BACKGROUND: To explore challenges of liquid-based cytology (LBC) specimens for next-generation sequencing (NGS) in lung adenocarcinoma and evaluate the efficacy of targeted therapy.
METHODS: A retrospective analysis was conducted on the NGS test of 357 cases of advanced lung adenocarcinoma LBC specimens and compared with results of histological specimens to assess the consistency. The impact of tumor cellularity on NGS test results was evaluated. The utility of epidermal growth factor receptor-tyrosine kinase inhibitors (EGFR-TKIs) was collected. Clinical efficacy evaluation was performed and survival curve analysis was conducted using the Kaplan-Meier method.
RESULTS: There were 275 TKI-naive and 82 TKI-treated specimens, the mutation rates of cancer-related genes detected in both groups were similar (86.2% vs. 86.6%). The EGFR mutation rate in the TKI treated group was higher than that in the TKI-naive group (69.5% > 54.9%, P = 0.019). There was no significant difference in the EGFR mutation frequency among different tumor cellularity in the TKI-naive group. However, in the TKI treated group, the frequency of EGFR sensitizing mutation and T790M resistance mutation in specimens with < 20% tumor cellularity was significantly lower than that in specimens with ≥ 20% tumor cellularity. Among 22 cases with matched histological specimens, 72.7% (16/22) of LBC specimens were completely consistent with results of histological specimens. Among 92 patients with EGFR-mutant lung adenocarcinoma treated with EGFR-TKIs in the two cohorts, 88 cases experienced progression, and the median progression-free survival (PFS) was 12.1 months.
CONCLUSIONS: Cytological specimens are important sources for gene detection of advanced lung adenocarcinoma. When using LBC specimens for molecular testing, it is recommended to fully evaluate the tumor cellularity of the specimens.

Background and aims: The percentage of tumor cells (tumor cellularity) in a cancerous tissue has been assumed to correlate with the variant allele fraction (VAF) of an identified pathogenic variant. Many laboratories use the tumor cellularity as part of a quality criteria for specimen processing and clinical reporting. However, a systematic study of such correlation has yet to be shown. We performed a relatively large-scale study to determine whether pathologist-estimated tumor cellularity is correlated with next-generation sequencing (NGS)-derived VAF. Materials and Methods: A total of 1511 non-small cell lung cancer (NSCLC) and colorectal cancer (CRC) specimens, including formalin-fixed paraffin-embedded (FFPE) and fine needle aspirated (FNA) tissues, were analyzed by cancer hotspot NGS. For a given specimen, pathogenic variants of BRAF, EGFR, KRAS, and NRAS were identified and the determined VAFs were correlated with the corresponding tissue tumor cellularity. Results: The coefficient of determination R-squared (R2) values were calculated for each correlation. All R2 values were lower than 0.25, indicating poor correlations. Pathogenic variants were found, not uncommonly, in tumor specimens that carried 10% or lower tumor cellularity. There were no apparent differences of R2 values between the FFPE and FNA specimens. Conclusion: In both NSCLC and CRC, the lack of linear relationship between tumor cellularity and VAF was found across a wide range of tumor cell percentages. Caution should be used when using tumor cellularity to triage specimens for NGS testing. The tumor cellularity should be considered in relation to the limit of detection of the specific assay for the proper interpretation of a negative test result.

In this work, we address the task of tumor cellularity (TC) estimation with a novel framework based on the label distribution learning (LDL) paradigm. We propose a self-ensemble label distribution learning framework (SLDL) to resolve the challenges of existing LDL-based methods, including difficulties for inter-rater ambiguity exploitation, proper and flexible label distribution generation, and accurate TC value recovery. The proposed SLDL makes four main contributions which have been demonstrated to be quite effective in numerous experiments. First, we propose an expertness-aware conditional VAE for diversified single-rater modeling and an attention-based multi-rater fusion strategy that enables effective inter-rater ambiguity exploitation. Second, we propose a template-based label distribution generation method that is tailored for the TC estimation task and constructs label distributions based on the annotation priors. Third, we propose a novel restricted distribution loss, significantly improving the TC value estimation by effectively regularizing the learning with unimodal loss and regression loss. Fourth, to the best of our knowledge, we are the first to simultaneously leverage inter-rater and intra-rater variability to address the label ambiguity issue in the breast tumor cellularity estimation tasks. The experimental results on the public BreastPathQ dataset demonstrate that the SLDL outperforms the existing methods by a large margin and achieves new state-of-the-art results in the TC estimation task. The code will be available from https://github.com/PerceptionComputingLab/ULTRA.

