.4 |
> 65 | 200 | 47.6 |
Tumour stage | | |
pT 1 | 50 | 11.9 |
pT 2 | 36 | 8.6 |
pT 3 | 322 | 76.8 |
pT 4 | 11 | 2.6 |
Lymph node status | | |
pN 0 | 161 | 38.3 |
pN + | 259 | 61.7 |
According to the suggestions of the international immuno-oncology working group for assessing TILs on solid tumours, we constructed a multi-spot tissue microarray (TMA) (Simon et al.
2005; Helbig
2016; Hendry et al.
2017a,
b). Construction of the multi-spot TMA and immunohistochemical staining procedures were performed as previously described (Simon et al.
2005; Helbig
2016). In brief, tissue cylinders, with a diameter of 1.2 mm each, were punched from selected tumour tissue blocks using a self-constructed, semi-automated precision instrument and embedded in empty recipient paraffin blocks. For the multi-spot TMA (165 patients), up to 8 tumour spots were punched out of the tumour, 4 spots each from the surface and the invasion front. The 165 patients evaluated using the multi-spot TMA were used as a test cohort. These data were statistically correlated with survival and molecular data, such as
TP53 mutational status.
We analysed 312 additional patients using a single-spot TMA. For this TMA, one tissue core from each tumour was randomly punched out and transferred into a TMA recipient block. 4 μm sections of the resulting TMA blocks were transferred to an adhesive coated slide system (Instrumedics Inc., Hackensack, NJ) for immunohistochemistry (IHC). All procedures performed for studies involving human participants were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration, and its later amendments, or comparable ethical standards.
Immunohistochemistry
IHC was performed on the TMA slides. The following antibodies were used for IHC studies: a rabbit IgG monoclonal antibody (clone D2G40; dilution 1:300; Cell Signaling Technology, Netherlands) was used for LAG3, a rabbit monoclonal antibody (clone SP7; dilution 1:50; Thermo Fisher Scientific, USA) was used for CD3 and a mouse monoclonal antibody (clone C8/144B, dilution 1:200; Dako/Agilent, USA) was used for CD8. All immunohistochemical staining was performed using the Leica BOND-MAX stainer (Leica Biosystems, Germany) according to the protocol of the manufacturer. The evaluation of immunohistochemical expression was assessed manually by two pathologists (AQ and HL). Discrepancies in the results, which occurred only in a small number of samples, were resolved by consensus review.
Multicolour immunohistochemical stainings were performed on a Ventana Discovery Ultra automatic staining system (Ventana/Roche, Basel, Switzerland) using following primary antibodies: rabbit anti-LAG3 IgG monoclonal antibody D2G40, mouse anti-CD8 monoclonal antibody C8/144B, mouse anti-FOXP3 monoclonal antibody 236A/E7 (Abcam, UK; dilution 1:100), rabbit anti-CD4 monoclonal antibody 4B12 (Roche, Switzerland, ready to use). After conjugation with an antibody-bound enzyme (horseradish peroxidase or alcalic phosphatase), detection was carried out using DISCOVERY Silver kit (LAG3), DISCOVERY Yellow kit (FOXP3), DISCOVERY Teal kit (CD8), DISCOVERY Red Kit (CD4; all Ventana/Roche, Switzerland)). Counterstaining was done with hematoxylin and bluing reagent.
Strategy of evaluation
LAG3: < 1% of lymphocytes was defined as negative, 1–2% of lymphocytes were assessed as “LAG3 low”, > 2% of lymphocytes was counted as “LAG3 high”. The reading strategy followed the assessment of LAG3 in clinical trials in malignant melanoma, where the response rates of the LAG3 blockade correlated with LAG3 expression of ≥ 1% (Ascierto and McArthur
2017). For statistical analysis, the cut off was determined as ≥ 1%, thus low and high LAG3 expression was assessed as positive and < 1% expression as negative.
CD3: CD3 expression in < 3 lymphocytes/mm2 was evaluated as negative, > 3–50 lymphocytes/mm2 were assessed as low positive and > 50 lymphocytes/ mm2 were defined as high positive, considering peritumoral and intratumoral distribution.
