Density Of Tumor-infiltrating Lymphocytes Research Articles

Heterogenous profiles between primary lung cancers and paired brain metastases reveal tumor evolution.

Brain metastases (BMs) are the most common central nervous system (CNS) malignant tumors, with rapid disease progression and extremely poor prognosis. The heterogeneity between primary lung cancers and BMs leads to the divergent efficacy of the adjuvant therapy response to primary tumors and BMs. However, the extent of heterogeneity between primary lung cancers and BMs, and the evolutionary process remains little known. To deeply insight into the extent of inter-tumor heterogeneity at a single-patient level and the process of these evolutions, we retrospectively analyzed a total of 26 tumor samples from 10 patients with matched primary lung cancers and BMs. One patient underwent four times brain metastatic lesion surgery with diverse locations and one operation for the primary lesion. The genomic and immune heterogeneity between primary lung cancers and BMs were evaluated by utilizing whole-exome sequencing (WESeq) and immunohistochemical analysis. In addition to inheriting genomic phenotype and molecular phenotype from the primary lung cancers, massive unique genomic phenotype and molecular phenotype were also observed in BMs, which revealed unimaginable complexity of tumor evolution and extensive heterogeneity among lesions at a single-patient level. By analysis of a multi-metastases case (Case 3) of cancer cells' subclonal composition, we found similar multiple subclonal clusters in the four spatial and temporal isolated brain metastatic focus, with the characteristics of polyclonal dissemination. Our study also verified that the expression level of immune checkpoints-related molecule Programmed Death-Ligand 1 (PD-L1) (P = 0.0002) and the density of tumor-infiltrating lymphocytes (TILs) (P = 0.0248) in BMs were significantly lower than that in paired primary lung cancers. Additionally, tumor microvascular density (MVD) also differed between primary tumors and paired BMs, indicating that temporal and spatial diversity profoundly contributes to the evolution of BMs heterogeneity. Our study revealed the significance of temporal and spatial factors to the evolution of tumor heterogeneity by multi-dimensional analysis of matched primary lung cancers and BMs, which also provided novel insight for formulating individualized treatment strategies for BMs.

Papillary thyroid cancer immune phenotypes via tumor-infiltrating lymphocyte spatial analysis.

Standard-of-care treatment options provide an excellent prognosis for papillary thyroid cancers (PTCs); however, approximately 10% of cases are advanced PTCs, resulting in less than 50% 5-year survival rates. Understanding the tumor microenvironment is essential for understanding cancer progression and investigating potential biomarkers for treatment, such as immunotherapy. Our study focused on tumor-infiltrating lymphocytes (TILs), which are the main effectors of antitumor immunity and related to the mechanism of immunotherapy. Using an artificial intelligence model, we analyzed the density of intratumoral and peritumoral TILs in the pathologic slides of The Cancer Genome Atlas PTC cohort. Tumors were classified into three immune phenotypes (IPs) based on the spatial distribution of TILs: immune-desert (48%), immune-excluded (34%), and inflamed (18%). Immune-desert IP was mostly characterized by RAS mutations, high thyroid differentiation score, and low antitumor immune response. Immune-excluded IP predominantly consisted of BRAF V600E-mutated tumors and had a higher rate of lymph node metastasis. Inflamed IP was characterized by a high antitumor immune response, as demonstrated by a high cytolytic score, immune-related cell infiltrations, expression of immunomodulatory molecules (including immunotherapy target molecules), and enrichment of immune-related pathways. This study is the first to investigate IP classification using TILs in PTC through a tissue-based approach. Each IP had unique immune and genomic profiles. Further studies are warranted to assess the predictive value of IP classification in advanced PTC patients treated with immunotherapy.

Association between radiomics features of DCE-MRI and CD8+ and CD4+ TILs in advanced gastric cancer.

Objective: The aim of this investigation was to explore the correlation between the levels of tumor-infiltrating CD8+ and CD4+ T cells and the quantitative pharmacokinetic parameters of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in patients with advanced gastric cancer. Methods: We retrospectively analyzed the data of 103 patients with histopathologically confirmed advanced gastric cancer (AGC). Three pharmacokinetic parameters, Kep, Ktrans, and Ve, and their radiomics characteristics were obtained by Omni Kinetics software. Immunohistochemical staining was used to determine CD4+ and CD8+ TILs. Statistical analysis was subsequently performed to assess the correlation between radiomics characteristics and CD4+ and CD8+ TIL density. Results: All patients included in this study were finally divided into either a CD8+ TILs low-density group (n = 51) (CD8+ TILs < 138) or a high-density group (n = 52) (CD8+ TILs ≥ 138), and a CD4+ TILs low-density group (n = 51) (CD4+ TILs < 87) or a high-density group (n = 52) (CD4+ TILs ≥ 87). ClusterShade and Skewness based on Kep and Skewness based on Ktrans both showed moderate negative correlation with CD8+ TIL levels (r = 0.630-0.349, p < 0.001), with ClusterShade based on Kep having the highest negative correlation (r = -0.630, p < 0.001). Inertia-based Kep showed a moderate positive correlation with the CD4+ TIL level (r = 0.549, p < 0.001), and the Correlation based on Kep showed a moderate negative correlation with the CD4+ TIL level, which also had the highest correlation coefficient (r = -0.616, p < 0.001). The diagnostic efficacy of the above features was assessed by ROC curves. For CD8+ TILs, ClusterShade of Kep had the highest mean area under the curve (AUC) (0.863). For CD4+ TILs, the Correlation of Kep had the highest mean AUC (0.856). Conclusion: The radiomics features of DCE-MRI are associated with the expression of tumor-infiltrating CD8+ and CD4+ T cells in AGC, which have the potential to noninvasively evaluate the expression of CD8+ and CD4+ TILs in AGC patients.

Predictive value of tumor-infiltrating lymphocyte (TIL) dynamics in the tumor microenvironment (TME) during preoperative chemoradiotherapy (CRT) on pathologic complete response (pCR) in microsatellite-stable (MSS) locally advanced rectal cancer (LARC).

