Next-Generation Morphometry for pathomics-data mining in histopathology.

Pathology diagnostics relies on the assessment of morphology by trained experts, which remains subjective and qualitative. Here we developed a framework for large-scale histomorphometry (FLASH) performing deep learning-based semantic segmentation and subsequent large-scale extraction of interpretable, quantitative, morphometric features in non-tumour kidney histology. We use two internal and three external, multi-centre cohorts to analyse over 1000 kidney biopsies and nephrectomies. By associating morphometric features with clinical parameters, we confirm previous concepts and reveal unexpected relations. We show that the extracted features are independent predictors of long-term clinical outcomes in IgA-nephropathy. We introduce single-structure morphometric analysis by applying techniques from single-cell transcriptomics, identifying distinct glomerular populations and morphometric phenotypes along a trajectory of disease progression. Our study provides a concept for Next-generation Morphometry (NGM), enabling comprehensive quantitative pathology data mining, i.e., pathomics.

The sodium-glucose co-transporter-2 inhibitor ertugliflozin modifies the signature of cardiac substrate metabolism and reduces cardiac mTOR signalling, endoplasmic reticulum stress and apoptosis.

AIM: To investigate cardiac signalling pathways connecting substrate utilization with left ventricular remodelling in a murine pressure overload model. METHODS: Cardiac hypertrophy was induced by transverse aortic constriction surgery in 20-week-old C57BL/6J mice treated with or without the sodium-glucose co-transporter 2 (SGLT2) inhibitor ertugliflozin (225 mg kg-1 chow diet) for 10 weeks. RESULTS: Ertugliflozin improved left ventricular function and reduced myocardial fibrosis. This occurred simultaneously with a fasting-like response characterized by improved glucose tolerance and increased ketone body concentrations. While cardiac insulin signalling was reduced in response to SGLT2 inhibition, AMP-activated protein kinase (AMPK) signalling was increased with induction of the fatty acid transporter cluster of differentiation 36 and phosphorylation of acetyl-CoA carboxylase (ACC). Further, enzymes responsible for ketone body catabolism (ß-hydroxybutyrate dehydrogenase, succinyl-CoA:3-oxoacid-CoA transferase and acetyl-CoA acetyltransferase 1) were induced by SGLT2 inhibition. Ertugliflozin led to more cardiac abundance of fatty acids, tricarboxylic acid cycle metabolites and ATP. Downstream mechanistic target of rapamycin (mTOR) pathway, relevant for protein synthesis, cardiac hypertrophy and adverse cardiac remodelling, was reduced by SGLT2 inhibition, with alleviation of endoplasmic reticulum (ER) stress and unfolded protein response (UPR) providing a potential mechanism for abundant reduced left ventricular apoptosis and fibrosis. CONCLUSION: SGLT2 inhibition reduced left ventricular fibrosis in a murine model of cardiac hypertrophy. Mechanistically, this was associated with reduced cardiac insulin and increased AMPK signalling as a potential mechanism for less cardiac mTOR activation with alleviation of downstream ER stress, UPR and apoptosis.

Deep learning-based classification of kidney transplant pathology: a retrospective, multicentre, proof-of-concept study.

