Bias reduction using combined stain normalization and augmentation for AI-based classification of histological images

Artificial intelligence (AI)-assisted diagnosis is an ongoing revolution in pathology. However, a frequent drawback of AI models is their propension to make decisions based rather on bias in training dataset than on concrete biological features, thus weakening pathologists' trust in these tools. Technically, it is well known that microscopic images are altered by tissue processing and staining procedures, being one of the main sources of bias in machine learning for digital pathology. So as to deal with it, many teams have written about color normalization and augmentation methods. However, only a few of them have monitored their effects on bias reduction and model generalizability. In our study, two methods for stain augmentation (AugmentHE) and fast normalization (HEnorm) have been created and their effect on bias reduction has been monitored. Actually, they have also been compared to previously described strategies. To that end, a multicenter dataset created for breast cancer histological grading has been used. Thanks to it, classification models have been trained in a single center before assessing its performance in other centers images. This setting led to extensively monitor bias reduction while providing accurate insight of both augmentation and normalization methods. AugmentHE provided an 81% increase in color dispersion compared to geometric augmentations only. In addition, every classification model that involved AugmentHE presented a significant increase in the area under receiving operator characteristic curve (AUC) over the widely used RGB shift. More precisely, AugmentHE-based models showed at least 0.14 AUC increase over RGB shift-based models. Regarding normalization, HEnorm appeared to be up to 78x faster than conventional methods. It also provided satisfying results in terms of bias reduction. Altogether, our pipeline composed of AugmentHE and HEnorm improved AUC on biased data by up to 21.7% compared to usual augmentations. Conventional normalization methods coupled with AugmentHE yielded similar results while being much slower. In conclusion, we have validated an open-source tool that can be used in any deep learning-based digital pathology project on H&E whole slide images (WSI) that efficiently reduces stain-induced bias and later on might help increase pathologists' confidence when using AI-based products.

BCL3 rearrangements in B-cell lymphoid neoplasms occur in two breakpoint clusters associated with different diseases

The t(14;19)(q32;q13) often juxtaposes BCL3 with immunoglobulin heavy chain (IGH) resulting in overexpression of the gene. In contrast to other oncogenic translocations, BCL3 rearrangement (BCL3-R) has been associated with a broad spectrum of lymphoid neoplasms. Here we report an integrative whole-genome sequence, transcriptomic, and DNA methylation analysis of 13 lymphoid neoplasms with BCL3-R. The resolution of the breakpoints at single base-pair revealed that they occur in two clusters at 5' (n=9) and 3' (n=4) regions of BCL3 associated with two different biological and clinical entities. Both breakpoints were mediated by aberrant class switch recombination of the IGH locus. However, the 5' breakpoints (upstream) juxtaposed BCL3 next to an IGH enhancer leading to overexpression of the gene whereas the 3' breakpoints (downstream) positioned BCL3 outside the influence of the IGH and were not associated with its expression. Upstream BCL3-R tumors had unmutated IGHV, trisomy 12, and mutated genes frequently seen in chronic lymphocytic leukemia (CLL) but had an atypical CLL morphology, immunophenotype, DNA methylome, and expression profile that differ from conventional CLL. In contrast, downstream BCL3-R neoplasms were atypical splenic or nodal marginal zone lymphomas (MZL) with mutated IGHV, complex karyotypes and mutated genes typical of MZL. Two of the latter four tumors transformed to a large B-cell lymphoma. We designed a novel fluorescence in situ hybridization assay that recognizes the two different breakpoints and validated these findings in 17 independent tumors. Overall, upstream or downstream breakpoints of BCL3-R are mainly associated with two subtypes of lymphoid neoplasms with different (epi)genomic, expression, and clinicopathological features resembling atypical CLL and MZL, respectively.