3608 Background: Preoperative CRT followed by consolidation nivolumab before surgery showed a promising pCR rate in MSS LARC; both high PD-L1 expression and an elevated CD8+ T cell/effector regulatory T cell ratio analyzed by pre-CRT samples were independently related to high pCR rate (Bando H, Clin Cancer Res. 2022). Here, we applied AI-powered spatial TIL analysis to the VOLTAGE study to investigate whether dynamic change of TIL in TME may predict pCR in MSS LARC. Methods: The VOLTAGE study is a multicenter phase I/II study to evaluate the efficacy of CRT followed by 5 cycles of nivolumab and surgery in patients (pts) with LARC. In this study, tumor samples were obtained at multiple time points; pre-CRT (B1), post-CRT/pre-nivolumab (B2), post-3 cycles of nivolumab (B3), and surgery. In this analysis, a total of 86 H&amp;E whole-slide images (WSI) harvested at B1 and B2 time points from all the pts enrolled in the VOLTAGE study were included in the analysis. Lunit SCOPE IO (Lunit, Republic of Korea), AI-powered H&amp;E-WSI analyzer developed based on 16,443 WSI with pathologists' annotations, was applied to quantify TIL density in TME. Results: In a total of 43 pts, pCR rate of MSS and microsatellite instability-high (MSI-H) were 28.9% (11/38) and 60% (3/5), respectively. In MSS subgroup (n = 38), which has a higher unmet need for pCR prediction, tumor samples with pCR (n = 11) had increased TIL density in TME (tTIL) in post-CRT (B2, median [IQR] 726 [249-1607] /mm2) compared to pre-CRT (B1, 598 [366-905] /mm2), whereas those without pCR (n = 27) had decreased tTIL in post-CRT (405 [148-748] /mm2), compared to pre-CRT (B1, 748 [460-1153] /mm2). Intratumoral TIL density changes were more prominent than stromal TIL density changes in MSS tumor samples with pCR (mean fold change x17.7 vs. x1.5). MSS Pts with tTIL change from B1 to B2 (during CRT) more than x1.8 times achieved 75% (6/8) pCR rate, whereas those with tTIL change less than x1.8 had 16.7% (5/30) pCR rate (p = 0.0035). Conclusions: Change of TIL density in TME during CRT was significantly correlated with favorable clinical outcome of preoperative CRT and consolidation nivolumab, resulting in high pCR rate in MSS LARC.

Machine learning (ML)-based quantification of tumor-infiltrating lymphocytes (TIL) and clinical outcomes of patients with melanoma treated with immune-checkpoint inhibitors (ICI).

9560 Background: TIL quantification has shown promising prognostic and predictive impact in various tumors treated with ICI. More recently, TIL therapy has become an emerging treatment agent for ICI-refractory melanoma. In this work, we studied the effect of intrinsic TILs on clinical outcomes of patients with melanoma treated with ICI and quantified its utility as a biomarker in combination with tumor mutational burden (TMB). Methods: We applied a previously developed ML model to process digital whole-slide images of hematoxylin and eosin slides to quantify TIL density (TIL/mm2) in 182 metastatic melanoma samples treated with ICI at the Dana-Farber Cancer Institute. All samples underwent next-generation targeted-panel sequencing and had available TMB data (mutations/megabase). Overall survival (OS) was calculated from the data of ICI initiation to the date of death. TTF was calculated from the data of ICI initiation to the date of next line treatment or death. Alive patients were censored at the date of last follow-up. Multivariable cox regression was used to estimate adjusted hazard ratios (adj-HR) and p-values (adj-p), with TMB, prior lines of treatment and regimen type (single vs. combination ICI) used as covariates. TMB-high (TMB-H) group was defined as samples with TMB≥10. TIL-high (TIL-H) group was defined as samples with TIL density higher than the median. Results: Of 182 patients, 60% (110/182) were male and the median age at ICI start was 65.5 years. 63% (115/182) of patients received single-agent ICI (anti-PD1 or anti-CTLA4). Of 182 tumor specimens assessed for TILs, 47 (25.8%) were obtained from the primary site. There was no difference in TIL density (medians 583 vs. 817 TILs/mm2, p=0.57; 709 TIL/mm2 for full cohort) or TMB (medians 11.4 vs 12.2 mutations/megabase) between primary and metastatic samples, respectively. Moreover, there was no correlation between TIL density and TMB (pearson’s correlation coefficient = 0.92, p=1) across all samples. Compared to patients in the TIL-low (TIL-L) category, patients in TIL-H had significantly longer median TTF (33.9 [15.9-NR] versus 13.0 [7.8-18.6] months, adj-HR: 0.53 (95% CI: 0.36-0.78); adj-P=0.0012), while there was no difference in OS (P=0.57). In fact, patients in the TMB-H/TIL-H group had the longest TTF, followed by TMB-H/TIL-L, TMB-L/TIL-H then TMB-L/TIL-L (table). Conclusions: Our results support the potential use of ML-based TIL scoring as a novel and independent biomarker to predict time to failure on ICI. There is ongoing effort to validate our findings in an external dataset. [Table: see text]

Artificial intelligence–powered tumor-infiltrating lymphocytes analyzer to reveal distinct immune landscapes in breast cancer by molecular subtype and HER2 score.

