ABSTRACT:Mycosis fungoides (MF) and Sezary syndrome (SS) are common entities among primary cutaneous lymphomas. Large cell transformation is challenging for diagnosis and therapy. Molecular mechanisms by which these lymphomas undergo this transformation are poorly defined. We studied the immunohistochemical status of p53 in these entities and assessed whether p53 expression could be a useful tool for diagnosis and assessment of transformation. We extracted patients with transformed and untransformed SS or MF from the French Study Group on Cutaneous Lymphoma database between 2014 and 2021, followed in the Rennes University Hospital. An immunohistochemical study of p53 expression was performed on the biopsies sampled as part of routine care. We compared p53 overexpression in the different groups. We included 25 patients with MF, 7 patients with transformed MF (T-MF), 11 patients with SS, and 5 patients with transformed SS (T-SS). Using a cut-off set at 30% expression of neoplastic cells, we noted an overexpression of p53 in T-MF and T-SS compared with nontransformed forms (47% vs. 12%, respectively, P < 0.01) and in MF compared with SS (23% vs. 7%, respectively, P < 0.01). Overexpression of p53 with a cut-off at 30% therefore seems to be a discriminating tool in the differential diagnosis of MF/SS versus their transformed forms as well as the differential diagnosis between MF and SS.

Background VEXAS (Vacuoles, E1 enzyme, X-linked, autoinflammatory, somatic) syndrome is caused by acquired somatic mutations in UBA1. Sweet-syndrome-like skin disorders [and especially histiocytoid Sweet syndrome (HSS)] may be associated with VEXAS syndrome.Objectives To characterize the clinical and histopathological features of HSS in patients with VEXAS syndrome.Methods Skin biopsies with a histological diagnosis of HSS at Rennes University Medical Center (Rennes, France) between October 2011 and January 2022 were reviewed in this study. Sanger sequencing and digital polymerase chain reaction were used to screen skin, blood and bone marrow samples for UBA1 variants, and thus classify patients as having VEXAS syndrome or not. We evaluated the clinical, histological and molecular (UBA1) characteristics of patients with or without VEXAS syndrome.Results We compared 15 skin biopsies from 7 patients found to have VEXAS syndrome and 19 skin biopsies from 15 patients without VEXAS syndrome. Persistent C-reactive protein elevation, macrocytosis, anaemia and haematological malignancies were more prevalent in patients with VEXAS syndrome [6/7 (86%), 6/7 (86%), 7/7 (100%) and 6/7 (86%), respectively] than in patients without [5/14 (36%), 6/15 (40%), 8/15 (53%) and 8/15 (53%), respectively]. These features sometimes appeared after the first skin manifestations, and a UBA1 mutation was found in the skin of five patients with VEXAS syndrome. Dermal infiltration by reniform histiocytoid cells (myeloperoxidase-positive and/or CD163-positive) and a periadnexal distribution were more frequently observed in VEXAS syndrome biopsies [15/15 (100%) and 3/15 (20%), respectively, vs. 11/19 (58%) and 0/19 (0%) in non-VEXAS syndrome biopsies, respectively].Conclusions Our findings might help pathologists to consider a diagnosis of VEXAS syndrome and to initiate early genetic testing. VEXAS (Vacuoles, E1 enzyme, X-linked, autoinflammatory, somatic) syndrome should be considered when the first suggestive clinical manifestations arise and skin histiocytoid Sweet syndrome might be an early manifestation. When associated with a haematological disorder, the observation of histiocytoid Sweet syndrome with infiltrated, immature myeloid cells should prompt genetic screening for UBA1 mutations.

The biopsy Gleason score is an important prognostic marker for prostate cancer patients. It is, however, subject to substantial variability among pathologists. Artificial intelligence (AI)-based algorithms employing deep learning have shown their ability to match pathologists’ performance in assigning Gleason scores, with the potential to enhance pathologists’ grading accuracy. The performance of Gleason AI algorithms in research is mostly reported on common benchmark datasets or within public challenges. In contrast, many commercial algorithms are evaluated in clinical studies, for which data is not publicly released. As commercial AI vendors typically do not publish performance on public benchmarks, comparison between research and commercial AI is difficult. The aim of this study is to evaluate and compare the performance of top-ranked public and commercial algorithms using real-world data.We curated a diverse dataset of whole-slide prostate biopsy images through crowdsourcing, containing images with a range of Gleason scores and from diverse sources. Predictions were obtained from five top-ranked public algorithms from the PANDA challenge and from two commercial Gleason grading algorithms. Additionally, ten pathologists evaluated the data set in a reader study.Overall, the pairwise quadratic weighted kappa among pathologists ranged from 0.777 to 0.916. Both public and commercial algorithms showed high agreement with pathologists, with quadratic kappa ranging from 0.617 to 0.900. Commercial algorithms performed on par or outperformed top public algorithms.