Real-world EGFR testing practices for non-small-cell lung cancer by thoracic pathology laboratories across Europe

BACKGROUND

Testing for epidermal growth factor receptor (EGFR) mutations is an essential recommendation in guidelines for metastatic non-squamous non-small-cell lung cancer, and is considered mandatory in European countries. However, in practice, challenges are often faced when carrying out routine biomarker testing, including access to testing, inadequate tissue samples and long turnaround times (TATs).

MATERIALS AND METHODS

To evaluate the real-world EGFR testing practices of European pathology laboratories, an online survey was set up and validated by the Pulmonary Pathology Working Group of the European Society of Pathology and distributed to 64 expert testing laboratories. The retrospective survey focussed on laboratory organisation and daily EGFR testing practice of pathologists and molecular biologists between 2018 and 2021.

RESULTS

TATs varied greatly both between and within countries. These discrepancies may be partly due to reflex testing practices, as 20.8% of laboratories carried out EGFR testing only at the request of the clinician. Many laboratories across Europe still favour single-test sequencing as a primary method of EGFR mutation identification; 32.7% indicated that they only used targeted techniques and 45.1% used single-gene testing followed by next-generation sequencing (NGS), depending on the case. Reported testing rates were consistent over time with no significant decrease in the number of EGFR tests carried out in 2020, despite the increased pressure faced by testing facilities during the COVID-19 pandemic. ISO 15189 accreditation was reported by 42.0% of molecular biology laboratories for single-test sequencing, and by 42.3% for NGS. 92.5% of laboratories indicated they regularly participate in an external quality assessment scheme.

CONCLUSIONS

These results highlight the strong heterogeneity of EGFR testing that still occurs within thoracic pathology and molecular biology laboratories across Europe. Even among expert testing facilities there is variability in testing capabilities, TAT, reflex testing practice and laboratory accreditation, stressing the need to harmonise reimbursement technologies and decision-making algorithms in Europe.

Single-cell RNA sequencing of human non-hematopoietic bone marrow cells reveals a unique set of inter-species conserved biomarkers for native mesenchymal stromal cells

BACKGROUND

Native bone marrow (BM) mesenchymal stem/stromal cells (BM-MSCs) participate in generating and shaping the skeleton and BM throughout the lifespan. Moreover, BM-MSCs regulate hematopoiesis by contributing to the hematopoietic stem cell niche in providing critical cytokines, chemokines and extracellular matrix components. However, BM-MSCs contain a heterogeneous cell population that remains ill-defined. Although studies on the taxonomy of native BM-MSCs in mice have just started to emerge, the taxonomy of native human BM-MSCs remains unelucidated.

METHODS

By using single-cell RNA sequencing (scRNA-seq), we aimed to define a proper taxonomy for native human BM non-hematopoietic subsets including endothelial cells (ECs) and mural cells (MCs) but with a focal point on MSCs. To this end, transcriptomic scRNA-seq data were generated from 5 distinct BM donors and were analyzed together with other transcriptomic data and with computational biology analyses at different levels to identify, characterize and classify distinct native cell subsets with relevant biomarkers.

RESULTS

We could ascribe novel specific biomarkers to ECs, MCs and MSCs. Unlike ECs and MCs, MSCs exhibited an adipogenic transcriptomic pattern while co-expressing genes related to hematopoiesis support and multilineage commitment potential. Furthermore, by a comparative analysis of scRNA-seq of BM cells from humans and mice, we identified core genes conserved in both species. Notably, we identified MARCKS, CXCL12, PDGFRA, and LEPR together with adipogenic factors as archetypal biomarkers of native MSCs within BM. In addition, our data suggest some complex gene nodes regulating critical biological functions of native BM-MSCs together with a preferential commitment toward an adipocyte lineage.

CONCLUSIONS

Overall, our taxonomy for native BM non-hematopoietic compartment provides an explicit depiction of gene expression in human ECs, MCs and MSCs at single-cell resolution. This analysis helps enhance our understanding of the phenotype and the complexity of biological functions of native human BM-MSCs.

