RESUMO
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 As commercial AI vendors typically do not publish performance on public benchmarks, comparison between research and commercial AI is difficult.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.
RESUMO
This study aimed to investigate obesity-related glomerulopathy (ORG) at cellular, structural, and transcriptomic levels. Thirty Wistar rats were randomized into two groups: 15 rats were fed with a standard diet (SD-rats), and 15 rats were fed with a high-fat diet (HFD-rats). After 10 weeks, the weight, kidney function, histological features, and transcriptomic changes were assessed. HFD-rats gained significantly more weight (55.8% vs. 29.2%; p < 0.001) and albuminuria (10,384.04 ng/mL vs. 5845.45 ng/mL; p < 0.001) compared to SD-rats. HFD-rats exhibited early stages of ORG, with predominant mesangial matrix increase and podocyte hypertrophy (PH). These lesions correlated with differentially expressed (DE) genes and miRNAs. Functional analysis showed that miR-205, which was DE in both the kidneys and urine of HFD-rats, negatively regulated the PTEN gene, promoting lipid endocytosis in podocytes. The downregulation of PTEN was proved through a higher PTEN/nephrin ratio in the SD-rats and the presence of lipid vacuoles in HFD-podocytes. This study has found a specific targetome of miRNAs and gene expression in early stages of ORG. Also, it emphasizes the potential value of miR-205 as a urinary biomarker for detecting podocyte injury in ORG, offering a tool for early diagnosis, and opening new avenues for future therapeutic research of obesity-related glomerulopathy.
Assuntos
Dieta Hiperlipídica , MicroRNAs , Obesidade , Podócitos , RNA Mensageiro , Ratos Wistar , Animais , MicroRNAs/genética , Obesidade/complicações , Obesidade/genética , Obesidade/metabolismo , Ratos , Dieta Hiperlipídica/efeitos adversos , Masculino , Podócitos/metabolismo , Podócitos/patologia , RNA Mensageiro/genética , RNA Mensageiro/metabolismo , Nefropatias/etiologia , Nefropatias/genética , Nefropatias/metabolismo , Nefropatias/patologia , Perfilação da Expressão Gênica/métodos , Regulação da Expressão Gênica , Transcriptoma , PTEN Fosfo-Hidrolase/genética , PTEN Fosfo-Hidrolase/metabolismoRESUMO
Risk of depression increased in the general population after the COVID-19 pandemic outbreak. By examining the interplay between genetics and individual environmental exposures during the COVID-19 lockdown, we have been able to gain an insight as to why some individuals are more vulnerable to depression, while others are more resilient. This study, conducted on a Spanish cohort of 9218 individuals (COVICAT), includes a comprehensive non-genetic risk analysis, the exposome, complemented by a genomics analysis in a subset of 2442 participants. Depression levels were evaluated using the Hospital Anxiety and Depression Scale. Together with Polygenic Risk Scores (PRS), we introduced a novel score; Poly-Environmental Risk Scores (PERS) for non-genetic risks to estimate the effect of each cumulative score and gene-environment interaction. We found significant positive associations for PERSSoc (Social and Household), PERSLife (Lifestyle and Behaviour), and PERSEnv (Wider Environment and Health) scores across all levels of depression severity, and for PRSB (Broad depression) only for moderate depression (OR 1.2, 95% CI 1.03-1.40). On average OR increased 1.2-fold for PERSEnv and 1.6-fold for PERLife and PERSoc from mild to severe depression level. The complete adjusted model explained 16.9% of the variance. We further observed an interaction between PERSEnv and PRSB showing a potential mitigating effect. In summary, stressors within the social and behavioral domains emerged as the primary drivers of depression risk in this population, unveiling a mitigating interaction effect that should be interpreted with caution.