Les carcinomes rénaux intéressent un groupe très hétérogène de tumeurs dont la classification a grandement évolué depuis 1981. La dernière mise à jour de 2022, classe les carcinomes rénaux en six catégories selon leur morphologie ou selon la présence d’altérations moléculaires spécifiques. Le démembrement moléculaire des cancers du rein de morphologie papillaire a permis d’isoler le carcinome rénal papillaire (CRP) d’autres entités de morphologie semblable qui sont maintenant caractérisées par des altérations moléculaires spécifiques comme le carcinome Fumarate Hydratase (FH) déficient, le carcinome avec réarrangement du gène TFE3 ou le carcinome avec altération du gène TFEB. Cette nouvelle classification permet un diagnostic plus précis mais impose de bien connaître les altérations génomiques à rechercher par des techniques immunohistochimiques ou de biologie moléculaire. Dans la nouvelle classification OMS 2022, il n’est plus recommandé de sous-typer les CRP en type 1 et 2 mais de les classer en CRP de bas grade (ISUP 1–2) ou de haut grade (ISUP 3–4) selon leur grade ISUP. Les autres facteurs pronostiques n’ont pas changé comme le stade pTNM, la présence d’emboles vasculaires ou d’un contingent dédifférencié sarcomatoïde ou rhabdoïde. La nécrose n’est actuellement pas reconnue comme un élément de mauvais pronostic pour ce carcinome. Le diagnostic de CRP de haut grade est dorénavant un diagnostic d’élimination et ne pourra être retenu qu’après avoir écarté un carcinome avec réarrangement du gène TFE3, un carcinome avec altération du gène TFEB et un carcinome FH déficient. Le diagnostic de CRP implique pour le clinicien, et ce quel que soit l’âge du patient, de proposer une consultation en oncogénétique afin de dépister un éventuel syndrome génétique associé.

Renal cell carcinomas (RCC) represent a group of heterogeneous tumors whose classification has greatly evolved since 1981. The latest update in 2022 classifies all renal cell carcinomas into six categories according to their morphology or the detection of specific molecular alterations. Molecular disassembly of renal cell carcinomas with papillary features has enabled the identification of new entities characterized by a specific molecular alteration, such as Fumarate Hydratase (FH) deficient RCC, TFE3- rearranged RCC or TFEB-altered RCC. This new classification allows for a more accurate diagnosis but requires a thorough knowledge of the genomic alterations to search for with immunohistochemical or molecular biology techniques. According to the new WHO 2022 classification, papillary renal cell carcinoma (PRC) type 1 or type 2 classification is no longer recommended. A classification based on nucleolar ISUP grade must be preferred: low-grade PRC (ISUP 1-2) or high-grade PRC (ISUP 3-4). The other prognostic factors remain the same: the pTNM stage, lymphovascular invasion, and the presence or absence of dedifferentiated areas referring to sarcomatoid or rhabdoid features. Of note, the presence of necrosis is not currently recognized as a poor prognostic element for this type of carcinoma. The diagnosis of high-grade PRC is from now on a diagnosis of exclusion. It can only be sustained after having ruled out TFE3- rearranged RCC, TFEB- altered RCC, and FH-deficient RCC. For clinicians, the diagnosis of PRC implies suggesting an oncogenetic consultation to screen for an associated genetic tumor syndrome regardless of the patient's age. (c) 2024 Published by Elsevier Masson SAS.