CD8: CD8 was analysed according to the CD3 evaluation criteria. For statistical analysis, high expression of CD3 or CD8 with > 50 lymphocytes/mm2 were assessed as positive.
Regarding the multi-spot TMA considering eight tumor spots in total, four spots each of the tumour surface and the invasive margin, were examined. We calculated the average of the scores and matched the four samples to one category based on limit values: 0, negative; 0–0.9, low; 1–2, high (e.g. LAG3 expression in spot 1 = 2, spot 2 = 1, spot 3 = 0, spot 4 = 2, average of the spots: 1.25 → category “high”). Discrepancies in the results were resolved by consensus review.
Immunofluorescence multi-colour staining
Immunofluorescence staining was performed on TMAs and whole section slides. Therefore, paraffin sections were deparaffinised and antigens were retrieved with EDTA at pH 8 (PT Module, Lab Vision Thermo Scientific). Slides were blocked using normal horse serum, for 30 min at room temperature (Vector Laboratories). Slides were incubated overnight at 4 °C with a master mix containing the primary antibodies (LAG3, 1:75, Cell Signaling; CD4, mouse monoclonal 4B12, 1:75, Thermo Fisher Scientific; CD3, rat monoclonal CD3-12, 1:50, Abcam; CD8, 1:100, Dako/Agilent). Slides were washed and stained with a master mix containing the corresponding secondary antibodies coupled to Alexa Fluor 555 (donkey anti-rabbit, Abcam), Alexa Fluor 594 (donkey anti-rat, Jackson Laboratories) and Alexa Fluor 647 (donkey anti-mouse, Jackson Laboratories) for 1 h at room temperature. Nuclei were visualised with DAPI (Sigma-Aldrich). Slides were mounted using an antifade solution (ProLong Diamond, Invitrogen) and scanned with a 40× objective (gSTED super-resolution confocal microscope, Leica). Images were adjusted for brightness and contrast using ImageJ (FIJI).
mRNA in-situ (RNAScope)
The RNAscope assay was performed according to manufacturer’s instructions (Bott et al.
2011). In brief, paraffin-embedded TMA blocks were cut into 5 μm sections, pre-treated according to an extended protocol (30 min for pre-treatment 2 and 3), digested and hybridised at 40 °C in the HybEZ oven with the human LAG3 mRNA probe, which was provided by Advanced Cell Diagnostics Europe. The samples were then incubation with haematoxylin for 10 s. Target expression was compared to both negative (dapB) and positive (PPIB) controls. The scoring of the signals was performed as recommend by the manufacturer, where no staining or less than one molecule per 10 cells, score 0; 1–3 dots/cell, score 1; 4–9 dots/cell, score 2; 10–15 dots/cell, score 3; > 15 dots/cell, score 4. The DapB score was 0 and the PPIB score was 2. Positivity was defined as a score > 0. The determination of protein expression using immunohistochemistry, as well as mRNA expression, for LAG3 was assessed independently.
Analysis of TP53 mutation status
The TP53 status was evaluated by immunohistochemistry on the single-spot TMA. The results were correlated with the
TP53-mutational status by parallel genomic sequencing. A detailed description of the analysis was described previously (Quaas
2019). In brief, tumor DNA extraction was followed by amplification with a customized GeneRead DNAseq Targeted Panel V2 (Qiagen, Hilden, Germany), library construction and quantification. Exons 5-8 of the
TP53 gene were sequenced on the MiSeq (illumina, Berlin, Germany) with a variance-cutoff of 5%. The results were only interpreted if the coverage was > 200×.
Statistical analysis
Clinical data were collected prospectively according to a standardised protocol. SPSS Statistics for Mac (Version 21, SPSS) was used for statistical analysis. Interdependence between staining and clinical data were calculated using the chi-squared and Fisher’s exact tests and displayed by cross-tables. Survival curves were plotted using the Kaplan–Meier method and analysed using the log-rank test. Univariate and multivariate analyses were performed for prognostic factors of overall survival using the Cox regression model. All tests were two-sided. p values < 0.05 were considered statistically significant.