The segmentation and classification of cell nuclei are pivotal steps in the pipelines for the analysis of bioimages. Deep learning (DL) approaches are leading the digital pathology field in the context of nuclei detection and classification. Nevertheless, the features that are exploited by DL models to make their predictions are difficult to interpret, hindering the deployment of such methods in clinical practice. On the other hand, pathomic features can be linked to an easier description of the characteristics exploited by the classifiers for making the final predictions. Thus, in this work, we developed an explainable computer-aided diagnosis (CAD) system that can be used to support pathologists in the evaluation of tumor cellularity in breast histopathological slides. In particular, we compared an end-to-end DL approach that exploits the Mask R-CNN instance segmentation architecture with a two steps pipeline, where the features are extracted while considering the morphological and textural characteristics of the cell nuclei. Classifiers that are based on support vector machines and artificial neural networks are trained on top of these features in order to discriminate between tumor and non-tumor nuclei. Afterwards, the SHAP (Shapley additive explanations) explainable artificial intelligence technique was employed to perform a feature importance analysis, which led to an understanding of the features processed by the machine learning models for making their decisions. An expert pathologist validated the employed feature set, corroborating the clinical usability of the model. Even though the models resulting from the two-stage pipeline are slightly less accurate than those of the end-to-end approach, the interpretability of their features is clearer and may help build trust for pathologists to adopt artificial intelligence-based CAD systems in their clinical workflow. To further show the validity of the proposed approach, it has been tested on an external validation dataset, which was collected from IRCCS Istituto Tumori \"Giovanni Paolo II\" and made publicly available to ease research concerning the quantification of tumor cellularity.

OBJECTIVE: To correlate MRI morphological response patterns with histopathological tumor regression grading system based on tumor cellularity in locally advanced breast cancer (LABC)-treated neoadjuvant with third-generation aromatase inhibitors.
METHODS: Fifty postmenopausal patients with ER-positive/HER-2-negative LABC treated with neoadjuvant letrozole and exemestane given sequentially in an intra-patient cross-over regimen for at least 4 months with MRI response monitoring at baseline as well as after at least 2 and 4 months on treatment. The MRI morphological response pattern was classified into 6 categories: 0/complete imaging response; I/concentric shrinkage; II/fragmentation; III/diffuse; IV/stable; and V/progressive. Histopathological tumor regression was assessed based on the recommendations from The Royal College of Pathologists regarding tumor cellularity.
RESULTS: Following 2 and 4 months with therapy, the most common MRI pattern was pattern II (24/50 and 21/50, respectively). After 4 months on therapy, the most common histopathological tumor regression grade was grade 3 (21/50). After 4 months an increasing correlation is observed between MRI patterns and histopathology. The overall correlation, between the largest tumor diameter obtained from MRI and histopathology, was moderate and positive (r = 0.50, P-value = 2e-04). Among them, the correlation was highest in type IV (r = 0.53).
CONCLUSIONS: The type II MRI pattern \"fragmentation\" was more frequent in the histopathological responder group; and types I and IV in the non-responder group. Type II pattern showed the best endocrine responsiveness and a relatively moderate correlation between sizes obtained from MRI and histology, whereas type IV pattern indicated endocrine resistance but the strongest correlation between MRI and histology.

Purpose: The breast pathology quantitative biomarkers (BreastPathQ) challenge was a grand challenge organized jointly by the International Society for Optics and Photonics (SPIE), the American Association of Physicists in Medicine (AAPM), the U.S. National Cancer Institute (NCI), and the U.S. Food and Drug Administration (FDA). The task of the BreastPathQ challenge was computerized estimation of tumor cellularity (TC) in breast cancer histology images following neoadjuvant treatment. Approach: A total of 39 teams developed, validated, and tested their TC estimation algorithms during the challenge. The training, validation, and testing sets consisted of 2394, 185, and 1119 image patches originating from 63, 6, and 27 scanned pathology slides from 33, 4, and 18 patients, respectively. The summary performance metric used for comparing and ranking algorithms was the average prediction probability concordance (PK) using scores from two pathologists as the TC reference standard. Results: Test PK performance ranged from 0.497 to 0.941 across the 100 submitted algorithms. The submitted algorithms generally performed well in estimating TC, with high-performing algorithms obtaining comparable results to the average interrater PK of 0.927 from the two pathologists providing the reference TC scores. Conclusions: The SPIE-AAPM-NCI BreastPathQ challenge was a success, indicating that artificial intelligence/machine learning algorithms may be able to approach human performance for cellularity assessment and may have some utility in clinical practice for improving efficiency and reducing reader variability. The BreastPathQ challenge can be accessed on the Grand Challenge website.

BACKGROUND: PDAC remains a tumor entity with poor prognosis and a 5-year survival rate below 10%. Recent research has revealed invasive biomarkers, such as distinct molecular subtypes, predictive for therapy response and patient survival. Non-invasive prediction of individual patient outcome however remains an unresolved task.
METHODS: Discrete cellularity regions of PDAC resection specimen (n = 43) were analyzed by routine histopathological work up. Regional tumor cellularity and CT-derived Hounsfield Units (HU, n = 66) as well as iodine concentrations were regionally matched. One-way ANOVA and pairwise t-tests were performed to assess the relationship between different cellularity level in conventional, virtual monoenergetic 40 keV (monoE 40 keV) and iodine map reconstructions.
RESULTS: A statistically significant negative correlation between regional tumor cellularity in histopathology and CT-derived HU from corresponding image regions was identified. Radiological differentiation was best possible in monoE 40 keV CT images. However, HU values differed significantly in conventional reconstructions as well, indicating the possibility of a broad clinical application of this finding.
CONCLUSIONS: In this study we establish a novel method for CT-based prediction of tumor cellularity for in-vivo tumor characterization in PDAC patients.