BACKGROUND: Histopathological assessment of transplant biopsies is currently the standard method to diagnose allograft rejection and can help guide patient management, but it is one of the most challenging areas of pathology, requiring considerable expertise, time, and effort. We aimed to analyse the utility of deep learning to preclassify histology of kidney allograft biopsies into three main broad categories (ie, normal, rejection, and other diseases) as a potential biopsy triage system focusing on transplant rejection. METHODS: We performed a retrospective, multicentre, proof-of-concept study using 5844 digital whole slide images of kidney allograft biopsies from 1948 patients. Kidney allograft biopsy samples were identified by a database search in the Departments of Pathology of the Amsterdam UMC, Amsterdam, Netherlands (1130 patients) and the University Medical Center Utrecht, Utrecht, Netherlands (717 patients). 101 consecutive kidney transplant biopsies were identified in the archive of the Institute of Pathology, RWTH Aachen University Hospital, Aachen, Germany. Convolutional neural networks (CNNs) were trained to classify allograft biopsies as normal, rejection, or other diseases. Three times cross-validation (1847 patients) and deployment on an external real-world cohort (101 patients) were used for validation. Area under the receiver operating characteristic curve (AUROC) was used as the main performance metric (the primary endpoint to assess CNN performance). FINDINGS: Serial CNNs, first classifying kidney allograft biopsies as normal (AUROC 0·87 [ten times bootstrapped CI 0·85-0·88]) and disease (0·87 [0·86-0·88]), followed by a second CNN classifying biopsies classified as disease into rejection (0·75 [0·73-0·76]) and other diseases (0·75 [0·72-0·77]), showed similar AUROC in cross-validation and deployment on independent real-world data (first CNN normal AUROC 0·83 [0·80-0·85], disease 0·83 [0·73-0·91]; second CNN rejection 0·61 [0·51-0·70], other diseases 0·61 [0·50-0·74]). A single CNN classifying biopsies as normal, rejection, or other diseases showed similar performance in cross-validation (normal AUROC 0·80 [0·73-0·84], rejection 0·76 [0·66-0·80], other diseases 0·50 [0·36-0·57]) and generalised well for normal and rejection classes in the real-world data. Visualisation techniques highlighted rejection-relevant areas of biopsies in the tubulointerstitium. INTERPRETATION: This study showed that deep learning-based classification of transplant biopsies could support pathological diagnostics of kidney allograft rejection. FUNDING: European Research Council; German Research Foundation; German Federal Ministries of Education and Research, Health, and Economic Affairs and Energy; Dutch Kidney Foundation; Human(e) AI Research Priority Area of the University of Amsterdam; and Max-Eder Programme of German Cancer Aid.

Non-invasive molecular imaging of kidney diseases.

In nephrology, differential diagnosis or assessment of disease activity largely relies on the analysis of glomerular filtration rate, urinary sediment, proteinuria and tissue obtained through invasive kidney biopsies. However, currently available non-invasive functional parameters, and most serum and urine biomarkers, cannot capture intrarenal molecular disease processes specifically. Moreover, although histopathological analyses of kidney biopsy samples enable the visualization of pathological morphological and molecular alterations, they only provide information about a small part of the kidney and do not allow longitudinal monitoring. These limitations not only hinder understanding of the dynamics of specific disease processes in the kidney, but also limit the targeting of treatments to active phases of disease and the development of novel targeted therapies. Molecular imaging enables non-invasive and quantitative assessment of physiological or pathological processes by combining imaging technologies with specific molecular probes. Here, we discuss current preclinical and clinical molecular imaging approaches in nephrology. Non-invasive visualization of the kidneys through molecular imaging can be used to detect and longitudinally monitor disease activity and can therefore provide companion diagnostics to guide clinical trials, as well as the safe and effective use of drugs.

Pro-cachectic factors link experimental and human chronic kidney disease to skeletal muscle wasting programs.

Skeletal muscle wasting is commonly associated with chronic kidney disease (CKD), resulting in increased morbidity and mortality. However, the link between kidney and muscle function remains poorly understood. Here, we took a complementary interorgan approach to investigate skeletal muscle wasting in CKD. We identified increased production and elevated blood levels of soluble pro-cachectic factors, including activin A, directly linking experimental and human CKD to skeletal muscle wasting programs. Single-cell sequencing data identified the expression of activin A in specific kidney cell populations of fibroblasts and cells of the juxtaglomerular apparatus. We propose that persistent and increased kidney production of pro-cachectic factors, combined with a lack of kidney clearance, facilitates a vicious kidney/muscle signaling cycle, leading to exacerbated blood accumulation and, thereby, skeletal muscle wasting. Systemic pharmacological blockade of activin A using soluble activin receptor type IIB ligand trap as well as muscle-specific adeno-associated virus-mediated downregulation of its receptor ACVR2A/B prevented muscle wasting in different mouse models of experimental CKD, suggesting that activin A is a key factor in CKD-induced cachexia. In summary, we uncovered a crosstalk between kidney and muscle and propose modulation of activin signaling as a potential therapeutic strategy for skeletal muscle wasting in CKD.

SARS-CoV-2 RNA screening in routine pathology specimens.