[Dasatinib-induced follicular lymphoid hyperplasia, an entity to know]

Follicular lymphoid hyperplasia induced by dasatinib is an entity recently described. It is sometimes difficult to rule out the diagnostic of small B-cell lymphoma. Usually, the node is swollen, with follicular architecture conserved, composed by germinal centers with variable size and shape, with a hight number of mitoses and tingible bodies macrophages inside. Follicular lymphoid hyperplasia is isolated or associated with multiple reactive patterns. The immunohistochemical profil of germinal centers is CD20+, CD10+, BCL6+, BCL2-. Swollen node disappears in a short time after dasatinib discontinuation. Clinicians and pathologists need to be aware of this entity, so as not to avoid mistakenly suspect lymphoma when lymphadenopathy occurs in a patient with chronic myeloid leukemia treated with dasatinib.

Ultrafast Gene Fusion Assessment for Nonsquamous NSCLC

Introduction

Gene fusion testing of ALK, ROS1, RET, NTRK, and MET exon 14 skipping mutations is guideline recommended in nonsquamous NSCLC (NS-NSCLC). Nevertheless, assessment is often hindered by the limited availability of tissue and prolonged next-generation sequencing (NGS) testing, which can protract the initiation of a targeted therapy. Therefore, the development of faster gene fusion assessment is critical for optimal clinical decision-making. Here, we compared two ultrafast gene fusion assays (UFGFAs) using NGS (Genexus, Oncomine Precision Assay, Thermo Fisher Scientific) and a multiplex reverse-transcriptase polymerase chain reaction (Idylla, GeneFusion Assay, Biocartis) approach at diagnosis in a retrospective series of 195 NS-NSCLC cases and five extrapulmonary tumors with a known NTRK fusion.

Methods

A total of 195 NS-NSCLC cases (113 known gene fusions and 82 wild-type tumors) were included retrospectively. To validate the detection of a NTRK fusion, we added five NTRK-positive extrathoracic tumors. The diagnostic performance of the two UFGFAs and standard procedures was compared.

Results

The accuracy was 92.3% and 93.1% for Idylla and Genexus, respectively. Both systems improved the sensitivity for detection by including a 5'-3' imbalance analysis. Although detection of ROS1, MET exon 14 skipping, and RET was excellent with both systems, ALK fusion detection was reduced with sensitivities of 87% and 88%, respectively. Idylla had a limited sensitivity of 67% for NTRK fusions, in which only an imbalance assessment was used.

Conclusions

UFGFA using NGS and reverse-transcriptase polymerase chain reaction approaches had an equal level of detection of gene fusion but with some technique-specific limitations. Nevertheless, UFGFA detection in routine clinical care is feasible with both systems allowing faster initiation of therapy and a broad degree of screening.

Histological Subtypes Drive Distinct Prognostic Immune Signatures in Classical Hodgkin Lymphoma

Despite the success of standard front-line chemotherapy, 20% of classical Hodgkin lymphoma (cHL) patients still relapse or have refractory disease (r/r), and a subset of them die due to disease progression. There is a critical lack of predictive factors for early identification of those r/r patients who may benefit from new therapeutic strategies. This study aimed to evaluate the dynamic expression of 586 immune-related genes in a cohort of 42 cHL patients including 30 r/r cHL after first-line chemotherapy. Gene expression profiling (GEP) using NanoString technology identified a 19-gene immune signature at diagnosis predictive of cHL relapse, but dependent on histological subtypes. Genes related to tumor survival were found upregulated while genes related to B-lineage were downregulated at diagnosis in r/r nodular sclerosis cHL. In contrast to the mixed-cellularity subtype, comparative GEP analyses between paired diagnosis/relapse biopsies of nodular sclerosis cHL showed 118 differentially expressed genes, supporting an immune contexture switch at relapse with upregulation of immunosuppressive cytokines, such as LGALS1 and TGFB1, and downregulation of the T-cell co-stimulatory receptor ICOS. These results indicate that the predictive value of immune signature in cHL is strongly influenced by histological subtype which should be considered when assessing new immunotherapy target strategies.