The present study aimed to elucidate the genetic features of multiple lung cancer (MLC) and identify effective molecular markers for diagnosis using next generation sequencing (NGS). The present data may also inform patient treatment and prognosis. A total of 35 lesions were obtained from 17 patients with MLC. Based on lesion histology and NGS, 13 cases of multiple primary lung cancer (MPLC) were identified and 4 cases were classified as intrapulmonary metastasis (IPM). All 4 patients with IPM exhibited an epidermal growth factor receptor (EGFR) mutation and synchronous mutation of at least one tumor suppressor gene. The frequency and percentage of EGFR mutations, accompanied with tumor suppressor genes, were significantly higher in patients with IPM compared with MPLC. Furthermore, a high EGFR-heterogeneity score and male sex were risk factors of IPM occurrence. There were significant differences in mean EGFR mutation abundance alone, mutations of tumor suppressor genes and mutations of EGFR combined with tumor suppressor genes between patients with adenocarcinoma (ADC) and adenocarcinoma in situ (AIS). In conclusion, histological characteristics combined with genetic alterations may be an effective method for the diagnosis of MPLC and IPM, and NGS may serve as a useful diagnostic tool. MLC exhibited unique molecular characteristics, including higher rates of EGFR mutations, EGFR driver mutations accompanied with tumor suppressor gene mutations and the absence of anaplastic lymphoma kinase mutations, which may help distinguish between patients with MPLC or IPM. The present study hypothesized that the mean frequency of EGFR mutations, mutations of tumor suppressor genes and mutations of both EGFR and tumor suppressor genes may serve an important role in the development of AIS to ADC. The results of the present study highlight the potential underlying mechanisms of lung ADC development, which may assist with future elucidation of effective treatments to prevent the progression of lung cancer.

Desmoplasia is a prominent feature of pancreatic ductal adenocarcinoma (PDA). Stromal desmoplasia reflects the low cellularity that is characteristic of PDA, and it may play a role in PDA chemoresistance. In this retrospective study, we evaluated the relationship between tumor cellularity in resected PDA specimens and long-term patient outcomes.
We retrospectively reviewed the data from 175 patients who underwent PDA resection between January 2010 and December 2015 at Seoul National University Bundang Hospital, and analyzed their clinicopathological features and the relationship between tumor cellularity (high vs low based on a cutoff of 30% cellularity) and patient outcomes.
The high-cellularity group had significantly shorter overall survival (OS) (18.7 months vs 26.6 months, p=0.006) and disease-free survival (11.0 months vs 16.9 months, p=0.031) than the low-cellularity group. Multivariate analysis revealed that high tumor cellularity was an independent risk factor for poor OS (hazard ratio, 2.008; 95% confidence interval, 1.361 to 2.962; p<0.001). Adjuvant therapy improved OS in the low-cellularity group (16.3 months vs 41.3 months, p=0.001) but not in the high-cellularity group (15.9 months vs 24.4 months, p=0.107).
Tumor cellularity in PDA specimens may be a prognostic and predictive biomarker that could aid in identifying patients who would benefit from adjuvant therapy for PDA.

Next-generation sequencing (NGS) has recently been rapidly adopted in the molecular diagnosis of cancer, but it still faces some obstacles. In this study, 665 lung adenocarcinoma samples (558 TKI-naive and 107 TKI-relapsed samples) were interrogated using NGS, and the challenges and possible solutions of subjecting appropriate tissue samples to NGS testing were explored. The results showed that lower frequencies of HER2/BRAF/PIK3CA and acquired EGFR T790M mutations were observed in biopsy samples with <20% tumor cellularity than in those with ≥20%, but there were no significant differences in the frequencies of EGFR or KRAS mutations. Moreover, tumor heterogeneity was assessed by heterogeneity score (HS), which was calculated through multiplying by 2 the mutant allele frequency (MAF) of tumor cells. In TKI-naive samples, intratumor heterogeneity could occur in EGFR, KRAS, HER2, BRAF, and PIK3CA mutant tumors, but the degree was variable. Higher EGFR, but lower BRAF and PIK3CA HS values were observed compared with KRAS HS. In TKI-relapsed samples, analysis of concomitant sensitizing EGFR and T790M MAFs showed that intratumor heterogeneity was common in acquired EGFR T790M mutant tumors. The mutational status between primary and metastatic tumors was usually concordant, but KRAS, HER2, and PIK3CA HS were significantly higher in metastatic tumors than in primary tumors. Additionally, the discordance rate of mutational status in multifocal lung adenocarcinomas diagnosed as equivocal or multiple primary tumors was high. Together, our findings demonstrate that a comprehensive quality assessment is necessary during tissue process to mitigate the challenges of poor tumor cellularity, tumor heterogeneity, and multifocal clonally independent tumors.