Tumor-infiltrating Lymphocytes Score Research Articles

Annotation-efficient deep learning for breast cancer whole-slide image classification using tumour infiltrating lymphocytes and slide-level labels

Tumour-Infiltrating Lymphocytes (TILs) are pivotal in the immune response against cancer cells. Existing deep learning methods for TIL analysis in whole-slide images (WSIs) demand extensive patch-level annotations, often requiring labour-intensive specialist input. To address this, we propose a framework named annotation-efficient segmentation and attention-based classifier (ANSAC). ANSAC requires only slide-level labels to classify WSIs as having high vs. low TIL scores, with the binary classes divided by an expert-defined threshold. ANSAC automatically segments tumour and stroma regions relevant to TIL assessment, eliminating extensive manual annotations. Furthermore, it uses an attention model to generate a map that highlights the most pertinent regions for classification. Evaluating ANSAC on four breast cancer datasets, we demonstrate substantial improvements over three baseline methods in identifying TIL-relevant regions, with up to 8% classification improvement on a held-out test dataset. Additionally, we propose a pre-processing modification to a well-known method, enhancing its performance up to 6%.

A panoptic segmentation dataset and deep-learning approach for explainable scoring of tumor-infiltrating lymphocytes

Tumor-Infiltrating Lymphocytes (TILs) have strong prognostic and predictive value in breast cancer, but their visual assessment is subjective. To improve reproducibility, the International Immuno-oncology Working Group recently released recommendations for the computational assessment of TILs that build on visual scoring guidelines. However, existing resources do not adequately address these recommendations due to the lack of annotation datasets that enable joint, panoptic segmentation of tissue regions and cells. Moreover, existing deep-learning methods focus entirely on either tissue segmentation or cell nuclei detection, which complicates the process of TILs assessment by necessitating the use of multiple models and reconciling inconsistent predictions. We introduce PanopTILs, a region and cell-level annotation dataset containing 814,886 nuclei from 151 patients, openly accessible at: sites.google.com/view/panoptils. Using PanopTILs we developed MuTILs, a neural network optimized for assessing TILs in accordance with clinical recommendations. MuTILs is a concept bottleneck model designed to be interpretable and to encourage sensible predictions at multiple resolutions. Using a rigorous internal-external cross-validation procedure, MuTILs achieves an AUROC of 0.93 for lymphocyte detection and a DICE coefficient of 0.81 for tumor-associated stroma segmentation. Our computational score closely matched visual scores from 2 pathologists (Spearman R = 0.58–0.61, p < 0.001). Moreover, computational TILs scores had a higher prognostic value than visual scores, independent of TNM stage and patient age. In conclusion, we introduce a comprehensive open data resource and a modeling approach for detailed mapping of the breast tumor microenvironment.

Tumor-infiltrating lymphocytes and immune-related adverse events in advanced melanoma

The predictive value of tumor-infiltrating lymphocytes (TILs) in immune-related adverse event (irAE) development remains unknown, although an association between tumor immunogenicity and irAEs has been suggested. We investigated the association between TIL abundance in pretreatment primary and metastasis specimens and the subsequent development of severe irAEs. We retrospectively identified patients with advanced cutaneous melanoma who received first-line anti-programmed cell death protein 1 (PD-1) with or without anti-cytotoxic T-lymphocyte associated protein 4 (anti-CTLA-4) from 10 hospitals in the Netherlands. TILs were scored on representative hematoxylin and eosin (H&E) stains of the primary melanoma and pretreatment melanoma metastasis as 'absent', 'nonbrisk', or 'brisk'. A univariable logistic regression analysis was carried out to assess the association between the TIL scores and the development of severe irAEs. Fine and Gray subdistribution hazard models were used to estimate the cumulative incidence of severe irAEs. Of the 1346 eligible patients, 536 patients had primary melanoma specimens available, and 613 patients had metastasis specimens available. Severe irAEs occurred in 15% of anti-PD-1-treated patients and 49% of anti-PD-1+ anti-CTLA-4-treated patients. The presence of TILs was not associated with the occurrence of grade ≥3 irAEs in primary melanoma specimens (P= 0.70) nor pretreatment metastasis specimens (P= 0.91). In the univariable analysis, patients with brisk TILs did not have a higher chance of developing severe irAEs compared with patients with absent TILs, for both primary specimen (odds ratio 1.15, 95% confidence interval 0.60-2.18) and metastasis specimen (odds ratio 0.77, 95% confidence interval 0.37-1.59). There was also no significant difference in the lifetime risk or timing of the development of severe irAEs in patients with TILs present compared with patients with TILs absent. There was no association between the TIL scores on H&E-stained slides from the primary melanoma or pretreatment metastasis and the development of grade 3 or higher irAEs. Additionally, no correlation was found between the presence of TILs and the timing of irAEs.