Phenotypic and Histological Distribution Analysis Identify Mast Cell Heterogeneity in Non-Small Cell Lung Cancer

Simple Summary

During the fight against tumor, some cells of the immune system such as cytotoxic lymphocytes eliminate tumoral cells while others such as tumor-associated macrophages favor tumor development. Mast cells (MCs) are multifaceted immune cells whose role in cancer is still poorly understood. Moreover, MCs are poorly characterized in the context of cancer and their presence in the tumor microenvironment has been reported to be either associated with good or bad prognosis. In this pilot study we characterized tumor-associated MCs (TAMCs) in lung cancer. We showed that TAMCs exhibited a typical phenotype and can be classified in two subsets according to alphaE integrin (CD103) expression. CD103⁺ TAMCs appeared more mature, more prone to interact with CD4⁺ T cells, and located closer to cancer cells than their CD103⁻ counterpart. This study revealed that a high frequency of total TAMC correlated with better overall survival and progression free survival in patients and underlined MC heterogeneity in cancer.

Abstract

Mast cells (MCs) are multifaceted innate immune cells often present in the tumor microenvironment (TME). However, MCs have been only barely characterized in studies focusing on global immune infiltrate phenotyping. Consequently, their role in cancer is still poorly understood. Furthermore, their prognosis value is confusing since MCs have been associated with good and bad (or both) prognosis depending on the cancer type. In this pilot study performed on a surgical cohort of 48 patients with Non-Small Cell Lung Cancer (NSCLC), we characterized MC population within the TME and in matching non-lesional lung areas, by multicolor flow cytometry and confocal microscopy. Our results showed that tumor-associated MCs (TAMCs) harbor a distinct phenotype as compared with MCs present in non-lesional counterpart of the lung. Moreover, we found two TAMCs subsets based on the expression of CD103 (also named alphaE integrin). CD103⁺ TAMCs appeared more mature, more prone to interact with CD4⁺ T cells, and located closer to cancer cells than their CD103⁻ counterpart. In spite of these characteristics, we did not observe a prognosis advantage of a high frequency of CD103⁺ TAMCs, while a high frequency of total TAMC correlated with better overall survival and progression free survival. Together, this study reveals that TAMCs constitute a heterogeneous population and indicates that MC subsets should be considered for patients’ stratification and management in future research.

"Accelerated phase" chronic lymphocytic leukemia: still an intermediate risk disease in the era of targeted therapies

N/A This article is protected by copyright. All rights reserved.

Current and Future Tools for Diagnosis of Kaposi's Sarcoma

Simple Summary

Kaposi’s sarcoma, a rare opportunistic tumor, is observed in four epidemiological conditions (AIDS-related, iatrogenic, endemic or classic KS). Although in most cases KS is an indolent disease, it can be locally aggressive and/or it can invade other organs than the skin, resulting in more severe presentations, especially in patients with severe immunosuppression. There is no consensus on the imaging workup that is necessary for either the initial staging of the disease or the follow-up. Future perspectives include the use of certain non-invasive imaging tools that may help to evaluate the clinical response to treatment, as well as certain new histological markers that may help in guiding the treatment planning for this atypical neoplasm.

Abstract

Kaposi’s sarcoma (KS) is a rare, atypical malignancy associated with immunosuppression and can be qualified as an opportunistic tumor, which responds to immune modulation or restoration. Four different epidemiological forms have been individualized (AIDS-related, iatrogenic, endemic or classic KS). Although clinical examination is sufficient to diagnose cutaneous lesions of KS, additional explorations are necessary in order to detect lesions involving other organs. New histological markers have been developed in recent years concerning the detection of HHV-8 latent or lytic proteins in the lesions, helping to confirm the diagnosis when it is clinically doubtful. More recently, the evaluation of the local immune response has also been shown to provide some guidance in choosing the appropriate therapeutic option when necessary. We also review the indication and the results of conventional radiological imaging and of non-invasive imaging tools such as ¹⁸F-fluoro-deoxy-glucose positron emission tomography, thermography and laser Doppler imaging for the diagnosis of KS and for the follow-up of therapeutic response in patients requiring systemic treatment.