1049 Background: Tumor-infiltrating lymphocytes (TIL) reflect an ongoing anti-tumor immune response in the tumor microenvironment (TME) and have been studied as a predictive as well as prognostic biomarker in breast cancer (BC). The immune profile differs depending on BC molecular subtypes, with triple-negative BC (TNBC) and HER2-positive BC more frequently infiltrated by higher numbers of TIL than hormone receptor (HR)-positive tumors. In this study, we analyze TIL and immune phenotype by molecular subtype in BC and explore differences based on HER2 scores using an AI-powered TIL analyzer. Methods: A total of 1,973 cases of BC with molecular subtype, HER2 immunohistochemistry (IHC) or in situ hybridization (ISH) results, and TIL analysis from H&amp;E slides from TCGA (n=1021), Samsung Medical Center (SMC, n=663), and KyungHee Medical Center (KHMC, n=289) were included in the analysis. The density of TIL were spatially analyzed from H&amp;E WSI, and the immune phenotype (IP) was analyzed for each 1 mm2-sized patch. Tumor microenvironment (TME) was categorized into three types: inflamed (TIL distributed intratumorally), Immune-excluded (TIL excluded, out of cancer stroma), immune-desert (scant TIL in TME), and the score for each IP was calculated as the percentage of the total patches in WSI. Results: TNBC showed the highest TIL density in both cancer epithelium (214.5 ± 299.7/mm2) and stroma (1133.3 ± 1245.9/mm2), and a high percentage of inflamed IP (35.7%), IS (28.4 ± 28.1) than other subtypes (inflamed IP: 17.5%, IS 16.2±20.2, both p&lt;0.001). HER2-positive BCs showed higher sTIL density (904.0 ± 901.7/mm2) compared to HR- positive BC (616.7 ± 794.1/mm2) but were associated with a higher percentage of immune-excluded IP (71.2%) and IES (57.6 ± 24.6). HR-positive BC had a high percentage of immune-desert IP (30.4%). In both HR-positive and negative BCs, higher IES was observed as the HER2 score increased. HER2-positive BC significantly showed a higher proportion of cancer stroma in the cancer region from WSI (Stromal ratio, 0.682 ± 0.149 vs. 0.578 ± 0.176, p&lt;0.001). Conclusions: Different TIL densities and IP were seen in BC for each molecular subtype, pointing to a distinct immune landscape. HER2 score was associated with immune-excluded IP despite having higher TIL density, which should be considered for immunotherapy in BC.

Artificial intelligence (AI) –powered spatial analysis of tumor-infiltrating lymphocytes (TILs) for prediction of prognosis in resectable pancreatic adenocarcinoma (PDAC).

4162 Background: The degree of TIL infiltration in the tumor microenvironment has been suggested as a prognostic factor in multiple cancer types, but the significance in PDAC is not well known. The aim of this study is to assess the prognostic role of AI-based spatial TIL density assessment and immune phenotype (IP) classification in resectable PDAC. Methods: We collected hematoxylin and eosin (H&amp;E)-stained whole slide images (WSI) of tumor tissues and clinical data from stage I – III PDAC patients who received curative surgery at Samsung Medical Center, in Seoul, Korea, between Jan 2017 and Dec 2018. For spatial TIL analysis, we used Lunit SCOPE IO, an AI-powered pathology slide analyzer that can identify TIL and segment tumor epithelium and stroma from H&amp;E-stained WSI. It then estimates intratumoral (iTIL) and stromal TIL (sTIL) densities, as well as IP classification of each case. IPs were defined as follows: inflamed IP (IIP) as high iTIL and sTIL; immune-excluded as low iTIL and high sTIL; immune-desert as low TIL overall. Predetermined cutoffs prior to this study were used for the determination of IPs. Results: A total of 209 patients were identified, 182 treated with upfront surgery and 27 with preoperative chemo(radio)therapies followed by surgery. The median recurrence-free survival (RFS) and overall survival (OS) of all patients was 16.8 months (95% CI 12.1 - 21.4), and 36.8 months (95% CI 31.2 - 42.4), respectively. In the 182 patients treated with upfront surgery, the median iTIL and sTIL densities were 85.3/mm2 (IQR 56.9 – 133.0) and 672.2/mm2 (IQR 445.9 – 912.9). 85.7% of the cases were immune-excluded, while 11.5% were IIP. The patients having iTIL higher than the median exhibited longer RFS (HR 0.62, 95% CI 0.42 - 0.92, p = 0.017). Also, the patients having IIP exhibited longer OS (p = 0.045, HR 2.05, 95% CI 1.00 – 4.22) and a marginally longer RFS (p = 0.051, HR 2.03, 95% CI 0.98 – 4.18) compared to those with non-IIP. In the 27 patients who were treated with preoperative therapies, the iTIL densities in the surgical specimens were significantly higher compared to the 182 untreated samples (mean iTIL 168.8/mm2 after chemo(radio)therapies vs. 112.6/mm2 in upfront surgery, p = 0.012), although the increase in the proportion of IIP was not significant. Using the iTIL median of the 182 upfront surgery cases, 17/27(63.0%) patients had high iTIL in WSIs obtained after preoperative therapies, and these patients were observed to have a significantly better RFS (HR 0.26. 95% CI 0.09 - 0.73, p = 0.006). Conclusions: While most cases of resectable PDAC were found to have TIL infiltration mainly in stroma, relatively higher iTIL infiltration was associated with better prognosis. An increase in iTIL after chemo(radio)therapies may be associated with better outcomes.

Artificial intelligence (AI) –powered H&amp;E whole-slide image (WSI) analysis of tertiary lymphoid structure (TLS) correlates with immune phenotype and related molecular signatures in non–small-cell lung cancer.

e20520 Background: Recent data suggested immune infiltrates in TLS around tumor bed may be a favorable predictive factor for immunotherapy in various cancers. TLS have been suggested to be correlated with inflamed immune phenotype and relevant inflammatory gene signatures of adaptive immunity. To analyze this relationship, here, we developed an AI model to assess TLS objectively in H&amp;E WSI, and assessed its correlation with immune phenotype and immunologic signatures. Methods: H&amp;E images, relevant gene expression profiles and clinical data from The Cancer Genome Atlas (TCGA) lung cancer dataset (LUAD and LUSC, N = 913) were used for the analysis. Lunit SCOPE TLS, an AI-powered H&amp;E WSI analyzer, was developed with 3.59 x 109 μm2 annotated area and 1,439 H&amp;E stained WSI of 18 cancer types to segment TLS within tumor microenvironment (TME). Based on spatial tumor-infiltrating lymphocyte (TIL) density, IPs were classified into inflamed IP (IIP) as high intratumoral TIL (iTIL), immune-excluded IP (IEP) as low iTIL and high sTIL, and immune-desert IP (IDP) as low TIL overall. The infiltration of immune cells, the activity of related pathways and biological processes, and the signature scores of interest gene sets were analyzed by using the CIBERSORT, DESeq2 and GSEA tools. Results: Of 913 samples, 69.7% (636/913) contain TLS. The median value of the proportion of TLS area within TME was 0.12%, which was applied for the cut point of TLS-high vs -low group. The proportion of TLS-high was not significantly different according to EGFR mutation (mutation vs wild type: 45.5% vs 50.2%), KRAS mutation (49.3% vs 50%), but numerically decreased in the merging set of other driver mutations including ALK, ROS1, RET, MET, NTRK1-3 (36% vs 50.3%, p = 0.223). Interestingly, TLS-high proportion was significantly different according to immune phenotype, as TLS were present in 57.6% of IIP (186/323), 49.8% of IEP (241/484), and 28.3% of IDP (30/106, p &lt; 0.001). TLS-high group was positively correlated with memory B cells (fold change [fc] 1.93, p &lt; 0.001), CD8+ T cells (fc 1.20, p &lt; 0.001), and M1 Macrophages (fc 1.17, p &lt; 0.001), and negatively correlated with neutrophils (fc 0.59, p &lt; 0.001) and M2 macrophages (fc 0.91, p = 0.006). This result was additionally supported by GSEA analysis of GO:BP or Hallmark gene sets which showed TLS-high was associated with B cell receptor signaling pathway (normalized enrichment score [NES] 3.16, p = 0.001), immunoglobulin production (NES 3.07, p = 0.002), and interferon-gamma response (NES 2.20, p = 0.002). Epithelial-mesenchymal transition was negatively associated with TLS-high (NES -1.54, p &lt; 0.001). Conclusions: TLS is associated with inflamed immune phenotype and infiltration of memory B cells as well as CD8+ T cells in non-small cell lung cancer.