Deep learning-based quantitative assessment of tumor-infiltrating lymphocytes from hematoxylin and eosin-stained slides in triple-negative breast cancer: A prognostic study.

553 Background: Tumor-infiltrating lymphocytes (TILs) are pivotal in predicting responses to preoperative chemotherapy in triple-negative breast cancer. Yet, their clinical adoption as a diagnostic marker is limited by the challenge of achieving high accuracy with objective measures. This study introduces a novel AI-based method to quantitatively assess stromal TILs within the tumor bed using hematoxylin and eosin (H&amp;E)-stained slides, examining their correlation with disease-free survival (DFS) and overall survival (OS). Methods: In this study, 68 cases of breast cancer with axillary lymph node metastasis at the time of resection were selected from a total of 3,902 resections conducted between 2002 and 2016. We then developed an AI-based model to detect the epithelium and lymphocytes (CD3/CD20) from H&amp;E-stained slides. This model was trained using annotated datasets, which included 26,509 images of the epithelium and 12,273 images of lymphocytes. The tumor bed areas were precisely defined by pathologists. Subsequently, the AI model identified epithelial and lymphocyte regions within these predefined areas. The TILs score was then calculated as the ratio of the lymphocyte area to the total stromal area within the tumor bed, excluding the epithelial areas identified by the AI model. Finally, we assessed the prognostic significance of the TILs score for DFS and OS through ROC analysis and log-rank tests. Results: ROC analysis identified an optimal TILs score threshold of 0.0042 for both DFS and OS, with an AUC of 0.743 (95% CI, 0.622-0.864) for DFS and 0.686 (95% CI, 0.559-0.812) for OS. This threshold significantly stratified patients into groups with distinct DFS and OS outcomes (P=0.003 and P=0.008, respectively). Conclusions: AI quantification of TILs offers a promising prognostic tool in triple-negative breast cancer, enhancing the objectivity and clinical utility of pathological analyses. Accurate TILs measurement could revolutionize prognostication and treatment planning.

Correlation of TILs indices generated by visual review and digital computational algorithm with outcomes from preoperative chemotherapy for TNBC in TBCRC 030.

555 Background: Predictive biomarkers for response to neoadjuvant chemotherapy (NAC) in early-stage triple negative breast cancer (TNBC) are needed. Stromal tumor infiltrating lymphocytes (TILs) are prognostic and predictive in TNBC, typically scored by visual examination (VE) of H&amp;E stained slides. Evaluation of image-based surrogate signatures for TILs and other immune profiles by computational assessment (CA) of slides may offer greater precision and more comprehensive assessment of the immune microenvironment. This study was designed to evaluate TILs by VE and CA methods and investigate through blinded assessment the predictive capacity of immune activation. Methods: The phase II study TBCRC 030 randomized patients (pts) with BRCA1/2-proficient stage I-III TNBC to cisplatin (C) or paclitaxel (T) NAC, with primary endpoint response at surgery. TILs were visually scored on a continuous scale by a single experienced pathologist on digitized H&amp;E images of pre-treatment core needle biopsies (CNB). CA used the 4D QPOR platform’s statistical physics and tumor biology based algorithm to quantify cell cycle aberrations and microenvironment dynamics including TILs, immune response (IHI), and combined immune response and cell cycle signature (CmbI) as continuous biomarkers from digitized CNB images. Evaluation of TILs by VE and CA were each correlated with response (RCB 0/1) or non-response (RCB 2/3) to NAC. Receiver operator characteristic (ROC) curves with area under the curve (AUC) measures and odds ratios (OR) with 95% confidence intervals were used to assess the predictive performance of each biomarker. Results: Of 139 evaluable pts, 121 had usable data for TILs by both VE and CA (59 received C, 62 received T). Descriptive statistics for biomarkers by response are shown in the Table. Notably TILs score by VE was significantly predictive of response, as was CA CmbI. TILs by VE had OR 1.86 (95% CI 1.24, 2.87, p=0.031) and AUC 0.69 (95% CI 0.57, 0.66), whereas CA CmbI had OR 1.54 (95% CI 1.00, 2.42, p=0.053) and AUC 0.62 (95% CI 0.51, 0.73). CA TILs and IHI were not predictive of response. TILs by VE predicted response to T (OR 2.91 95% CI 1.56, 6.16, p=0.002) but not C (OR 1.22 95% CI 0.67, 2.18 p=0.49). Conclusions: In this analysis of TBCRC 030, VE of baseline TILs and CA CmbI were predictive of response to NAC for TNBC, however CA did not outperform VE. TILs predicted response to specific chemotherapy. Automated computational evaluation of TILs or the broader immune microenvironment is an emerging field which may allow accurate and reproducible assessment of these important biomarkers. Clinical trial information: NCT01982448 . [Table: see text]

Association between stromal tumor-infiltrating lymphocytes (TILs) and pathologic complete response (pCR) in patients with early breast cancer (BC) treated with neoadjuvant chemotherapy and HER2-directed therapies in NSABP B-41.