Assuntos
COVID-19 , Depressão , Expossoma , Interação Gene-Ambiente , Humanos , COVID-19/epidemiologia , COVID-19/psicologia , COVID-19/virologia , Depressão/epidemiologia , Depressão/etiologia , Masculino , Feminino , Pessoa de Meia-Idade , Adulto , Espanha/epidemiologia , SARS-CoV-2/isolamento & purificação , SARS-CoV-2/genética , Idoso , Fatores de Risco , Pandemias , Quarentena/psicologia , Estudos de CoortesRESUMO
Human tissue, which is inherently three-dimensional (3D), is traditionally examined through standard-of-care histopathology as limited two-dimensional (2D) cross-sections that can insufficiently represent the tissue due to sampling bias. To holistically characterize histomorphology, 3D imaging modalities have been developed, but clinical translation is hampered by complex manual evaluation and lack of computational platforms to distill clinical insights from large, high-resolution datasets. We present TriPath, a deep-learning platform for processing tissue volumes and efficiently predicting clinical outcomes based on 3D morphological features. Recurrence risk-stratification models were trained on prostate cancer specimens imaged with open-top light-sheet microscopy or microcomputed tomography. By comprehensively capturing 3D morphologies, 3D volume-based prognostication achieves superior performance to traditional 2D slice-based approaches, including clinical/histopathological baselines from six certified genitourinary pathologists. Incorporating greater tissue volume improves prognostic performance and mitigates risk prediction variability from sampling bias, further emphasizing the value of capturing larger extents of heterogeneous morphology.
Assuntos
Imageamento Tridimensional , Neoplasias da Próstata , Aprendizado de Máquina Supervisionado , Humanos , Masculino , Aprendizado Profundo , Imageamento Tridimensional/métodos , Prognóstico , Neoplasias da Próstata/patologia , Neoplasias da Próstata/diagnóstico por imagem , Microtomografia por Raio-X/métodosRESUMO
In this study, we performed genomic analyses of cell cycle and tumor microenvironment changes during and after ribociclib and letrozole or chemotherapy in the CORALLEEN trial. 106 women with untreated PAM50-defined Luminal B early breast cancers were randomly assigned to receive neoadjuvant ribociclib and letrozole or standard-of-care chemotherapy. Ki67 immunohistochemistry, tumor-infiltrating lymphocytes quantification, and RNA sequencing were obtained from tissue biopsies pre-treatment, on day 14 of treatment, and tumor specimens from surgical resection. Results showed that at surgery, Ki67 and the PAM50 proliferation scores were lower after ribociclib compared to chemotherapy. However, consistent reactivation of tumor cell proliferation from day 14 to surgery was only observed in the ribociclib arm. In tumors with complete cell cycle arrest (CCCA) at surgery, PAM50 proliferation scores were lower in the ribociclib arm compared to chemotherapy (p < 0.001), whereas the opposite was observed with tumor cellularity (p = 0.002). Gene expression signatures (GES) associated with antigen-presenting cells (APCs) and innate immune system activity showed increased expression post-chemotherapy but decreased expression post-ribociclib. Interferon-associated GES had decreased expression with CCCA and increased expression with non-CCCA. Our findings suggest that while both treatment strategies decreased proliferation, the depth and the patterns over time differed by treatment arm. Immunologically, ribociclib was associated with downregulated GES associated with APCs and the innate immune system in Luminal B tumors, contrary to existing preclinical data. Further studies are needed to understand the effect of CDK4/6 inhibition on the tumor cells and microenvironment, an effect which may vary according to tumor subtypes.
RESUMO
BACKGROUND: The immune system has a central role in preventing carcinogenesis. Alteration of systemic immune cell levels may increase cancer risk. However, the extent to which common genetic variation influences blood traits and cancer risk remains largely undetermined. Here, we identify pleiotropic variants and predict their underlying molecular and cellular alterations. METHODS: Multivariate Cox regression was used to evaluate associations between blood traits and cancer diagnosis in cases in the UK Biobank. Shared genetic variants were identified from the summary statistics of the genome-wide association studies of 27 blood traits and 27 cancer types and subtypes, applying the conditional/conjunctional false-discovery rate approach. Analysis of genomic positions, expression quantitative trait loci, enhancers, regulatory marks, functionally defined gene sets, and bulk- and single-cell expression profiles predicted the biological impact of pleiotropic variants. Plasma small RNAs were sequenced to assess association with cancer diagnosis. RESULTS: The study identified 4093 common genetic variants, involving 1248 gene loci, that contributed to blood-cancer pleiotropism. Genomic hotspots of pleiotropism include chromosomal regions 5p15-TERT and 6p21-HLA. Genes whose products are involved in regulating telomere length are found to be enriched in pleiotropic variants. Pleiotropic gene candidates are frequently linked to transcriptional programs that regulate hematopoiesis and define progenitor cell states of immune system development. Perturbation of the myeloid lineage is indicated by pleiotropic associations with defined master regulators and cell alterations. Eosinophil count is inversely associated with cancer risk. A high frequency of pleiotropic associations is also centered on the regulation of small noncoding Y-RNAs. Predicted pleiotropic Y-RNAs show specific regulatory marks and are overabundant in the normal tissue and blood of cancer patients. Analysis of plasma small RNAs in women who developed breast cancer indicates there is an overabundance of Y-RNA preceding neoplasm diagnosis. CONCLUSIONS: This study reveals extensive pleiotropism between blood traits and cancer risk. Pleiotropism is linked to factors and processes involved in hematopoietic development and immune system function, including components of the major histocompatibility complexes, and regulators of telomere length and myeloid lineage. Deregulation of Y-RNAs is also associated with pleiotropism. Overexpression of these elements might indicate increased cancer risk.
