A new AI-assisted scoring system for PD-L1 expression in NSCLC

Crossing the Andes: Challenges and opportunities for digital pathology in Latin America

The most widely accepted and used type of digital pathology (DP) is whole-slide imaging (WSI). The USFDA granted two WSI system approvals for primary diagnosis, the first in 2017. In Latin America, DP has the potential to reshape healthcare by enhancing diagnostic capabilities through artificial intelligence (AI) and standardizing pathology reports. Yet, we must tackle regulatory hurdles, training, resource availability, and unique challenges to the region. Collectively addressing these hurdles can enable the region to harness DP's advantages-enhancing disease diagnosis, medical research, and healthcare accessibility for its population. Americas Health Foundation assembled a panel of Latin American pathologists who are experts in DP to assess the hurdles to implementing it into pathologists' workflows in the region and provide recommendations for overcoming them. Some key steps recommended include creating a Latin American Society of Digital Pathology to provide continuing education, developing AI models trained on the Latin American population, establishing national regulatory frameworks for protecting the data, and standardizing formats for DP images to ensure that pathologists can collaborate and validate specimens across the various DP platforms.

Pan-cancer evaluation of tumor-infiltrating lymphocytes and programmed cell death protein ligand-1 in metastatic biopsies and matched primary tumors

Tumor immunological characterization includes evaluation of tumor-infiltrating lymphocytes (TILs) and programmed cell death protein ligand-1 (PD-L1) expression. This study investigated TIL distribution, its prognostic value, and PD-L1 expression in metastatic and matched primary tumors (PTs). Specimens from 550 pan-cancer patients of the SHIVA01 trial (NCT01771458) with available metastatic biopsy and 111 matched PTs were evaluated for TILs and PD-L1. Combined positive score (CPS), tumor proportion score (TPS), and immune cell (IC) score were determined. TILs and PD-L1 were assessed according to PT organ of origin, histological subtype, and metastatic biopsy site. We found that TIL distribution in metastases did not vary according to PT organ of origin, histological subtype, or metastatic biopsy site, with a median of 10% (range: 0-70). TILs were decreased in metastases compared to PT (20% [5-60] versus 10% [0-40], p < 0.0001). CPS varied according to histological subtype (p = 0.02) and biopsy site (p < 0.02). TPS varied according to PT organ of origin (p = 0.003), histological subtype (p = 0.0004), and metastatic biopsy site (p = 0.00004). TPS was higher in metastases than in PT (p < 0.0001). TILs in metastases did not correlate with overall survival. In conclusion, metastases harbored fewer TILs than matched PT, regardless of PT organ of origin, histological subtype, and metastatic biopsy site. PD-L1 expression increased with disease progression. © 2024 The Author(s). The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Whole slide image-based weakly supervised deep learning for predicting major pathological response in non-small cell lung cancer following neoadjuvant chemoimmunotherapy: a multicenter, retrospective, cohort study

Objective

Develop a predictive model utilizing weakly supervised deep learning techniques to accurately forecast major pathological response (MPR) in patients with resectable non-small cell lung cancer (NSCLC) undergoing neoadjuvant chemoimmunotherapy (NICT), by leveraging whole slide images (WSIs).

Methods

This retrospective study examined pre-treatment WSIs from 186 patients with non-small cell lung cancer (NSCLC), using a weakly supervised learning framework. We employed advanced deep learning architectures, including DenseNet121, ResNet50, and Inception V3, to analyze WSIs on both micro (patch) and macro (slide) levels. The training process incorporated innovative data augmentation and normalization techniques to bolster the robustness of the models. We evaluated the performance of these models against traditional clinical predictors and integrated them with a novel pathomics signature, which was developed using multi-instance learning algorithms that facilitate feature aggregation from patch-level probability distributions.

Results

Univariate and multivariable analyses confirmed histology as a statistically significant prognostic factor for MPR (P-value< 0.05). In patch model evaluations, DenseNet121 led in the validation set with an area under the curve (AUC) of 0.656, surpassing ResNet50 (AUC = 0.626) and Inception V3 (AUC = 0.654), and showed strong generalization in external testing (AUC = 0.611). Further evaluation through visual inspection of patch-level data integration into WSIs revealed XGBoost's superior class differentiation and generalization, achieving the highest AUCs of 0.998 in training and robust scores of 0.818 in validation and 0.805 in testing. Integrating pathomics features with clinical data into a nomogram yielded AUC of 0.819 in validation and 0.820 in testing, enhancing discriminative accuracy. Gradient-weighted Class Activation Mapping (Grad-CAM) and feature aggregation methods notably boosted the model's interpretability and feature modeling.

Conclusion

The application of weakly supervised deep learning to WSIs offers a powerful tool for predicting MPR in NSCLC patients treated with NICT.