Baseline SUVmax is related to tumor cell proliferation and patient outcome in follicular lymphoma

Follicular lymphoma (FL) is the most common indolent lymphoma. Despite the clear benefit of CD20-based therapy, a subset of FL patients still progress to aggressive lymphoma. Thus, identifying early biomarkers that incorporate PET metrics could be helpful to identify patients with a high risk of treatment failure with Rituximab. We retrospectively included a total of 132 untreated FL patients separated into training and validation cohorts. Optimal threshold of baseline SUVmax was first determined in the training cohort (n=48) to predict progression-free survival (PFS). The PET results were investigated along with the tumor and immune microenvironment, which were determined by immunochemistry and transcriptome studies involving gene set enrichment analyses and immune cell deconvolution, together with the tumor mutation profile. We report that baseline SUVmax >14.5 was associated with poorer PFS than baseline SUVmax ≤14.5 (HR=0.28; p=0.00046). Neither immune T-cell infiltration nor immune checkpoint expression were associated with baseline PET metrics. By contrast, FL samples with Ki-67 staining ≥10% showed enrichment of cell cycle/DNA genes (p=0.013) and significantly higher SUVmax values (p=0.007). Despite similar oncogenic pathway alterations in both SUVmax groups of FL samples, 4 out of 5 cases harboring the infrequent FOXO1 transcription factor mutation were seen in FL patients with SUVmax >14.5. Thus, high baseline SUVmax reflects FL tumor proliferation and, together with Ki-67 proliferative index, can be used to identify patients at risk of early relapse with R-chemotherapy.

Primary B-Cell lymphoma of the uterine cervix presenting with right ureter hydronephrosis: A case report

Primary lymphoma of the uterine cervix is a very infrequent disease, usually affecting perimenopausal women. Symptoms are very similar to other gynecological malignancies, but treatment and prognosis completely differ, as most of these patients have a better survival. This condition has to be suspected in women with recent normal Pap smear test, rapidly growing tumor and initially non-contributory biopsies. We report a case of primary diffuse large-B-cell lymphoma of the uterine cervix mimicking a locally advanced cervical cancer with right ureter hydronephrosis at diagnosis. She was medically managed with a combination of rituximab and chemotherapy with cyclophosphamide, hydroxydaunorubicin, oncovin, and prednisone, associated to methotrexate for central nervous system prophylaxis. We will discuss about the role of combined treatments with surgery and radiotherapy, and the fertility sparing management in young women.

Immunohistochemistry with anti-MAL antibody and RNAscope with MAL probes are complementary techniques for diagnosis of primary mediastinal large B-cell lymphoma

AIMS

Primary mediastinal large B-cell lymphoma (PMBL) diagnosis can be challenging on needle biopsies. Robust techniques are needed to ensure diagnosis of this lymphoma which is highly sensitive to recently developed therapy protocols.

METHODS

In this study, we sought to determine precise PMBL phenotype, compared with diffuse large B-cell lymphoma not otherwise specified, by combining immunohistochemistry with anti-MAL antibody and RNA in situ hybridisation (RNAscope) with specific MAL probes.

RESULTS

The overall MAL positivity level reached 93% (14/15) of cases of PMBL. Among the 15 cases enrolled in the study, 11 were undoubtedly positive for MAL immunostaining whereas 13 were positive by RNA in situ hybridisation. Interestingly, one case that was negative by in situ hybridisation turned out to be positive by immunohistochemistry.

CONCLUSIONS

Taken together, our results demonstrate that in situ detection of both MAL transcripts and protein are complementary and increase the sensitivity and specificity of PMBL diagnosis.