Virus detection methods are important to cope with the SARS-CoV-2 pandemics. Apart from the lung, SARS-CoV-2 was detected in multiple organs in severe cases. Less is known on organ tropism in patients developing mild or no symptoms, and some of such patients might be missed in symptom-indicated swab testing. Here, we tested and validated several approaches and selected the most reliable RT-PCR protocol for the detection of SARS-CoV-2 RNA in patients' routine diagnostic formalin-fixed and paraffin-embedded (FFPE) specimens available in pathology, to assess (i) organ tropism in samples from COVID-19-positive patients, (ii) unrecognized cases in selected tissues from negative or not-tested patients during a pandemic peak, and (iii) retrospectively, pre-pandemic lung samples. We identified SARS-CoV-2 RNA in seven samples from confirmed COVID-19 patients, in two gastric biopsies, one small bowel and one colon resection, one lung biopsy, one pleural resection and one pleural effusion specimen, while all other specimens were negative. In the pandemic peak cohort, we identified one previously unrecognized COVID-19 case in tonsillectomy samples. All pre-pandemic lung samples were negative. In conclusion, SARS-CoV-2 RNA detection in FFPE pathology specimens can potentially improve surveillance of COVID-19, allow retrospective studies, and advance our understanding of SARS-CoV-2 organ tropism and effects.

A collagen-binding protein enables molecular imaging of kidney fibrosis in vivo.

Pathological deposition of collagen is a hallmark of kidney fibrosis. To illustrate this process we employed multimodal optical imaging to visualize and quantify collagen deposition in murine models of kidney fibrosis (ischemia-reperfusion or unilateral ureteral obstruction) using the collagen-binding adhesion protein CNA35. For in vivo imaging, we used hybrid computed tomography-fluorescence molecular tomography and CNA35 labeled with the near-infrared fluorophore Cy7. Upon intravenous injection, CNA35-Cy7 accumulation was significantly higher in fibrotic compared to non-fibrotic kidneys. This difference was not detected for a non-specific scrambled version of CNA35-Cy7. Ex vivo, on kidney sections of mice and patients with renal fibrosis, CNA35-FITC co-localized with fibrotic collagen type I and III, but not with the basement membrane collagen type IV. Following intravenous injection, CNA35-FITC bound to both interstitial and perivascular fibrotic areas. In line with this perivascular accumulation, we observed significant perivascular fibrosis in the mouse models and in biopsy sections from patients with chronic kidney disease using computer-based morphometry quantification. Thus, molecular imaging of collagen using CNA35 enabled specific non-invasive quantification of kidney fibrosis. Collagen imaging revealed significant perivascular fibrosis as a consistent component next to the more commonly assessed interstitial fibrosis. Our results lay the basis for further probe and protocol optimization towards the clinical translation of molecular imaging of kidney fibrosis.

Cellular Origin and Functional Relevance of Collagen I Production in the Kidney.

Background Interstitial fibrosis is associated with chronic renal failure. In addition to fibroblasts, bone marrow-derived cells and tubular epithelial cells have the capacity to produce collagen. However, the amount of collagen produced by each of these cell types and the relevance of fibrosis to renal function are unclear.Methods We generated conditional cell type-specific collagen I knockout mice and used (reversible) unilateral ureteral obstruction and adenine-induced nephropathy to study renal fibrosis and function.Results In these mouse models, hematopoietic, bone marrow-derived cells contributed to 38%-50% of the overall deposition of collagen I in the kidney. The influence of fibrosis on renal function was dependent on the type of damage. In unilateral ureteral obstruction, collagen production by resident fibroblasts was essential to preserve renal function, whereas in the chronic model of adenine-induced nephropathy, collagen production was detrimental to renal function.Conclusions Our data show that hematopoietic cells are a major source of collagen and that antifibrotic therapies need to be carefully considered depending on the type of disease and the underlying cause of fibrosis.

Treatment of Renal Fibrosis-Turning Challenges into Opportunities.