Discussion
In a large set of 421 patients with EAC, we report the impact of the checkpoint inhibitor LAG3 considering the protein and mRNA expression, as well as the distribution pattern within the tumour, in correlation with clinical and molecular data. In our cohort, LAG3 was positively correlated with subset of T-cells (predominant CD8 positive T-cells). Additionally, elevated LAG3 expression was linked to a significantly better outcome for patients in the advanced tumour stages subgroup.
The construction of the multi-spot TMA for the analysis of tumour heterogeneity and associated TILs followed the recommendations of the International Immuno-Oncology Biomarkers Working Group (Hendry et al.
2017a,
b). We started our analysis with a test-cohort of 165 patients using the multi-spot TMA, followed by a much larger cohort considering additional patients to reassess the results found before. The multi-spot TMA demonstrated a heterogenic LAG3 distribution within the eight tumour spots; however, by comparing the tumour surface with the infiltration margin, we found a consistent expression pattern of LAG3. Thus, in the case of endoscopically taken biopsies LAG3 expression in EAC can serve as a reliable predictor regarding the overall LAG3 expression within the tumour.
In contrast to our initial hypothesis, we are able to show a favourable outcome of patients with elevated LAG3 in TILs. However, most probably due to a correlation with pT-stage and lymph node metastasis, the LAG3 expression failed to serve as an independent prognostic marker, since the survival benefit correlated with LAG3 expression was predominantly detectable in pT3/4 tumours. This is fully in line with recent studies on breast cancer patients but in contrast to previous results in other solid tumour entities, such as melanoma and hepatocellular carcinoma (Li
2013; Hemon
2011; Sidaway
2017). However, divergent descriptions concerning the prognostic impact of a single immune checkpoint between different tumour entities are well known and likewise given for PD-1/PD-L1 (Bertucci
2015,
2017). This phenomenon can be explained by various analysis techniques and antibodies or by different organic systems. Due to technical variabilities, we analysed the mRNA expression profile of LAG3 for 77 patients and compared it to the protein expression detected using IHC in a double-blinded examination. Since there was a distinct correlation between the mRNA and protein expression profiles of LAG3, we confirm the viability of our antibody and analysis techniques.
The favourable outcome in context with LAG3 expression appears to be counterintuitive, initially, considering the suppressive effect of this immune checkpoint. Furthermore, our distinct results require exploratory approaches to clarify the role of immunomodulation in EAC. We, therefore, developed hypotheses to explain how LAG3 overexpression could be related to a better outcome.
An elevated expression may be part of a powerful immune reaction leading to an effective antitumoral response. Thus, we assume that pro-inflammatory signalling pathways against the tumour could be compensatory regulation by the suppressive effect of LAG3. We found a strong correlation between the existence of T-cells (CD3) and LAG3, supposing that an inflammatory microenvironment is attended by increased LAG3. Since tumours with elevated T-cells showed a significantly better outcome in our cohort, we underline the importance of an inflammatory antitumoral response, which is in accordance with former studies on EAC (Noble
2016). We therefore presume that LAG3 serves as a biomarker for a strong immune response. The initially assumed inhibitory immunomodulatory effect does not seem to be reflected by our current analysis to the extent expected. As described above, similar effects have been demonstrated for PD-L1 expression in different tumour entities. Whether a pharmacological LAG3 inhibition will prove to be effective in EAC cannot be conclusively answered on the basis of the available data. In our opinion, however, a positive effect of LAG3 inhibitor therapy can be assumed. However, further functional investigations with regard to LAG3 expression and interactions with tumour cells are necessary.
Taken together, we presume that LAG3 serves as a surrogate parameter in immunogenic tumour biology; although, we must admit that the limited knowledge of downstream mechanisms and the interaction with other intracellular pathways only allows us to hypothesise about a possible function of the protein in EAC. In addition, it remains obscure why only advanced tumour stages revealed a prognostic impact of LAG3 in contrast to early stages, underlining the urgent need for further research on immunomodulation.
Nevertheless, our assumptions support the use of LAG3 inhibition for EAC, since the loss of regulatory survival mechanisms would lead to a stronger immune response via CD3/8. Therapeutic studies investigating the efficacy of LAG3 inhibition in EAC will also provide predictive evidence on the determination of the LAG3 expression in the tumour microenvironment, as we have learned in the past for the determination of PD-L1 in lung cancer.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.