Neoadjuvant pembrolizumab, GX-188E, and GX-I7 in patients with human papillomavirus-16- and/or 18-positive head and neck squamous cell carcinoma: Single-arm, phase 2 trial with single cell transcriptomic analysis and artificial intelligence-powered spatial analysis.

6075 Background: With the increasing prevalence of head and neck squamous cell carcinoma (HNSCC) associated with human papilloma virus (HPV) infection, effective treatment strategies for HPV-positive HNSCC are urgently needed. Here, we present the results of neoadjuvant checkpoint blockade of pembrolizumab, therapeutic HPV DNA vaccine of GX-188E, and long-acting interleukin-7 of GX-I7 for patients with resectable HPV-16 and/or 18-positive HNSCC. Methods: In this single-arm, phase 2 trial, patients with resectable HPV-16 and/or 18-positive HNSCC were enrolled. Patients were given pembrolizumab 200mg on day 1 and day 22; GX-188E 2mg on day 1, 8, and 22; and GX-I7 1200 ug/kg on day 8 before surgical resection. Major pathologic response (MPR; defined as residual viable tumor of less than or equal to 10%) was primary endpoint with null and alternative hypothesis of MPR rate ≤5% and ≥35%, respectively. Secondary objectives were safety, recurrence, and survival. Results: Of the 11 patients included, all underwent surgical resection after neoadjuvant treatment without delays in surgery or increased surgical complications. MPR was achieved in seven patients (63.6%) and pathologic complete response was achieved in four patients (36.3%), which met the primary endpoint. Single cell RNA sequencing of baseline and post-treatment paired specimens revealed that the proportion of follicular helper T cell cluster, featured with CXCR5 and BCL6 expression, was significantly increased among tumor-infiltrating immune cells, accompanied with the reinvigoration of CD8 T cell cluster toward less exhausted phenotypes, represented with the upregulation of TCF7 and CD28 expression. In addition, artificial intelligence-powered spatial analysis revealed that triple combination significantly increased the density of tumor-infiltrating lymphocytes, completely converting immune-desert and immune-excluded type of tumor to inflamed immune phenotype. Conclusions: Neoadjuvant pembrolizumab, GX-188E, and GX-I7 showed promising activity and manageable safety profile in patients with resectable HPV-16 and/or 18-positive HNSCC. Therapeutic induction of brisk immune responses significantly reshaped the tumor microenvironment and associated with pathologic regression, warranting further investigation of pembrolizumab, GX-188E, and GX-I7 for patients with HPV-positive HNSCC. Clinical trial information: NCT05286060 .

Dynamic change of immune phenotype assessed by artificial intelligence (AI)-powered analysis of tumor-infiltrating lymphocytes (TILs) during neoadjuvant durvalumab with or without tremelimumab (D+/-T) in head and neck squamous cell carcinoma (HNSCC).

2578 Background: The results of a phase II trial evaluating preoperative durvalumab with or without tremelimumab (D+/-T) in resectable HNSCC (NCT03737968) suggest that the neoadjuvant immunotherapy is associated with pathologic tumor regression with manageable safety profiles. Here, we applied AI-powered spatial TIL analyzer, Lunit SCOPE IO, to tumor specimens of the patients included in the trial to assess the impact of spatial TIL density and inflamed immune phenotype (IIP) on immunotherapy responses. Methods: H&amp;E-stained whole-slide images (WSIs) of pre- and post-treatment tumor specimens from HNSCC pts enrolled in the aforementioned D+/-T trial were collected. Patients with locally advanced but operable HNSCC were randomly assigned in 1:1 to receive a single dose of D (1,500mg) or D+T (1,500mg+75mg) followed by surgery, at Severance Hospital in Seoul, Korea. Lunit SCOPE IO was used to segment tumor epithelium and stroma to identify and quantify intratumoral TIL (iTIL) and stromal TIL. IIP was defined as the proportion of high iTIL area more than 33.3% of analyzable area. Clinical data and the combined positive scores (CPS) of PD-L1 were also collected. Results: A total of 39 paired tumor samples of pre- and post-treatment WSI (23 from D+T and 16 from D) were included. Overall, tumor regression grade 2 (TRG2, ≥ 50% of tumor regression) was achieved in 15.4% (6/39) pts, and downstaging after treatment was achieved in 35.9% (14/39) pts. By pre-treatment WSIs, 38.5% (15/39) samples were classified as IIP, and subgroups of D+T and D had 34.8% (8/23) and 43.8% (7/16), respectively. In the D+T arm, the proportions of TRG2 were 25% (2/8) in IIP versus 13.3% (2/15) in non-IIP, whereas D arm had 28.6% (2/7) TRG2 in IIP, but no TRG2 in non-IIP (0/9), respectively. Comparing the WSI of pre- and post- D+/-T samples, the proportion of IIP was significantly increased to 66.7% from 38.5 ( p = 0.015). The increase in IIP proportion was prominent in the D+T arm (34.8% to 69.6%, p = 0.043), but not in the D arm (43.8% to 62.5%, p = 0.37). Similarly, PD-L1 CPS increased after therapy, especially in the D+T group (mean CPS 6.0 to 20.7, p = 0.009). Interestingly, pts who remained non-IIP in both pre- and post-treatment (25.6%, 10/39) had significantly poor prognosis compared to the others, as the 12-month disease-free survival (DFS) rate was 50.0% vs 93.1% (median DFS 13.4m vs not reached, hazard ratio [HR] 4.26, 95% confidence interval [CI] 1.32-13.8, p = 0.009). This prognostic impact was prominent in the D+T arm (HR 6.75, 95% CI 1.49-30.6, p = 0.004). Conclusions: Neoadjuvant durvalumab +/- tremelimumab promotes dynamic change toward inflamed immune phenotype, resulting in favorable clinical outcomes. Resistance mechanism of stand-still-non-inflamed patients even on D+/-T should be further investigated. Clinical trial information: NCT03737968 .