Assuntos
Estudo de Associação Genômica Ampla , Neoplasias , Humanos , Feminino , Fenótipo , Locos de Características Quantitativas , Pleiotropia Genética , Neoplasias/genética , Polimorfismo de Nucleotídeo Único , Predisposição Genética para DoençaRESUMO
In recent years, technological advances in tissue preparation, high-throughput volumetric microscopy, and computational infrastructure have enabled rapid developments in nondestructive 3D pathology, in which high-resolution histologic datasets are obtained from thick tissue specimens, such as whole biopsies, without the need for physical sectioning onto glass slides. While 3D pathology generates massive datasets that are attractive for automated computational analysis, there is also a desire to use 3D pathology to improve the visual assessment of tissue histology. In this perspective, we discuss and provide examples of potential advantages of 3D pathology for the visual assessment of clinical specimens and the challenges of dealing with large 3D datasets (of individual or multiple specimens) that pathologists have not been trained to interpret. We discuss the need for artificial intelligence triaging algorithms and explainable analysis methods to assist pathologists or other domain experts in the interpretation of these novel, often complex, large datasets.
Assuntos
Inteligência Artificial , Microscopia , Humanos , Microscopia/métodos , BiópsiaRESUMO
BACKGROUND: Prostate cancer (PCa) was the second most frequent cancer and the fifth leading cause of cancer death among men in 2020. The aim of this study was to analyze trends in the incidence, mortality and survival of PCa in Girona, Spain, over 25 years. METHODS: Population-based study of PCa collected in the Girona Cancer Registry, 1994-2018. Age-adjusted incidence and mortality rates were calculated per 100,000 men-year. Joinpoint regression models were used for trends, calculating the annual percentage changes (APC). Observed and net survival were analyzed using Kaplan-Meier and Pohar-Perme estimations, respectively. RESULTS: A total of 9,846 cases of PCa were registered between 1994-2018. The age-adjusted incidence and mortality rates were 154.7 (95%CI: 151.7 157.8) and 38.9 (95%CI: 37.3 -40.6), respectively. An increased incidence of 6.2% was observed from 1994 to 2003 (95%CI: 4.4 -8.1), and a decrease of -2.7% (95%CI: -3.5 -;-1.9) between 2003 and 2018. Mortality APC was -2.6% (95%CI: -3.3 --2.0). Five-year observed and net survival were 72.8% (95%CI: 71.8 - 73.7) and 87.2% (95%CI: 85.9 - 88.4), respectively. Five-year net survival increased over time from 72.9% (1994-1998) to 91.3% (2014-2018). CONCLUSIONS: The analyses show a clear reduction in PCa incidence rates from 2003 on, along with an increase in overall survival when comparing the earlier period with more recent years.