Breast cancer is one of the most prevalent cancers worldwide and pathologists are closely involved in establishing a diagnosis. Tools to assist in making a diagnosis are required to manage the increasing workload. In this context, artificial intelligence (AI) and deep-learning based tools may be used in daily pathology practice. However, it is challenging to develop fast and reliable algorithms that can be trusted by practitioners, whatever the medical center. We describe a patch-based algorithm that incorporates a convolutional neural network to detect and locate invasive carcinoma on breast whole-slide images. The network was trained on a dataset extracted from a reference acquisition center. We then performed a calibration step based on transfer learning to maintain the performance when translating on a new target acquisition center by using a limited amount of additional training data. Performance was evaluated using classical binary measures (accuracy, recall, precision) for both centers (referred to as "test reference dataset" and "test target dataset") and at two levels: patch and slide level. At patch level, accuracy, recall, and precision of the model on the reference and target test sets were 92.1% and 96.3%, 95% and 87.8%, and 73.9% and 70.6%, respectively. At slide level, accuracy, recall, and precision were 97.6% and 92.0%, 90.9% and 100%, and 100% and 70.8% for test sets 1 and 2, respectively. The high performance of the algorithm at both centers shows that the calibration process is efficient. This is performed using limited training data from the new target acquisition center and requires that the model is trained beforehand on a large database from a reference center. This methodology allows the implementation of AI diagnostic tools to help in routine pathology practice.

To elucidate the spectrum of metastatic solid tumors to the testis and their clinicopathologic features. The databases and files of 26 pathology departments from 9 countries on 3 continents were surveyed to identify metastatic solid tumors to the testis and to characterize their clinicopathologic features in detail. We compiled a series of 157 cases of metastatic solid tumors that secondarily involved the testis. The mean patient age at diagnosis was 64 years (range, 12–93 years). Most patients (127/144; 88%) had clinical manifestation of the disease, with testicular mass/nodule (89/127; 70%) being the most common finding. The main mechanism of testicular involvement was metastasis in 154/157 (98%) cases. Bilateral testicular involvement was present in 12/157 (8%) patients. Concurrent or prior extratesticular metastases were present in 78/101 (77%) patients. The diagnosis was made mainly in orchiectomy specimens (150/157; 95%). Different types of carcinomas (138/157; 87%), most commonly adenocarcinoma (72/157; 46%), were the most common malignancies. The most common primary carcinomas included prostatic (51/149; 34%), renal (29/149; 20%), and colorectal (13/149; 9%). Intratubular growth was identified in 13/124 (11%) cases and paratesticular involvement was found in 73/152 (48%) cases. In patients with available follow-up (110/157; 70%), more than half (58/110; 53%) died of disease. In this largest series compiled to date, we found that most secondary tumors of the testis represent metastases from the genitourinary and gastrointestinal tract carcinomas and typically occur in the setting of disseminated disease.

Introduction Cases of osteonecrosis of the jaw have been reported by dental surgeons to the pharmacovigilance center in Rennes, France, occurring among patients treated with palbociclib, a cyclin-dependent kinase 4/6 inhibitor. Although this event was not expected with the drug, a safety signal was raised. Describing a local case series, the aim of our study was to identify specific patterns that might suggest a triggering role for these drugs, and to discuss pathophysiological hypotheses. Materials and methods A retrospective case series of patients exposed to cyclin-dependent kinase 4/6 inhibitors between 2016 and 2020 with a diagnosis of osteonecrosis of the jaw at the Rennes Dental Care Center was analyzed. The descriptive analysis was conducted on patient demographics, breast cancer characteristics, osteonecrosis of the jaw, biological data, and exposure to cyclin-dependent kinase 4/6 inhibitors. Results We identified eight cases, most of them at stages 0–1 (62.5%). Four patients were still exposed to palbociclib at the time of diagnosis and four had discontinued the treatment before the diagnosis. Chronological imputability could not be excluded given the drug’s half-life and the variable intervals of dental monitoring from one patient to another. All patients had at least one dental osteonecrosis risk factor (including dental extraction, dentures, and denosumab exposure at the time of diagnosis). Neutropenia and mucositis were not systematically reported at the time of diagnosis. The anatomopathological characteristics were nonspecific. Conclusion We did not identify a specific pattern that could suggest a triggering role of palbociclib in the development of ONJ.

Deep learning (DL), often called artificial intelligence (AI), has been increasingly used in Pathology thanks to the use of scanners to digitize slides which allow us to visualize them on monitors and process them with AI algorithms. Many articles have focused on DL applied to prostate cancer (PCa). This systematic review explains the DL applications and their performances for PCa in digital pathology. Article research was performed using PubMed and Embase to collect relevant articles. A Risk of Bias (RoB) was assessed with an adaptation of the QUADAS-2 tool. Out of the 77 included studies, eight focused on pre-processing tasks such as quality assessment or staining normalization. Most articles (n = 53) focused on diagnosis tasks like cancer detection or Gleason grading. Fifteen articles focused on prediction tasks, such as recurrence prediction or genomic correlations. Best performances were reached for cancer detection with an Area Under the Curve (AUC) up to 0.99 with algorithms already available for routine diagnosis. A few biases outlined by the RoB analysis are often found in these articles, such as the lack of external validation. This review was registered on PROSPERO under CRD42023418661.