Relationship between tumor microenvironment (TME)-based histomic TGFβ signature (TGFBs), stromal fibroblast recruitment, and exclusion of immune cells as immunotherapy resistance mechanisms.

2585 Background: TGF-beta activates fibroblasts within the TME and may be associated with poor response to immunotherapy.However, objective and reliable assessment of the TGFBs by gene expression profiling (GEP) is challenging. Using H&amp;E whole-slide images (WSI) from The Cancer Genome Atlas (TCGA), we developed an artificial intelligence (AI) model to predict a normalized TGFBs, applying it to real-world datasets (RWDs) of advanced solid tumor patients treated with immunotherapy. Methods: The TGFBs was defined using the mean GEP values of the 'Hallmark TGF-beta signaling' gene set. H&amp;E WSIs of 23 carcinoma types from TCGA (n=6,945) were used to discover the association of the TGFBs with various cell types, including fibroblasts and tumor-infiltrating lymphocytes (TILs) detected by an AI model, Lunit SCOPE IO (AI-1). Another custom AI model (AI-2) to predict TGFBs was developed by integrating self-supervised and supervised features and other semantic content extracted by AI-1 with multilayered perceptron classifiers. TGFBs prediction scores from the AI-2 were applied to 5 different RWDs containing 1,792 immunotherapy-treated patients (&gt; 16 primary tumor types). The immune-excluded phenotype (IEP) was defined by high stromal TIL and low intratumoral TIL density. Results: In the TCGA dataset, TGFBs and fibroblast density detected by AI-1 were closely correlated (R=0.19, p&lt;0.001) compared to other cell types, with both highest in pancreatic adenocarcinoma (PAC)(median [interquartile range, IQR] 0.73 [0.66-0.80], 1306 [985-1588]/mm2, respectively) and CMS4 colorectal cancer (0.61 [0.54-0.67], 868 [582-1128]/mm2) compared to other cancers (0.57 [0.47-0.66], 577 [295-890]/mm2). The AI-2 predicted TGFBs with area-under-the-curve of 0.781 on cross-validation within the TCGA dataset. The TGFBs of immunotherapy-treated RWDs were inferred from their corresponding H&amp;E WSI, with scores comparable to those from the TCGA (0.42 [0.16-0.60] vs 0.58 [0.48-0.67]), with PAC having the highest TGFBs (0.65 [0.45-0.77]) in the RWD as well. Interestingly, TGFBs inferred scores and fibroblast density were significantly correlated (R=0.7, p&lt;0.001) in the RWD and also significantly correlated with the IEP (0.52 [0.30-0.66]) compared to other immune phenotypes (0.42 [0.16-0.60], p&lt;0.001). In the RWD, patients with inferred TGFBs in the upper 75% quartile had a poor clinical response to immunotherapy (objective response rate: 15.1% vs 20.7%, median progression-free survival 3.0 m vs 4.0 m, hazard ratio 1.21, 95% confidence interval 1.07-1.36, p=0.002). Conclusions: TGFBs is closely correlated with fibroblast density within the TME and may confer resistance to immunotherapy. Unique morphologic features of the TGFBs can be captured by AI, enabling scalable application to various datasets for clinical and translational research.

Artificial intelligence (AI) –powered spatial analysis of macrophages in tumor microenvironment and its association with interferon-gamma (IFNG) signature and immune phenotype (IP) in pan-cancer dataset.

2621 Background: Macrophages are known to play a crucial role in cancer progression, but few studies have reported macrophage spatial distribution and its association with genomic signatures and tumor-infiltrating lymphocytes (TIL). Here, we applied an AI-powered Whole Slide Image (WSI) analyzer in The Cancer Genome Atlas (TCGA) pan-cancer dataset to show correlation between specific localization of macrophage and anti-tumor activity, which is reflected by IFNG signature and IP based on spatial TIL density. Methods: For spatial macrophage analysis in TCGA pan-cancer dataset, we used Lunit SCOPE IO, an AI-powered H&amp;E WSI analyzer, which identifies and quantifies various classes of cells such as tumor cell, macrophage, and TIL within cancer or stroma area. IPs were defined as follows: inflamed IP (IIP) as high intratumoral TIL (iTIL) and stromal TIL (sTIL); immune-excluded IP (IEP) as low iTIL and high sTIL; immune-desert IP (IDP) as low TIL overall. The immune cell profiling and quantification of the IFNG signature score were analyzed by using CIBERSORT and RSEM. Results: In TCGA dataset (N = 7,402), median macrophage density in total TME area (tMP) was 19.9 /mm2 (interquartile range [IQR] 9.0-44.9), and it was highly enriched in lung adenocarcinoma (median 65.1, IQR 34.3-110.0), lung squamous cell carcinoma (61, 32.4-101.3), and head and neck squamous cell carcinoma (39.4, 19.6-77.3). Based on IP classification, tMP was highly enriched in IIP (37.8, 18.7-73.8) compared to IEP (25.5, 13.8-48.6) and IDP (8.9, 4.4-17.1). Interestingly, spatial analysis showed that intratumoral macrophage density (iMP) in IIP was highly enriched compared to the others (mean fold change 2.57) and the difference of stromal macrophage density (sMP) among IPs was relatively smaller (mean fold change x1.85). Classified into four groups based on median values of iMP and sMP, high-iMP/high-sMP group had the highest proportion of IIP (47.3% vs 18.7%, p &lt; 0.001), whereas low-iMP/high-sMP group had the highest one of IEP (54.9% vs 32.3%, p &lt; 0.001) compared to the others, respectively. Moreover, the high-iMP/high-sMP group had the highest level of IFNG signature score among 4 groups (x2.13 times higher, p &lt; 0.001). Notably, both iMP and sMP were significantly correlated with CD68 expression (rho 0.292 and 0.287) but CSF1R expression was more strongly correlated with iMP (rho 0.259) compared to sMP (rho 0.078). Conclusions: IIP, IFNG signature, and CSF1R expression are more strongly correlated with intratumoral macrophage density versus stromal macrophage density, suggesting a relationship between macrophage location and function in the TME warranting further investigation.