The rapidly changing field of predictive biomarkers of non-small cell lung cancer

Lung cancer is a leading cause of cancer-related death worldwide in both men and women, however mortality in the US and EU are recently declining in parallel with the gradual cut of smoking prevalence. Consequently, the relative frequency of adenocarcinoma increased while that of squamous and small cell carcinomas declined. During the last two decades a plethora of targeted drug therapies have appeared for the treatment of metastasizing non-small cell lung carcinomas (NSCLC). Personalized oncology aims to precisely match patients to treatments with the highest potential of success. Extensive research is done to introduce biomarkers which can predict the effectiveness of a specific targeted therapeutic approach. The EGFR signaling pathway includes several sufficient targets for the treatment of human cancers including NSCLC. Lung adenocarcinoma may harbor both activating and resistance mutations of the EGFR gene, and further, mutations of KRAS and BRAF oncogenes. Less frequent but targetable genetic alterations include ALK, ROS1, RET gene rearrangements, and various alterations of MET proto-oncogene. In addition, the importance of anti-tumor immunity and of tumor microenvironment has become evident recently. Accumulation of mutations generally trigger tumor specific immune defense, but immune protection may be upregulated as an aggressive feature. The blockade of immune checkpoints results in potential reactivation of tumor cell killing and induces significant tumor regression in various tumor types, such as lung carcinoma. Therapeutic responses to anti PD1-PD-L1 treatment may correlate with the expression of PD-L1 by tumor cells. Due to the wide range of diagnostic and predictive features in lung cancer a plenty of tests are required from a single small biopsy or cytology specimen, which is challenged by major issues of sample quantity and quality. Thus, the efficacy of biomarker testing should be warranted by standardized policy and optimal material usage. In this review we aim to discuss major targeted therapy-related biomarkers in NSCLC and testing possibilities comprehensively.

Clinical Validation of Artificial Intelligence-Powered PD-L1 Tumor Proportion Score Interpretation for Immune Checkpoint Inhibitor Response Prediction in Non-Small Cell Lung Cancer

PURPOSE

Evaluation of PD-L1 tumor proportion score (TPS) by pathologists has been very impactful but is limited by factors such as intraobserver/interobserver bias and intratumor heterogeneity. We developed an artificial intelligence (AI)-powered analyzer to assess TPS for the prediction of immune checkpoint inhibitor (ICI) response in advanced non-small cell lung cancer (NSCLC).

MATERIALS AND METHODS

The AI analyzer was trained with 393,565 tumor cells annotated by board-certified pathologists for PD-L1 expression in 802 whole-slide images (WSIs) stained by 22C3 pharmDx immunohistochemistry. The clinical performance of the analyzer was validated in an external cohort of 430 WSIs from patients with NSCLC. Three pathologists performed annotations of this external cohort, and their consensus TPS was compared with AI-based TPS.

RESULTS

In comparing PD-L1 TPS assessed by AI analyzer and by pathologists, a significant positive correlation was observed (Spearman coefficient = 0.925; P < .001). The concordance of TPS between AI analyzer and pathologists according to TPS ≥50%, 1%-49%, and <1% was 85.7%, 89.3%, and 52.4%, respectively. In median progression-free survival (PFS), AI-based TPS predicted prognosis in the TPS 1%-49% or TPS <1% group better than the pathologist's reading, with the TPS ≥50% group as a reference (hazard ratio [HR], 1.49 [95% CI, 1.19 to 1.86] v HR, 1.36 [95% CI, 1.08 to 1.71] for TPS 1%-49% group, and HR, 2.38 [95% CI, 1.69 to 3.35] v HR, 1.62 [95% CI, 1.23 to 2.13] for TPS <1% group).

CONCLUSION

PD-L1 TPS assessed by AI analyzer correlates with that of pathologists, with clinical performance also being comparable when referenced to PFS. The AI model can accurately predict tumor response and PFS of ICI in advanced NSCLC via assessment of PD-L1 TPS.

Prognostic significance of immunohistochemical classification utilizing biopsy specimens in patients with extensive-disease small-cell lung cancer treated with first-line chemotherapy and immune checkpoint inhibitors

PURPOSE

Although immune checkpoint inhibitors (ICIs), together with cytotoxic chemotherapy (chemoimmunotherapy), have been adapted for the initial treatment of extensive-disease small-cell lung cancer (ED-SCLC), they have achieved limited success. In ED-SCLC, a subtype of SCLC, the expression of immune-related molecules and clinical data are not well understood in relation to ICI treatment efficiency.

METHODS

We examined lung biopsy specimens from patients diagnosed with ED-SCLC treated with chemoimmunotherapy or chemotherapy. SCLC subtype, expression of HLA class I, and infiltration of CD8-positive cells were examined using immunohistochemistry (IHC). Subsequently, the association between clinical factors, IHC results, and progression-free survival or overall survival was assessed.