Fumarate hydratase deficient renal cell carcinoma (FH-RCC) is a rare malignant neoplasia caused by constitutive or somatic mutations in the FH gene whose diagnosis is primordial, requiring genetic counselling. Because of histological heterogeneity, such tumors have been in the past misclassified as "type 2 papillary carcinoma", "tubulo-cystic renal cell carcinoma" or "high grade papillary carcinoma". We report here a case of FH deficient renal cell carcinoma (FH-RCC) in a 69 years old patient. Through this observation, we precise the epidemiological and histological aspects and diagnosis criteria of this rare tumor. (c) 2023 Elsevier Masson SAS. All rights reserved.

Context: The advent of immune check inhibitors (ICIs) has tremendously changed the prognosis of metastatic renal cell carcinoma (mRCC), adding an unseen sub-stantial overall survival benefit. These agents could be administered alone or in combination with anti-vascular endothelial growth factor (anti-VEGF) therapies. So far, treatment allocation is based only on clinical stratification risk models.Objective: Herein, we aimed to report the different molecular classifications reported in the first-line treatment of mRCC and discuss the awaited clinical impli-cations in terms of treatment selection.Evidence acquisition: Medline database as well as European Society for Medical Oncology (ESMO)/American Society of Clinical Oncology (ASCO) conference pro-ceedings were searched to identify biomarker studies. Inclusion criteria comprised randomized and nonrandomized clinical trials that included patients treated in the first line of mRCC setting, patients treated with anti-VEGF therapies or ICIs, biolog-ical modeling, and available survival outcomes. Evidence synthesis: Four classification models were identified with subsequent clin-ical implications: Beuselinck model (34 gene signatures), IMmotion150, Hakimi, and JAVELIN 101 model. Tumor profiling shows distinct outcomes when treated with one or other combination. Patients are clustered into two gene signatures: angiogenic and proinflammatory (as per JAVELIN). The first is more likely to respond to therapy that includes anti-VEGF agents, while the best outcomes are obtained with an ICI combination with the second.Conclusions: The findings presented here were mostly derived from ancillary regis-tered studies of new drugs in the setting of mRCC. Further validation is needed, which sets new paradigms for investigation in clinical research based on tumor biology for treatment allocation and not only on clinical stratification tools.

BRCA1 and BRCA2 genes play a crucial role in repairing DNA double-strand breaks through homologous recombination. Their mutations represent a significant proportion of homologous recombination deficiency and are a reliable effective predictor of sensitivity of high-grade ovarian cancer (HGOC) to poly(ADP-ribose) polymerase inhibitors. However, their testing by next-generation sequencing is costly and time-consuming and can be affected by various preanalytical factors. In this study, we present a deep learning classifier for BRCA mutational status prediction from hematoxylin-eosin-safran–stained whole slide images (WSI) of HGOC. We constituted the OvarIA cohort composed of 867 patients with HGOC with known BRCA somatic mutational status from 2 different pathology departments. We first developed a tumor segmentation model according to dynamic sampling and then trained a visual representation encoder with momentum contrastive learning on the predicted tumor tiles. We finally trained a BRCA classifier on more than a million tumor tiles in multiple instance learning with an attention-based mechanism. The tumor segmentation model trained on 8 WSI obtained a dice score of 0.915 and an intersection-over-union score of 0.847 on a test set of 50 WSI, while the BRCA classifier achieved the state-of-the-art area under the receiver operating characteristic curve of 0.739 in 5-fold cross-validation and 0.681 on the testing set. An additional multiscale approach indicates that the relevant information for predicting BRCA mutations is located more in the tumor context than in the cell morphology. Our results suggest that BRCA somatic mutations have a discernible phenotypic effect that could be detected by deep learning and could be used as a prescreening tool in the future.