3017 Background: In the phase 3 NSABP B-41 trial, there was an association between pCR and survival in 519 women with early HER2-positive BC treated with neoadjuvant chemotherapy in combination with trastuzumab (T), lapatinib (L), or both (TL). There is a need to identify prognostic biomarkers to tailor patients’ treatment, and the quantification of stromal TILs represents a promising, accessible, and reproducible option. In this analysis, we studied the association between TILs and pCR. Methods: Eligible patients had a baseline core biopsy sample, known pCR status, and no consent withdrawal for the use of specimens. Slides were scanned using an Aperio GT 450 automated scanner to create whole slide images. TIL analysis was then completed based on the international TIL working group guidelines (RS). Patients were grouped into three TIL score categories based on tumor TIL %: low (1-5%), intermediate (6-30%), and high (31-95%). The dose-response relationship between TIL percentage and pCR was explored via generalized linear models with splines. Results: 257 patients were included. 89 (35%) received T, 92 (36%) L, and 76 (29%) TL. 149 (58%) had ER-positive BC. TIL categories in this cohort: 49% low, 22% intermediate, and 29% high. TIL expression was similar across BC intrinsic subtypes and by ER. There was no association between the TIL group and pCR, irrespective of the treatment group (p=0.98). Among tumors with low TILs, T or TL had similar pCR compared to L alone (51.3 v 49%; p=0.95). Among tumors with high TILs, T or TL had higher pCR than L (61.2 v 40.4%; p=0.04). Among patients with ER-positive BC, high TILs were associated with a numerical increase in pCR rate (56% vs 39.6%; p=0.11). Among patients with ER-negative tumors, high TILs were associated with a numerical decreased rate of pCR (50% v. 66.7%; p=0.15). The p-value of the Breslow test of homogeneity of the odds ratios across ER status was 0.016, suggesting that the association between TILs (0-30 v. &gt;30) and pCR varies across ER status.No interaction between the TIL percentage and age, nodal status, menopausal status, or histologic grade was noted. Conclusions: This analysis revealed that patients with a high TIL percentage treated with T or TL compared to L were more likely to achieve a pCR. These findings can be partially explained by the mechanism of action of the HER2-directed agents. T leads to antibody-dependent cell-mediated cytotoxicity, and a higher percentage of TILs may enhance this mechanism, whereas L has intracellular tyrosine kinase disruption. The association between TILs and pCR varied across ER status. Our next steps include additional studies assessing the interaction between TILs and long-term outcomes, as well as gene expression and TILs to validate this biomarker for future clinical trials in HER2+ early BC. Clinical trial information: NCT00486668 .

Abstract PO5-25-03: Immune cell infiltrate profiles in primary Triple Negative Breast Cancer and co-relation with treatment response and survival outcomes

Abstract Introduction Tumor microenvironment (TME) in breast cancer is a heterogeneous landscape with multiple cell types, each modulating the tumor progression and the disease prognosis. These cell types include lymphocytes, RBCs, stromal cells, macrophages, cancer-associated fibroblasts (CAFs), and cancer stem cells. Each cell type in the TME influences tumor progression and clinical outcomes for patients in a good or bad way, depending on the interaction with tumor cells. Tumor-infiltrating lymphocytes (TILs) are shown to be prognostic for therapy response in many cancers in the western cohorts. Our recent study showed that a higher proportion of Indian TNBC patients had intra-tumoral TILs compared to Western cohorts, and these are predictive of better response to NACT and longer disease-free survival in patients, making it a distinct finding from Western reports. In the follow-up study, we are characterizing infiltrating immune cell subtypes and their comparative association with treatment response in TNBC patients. Results from preliminary analysis and comparative analysis between responders and non-responders will be presented here. We speculate that the pro- and anti-tumorigenic immune cell population and its spatial information would help us understand TME and its precise role in response to therapy. Methods A retrospective cohort of breast cancer patients from a single one-surgeon breast cancer clinic with a uniform treatment strategy was evaluated for this study. FFPE primary tissue samples with variable TILs scores (moderate to high), taken from patients who showed complete response to the NACT treatment (n=6) and non-responding group (n=8) determined by H&amp;E stained slides from our cohort, were further stained for immune cells (Pan-B cell, Pan-T cells, Pan-macrophages) and T cells (CD4; Helper T cell marker, CD8; Cytotoxic T cell marker), FOXP; Regularly CD4+ cell marker) and PanCK (Pan-tumor marker) and counterstained with DAPI for multiplex imaging. Multispectral images for these slides were obtained using the Leica Aperio Versa system. Parameters such as individual cell types of identification and scoring, spatial proximity to tumor cells, and individual cell types will be determined by HALO software using the High plex module. Results A total of 14 TNBC samples were stained for the T-cell subtypes panel and Immune cell subtype s panel to determine the TIME. DAPI stain was used for nuclear counting based the parameters such as threshold intensity, nuclear size, roundness, and aggressiveness. The cytoplasmic size was selected as 2um beyond the outside nuclear boundary. Using Halo-Indica software, around 50000-300000 cells/tissue were evaluated across all samples from mIF whole slide scans of biopsy samples. Cell count was validated by cross-checking individual ROIs and manual cell count using Fiji. A Highplex module was used to identify individual signal positivity for each marker based on threshold intensity and cell completeness percent. Preliminary data showed a higher proportion of CD8+ cells and macrophages population are spatially closer to the tumor cells in Responders compared to non-responders. Conclusion Immune profiles are distinct among responding and non-responding patients in the Indian TNBC cohort. The association of these profiles with long-term survival in patients and various clinical and pathological parameters will help us determine the clinical relevance of these profiles and identify distinct immune subgroups as prognostic in Indian TNBCs. Citation Format: Pooja Vaid, Devaki Kelkar, LS Shashidhara, C B Koppiker, Madhura Kulkarni. Immune cell infiltrate profiles in primary Triple Negative Breast Cancer and co-relation with treatment response and survival outcomes [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-25-03.