Assuntos
Segunda Neoplasia Primária , Neoplasias da Próstata , Masculino , Humanos , Incidência , Neoplasias da Próstata/epidemiologia , Sistema de Registros , Espanha/epidemiologia , Taxa de SobrevidaRESUMO
Background: There is little literature comparing intraoperative ultrasound (IOUS) with radio-guided occult lesions localization (ROLL) in nonpalpable invasive tumors in breast conserving surgery (BCS). There is a need to compare these two methods in terms of safety and efficacy. Methods: This is an observational cohort study. All patients treated with BCS for nonpalpable invasive breast cancer using IOUS from March 2016 to March 2020 were included and compared with a historical reference control group operated on using ROLL from March 2013 to March 2017. For each detection method, the ability to locate tumors intraoperatively, tumor and surgical specimen sizes, total resection volume (TRV), optimal resection volume, excess of healthy tissue resected (ETR), margin status, re-excision rate, surgical time, complications and costs were studied. Results: One hundred and fifty-eight were included, 83 with IOUS and 75 with ROLL. The mean tumor size is equivalent in both groups (11.88 mm IOUS vs. 12.29 mm ROLL, P=0.668). TRV is significantly lower with IOUS (24.92 vs. 60.32 cm3, P<0.001), and the ETR is also significantly lower in the IOUS group (21.74 vs. 58.37 cm3, P<0.001). The rate of positive margins did not differ (10.98% vs. 12.16%, P=1), nor did re-excision rate (10.98% vs. 8.11%, P=0.597). Complication rate did not differ (12.2% IOUS vs. 10.81% ROLL, P=0.808). Surgical time was shorter in IOUS (45.5 vs. 57 min, P>0.05). Conclusions: IOUS in BCS for nonpalpable invasive breast cancer is more accurate than ROLL because it decreases excision volumes with the same rate of free margins and re-excision. Also, IOUS is a more efficient and comfortable technique, and just as safe as ROLL.
RESUMO
Prostate cancer prognostication largely relies on visual assessment of a few thinly sectioned biopsy specimens under a microscope to assign a Gleason grade group (GG). Unfortunately, the assigned GG is not always associated with a patient's outcome in part because of the limited sampling of spatially heterogeneous tumors achieved by 2-dimensional histopathology. In this study, open-top light-sheet microscopy was used to obtain 3-dimensional pathology data sets that were assessed by 4 human readers. Intrabiopsy variability was assessed by asking readers to perform Gleason grading of 5 different levels per biopsy for a total of 20 core needle biopsies (ie, 100 total images). Intrabiopsy variability (Cohen κ) was calculated as the worst pairwise agreement in GG between individual levels within each biopsy and found to be 0.34, 0.34, 0.38, and 0.43 for the 4 pathologists. These preliminary results reveal that even within a 1-mm-diameter needle core, GG based on 2-dimensional images can vary dramatically depending on the location within a biopsy being analyzed. We believe that morphologic assessment of whole biopsies in 3 dimension has the potential to enable more reliable and consistent tumor grading.
Assuntos
Próstata , Neoplasias da Próstata , Masculino , Humanos , Próstata/patologia , Biópsia , Neoplasias da Próstata/patologia , Biópsia com Agulha de Grande Calibre , Gradação de TumoresRESUMO
MOTIVATION: Coincidence of Convergent Amino Acid Substitutions (CAAS) with phenotypic convergences allow pinpointing genes and even individual mutations that are likely to be associated with trait variation within their phylogenetic context. Such findings can provide useful insights into the genetic architecture of complex phenotypes. RESULTS: Here we introduce CAAStools, a set of bioinformatics tools to identify and validate CAAS in orthologous protein alignments for predefined groups of species representing the phenotypic values targeted by the user. AVAILABILITY AND IMPLEMENTATION: CAAStools source code is available at http://github.com/linudz/caastools, along with documentation and examples.
Assuntos
Biologia Computacional , Software , Filogenia , Substituição de Aminoácidos , DocumentaçãoRESUMO
Human pigmentation has largely been associated with different disease prevalence among populations, but most of these studies are observational and inconclusive. Known to be genetically determined, pigmentary traits have largely been studied by Genome-Wide Association Study (GWAS), mostly in Caucasian ancestry cohorts from North Europe, identifying robustly, several loci involved in many of the pigmentary traits. Here, we conduct a detailed analysis by GWAS and Polygenic Risk Score (PRS) of 13 pigmentary-related traits in a South European cohort of Caucasian ancestry (n = 20,000). We observed fair phototype strongly associated with non-melanoma skin cancer and other dermatoses and confirmed by PRS-approach the shared genetic basis with skin and eye diseases, such as melanoma (OR = 0.95), non-melanoma skin cancer (OR = 0.93), basal cell carcinoma (OR = 0.97) and darker phototype with vitiligo (OR = 1.02), cataracts (OR = 1.04). Detailed genetic analyses revealed 37 risk loci associated with 10 out of 13 analyzed traits, and 16 genes significantly associated with at least two pigmentary traits. Some of them have been widely reported, such as MC1R, HERC2, OCA2, TYR, TYRP1, SLC45A2, and some novel candidate genes C1QTNF3, LINC02876, and C1QTNF3-AMACR have not been reported in the GWAS Catalog, with regulatory potential. These results highlight the importance of the assess phototype as a genetic proxy of skin functionality and disease when evaluating open mixed populations.