Les tumeurs testiculaires sont des entités rares en oncologie, et les localisations secondaires de cancers solides le sont plus encore. Nous présentons deux cas issus de notre expérience au service d’anatomie pathologique du CHU de Rennes. Nous avons aussi colligé l’intégralité des cas rapportés dans les bases de données Medline/PubMed de tumeurs testiculaires secondaires non hématopoïètiques, survenues chez l’adulte, hors séries autopsiques, afin de réaliser une étude intégrant les données de la littérature existante sur le sujet. Au total, nous rapportons 98 cas de lésions secondaires testiculaires, issues de primitifs prostatiques (n=38, 38,77 %), colorectaux (n=19, 19,39 %), gastriques (n=12, 12,24 %), rénaux (n=7, 7,14 %), pulmonaires (n=6, 6,12 %) et autres. L’âge médian de survenue est de 66,5 ans. L’incidence des primitifs en cause varie selon le mode de découverte de la lésion testiculaire (prévalence significativement plus importante d’adénocarcinomes prostatiques en cas de primitif connu (p<0,01), prévalence significativement plus importante d’adénocarcinomes colorectaux (p=0,035) et pancréatiques (p=0,002) en cas de primitif inconnu), de même que l’âge au diagnostic supérieur en cas de primitif connu (p=0,007). Les lésions testiculaires secondaires peuvent poser des difficultés diagnostiques au pathologiste en cas de primitif inconnu, et nous proposons des éléments d’orientation diagnostique. Enfin, nous discutons des potentielles voies de dissémination tumorale vers une localisation testiculaire secondaire, illustrées par les deux cas que nous présentons.

Context Computational pathology is a new interdisciplinary field that combines traditional pathology with modern technologies such as digital imaging and machine learning to better understand the diagnosis, prognosis, and natural history of many diseases. Objective To provide an overview of digital and computational pathology and its current and potential applications in renal cell carcinoma (RCC). Evidence acquisition A systematic review of the English-language literature was conducted using the PubMed, Web of Science, and Scopus databases in December 2022 according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (PROSPERO ID: CRD42023389282). Risk of bias was assessed according to the Prediction Model Study Risk of Bias Assessment Tool. Evidence synthesis In total, 20 articles were included in the review. All the studies used a retrospective design, and all digital pathology techniques were implemented retrospectively. The studies were classified according to their primary objective: detection, tumor characterization, and patient outcome. Regarding the transition to clinical practice, several studies showed promising potential. However, none presented a comprehensive assessment of clinical utility and implementation. Notably, there was substantial heterogeneity for both the strategies used for model building and the performance metrics reported. Conclusions This review highlights the vast potential of digital and computational pathology for the detection, classification, and assessment of oncological outcomes in RCC. Preliminary work in this field has yielded promising results. However, these models have not yet reached a stage where they can be integrated into routine clinical practice. Patient summary Computational pathology combines traditional pathology and technologies such as digital imaging and artificial intelligence to improve diagnosis of disease and identify prognostic factors and new biomarkers. The number of studies exploring its potential in kidney cancer is rapidly increasing. However, despite the surge in research activity, computational pathology is not yet ready for widespread routine use.

Background: Intratumor heterogeneity (ITH) is a key feature in clear cell renal cell car-cinomas (ccRCCs) that impacts outcomes such as aggressiveness, response to treatments, or recurrence. In particular, it may explain tumor relapse after surgery in clinically low-risk patients who did not benefit from adjuvant therapy. Recently, single-cell RNA sequencing (scRNA-seq) has emerged as a powerful tool to unravel expression ITH (eITH) and might enable better assessment of clinical outcomes in ccRCC.Objective: To explore eITH in ccRCC with a focus on malignant cells (MCs) and assess its relevance to improve prognosis for low-risk patients.Design, setting, and participants: We performed scRNA-seq on tumor samples from five untreated ccRCC patients ranging from pT1a to pT3b. Data were complemented with a published dataset composed of pairs of matched normal and ccRCC samples.Intervention: Radical or partial nephrectomy on untreated ccRCC patients.Outcome measurements and statistical analysis: Viability and cell type proportions were determined by flow cytometry. Following scRNA-seq, a functional analysis was per-formed and tumor progression trajectories were inferred. A deconvolution approach was applied on an external cohort, and Kaplan-Meier survival curves were estimated with respect to the prevalence of malignant clusters.Results and limitations: We analyzed 54 812 cells and identified 35 cell subpopulations. The eITH analysis revealed that each tumor contained various degrees of clonal diversity. The transcriptomic signatures of MCs in one particularly heterogeneous sample were used to design a deconvolution-based strategy that allowed the risk stratification of 310 low-risk ccRCC patients.Conclusions: We described eITH in ccRCCs, and used this information to establish signif-icant cell population-based prognostic signatures and better discriminate ccRCC patients. This approach has the potential to improve the stratification of clinically low-risk patients and their therapeutic management.Patient summary: We sequenced the RNA content of individual cell subpopulations composed of clear cell renal cell carcinomas and identified specific malignant cells the genetic information of which can be used to predict tumor progression.(c) 2023 European Association of Urology. Published by Elsevier B.V. All rights reserved.