Predictive Factors of Neoadjuvant Chemotherapy Response in Breast Cancer Validated by Three Anatomopathological Scores: Residual Cancer Burden, Chevallier System, and Tumor-Infiltrating Lymphocytes.

The aim of this retrospective, observational, descriptive study was to identify predictors of response to neoadjuvant therapy in breast cancer patients and to validate them using three anatomopathological scores, such as residual cancer burden (RCB) score, Chevallier system, and tumor-infiltrating lymphocytes (TIL) score. We conducted a study on 88 female patientsaged 37 to 78 years with confirmed breast cancerwho were indicated for neoadjuvant chemotherapy. We analyzed different individual variables (such as age, menarche, and menopause), clinical/imaging characteristics of the breast tumor and axillary nodes, immunohistochemical biomarkers (such as ER/PR/HER2 and Ki67), and histopathological features (such as subtype and grading) in relation to three anatomopathological scores calculated based on the surgical resection specimen. The percentage of patients who could have benefited from conservative surgery increased from 6% at the time of diagnosis to 20% post-primary systemic therapy (PST). Age under 49 (p = 0.01), premenopausal status (p < 0.01), no special type (NST) (p = 0.04), high Ki67 (p < 0.01), triple-negative breast cancer (TNBC) (p = 0.02) are positive predictive factors of neoadjuvant therapy, whilelobular/mixt carcinoma-type (p = 0.04), luminal A (p = 0.01), positive lymph node (p < 0.01), and low differentiation grade (p = 0.01) are negative predictive factors for the response to PST. There is a strong correlation between the RCB score and the Chevallier system for quantifying the response to PST, with most predictive factors being confirmed through appropriate statistical analysis for both.TIL score values correlated with only some of the predictors, most likely due to the importance of calculating this score on both biopsy specimens at diagnosis and resection specimens after chemotherapy.

Tumor-infiltrating lymphocytes in HER2-positive breast cancer treated with neoadjuvant chemotherapy and dual HER2-blockade.

Tumor-infiltrating lymphocytes (TILs) have been associated with outcomes in HER2-positive breast cancer patients treated with neoadjuvant chemotherapy and trastuzumab. However, it remains unclear if TILs could be a prognostic and/or predictive biomarker in the context of dual HER2-targeting treatment. In this study, we evaluated the association between TILs and pathological response (pCR) and invasive-disease free survival (IDFS) in 389 patients with stage II-III HER2 positive breast cancer who received neoadjuvant anthracycline-containing or anthracycline-free chemotherapy combined with trastuzumab and pertuzumab in the TRAIN-2 trial. Although no significant association was seen between TILs and pCR, patients with TIL scores ≥60% demonstrated an excellent 3-year IDFS of 100% (95% CI 100-100), regardless of hormone receptor status, nodal stage and attainment of pCR. Additionally, in patients with hormone receptor positive disease, TILs as a continuous variable showed a trend to a positive association with pCR (adjusted Odds Ratio per 10% increase in TILs 1.15, 95% CI 0.99-1.34, p = 0.070) and IDFS (adjusted Hazard Ratio per 10% increase in TILs 0.71, 95% CI 0.50-1.01, p = 0.058). We found no interactions between TILs and anthracycline treatment. Our results suggest that high TIL scores might be able to identify stage II-III HER2-positive breast cancer patients with a favorable prognosis.