Assuntos
Estudo de Associação Genômica Ampla , Neoplasias Cutâneas , Humanos , Pigmentação da Pele/genética , Polimorfismo de Nucleotídeo Único/genética , Neoplasias Cutâneas/genética , Fatores de RiscoRESUMO
DESTINY-Breast01 (NCT03248492) evaluated trastuzumab deruxtecan (T-DXd; DS-8201) in patients with heavily pretreated HER2-positive metastatic breast cancer (mBC). We present a subgroup of 24 patients with a history of treated brain metastases (BM), a population with limited treatment options. In patients with BMs, the confirmed objective response rate (cORR) was 58.3% [95% confidence interval (CI), 36.6%-77.9%], and the median progression-free survival (mPFS) was 18.1 months (95% CI, 6.7-18.1 months). In patients without BMs (n = 160), cORR was 61.3% and mPFS was 16.4 months. Eight patients (47.1%) experienced a best overall intracranial response of partial response or complete response. Seven patients (41.2%) had a best percentage change in brain lesion diameter from baseline consistent with stable disease. Two patients (8.3%) with BMs and two (1.3%) without BMs experienced progression in the brain. The safety profile of T-DXd was consistent with previous studies. The durable clinical activity of T-DXd in this population warrants further investigation. SIGNIFICANCE: Advances in treating HER2-positive metastatic breast cancer have greatly improved patient outcomes, but intracranial progression remains an important risk for which few therapeutic options are currently available. T-DXd demonstrated durable efficacy in patients with stable, treated BMs. This article is highlighted in the In This Issue feature, p. 2711.
Assuntos
Neoplasias Encefálicas , Neoplasias da Mama , Imunoconjugados , Humanos , Feminino , Neoplasias da Mama/patologia , Receptor ErbB-2 , Trastuzumab/uso terapêutico , Imunoconjugados/uso terapêutico , Camptotecina/uso terapêutico , Neoplasias Encefálicas/tratamento farmacológico , Neoplasias Encefálicas/secundárioRESUMO
The combined analysis of haplotype panels with phenotype clinical cohorts is a common approach to explore the genetic architecture of human diseases. However, genetic studies are mainly based on single nucleotide variants (SNVs) and small insertions and deletions (indels). Here, we contribute to fill this gap by generating a dense haplotype map focused on the identification, characterization, and phasing of structural variants (SVs). By integrating multiple variant identification methods and Logistic Regression Models (LRMs), we present a catalogue of 35 431 441 variants, including 89 178 SVs (≥50 bp), 30 325 064 SNVs and 5 017 199 indels, across 785 Illumina high coverage (30x) whole-genomes from the Iberian GCAT Cohort, containing a median of 3.52M SNVs, 606 336 indels and 6393 SVs per individual. The haplotype panel is able to impute up to 14 360 728 SNVs/indels and 23 179 SVs, showing a 2.7-fold increase for SVs compared with available genetic variation panels. The value of this panel for SVs analysis is shown through an imputed rare Alu element located in a new locus associated with Mononeuritis of lower limb, a rare neuromuscular disease. This study represents the first deep characterization of genetic variation within the Iberian population and the first operational haplotype panel to systematically include the SVs into genome-wide genetic studies.