Patterns of Tumor-Infiltrating Lymphocytes Used to Distinguish Primary Cutaneous Squamous Cell Carcinomas That Metastasize

This case-control study examines the type, location, and density of tumor-infiltrating lymphocytes in adult patients with vs without metastatic cutaneous squamous cell carcinoma.

The Prognostic Role of the Immune Microenvironment in Sinonasal Intestinal-Type Adenocarcinoma: A Computer-Assisted Image Analysis of CD3+ and CD8+ Tumor-Infiltrating Lymphocytes.

The prognostic value of conventional histopathological parameters in the sinonasal intestinal-type adenocarcinoma (ITAC) has been debated and novel variables should be investigated. Increasing evidence demonstrated that the evolution of cancer is strongly dependent upon the complex interactions within tumor microenvironment. The aim of this retrospective study was to assess the features of immune microenvironment in terms of CD3+ and CD8+ cells in a series of ITAC and explore their prognostic role, as well as their relations with clinicopathological variables. A computer-assisted image analysis of CD3+ and CD8+ tumor-infiltrating lymphocytes (TIL) density was conducted on surgical specimens of 51 patients with ITAC that underwent a curative treatment including surgery. ITAC displays variable TIL density, which is associated with OS. In a univariate model, the density of CD3+ TIL was significantly related to OS (p = 0.012), whereas the association with CD8+ TIL density resulted in being non-significant (p = 0.056). Patients with intermediate CD3+ TIL density were associated with the best outcome, whereas 5-year OS was the lowest for intermediate CD8+ TIL density. CD3+ TIL density maintained a significant association with OS in the multivariable analysis. TIL density was not significantly related to demographic and clinicopathological variables. CD3+ TIL density was independently associated with OS in a non-linear fashion and patients with intermediate CD3+ TIL density had the best outcome. Though based on a preliminary analysis on a relatively small series of patients, this finding makes TIL density a potential independent prognostic factor of ITAC.

HIF1α-dependent and independent pathways regulate the expression of PD-L1 in prostate cancer

PD-L1/PD-1 pathway is a major pathway exploited by human cancer types, which is a target for current immunotherapy. We investigated tumor microenvironmental factors involved in PD-L1 induction in prostate cancer (PC). We studied the expression of PD-L1 in a series of 66 PCs, in parallel with the expression of hypoxia- and acidity-related immunohistochemical markers (Hypoxia-inducible factor HIF1α, and lactate dehydrogenase LDHA) and tumor-infiltrating lymphocyte TIL density. Experiments with three PC cell lines, the 22Rv1, DU145, and PC3 were conducted focusing on the inducibility of PD-L1 by hypoxia, acidity, lymphocyte interactions, and radiation. In tissues, PD-L1 expression by cancer cells was directly related to PD-L1 expression by TILs and macrophages (p < 0.05), and the overexpression of HIF1α and LDH5 (p < 0.05). TIL density was inversely related to ΗΙF1α (p = 0.02). Exposure of PC cell lines to hypoxia strongly induced PD-L1 and protein and mRNA levels, directly controlled by HIF1α function (p < 0.001). Irradiation with 20 Gy had no apparent effect on PD-L1 expression. Culturing PC cell lines with culture medium (CM) from PBMCs strongly induced PD-L1 at protein and mRNA levels, independently from HIF1α, which was also confirmed when cells were incubated with Interferon-γ (p < 0.001). It is concluded that the combination of anti-PD-L1/PD-1 immunotherapy with hypoxia/HIF-targeting may be important in the treatment of specific subgroups of PC patients.

Regional Variation in the Tumor Microenvironment, Immune Escape and Prognostic Factors in Breast Cancer in Sub-Saharan Africa.

The low overall survival rates of patients with breast cancer in sub-Saharan Africa (SSA) are driven by regionally differing tumor biology, advanced tumor stages at diagnosis, and limited access to therapy. However, it is not known whether regional differences in the composition of the tumor microenvironment (TME) exist and affect patients' prognosis. In this international, multicentre cohort study, 1,237 formalin-fixed, paraffin-embedded breast cancer samples, including samples of the "African Breast Cancer-Disparities in Outcomes (ABC-DO) Study," were analyzed. The immune cell phenotypes, their spatial distribution in the TME, and immune escape mechanisms of breast cancer samples from SSA and Germany (n = 117) were investigated using histomorphology, conventional and multiplex IHC, and RNA expression analysis. The data revealed no regional differences in the number of tumor-infiltrating lymphocytes (TIL) in the 1,237 SSA breast cancer samples, while the distribution of TILs in different breast cancer IHC subtypes showed regional diversity, particularly when compared with German samples. Higher TIL densities were associated with better survival in the SSA cohort (n = 400), but regional differences concerning the predictive value of TILs existed. High numbers of CD163+ macrophages and CD3+CD8+ T cells accompanied by reduced cytotoxicity, altered IL10 and IFNγ levels and downregulation of MHC class I components were predominantly detected in breast cancer samples from Western SSA. Features of nonimmunogenic breast cancer phenotypes were associated with reduced patient survival (n = 131). We therefore conclude that regional diversity in the distribution of breast cancer subtypes, TME composition, and immune escape mechanisms should be considered for therapy decisions in SSA and the design of personalized therapies. See related Spotlight by Bergin et al., p. 705.