Abstract 7515: Enhancing immune checkpoint inhibitor efficacy with fulvestrant in estrogen receptor-positive breast cancer

Abstract Introduction: Breast cancer (BC) recurs decades after initial treatment that consists majority of the annual 43,000 BC deaths. Thus, Immune Checkpoint Inhibitor (ICI) therapy that bolster anti-cancer immunity is expected to improve long-term survival with reduced recurrence. Neoadjuvant ICI revolutionized management of triple negative breast cancer (TNBC); however, this represents only 5-7% of all BC cases, where estrogen receptor (ER)-positive subtype that consists of 70% of all BCs is not indicated due to less tumor-infiltrating lymphocytes (TILs). Using I-SPY2 trial we aim to investigate the potency of ER-alpha gene (ESR1) expression as a predictive biomarker, and whether the inhibition of the ER with fulvestrant enhances the efficacy of ICIs in ER+ BC. Methods:Total of 1157 BC patients from two independent cohorts (ISPY2 trial; n= 73, TCGA; n = 1084) were included, encompassing both clinical and transcriptomic profiles. The TIL score was determined using the xCell algorithm. Results:Using the transcriptome of the ISPY2 trial, we found that compared to ESR1 low expression patients, ESR1 high patients were associated with significantly lower pCR rates after neoadjuvant immunotherapy with ICI in whole cohort (n=71, pCR rate=30%, p&amp;lt;0.05) and HR+ subtype (n=50, pCR rate=20%, p&amp;lt;0.001), but not in triple-negative breast cancer (TNBC). ESR1 expression outperformed PD-1, PD-L1, and CTLA4 in Area Under the Curve (AUC) of the whole ISPY2 cohort (AUC=0.75, 0.68, 0.48, and 0.62, respectively), and further pronounced in HR+ BC (AUC=0.88, 0.73, 0.55, and 0.65, respectively), suggesting the potency of ESR1 expression as a predictive biomarker for ICI in HR+ BC. In agreement, ESR1 expression inversely correlated with the amount of intratumoral TILs consistently in both ISPY2 and TCGA cohorts in HR+ BC, but only in TCGA in whole cohort (all p&amp;lt;0.05). Among the intratumoral TILs, CD8+ T cell and M1 Macrophages were found to be less infiltrated in HR+ BC in TCGA. Given that ESR1 expression was associated with less intratumoral TILs and a poor response to ICI, we hypothesized that suppression of ER-alpha would enhance the effectiveness of ICI in HR+ BC in vitro. ER-alpha positive MCF7 cells were incubated with activated peripheral blood mononuclear cells with 1:1 ratio, and treated with 0.1 microgram of anti-PDL1, and 1nM of ER antagonist, fulvestrant, which demonstrated minimum cell killing. We found that ICI therapy with anti-PDL1 significantly decreased the cell viability of the MCF7 cells when ER-alpha expression was suppressed by fulvestrant, while activated peripheral blood mononuclear cells or anti-PDL1 alone or the combination did not exhibit any significant cell killing. Conclusion: We found that ESR1 expression is a predictive biomarker of ICI, and ER antagonist fulvestrant enhances the response of HR+ BC to ICI, suggesting that this combination could expand ICI indications to HR+ BC. Citation Format: Jun Arima, Kohei Chida, Masanori Oshi, Arya Mariam Roy, Kohei Taniguchi, Sang-Woong Lee, Kazuaki Takabe. Enhancing immune checkpoint inhibitor efficacy with fulvestrant in estrogen receptor-positive breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7515.

Abstract 3437: Stromal tumor-infiltrating lymphocytes in association with body size and composition among Black and White women with breast cancer