Current treatment modalities are not effective in halting the progression of most CKD. Renal fibrosis is a pathological process common to all CKD and thereby represents an excellent treatment target. A large number of molecular pathways involved in renal fibrosis were identified in preclinical studies, some of them being similar among different organs and some with available drugs in various phases of clinical testing. Yet only few clinical trials with antifibrotic drugs are being conducted in CKD patients. Here we review those clinical trials, focusing on agents with direct antifibrotic effects, with particular focus on pirfenidone and neutralizing antibodies directed against profibrotic growth factors and cell connection proteins. We discuss the potential reasons for the poor translation in treatment of renal fibrosis and propose possible approaches and future developments to improve it, eg, patient selection and companion diagnostics, specific and sensitive biomarkers as novel end points for clinical trials, and drug-targeting and theranostics.

Regardless of etiology, progressive renal disease causes ultrastructural and functional alterations of peritubular capillaries.

Progressive renal diseases are associated with rarefaction of peritubular capillaries, but the ultrastructural and functional alterations of the microvasculature are not well described. To study this, we analyzed different time points during progressive kidney damage and fibrosis in 3 murine models of different disease etiologies. These models were unilateral ureteral obstruction, unilateral ischemia-reperfusion injury, and Col4a3-deficient mice, we analyzed ultrastructural alterations in patient biopsy specimens. Compared with kidneys of healthy mice, we found a significant and progressive reduction of peritubular capillaries in all models analyzed. Ultrastructurally, compared with the kidneys of control mice, focal widening of the subendothelial space and higher numbers of endothelial vacuoles and caveolae were found in fibrotic kidneys. Quantitative analysis showed that peritubular capillary endothelial cells in fibrotic kidneys had significantly and progressively reduced numbers of fenestrations and increased thickness of the cell soma and lamina densa of the capillary basement membrane. Similar ultrastructural changes were also observed in patient's kidney biopsy specimens. Compared with healthy murine kidneys, fibrotic kidneys had significantly increased extravasation of Evans blue dye in all 3 models. The extravasation could be visualized using 2-photon microscopy in real time in living animals and was mainly localized to capillary branching points. Finally, fibrotic kidneys in all models exhibited a significantly greater degree of interstitial deposition of fibrinogen. Thus, peritubular capillaries undergo significant ultrastructural and functional alterations during experimental progressive renal diseases, independent of the underlying injury. Analyses of these alterations could provide read-outs for the evaluation of therapeutic approaches targeting the renal microvasculature.

The role of PDGF-D in healthy and fibrotic kidneys.

Platelet-derived growth factor (PDGF)-D, a specific PDGF receptor ß (PDGFR-ß) ligand, mediates mesangial proliferation in vitro and in vivo. However, its role in renal development, physiology, and fibrosis is relatively unknown. In healthy murine kidneys, PDGF-D was found to be expressed on renal mesenchymal cells (mesangial cells, fibroblasts, and vascular smooth muscle cells). During renal fibrosis, PDGF-D and its receptor PDGFR-ß were markedly and similarly upregulated in both human and murine kidneys on activated mesenchymal cells, but PDGF-D was also expressed de novo in injured renal tubular cells. The functional role of PDGF-D was studied in Pdgfd-/- mice, which showed no obvious spontaneous renal phenotype at a young age or during aging. Compared with wild-type littermates, Pdgfd-/- mice had significantly reduced renal interstitial fibrosis in two models of renal scarring: unilateral ureteral obstruction and unilateral ischemia/reperfusion injury. This was associated with reduced phosphorylation of PDGFR-ß and its downstream mediator p38. Systemic adenoviral overexpression of PDGF-D in healthy mice resulted in increased collagen deposition in the kidney interstitium. Thus, PDGF-D is upregulated in murine and human kidney fibrosis, may mediate renal scarring, and is dispensable for normal kidney development and physiological functions. PDGF-D may be a suitable therapeutic target to combat kidney fibrosis.

IL-6 Trans-Signaling Drives Murine Crescentic GN.