Mismatch repair deficiency (d-MMR)/microsatellite instability (MSI), KRAS, and BRAF mutational status are crucial for treating advanced colorectal cancer patients. Traditional methods like immunohistochemistry or polymerase chain reaction (PCR) can be challenged by artificial intelligence (AI) based on whole slide images (WSI) to predict tumor status. In this systematic review, we evaluated the role of AI in predicting MSI status, KRAS, and BRAF mutations in colorectal cancer. Studies published in PubMed up to June 2023 were included (n = 17), and we reported the risk of bias and the performance for each study. Some studies were impacted by the reduced number of slides included in the data set and the lack of external validation cohorts. Deep learning models for the d-MMR/MSI status showed a good performance in training cohorts (mean AUC = 0.89, [0.74–0.97]) but slightly less than expected in the validation cohort when available (mean AUC = 0.82, [0.63–0.98]). Contrary to the MSI status, the prediction of KRAS and BRAF mutations was less explored with a less robust methodology. The performance was lower, with a maximum of 0.77 in the training cohort, 0.58 in the validation cohort for KRAS, and 0.82 AUC in the training cohort for BRAF.

Le carcinome à cellules rénales fumarate hydratase (FH) déficient est une néoplasie maligne rare causée par des mutations constitutionnelles ou somatiques du gène FH, dont le diagnostic est primordial compte tenu de la nécessité d’un conseil génétique. Du fait d’une hétérogénéité histologique, de telles tumeurs ont pu par le passé être classées à tort en « carcinome papillaire de type 2 », en « carcinome tubulo-kystique » ou en « carcinome papillaire de haut grade ». Nous rapportons ici un cas de carcinome à cellules rénales FH déficient survenu chez une patiente de 69ans. À travers cette observation, nous préciserons les aspects épidémiologiques et histologiques ainsi que la démarche diagnostique à effectuer devant ce type de tumeur rare.

Testis tumors are uncommon in oncology, and testicular metastasis from distant solid tumors are even rarer. We present two cases encountered in our department of pathology in CHU de Rennes, France. Moreover, we collected all reported cases in the Medline/PubMed databases of non-hematopoietic secondary testis tumors in adults, excluding autopsy studies, to propose an integrative study on this topic. In total, we report 98 cases of secondary testis lesions to prostate (n = 38, 38.77 %), colorectal (n = 19, 19.39%), gastric (n = 12, 12.24%), kidney (n = 7, 7.14%), lung (n = 6, 6.12%) and other primary cancers. The median age at diagnosis was 66.5 years. We identified significantly more prostate adenocarcinoma (P < 0.0001) when the primary tumor was known and significantly more colorectal adenocarcinoma (P = 0.035) and pancreatic adenocarcinoma (P = 0.002) when the primary tumor was unknown. The age at diagnosis was older when the primary tumor was known (P = 0.007). We present the challenges for the diagnosis and propose some elements for diagnosis orientation. Finally, we discuss the possible ways of metastatic dissemination from primary site to testis, as illustrated by the two cases we present.(c) 2023 Elsevier Masson SAS. All rights reserved.

Background: Artificial Intelligence (AI)-based Deep Neural Networks (DNNs) can handle a wide range of applications in image analysis, ranging from automated segmentation to diagnostic and prediction. As such, they have revolutionized healthcare, including in the liver pathology field. Objective: The present study aims to provide a systematic review of applications and performances provided by DNN algorithms in liver pathology throughout the Pubmed and Embase databases up to December 2022, for tumoral, metabolic and inflammatory fields. Results: 42 articles were selected and fully reviewed. Each article was evaluated through the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool, highlighting their risks of bias. Conclusions: DNN-based models are well represented in the field of liver pathology, and their applications are diverse. Most studies, however, presented at least one domain with a high risk of bias according to the QUADAS-2 tool. Hence, DNN models in liver pathology present future opportunities and persistent limitations. To our knowledge, this review is the first one solely focused on DNN-based applications in liver pathology, and to evaluate their bias through the lens of the QUADAS2 tool.