Abstract Background: Obesity is a major risk factor for breast cancer and negatively affects disease prognosis. Tumor-infiltrating lymphocytes (TILs) in the stroma of breast tissue play a major role in eliminating cancer cells and modulating immunotherapy efficacy. We aim to study the association between body size and composition with stromal TILs in breast tumor tissue by race. Methods: The study included 920 Black and 395 White women with newly diagnosed invasive breast cancer who have stromal TIL scores from the Women’s Circle of Health Study (WCHS). Body mass index (BMI), waist circumference (WC), and waist-to-hip ratio (WHR) were measured during the home interview. Pre-treatment H&amp;E-stained tumor tissue sections were reviewed by our study pathologist and stromal TILs were scored on a scale of 0-100% at 10% increment according to the recommendations of the International TILs Working Group. Linear regressions adjusting for age at diagnosis, tumor grade, stage, and molecular subtype were used to test the differences in TILs across the categories of body size and composition indicators separately for Black and White women. Results: A higher stromal TILs score was associated with more aggressive breast cancer features, including higher grade, larger tumor size, more advanced stage, hormone receptor (HR) negativity, and human epidermal growth factor receptor 2 (HER2) positivity in both Black and White women (all P&amp;lt;0.05). Among Black women, WHR &amp;gt;0.90 vs ≤0.85 was associated with lower TILs overall (β = -3.5%, P = 0.045) and in triple-negative tumors (β = -12.3%, P = 0.003). Among White women, BMI 25.0 - 29.9 vs &amp;lt;25 was associated with higher TILs in HR-/HER2+ tumors (β = 28.1%, P = 0.031). A pattern suggested that BMI ≥30 vs &amp;lt;25 was associated with lower TILs in both HR-/HER2+ (β = -10.2, P=0.52) and triple-negative (β = -12.8, P=0.14) tumors, although the associations were not significant. No association was observed between TILs and any of the body composition parameters in women with HR+ tumors. Conclusions and Relevance: Body size and composition correlate with stromal TILs mainly in patients with HR- tumors, and this relation may be race-dependent. These findings suggest that body size and composition potentially modulate breast cancer treatment outcomes at least in part through TILs-dependent mechanisms. Funding: This work was in part supported by grants from the US National Institutes of Health (P01CA151135, R01CA100598, R01CA185623, P30CA016056, P30CA072720, K07CA201334, and R37CA248371), US Army Medical Research and Material Command (DAMD-17-01-1-0334), the Breast Cancer Research Foundation (CBA), and the Philip L. Hubbell family. Citation Format: Rand T. Akasheh, Thaer Khoury, Song Yao, Angela Omilian, Chi-Chen Hong, Elisa V. Bandera, Bonnie Qin, Nur Zeinomar, Christine Ambrosone, Ting-Yuan David Cheng. Stromal tumor-infiltrating lymphocytes in association with body size and composition among Black and White women with breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3437.

Intraepithelial tumor-infiltrating lymphocytes shape loco-regional PET/CT spread of locally advanced cervical cancer

BackgroundData suggest an association between positron emission tomography/CT (PET/CT) metabolic metrics and tumor microenvironment in several malignancies, and a potential role of PET/CT to monitor response to immunotherapy.ObjectiveTo evaluate the...

Exploration of the relationship between tumor-infiltrating lymphocyte score and histological grade in breast cancer

BackgroundThe histological grade is an important factor in the prognosis of invasive breast cancer and is vital to accurately identify the histological grade and reclassify of Grade2 status in breast cancer patients.MethodsIn this study, data were collected from 556 invasive breast cancer patients, and then randomly divided into training cohort (n = 335) and validation cohort (n = 221). All patients were divided into actual low risk group (Grade1) and high risk group (Grade2/3) based on traditional histological grade, and tumor-infiltrating lymphocyte score (TILs-score) obtained from multiphoton images, and the TILs assessment method proposed by International Immuno-Oncology Biomarker Working Group (TILs-WG) were also used to differentiate between high risk group and low risk group of histological grade in patients with invasive breast cancer. Furthermore, TILs-score was used to reclassify Grade2 (G2) into G2 /Low risk and G2/High risk. The coefficients for each TILs in the training cohort were retrieved using ridge regression and TILs-score was created based on the coefficients of the three kinds of TILs.ResultsStatistical analysis shows that TILs-score is significantly correlated with histological grade, and is an independent predictor of histological grade (odds ratio [OR], 2.548; 95%CI, 1.648–3.941; P < 0.0001), but TILs-WG is not an independent predictive factor for grade (P > 0.05 in the univariate analysis). Moreover, the risk of G2/High risk group is higher than that of G2/Low risk group, and the survival rate of patients with G2/Low risk is similar to that of Grade1, while the survival rate of patients with G2/High risk is even worse than that of patients with G3.ConclusionOur results suggest that TILs-score can be used to predict the histological grade of breast cancer and potentially to guide the therapeutic management of breast cancer patients.

Improving Tumor-Infiltrating Lymphocytes Score Prediction in Breast Cancer with Self-Supervised Learning.

Tumor microenvironment (TME) plays a pivotal role in immuno-oncology, which investigates the intricate interactions between tumors and the human immune system. Specifically, tumor-infiltrating lymphocytes (TILs) are crucial biomarkers for evaluating the prognosis of breast cancer patients and have the potential to refine immunotherapy precision and accurately identify tumor cells in specific cancer types. In this study, we conducted tissue segmentation and lymphocyte detection tasks to predict TIL scores by employing self-supervised learning (SSL) model-based approaches capable of addressing limited labeling data issues. Our experiments showed a 1.9% improvement in tissue segmentation and a 2% improvement in lymphocyte detection over the ImageNet pre-training model. Using these SSL-based models, we achieved a TIL score of 0.718 with a 4.4% improvement. In particular, when trained with only 10% of the entire dataset, the SwAV pre-trained model exhibited a superior performance over other models. Our work highlights improved tissue segmentation and lymphocyte detection using the SSL model with less labeled data for TIL score prediction.