Extracellular ATP and CD39 Activate cAMP-Mediated Mitochondrial Stress Response to Promote Cytarabine Resistance in Acute Myeloid Leukemia

Relapses driven by chemoresistant leukemic cell populations are the main cause of mortality for patients with acute myeloid leukemia (AML). Here, we show that the ectonucleotidase CD39 (ENTPD1) is upregulated in cytarabine-resistant leukemic cells from both AML cell lines and patient samples in vivo and in vitro. CD39 cell-surface expression and activity is increased in patients with AML upon chemotherapy compared with diagnosis, and enrichment in CD39-expressing blasts is a marker of adverse prognosis in the clinics. High CD39 activity promotes cytarabine resistance by enhancing mitochondrial activity and biogenesis through activation of a cAMP-mediated adaptive mitochondrial stress response. Finally, genetic and pharmacologic inhibition of CD39 ecto-ATPase activity blocks the mitochondrial reprogramming triggered by cytarabine treatment and markedly enhances its cytotoxicity in AML cells in vitro and in vivo. Together, these results reveal CD39 as a new residual disease marker and a promising therapeutic target to improve chemotherapy response in AML. SIGNIFICANCE: Extracellular ATP and CD39-P2RY13-cAMP-OxPHOS axis are key regulators of cytarabine resistance, offering a new promising therapeutic strategy in AML.This article is highlighted in the In This Issue feature, p. 1426.

Accurate diagnosis of lymphoma on whole-slide histopathology images using deep learning

Histopathological diagnosis of lymphomas represents a challenge requiring either expertise or centralised review, and greatly depends on the technical process of tissue sections. Hence, we developed an innovative deep-learning framework, empowered with a certainty estimation level, designed for haematoxylin and eosin-stained slides analysis, with special focus on follicular lymphoma (FL) diagnosis. Whole-slide images of lymph nodes affected by FL or follicular hyperplasia were used for training, validating, and finally testing Bayesian neural networks (BNN). These BNN provide a diagnostic prediction coupled with an effective certainty estimation, and generate accurate diagnosis with an area under the curve reaching 0.99. Through its uncertainty estimation, our network is also able to detect unfamiliar data such as other small B cell lymphomas or technically heterogeneous cases from external centres. We demonstrate that machine-learning techniques are sensitive to the pre-processing of histopathology slides and require appropriate training to build universal tools to aid diagnosis.

Exclusive B-cell phenotype of primary prostatic lymphomas: a potential role of chronic prostatitis

AIMS

Primary prostatic lymphomas (PPL) is exceedingly rare. The aim of this study was to investigate the largest series of PPL obtained from a nationwide expert pathologist network, and thus try to understand the pathophysiology of these tumours.

METHODS AND RESULTS

Up to 66 000 lymphoma cases have been collected and submitted for central expert review by the French Lymphopath network. We confirm the low frequency of PPL (n = 77; 0.12%), all cases being of B-cell origin. Diffuse large B-cell lymphoma and small lymphocytic lymphoma were the most frequent subtypes, comprising 31% and 26% of cases respectively, followed by mucosa-associated lymphoid tissue (MALT)/lymphoplasmacytic lymphoma (19%), follicular lymphoma (12%), mantle cell lymphoma (6%), Burkitt lymphoma (4%), and unclassified lymphoma (1%). Clinical data obtained in 25 cases suggests that PPLs are rather indolent tumours. Our hypothesis for B-cell recruitment in the prostatic tissue was derived from the observation in chronic inflammation (prostatitis) of frequent heterotopic proliferation of high endothelial venules (HEVs). The latter are dedicated to lymphocyte entry into secondary lymphoid organs, here putatively driving circulating clonal B-lymphocytes from the blood into the inflamed prostate. This may account for the relatively high incidence of small lymphocytic lymphoma consistently reported in series of primary or secondary prostatic lymphoma. As in other organs or glands, chronic inflammation may promote antigen-dependent intraprostatic MALT lymphoma and diffuse large B-cell lymphoma development.

CONCLUSIONS

PPLs are exclusively of B-cell origin, and chronic inflammation resulting from the proliferation of high endothelial venules could play some role in their development.