Assuntos
Genoma Humano , Haplótipos , Mutação INDEL , Aciltransferases , Europa (Continente) , Sequenciamento de Nucleotídeos em Larga Escala , Humanos , Lipase , Polimorfismo de Nucleotídeo Único , Sequenciamento Completo do Genoma/métodosRESUMO
INTRODUCTION: Lymphangioleiomyomatosis (LAM) is a rare low-grade metastasising disease characterised by cystic lung destruction. The genetic basis of LAM remains incompletely determined, and the disease cell-of-origin is uncertain. We analysed the possibility of a shared genetic basis between LAM and cancer, and LAM and pulmonary function. METHODS: The results of genome-wide association studies of LAM, 17 cancer types and spirometry measures (forced expiratory volume in 1â s (FEV1), forced vital capacity (FVC), FEV1/FVC ratio and peak expiratory flow (PEF)) were analysed for genetic correlations, shared genetic variants and causality. Genomic and transcriptomic data were examined, and immunodetection assays were performed to evaluate pleiotropic genes. RESULTS: There were no significant overall genetic correlations between LAM and cancer, but LAM correlated negatively with FVC and PEF, and a trend in the same direction was observed for FEV1. 22 shared genetic variants were uncovered between LAM and pulmonary function, while seven shared variants were identified between LAM and cancer. The LAM-pulmonary function shared genetics identified four pleiotropic genes previously recognised in LAM single-cell transcriptomes: ADAM12, BNC2, NR2F2 and SP5. We had previously associated NR2F2 variants with LAM, and we identified its functional partner NR3C1 as another pleotropic factor. NR3C1 expression was confirmed in LAM lung lesions. Another candidate pleiotropic factor, CNTN2, was found more abundant in plasma of LAM patients than that of healthy women. CONCLUSIONS: This study suggests the existence of a common genetic aetiology between LAM and pulmonary function.
RESUMO
The enormous mammal's lifespan variation is the result of each species' adaptations to their own biological trade-offs and ecological conditions. Comparative genomics have demonstrated that genomic factors underlying both, species lifespans and longevity of individuals, are in part shared across the tree of life. Here, we compared protein-coding regions across the mammalian phylogeny to detect individual amino acid (AA) changes shared by the most long-lived mammals and genes whose rates of protein evolution correlate with longevity. We discovered a total of 2,737 AA in 2,004 genes that distinguish long- and short-lived mammals, significantly more than expected by chance (P = 0.003). These genes belong to pathways involved in regulating lifespan, such as inflammatory response and hemostasis. Among them, a total 1,157 AA showed a significant association with maximum lifespan in a phylogenetic test. Interestingly, most of the detected AA positions do not vary in extant human populations (81.2%) or have allele frequencies below 1% (99.78%). Consequently, almost none of these putatively important variants could have been detected by genome-wide association studies. Additionally, we identified four more genes whose rate of protein evolution correlated with longevity in mammals. Crucially, SNPs located in the detected genes explain a larger fraction of human lifespan heritability than expected, successfully demonstrating for the first time that comparative genomics can be used to enhance interpretation of human genome-wide association studies. Finally, we show that the human longevity-associated proteins are significantly more stable than the orthologous proteins from short-lived mammals, strongly suggesting that general protein stability is linked to increased lifespan.
Assuntos
Estudo de Associação Genômica Ampla , Longevidade , Envelhecimento/genética , Animais , Genômica , Humanos , Longevidade/genética , Mamíferos/genética , FilogeniaRESUMO
Existing tools for post-radical prostatectomy (RP) prostate cancer biochemical recurrence (BCR) prognosis rely on human pathologist-derived parameters such as tumor grade, with the resulting inter-reviewer variability. Genomic companion diagnostic tests such as Decipher tend to be tissue destructive, expensive, and not routinely available in most centers. We present a tissue non-destructive method for automated BCR prognosis, termed "Histotyping", that employs computational image analysis of morphologic patterns of prostate tissue from a single, routinely acquired hematoxylin and eosin slide. Patients from two institutions (n = 214) were used to train Histotyping for identifying high-risk patients based on six features of glandular morphology extracted from RP specimens. Histotyping was validated for post-RP BCR prognosis on a separate set of n = 675 patients from five institutions and compared against Decipher on n = 167 patients. Histotyping was prognostic of BCR in the validation set (p < 0.001, univariable hazard ratio [HR] = 2.83, 95% confidence interval [CI]: 2.03-3.93, concordance index [c-index] = 0.68, median years-to-BCR: 1.7). Histotyping was also prognostic in clinically stratified subsets, such as patients with Gleason grade group 3 (HR = 4.09) and negative surgical margins (HR = 3.26). Histotyping was prognostic independent of grade group, margin status, pathological stage, and preoperative prostate-specific antigen (PSA) (multivariable p < 0.001, HR = 2.09, 95% CI: 1.40-3.10, n = 648). The combination of Histotyping, grade group, and preoperative PSA outperformed Decipher (c-index = 0.75 vs. 0.70, n = 167). These results suggest that a prognostic classifier for prostate cancer based on digital images could serve as an alternative or complement to molecular-based companion diagnostic tests.