Abstract A121: Temsirolimus in combination with metformin in patients with advanced or recurrent endometrial cancer

Abstract Background: Molecular alterations in the PI3K/AKT and Ras/Raf/MAPK are frequently seen in endometrial cancers and mTOR is a downstream target activated by both pathways. Blocking mTOR with temsirolimus produced modest response. Studies have shown that metformin, in addition to producing metabolic changes, also causes activation of AMPK, which in turn inhibits the mTOR pathway. Therefore, we enrolled patients with advanced or recurrent endometrial cancer in the expansion cohort of a phase I study of temsirolimus in combination with metformin. Methods: Patients received intravenous (IV) temsirolimus 25 mg weekly and oral metformin 2000 mg daily in 28-day cycles. The objective of this study was to evaluate the efficacy of this combination and assessment of correlatives of target inhibition in tumor biopsies from patients with known mutations. Response was assessed every 2 cycles by clinical evaluation, tumor markers, and imaging per RECIST 1.1. All toxicities were graded using NCI CTCAE, version 4.0. TIL (tumor infiltrating lymphocyte) percentage was assessed in tumor tissue samples collected at baseline and on treatment. Results: A total of 40 patients were included in this study. Among the 33 patients evaluable for response, 2 patients had partial response (PR) and 13 had stable disease (SD), including 11 with SD ≥4 months. The objective response rate was 6% and clinical benefit rate 39%. Molecular alteration in the PI3K and/or RAS pathway was seen in 25 of 33 patients. Eleven of the 25 patients (44%) had either PR or SD ≥4 months. Interestingly, all 3 patients with molecular alteration in both the pathways had SD ≥4 months. The most common treatment-related adverse events included hypertriglyceridemia (n=14), diarrhea (n=13), mucositis (n=13), anorexia (n=12) and anemia (n=10). The grade 3 AEs were anemia and thrombocytopenia (n=2 each) and mucositis, fatigue, weight loss, hypokalemia, hypophosphatemia, and increase in aspartate aminotransferase and alanine transaminase (n=1 each). Although baseline TIL scores did not distinguish responder from non-responder (p=0.25), on-treatment increase in TIL score was significantly higher among responders compared to non-responders (p-value 0.034). Metabolomics portion of the study is ongoing. Conclusions: In conclusion, the combination of temsirolimus with metformin had modest anti-tumor activity in patients with advanced or recurrent endometrial cancer without added safety concerns. Correlative studies may provide further insight about this combination. Citation Format: Jibran Ahmed, Bettzy Stephen, Samir Beyaz, Charlie Chung, Ece Kilic, Muhammad R Khawaja, Yali Yang, Alpa M Nick, Maria G Raso, Elizve B Toro, Daniel D Karp, Sarina A Piha-Paul, Anil Sood, Chaan Ng, Amber M Johnson, Pamela T Soliman, Funda Meric-Bernstam, Karen H Lu, Aung Naing. Temsirolimus in combination with metformin in patients with advanced or recurrent endometrial cancer [abstract]. In: Proceedings of the AACR-NCI-EORTC Virtual International Conference on Molecular Targets and Cancer Therapeutics; 2023 Oct 11-15; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2023;22(12 Suppl):Abstract nr A121.

197P Computational pathology pipeline enables quantification of intratumor heterogeneity and tumor-infiltrating lymphocyte score

197P Computational pathology pipeline enables quantification of intratumor heterogeneity and tumor-infiltrating lymphocyte score

The levels of immunosuppressive checkpoint protein PD-L1 and tumor-infiltrating lymphocytes were integrated to reveal the glioma tumor microenvironment.

In spite of significant strides in the realm of cancer biology and therapeutic interventions, the clinical prognosis for patients afflicted with glioblastoma (GBM) remains distressingly dismal. The tumor immune microenvironment (TIME), a crucial player in the progression, treatment response, and prognostic trajectory of glioma, warrants thorough exploration. Within this intricate microcosm, the immunosuppressive checkpoint protein PD-L1 and tumor-infiltrating lymphocytes (TILs) emerge as pivotal constituents, underscoring their potential role in deciphering glioma biology and informing treatment strategies. However, prognostic models based on the association between PD-L1 expression and TIL infiltration in the tumor immune microenvironment have not been established. The aim of this study was to explore TIME genes associated with PD-L1 expression and TIL invasion and to construct a risk score for predicting the overall survival (OS) of GBM patients based on these genes. The samples were separately classified according to the PD-L1 expression level and TIL score and TIME-related genes were identified using differential expression and weighted gene co-expression network analysis. The DEGs were subjected to least absolute contraction and selection operator (LASSO) -Cox regression to construct TIME associated risk score (TIMErisk). A TIMErisk was developed based on STEAP3 and CXCL13 genes. The STLEAP3 was demonstrated to be involved in glioma progression. The results showed that the patients in the high TIMErisk group had poor OS compared with subjects in the low TIMErisk group. The biological phenotypes associated with TIMErisk were analyzed in terms of functional enrichment, tumor immune profile, and tumor mutation profile. The results on tumor immune dysfunction and exclusion dysfunction (TIDE) score and immune surface score (IPS) showed that GBM patients with different TIME risks had different responses to immunotherapy. Tumor purity analysis indicated that PD-L1 and TIL scores were positively correlated with TIMErisk score and negatively correlated with tumor purity. These results show that the TIMErisk-based prognostic model had high predictive value for the prognosis and immune characteristics of GBM patients. Immunohistochemical staining images of patients in the high and low TIMErisk groups were analyzed, showing that the degree of immune cell infiltration was higher in the high TIMErisk group relative to the low TIMErisk group. The present study provides a basis for understanding glioma tumor microenvironment and a foundation for conducting comprehensive immunogenomic analysis.