RESULTS

Most of the cases showed the achaete-scute homolog 1 (ASCL1) subtype. Among the 75 SCLC cases, 29 expressed high levels of HLA class I, while 46 showed low levels or a negative result; 33 patients were characterized as CD8-high, whereas 42 were CD8-low. In the chemoimmunotherapy cohort, multivariate analysis revealed a correlation between CD8-high and improved survival. Specifically, patients in the CD8-high group of the chemoimmunotherapy cohort experienced enhanced survival compared to those in the chemotherapy cohort, which was attributed to ICI addition. IHC subtype analysis demonstrated a survival advantage in the SCLC-I and SCLC-A groups when ICI was combined with chemotherapy compared to chemotherapy alone.

CONCLUSION

Our study highlights the predictive value of IHC-classified subtypes and CD8-positive cell infiltration in estimating outcomes for patients with ED-SCLC treated with chemoimmunotherapy as a first-line therapy. These findings have practical implications for daily clinical assessments and treatment decisions.

Review of Molecular Technologies for Investigating Canine Cancer

Genetic molecular testing is starting to gain traction as part of standard clinical practice for dogs with cancer due to its multi-faceted benefits, such as potentially being able to provide diagnostic, prognostic and/or therapeutic information. However, the benefits and ultimate success of genomic analysis in the clinical setting are reliant on the robustness of the tools used to generate the results, which continually expand as new technologies are developed. To this end, we review the different materials from which tumour cells, DNA, RNA and the relevant proteins can be isolated and what methods are available for interrogating their molecular profile, including analysis of the genetic alterations (both somatic and germline), transcriptional changes and epigenetic modifications (including DNA methylation/acetylation and microRNAs). We also look to the future and the tools that are currently being developed, such as using artificial intelligence (AI) to identify genetic mutations from histomorphological criteria. In summary, we find that the molecular genetic characterisation of canine neoplasms has made a promising start. As we understand more of the genetics underlying these tumours and more targeted therapies become available, it will no doubt become a mainstay in the delivery of precision veterinary care to dogs with cancer.

Deep Learning for Lung Cancer Diagnosis, Prognosis and Prediction Using Histological and Cytological Images: A Systematic Review

Lung cancer is one of the deadliest cancers worldwide, with a high incidence rate, especially in tobacco smokers. Lung cancer accurate diagnosis is based on distinct histological patterns combined with molecular data for personalized treatment. Precise lung cancer classification from a single H&E slide can be challenging for a pathologist, requiring most of the time additional histochemical and special immunohistochemical stains for the final pathology report. According to WHO, small biopsy and cytology specimens are the available materials for about 70% of lung cancer patients with advanced-stage unresectable disease. Thus, the limited available diagnostic material necessitates its optimal management and processing for the completion of diagnosis and predictive testing according to the published guidelines. During the new era of Digital Pathology, Deep Learning offers the potential for lung cancer interpretation to assist pathologists' routine practice. Herein, we systematically review the current Artificial Intelligence-based approaches using histological and cytological images of lung cancer. Most of the published literature centered on the distinction between lung adenocarcinoma, lung squamous cell carcinoma, and small cell lung carcinoma, reflecting the realistic pathologist's routine. Furthermore, several studies developed algorithms for lung adenocarcinoma predominant architectural pattern determination, prognosis prediction, mutational status characterization, and PD-L1 expression status estimation.

Atlas of PD-L1 for Pathologists: Indications, Scores, Diagnostic Platforms and Reporting Systems

Background. Innovative drugs targeting the PD1/PD-L1 axis have opened promising scenarios in modern cancer therapy. Plenty of assays and scoring systems have been developed for the evaluation of PD-L1 immunohistochemical expression, so far considered the most reliable therapeutic predictive marker. Methods. By gathering the opinion of acknowledged experts in dedicated fields of pathology, we sought to update the currently available evidence on PD-L1 assessment in various types of tumors. Results. Robust data were progressively collected for several anatomic districts and leading international agencies to approve specific protocols: among these, TPS with 22C3, SP142 and SP263 clones in lung cancer; IC with SP142 antibody in breast, lung and urothelial tumors; and CPS with 22C3/SP263 assays in head and neck and urothelial carcinomas. On the other hand, for other malignancies, such as gastroenteric neoplasms, immunotherapy has been only recently introduced, often for particular histotypes, so specific guidelines are still lacking. Conclusions. PD-L1 immunohistochemical scoring is currently the basis for allowing many cancer patients to receive properly targeted therapies. While protocols supported by proven data are already available for many tumors, dedicated studies and clinical trials focusing on harmonization of the topic in other still only partially explored fields are surely yet advisable.