The expression of programmed death-ligand 1 and programmed death-ligand 2 in endometrial carcinosarcoma: Correlation with mismatch repair protein expression status, tumor-infiltrating lymphocyte infiltration, and clinical outcomes

BackgroundEndometrial carcinosarcomas have high malignant potential with a high recurrence rate and poor prognosis. Immunotherapy may be a promising treatment option. The aim of this study is to evaluate the expression of PD-L1/PD-L2 and its relationship to mismatch repair (MMR) protein status and tumor-infiltrating lymphocyte (TIL) density. MethodsWe performed immunohistochemical analyses of PD-L1 (clone 22C3), PD-L2 (clone TY25), MSH-2, MSH-6, PMS-2, and MLH-1 in 77 tumors. We count TILs using CD8 antibody. Clinicopathologic features were recorded and statistically correlated with immunohistochemical results. Kaplan-Meier analyses were used to analyze the prognosis. ResultsWhile PD-L1 positivity was seen more commonly in MMR protein deficient tumors (p = 0.010), PD-L2 positivity was seen more commonly in MMR protein proficient tumors (p = 0.003). PD-L1 positivity was also found to be more common in carcinosarcoma with high TIL infiltration. PD-L2 positivity was associated with decreased overall survival (OS) rates (p = 0.043, p = 0.043, respectively), whereas the PD-L1 positivity and TIL density were not significantly associated with OS rate. The OS rate of patients with MMR protein proficient tumors was significantly lower compared with those with MMR protein deficient tumors (p = 0.042). The lower TILs infiltration was associated with a shorter disease-free survival (DFS) rate. PD-L1 and PD-L2 positivity did not affect the DFS rate. ConclusionsPD-L1/PD-L2 might be a better target for immunotherapy in endometrial carcinosarcoma. PD-L2 positivity was also associated with a worse clinical outcome in patients with endometrial carcinosarcoma, suggesting that PD-L2 status can be used to predict clinical behavior. Further studies are needed to elucidate the relationship between PD-L1/PD-L2 expression and therapeutic response.

Abstract 2266: Tumor associated neutrophils (TAN) may produce neutrophil extracellular traps (NETs) which can suppress the infiltration of activated T cells in tumor tissue

Abstract Objective: High density of tumor-infiltrating lymphocytes (TIL) in the tumor is associated with favorable prognosis of cancer patients. It has been reported that many neutrophils also infiltrate in various tumor tissues as tumor-associated neutrophils (TAN) which produce considerable amounts of neutrophil extracellular traps (NETs). However, the crosstalk between TAN and TIL and its impact on tumor behavior have not been fully examined. Methods: Neutrophils and mononuclear cells (PBMC) were collected from the blood of healthy donor. Neutrophils were stimulated with PMA or LPS for 15 min, washed extensively, and incubated for another 4 hours to produce NETs. PBMC were cultured on anti-CD3 mAb-coated plate with recombinant-IL-2 for 7-14days. The migration of activated T cells through 3μm pore to CXCL-11 was examined in vitro. The localization and densities of TAN and TIL were immunohistochemically evaluated in 38 human serous ovarian cancer tissue. NETs were quantified with the density of CD66b(+) and Citrullinated Histon H3(+) cells. Results: Chemotactic migration of T cells to CXCL-11 was inhibited by 99%±1.6% (p&amp;lt;0.001) in the presence of PMA-activated neutrophils in lower chamber. The inhibition was similarly detected when NET components were removed by ultracentrifugation. Timelapse analysis also showed the NETs-depleted supernatant inhibited random migration, which was totally canceled by the pretreatment with catalase. When neutrophils were activated by LPS, T cell migration was also inhibited by 49%±19.9% (p&amp;lt;0.01), but not by NETs-depleted supernatant. In western blotting, CXCL11 was totally degraded by LPS-stimulated neutrophils, which was greatly reduced by the depletion of NET component and restored by DNase I as well as phenylmethylsulfonyl fluoride (PMSF) or Neutrophil elastase inhibitor (NEI). The reduced migration induced by LPS-stimulated neutrophils was also restored with these reagents. In ovarian cancer tissues, density of TIL was greater in tumors with diameter ≥10cm than that of tumors &amp;lt;10cm (p=0.04). However, the densities of TIL and NETs were irrelevant with age, tumor stage and neutrophil-to-lymphocyte ratio (NLR) in peripheral blood. Multicolor immunohistochemistry showed that TAN and TIL tended to be reciprocally distributed in tumor stroma in most of the ovarian cancer tissues. Cit-H3 was positively detected in 2.1%~75% of CD66b(+) TANs, and the proportion of NET in TAN greatly varied among tumors. In all cases, the densities of CD4(+) and CD8(+) T cells inversely correlated with that of NETs (r = -0.5182, p = 0.0012; r = -0.5684, p = 0.0003, respectively). Conclusion: Activated neutrophils negatively regulate T cell migration partially through NET formation. TAN may suppress T cell traffic into tumor tissue, which may attenuate cell-mediated immunity in tumor microenvironment and promote tumor progression. Citation Format: Kohei Tamura, Hideyo Miyato, Misaki Matsumiya, Rei Takahashi, Yuki Kaneko, Yurie Futoh, Kazuya Takahashi, Yuki Kimura, Akira Saito, Hideyuki Ohzawa, Yasushi Saga, Yuji Takei, Hiroyuki Fujiwara, Joji Kitayama. Tumor associated neutrophils (TAN) may produce neutrophil extracellular traps (NETs) which can suppress the infiltration of activated T cells in tumor tissue [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 2266.