Lag-3 expression and clinical outcomes in metastatic melanoma patients treated with combination anti-lag-3 + anti-PD-1-based immunotherapies

ABSTRACT Lymphocyte-activation gene-3 (LAG-3), an immune checkpoint receptor, negatively regulates T-cell function and facilitates immune escape of tumors. Dual inhibition of LAG-3 and programmed cell death receptor-1 (PD-1) significantly improved progression-free survival (PFS) in metastatic melanoma patients compared to anti-PD-1 therapy alone. Investigating the utility of LAG-3 expression as a biomarker of response to anti-LAG-3 + anti-PD-1 immunotherapy is of great clinical relevance. This study sought to evaluate the association between baseline LAG-3 expression and clinical outcomes following anti-LAG-3 and anti-PD-1-based immunotherapy in metastatic melanoma. LAG-3 immunohistochemistry (clone D2G4O) was performed on pre-treatment formalin-fixed, paraffin-embedded metastatic melanoma specimens from 53 patients treated with combination anti-LAG-3 + anti-PD-1-based therapies. Eleven patients had received prior anti-PD-1-based treatment. Patients were categorized as responders (complete/partial response; n = 36) or non-responders (stable/progressive disease; n = 17) based on the Response Evaluation Criteria in Solid Tumours (RECIST). Tumor-infiltrating lymphocytes (TILs) were scored on hematoxylin and eosin-stained sections. LAG-3 expression was observed in 81% of patients, with staining in TILs and dendritic cells. Responders displayed significantly higher proportions of LAG-3+ cells compared to non-responders (P = .0210). LAG-3 expression positively correlated with TIL score (P < .01). There were no significant differences in LAG-3 expression between different sites of metastases (P > .05). Patients with ≥ 1% LAG-3+ cells in their tumors had significantly longer PFS compared to patients with < 1% LAG-3 expression (P = .0037). No significant difference was observed in overall survival between the two groups (P = .1417). Therefore, the assessment of LAG-3 expression via IHC warrants further evaluation to determine its role as a predictive marker of response and survival in metastatic melanoma.

2205P Tumor-infiltrating lymphocytes score possesses a relation with adjuvant chemo-immunotherapy benefit and cellular morphology in large-cell neuroendocrine carcinoma

2205P Tumor-infiltrating lymphocytes score possesses a relation with adjuvant chemo-immunotherapy benefit and cellular morphology in large-cell neuroendocrine carcinoma

Tumor-Infiltrating Lymphocyte Recognition in Primary Melanoma by Deep Learning Convolutional Neural Network

The presence of tumor-infiltrating lymphocytes (TILs) is associated with a favorable prognosis of primary melanoma (PM). Recently, artificial intelligence (AI)-based approach in digital pathology was proposed for the standardized assessment of TILs on hematoxylin and eosin-stained whole slide images (WSIs). Herein, the study applied a new convolution neural network (CNN) analysis of PM WSIs to automatically assess the infiltration of TILs and extract a TIL score. A CNN was trained and validated in a retrospective cohort of 307 PMs including a training set (237 WSIs, 57,758 patches) and an independent testing set (70 WSIs, 29,533 patches). An AI-based TIL density index (AI-TIL) was identified after the classification of tumor patches by the presence or absence of TILs. The proposed CNN showed high performance in recognizing TILs in PM WSIs, showing 100% specificity and sensitivity on the testing set. The AI-based TIL index correlated with conventional TIL evaluation and clinical outcome. The AI-TIL index was an independent prognostic marker associated directly with a favorable prognosis. A fully automated and standardized AI-TIL appeared to be superior to conventional methods at differentiating the PM clinical outcome. Further studies are required to develop an easy-to-use tool to assist pathologists to assess TILs in the clinical evaluation of solid tumors.