RESUMO
BACKGROUND: The presence of invasive cribriform adenocarcinoma (ICC), an expanse of cells containing punched-out lumina uninterrupted by stroma, in radical prostatectomy (RP) specimens has been associated with biochemical recurrence (BCR). However, ICC identification has only moderate inter-reviewer agreement. OBJECTIVE: To investigate quantitative machine-based assessment of the extent and prognostic utility of ICC, especially within individual Gleason grade groups. DESIGN, SETTING, AND PARTICIPANTS: A machine learning approach was developed for ICC segmentation using 70 RP patients and validated in a cohort of 749 patients from four sites whose median year of surgery was 2007 and with median follow-up of 28 mo. ICC was segmented on one representative hematoxylin and eosin RP slide per patient and the fraction of tumor area composed of ICC, the cribriform area index (CAI), was measured. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: The association between CAI and BCR was measured in terms of the concordance index (c index) and hazard ratio (HR). RESULTS AND LIMITATIONS: CAI was correlated with BCR (c index 0.62) in the validation set of 411 patients with ICC morphology, especially those with Gleason grade group 2 cancer (n = 192; c index 0.66), and was less prognostic when patients without ICC were included (c index 0.54). A doubling of CAI in the group with ICC morphology was prognostic after controlling for Gleason grade, surgical margin positivity, preoperative prostate-specific antigen level, pathological T stage, and age (HR 1.19, 95% confidence interval 1.03-1.38; p = 0.018). CONCLUSIONS: Automated image analysis and machine learning could provide an objective, quantitative, reproducible, and high-throughput method of quantifying ICC area. The performance of CAI for grade group 2 cancer suggests that for patients with little Gleason 4 pattern, the ICC fraction has a strong prognostic role. PATIENT SUMMARY: Machine-based measurement of a specific cell pattern (cribriform; sieve-like, with lots of spaces) in images of prostate specimens could improve risk stratification for patients with prostate cancer. In the future, this could help in expanding the criteria for active surveillance.
Assuntos
Próstata , Neoplasias da Próstata , Amarelo de Eosina-(YS) , Hematoxilina , Humanos , Masculino , Recidiva Local de Neoplasia/patologia , Prognóstico , Próstata/patologia , Prostatectomia/métodos , Neoplasias da Próstata/patologia , Neoplasias da Próstata/cirurgiaRESUMO
INTRODUCTION: Sentinel lymph node biopsy (SLNB) in ductal carcinoma in situ (DCIS) is not indicated. However, in certain cases (size >3 cm, high grade, mass effect on mammography, or palpable mass), it may be possible to find incidental invasive carcinoma (IC) that requires an SLNB. We studied the correlation of the aforesaid factors with the probability of finding IC in the surgical specimen. METHODS: Data was collected from 3 different institutions between 2010 and 2016, recording characteristics such as, but not limited to: high grade, size >3 cm, mass effect on mammography, and palpable mass. RESULTS: On the whole, 468 "high-risk" DCIS cases were identified, 139 (29%) of which had IC. When the DCIS was high grade or the size was >3 cm, there was no significant difference in the probability of finding IC in the surgical specimen (OR = 1.13; 95% CI 0.84-1.51; OR = 1.2; 95% CI 0.85-1.40). Nevertheless, when a high grade and size (>3 cm) were combined, IC was more likely to exist (72.7 vs. 27.3%; p = 0.001). In addition, mass effect and palpation were independently associated with a significantly greater degree of IC (OR = 12.76; 95% CI 6.93-23.52). CONCLUSIONS: The results suggest that high-grade DCIS or DCIS with a size >3 cm, independently, does not require SLNB. Nonetheless, in the event that both factors are found in the same case, SLNB may be indicated. Additionally, SLNB is advisable for DCIS cases that are palpable or show a mass effect on mammography.