Abstract 5389: Micron-resolution spatial analysis near the tumor-stromal border reveals a distinct density distribution of tumor-infiltrating lymphocytes and related genomic features

Abstract Tumor-infiltrating lymphocytes (TILs) play a major role in predicting response to immunotherapy in solid tumors. However, few studies have analyzed TIL distribution and associated genomic signatures based on proximity to the tumor-stromal border (TSB). Here, we describe TIL density, IFN-γ signature, and tumor mutational burden (TMB), according to set distances from the TSB, in The Cancer Genome Atlas (TCGA) pan-carcinoma dataset. For spatial TIL analysis, we used Lunit SCOPE IO, an AI-powered H&amp;E whole-slide image analyzer, which identifies and quantifies tumor and lymphocyte cells within cancer or stroma area. In TCGA pan-carcinoma (n=7,468) dataset, the median values of cancer area and cancer stroma were 46.8 (interquartile range [IQR] 21.6-87.0) and 26.6 (IQR 10.9-54.5) mm2, respectively, and those of TIL density in each area were 72.1 (IQR 34.1-160.1) and 747.2 (IQR 330.4-1575.9)/mm2, showing 10-fold enrichment in stromal TIL (sTIL) compared to intratumoral TIL (iTIL). Subdividing the cancer area in 10 micron-wide steps from the TSB toward the tumor core, the median proportions of 0-10, 10-20, 20-30 (IT 0-10, IT 10-20, IT 20-30), and greater than 30 microns (IT 30-core) were 17.7%, 20.8%, 13.6%, and 47.9%, respectively. For subdividing the cancer stroma outwards from the TSB, the median proportions of 0-10, 10-20, 20-30 (ST 0-10, ST 10-20, ST 20-30), and greater than 30 microns (ST 30-outside) were 13.6%, 19.4%, 13.5%, and 53.5%, respectively. From the tumor core to TSB, median values of TIL density were 53.4 (IT 30-core), 59.3 (IT 20-30), 69.3 (IT 10-20), 127.4 (IT 0-10)/mm2, respectively. From the TSB to the remaining stroma, median values of TIL density were 98.7 (ST 0-10), 567.6 (ST 10-20), 708.8 (ST 20-30), and 896.1 (ST 30-outside)/mm2, respectively. Hence, the density of TIL generally increased from the tumor core to the outside of cancer stroma. Of note, IFN-γ signature showed the highest level of correlation with TIL density in IT 10-20 (Spearman’s rho [ρ] 0.5146) or IT 0-10 (ρ 0.5143), compared with other subdivided cancer areas (ρ 0.4567-0.4924), and cancer stromas (ρ 0.1454-0.4529). However, TMB was highly correlated with TIL density in ST 20-30 (ρ 0.3162) or ST 30-outside (ρ 0.3278), compared with other subdivided cancer stromas (ρ 0.0588-0.2843) and cancer areas (ρ 0.0784-0.1208). In conclusion, AI-powered analysis of the tumor micro-environment reveals that iTIL density increases from the tumor core to the outside of tumor-stromal border. IFN-γ signature is high in the area of the tumor-stromal border on the tumor side, while tumor mutational burden is associated with sTIL far from the tumor-stromal border. Citation Format: Sanghoon Song, Gahee Park, Sukjun Kim, Sangjoon Choi, Seokhwi Kim, Wonkyung Jung, Mingu Kang, Mohammad Mostafavi, Heon Song, Aaron Valero, Sérgio Pereira, Donggeun Yoo, Seulki Kim, Seunghwan Shin, Ken Nesmith, Chan-Young Ock. Micron-resolution spatial analysis near the tumor-stromal border reveals a distinct density distribution of tumor-infiltrating lymphocytes and related genomic features. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5389.

Abstract 6623: Spatial organization of tumor-infiltrating lymphocytes (TILs) is prognostic for survival in triple negative breast cancer (TNBC)

Abstract Introduction: TILs quantification is a predictor of response to neoadjuvant chemotherapy and prognostic for long term outcomes in TNBC. TILs are typically quantified using standard hematoxylin/eosin (H&amp;E) slides manually by trained pathologists; limitations include interobserver variability and need for cutoffs that aid in clinical decision making. Computational approaches utilizing H&amp;E slides have potential to improve reproducibility and refine predictive/prognostic utility by evaluation of additional metrics beyond quantification. Methods: H&amp;E whole slide images digitized at 20x magnification from patients with stage I-III TNBC (ER/PR ≤10%, HER2 negative) treated with neo/adjuvant chemotherapy at two institutions were utilized. All cell nuclei were automatically segmented using a deep learning model (Hover-Net) and classified as TIL or non-TIL based on morphological features. Features related to TIL density, spatial distribution, and morphological features were extracted. The top 3 features determined by least absolute shrinkage and selection operator were used to train a Cox proportional hazards regression model (SpaTILs) that assigned a recurrence free survival (RFS) event risk score to each patient. First quartile risk score in the training cohort was used as a cutoff for SpaTILs high vs low risk. Model performance for prediction of RFS and overall survival (OS) was evaluated in the testing cohort by Kaplan Meier method, with SpaTILs high vs low risk groups compared by log rank test and Cox regression analysis. Results: In the training cohort (n=26) and testing cohort (n=62), median age was 51 and 53 years, 77% and 36% were node positive, and 81% and 45% received neoadjuvant chemotherapy, respectively. Median follow up in testing and training cohorts was 2.3 and 6.8 y, respectively. In the testing cohort, 31% and 69% of patients were categorized as SpaTILs high and low risk, respectively. Baseline characteristics were not significantly different between SpaTILs high and low risk patients. In the SpaTILs high and low risk groups, 5y RFS was 53% and 90%, respectively (HR 4.85 [95% CI: 1.62-14.52], p=0.002) and 5y OS was 78% and 98%, respectively (HR 13.91 [95% CI 1.67-115.66], p=0.001). SpaTIL high risk remained a significant predictor for RFS and OS after adjusting for T stage and nodal status: HR 7.25 (95% CI 2.14-24.39), p=0.001 for RFS; HR 4.80 (95% CI 1.31-131.06), p=0.028 for OS. Conclusion: The computational model based on TILs density, spatial, and morphological features (SpaTILs) was independently prognostic for RFS and OS in early stage TNBC. This model is being validated in larger TNBC cohorts and warrants prospective evaluation in clinical trials. Citation Format: Sahil H. Patel, Germán Corredor, Rachel Yoder, Cristian Barrera, Miluska Castillo, Luis Bernabe, Joshua Staley, Shane Stecklein, Satish E. Viswanath, Carlos A. Castaneda, Priyanka Sharma, Anant Madabhushi. Spatial organization of tumor-infiltrating lymphocytes (TILs) is prognostic for survival in triple negative breast cancer (TNBC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6623.