The role of IL-6 signaling in renal diseases remains controversial, with data describing both anti-inflammatory and proinflammatory effects. IL-6 can act via classic signaling, engaging its two membrane receptors gp130 and IL-6 receptor (IL-6R). Alternatively, IL-6 trans-signaling requires soluble IL-6R (sIL-6R) to act on IL-6R-negative cells that express gp130. Here, we characterize the role of both pathways in crescentic nephritis. Patients with crescentic nephritis had significantly elevated levels of IL-6 in both serum and urine. Similarly, nephrotoxic serum-induced nephritis (NTN) in BALB/c mice was associated with elevated serum IL-6 levels. Levels of serum sIL-6R and renal downstream signals of IL-6 (phosphorylated signal transducer and activator of transcription 3, suppressor of cytokine signaling 3) increased over time in this model. Simultaneous inhibition of both IL-6 signaling pathways using anti-IL-6 antibody did not have a significant impact on NTN severity. In contrast, specific inhibition of trans-signaling using recombinant sgp130Fc resulted in milder disease. Vice versa, specific activation of trans-signaling using a recombinant IL-6-sIL-6R fusion molecule (Hyper-IL-6) significantly aggravated NTN and led to increased systolic BP in NTN mice. This correlated with increased renal mRNA synthesis of the Th17 cell cytokine IL-17A and decreased synthesis of resistin-like alpha (RELMalpha)-encoding mRNA, a surrogate marker of lesion-mitigating M2 macrophage subtypes. Collectively, our data suggest a central role for IL-6 trans-signaling in crescentic nephritis and offer options for more effective and specific therapeutic interventions in the IL-6 system.

Serum and urine markers of collagen degradation reflect renal fibrosis in experimental kidney diseases.

BACKGROUND: The extent of renal fibrosis in chronic kidney disease (CKD) is the best predictor for progression of most renal diseases. To date, no established biomarkers of renal fibrosis exist. METHODS: We measured circulating and urinary-specific matrix metalloproteinase (MMP)-generated collagen type I and III degradation fragments (C1M and C3M) and an N-terminal propeptide of collagen III (Pro-C3), as markers of collagen type III production, in three rat models of CKD and fibrosis: renal mass reduction (5/6 nephrectomy), progressive glomerulonephritis (chronic anti-Thy1.1 nephritis) and adenine crystal-induced nephropathy. Healthy rats served as controls. RESULTS: In all three models, the animals developed significant CKD and renal fibrosis. Compared with healthy rats, serum C1M and C3M significantly increased in rats with 5/6 nephrectomy and adenine nephropathy (2- to 3-fold), but not with chronic anti-Thy1.1 nephritis. Urinary C1M and C3M levels increased 9- to 100-fold in all three models compared with controls. Urinary degradation markers correlated closely with renal deposition of collagen type I and type III. Pro-C3 was significantly increased only in the urine of 5/6 nephrectomy rats. CONCLUSIONS: In particular, urinary markers of MMP-driven collagen degradation, rather than collagen production markers, may represent a novel, specific and non-invasive diagnostic approach to assess kidney fibrosis.

Gp130-dependent signaling in the podocyte.

Renal inflammation, in particular glomerular, is often characterized by increased IL-6 levels. The in vivo relevance of IL-6 signaling in glomerular podocytes, which play central roles in most glomerular diseases, is unknown. Here, we show that in normal mice, podocytes express gp130, the common signal-transducing receptor subunit of the IL-6 family of cytokines. Following systemic IL-6 or LPS injection in mice, podocyte IL-6 signaling was evidenced by downstream STAT3 phosphorylation. Next, we generated mice deficient for gp130 in podocytes. Expectedly, these mice exhibited abrogated IL-6 downstream signaling in podocytes. At the age of 40 wk, they did not show spontaneous renal pathology or abnormal renal function. The mice were then challenged using two LPS injury models as well as nephrotoxic serum to induce crescentic nephritis. Under all conditions, circulating IL-6 levels increased markedly and the mice developed the pathological hallmarks of the corresponding injury models such as proteinuria and development of glomerular crescents, respectively. However, despite the capacity of normal podocytes to transduce IL-6 family signals downstream, there were no significant differences between mice bearing the podocyte-specific gp130 deletion and their control littermates in any of these models. In conclusion, under the different conditions tested, gp130 signaling was not a critical component of the (patho-)biology of the podocyte in vivo.

Osteogenesis of heterotopically transplanted mesenchymal stromal cells in rat models of chronic kidney disease.

The current study is based on the hypothesis of mesenchymal stromal cells (MSCs) contributing to soft-tissue calcification and ectopic osteogenesis in chronic kidney disease (CKD). Rat MSCs were transplanted intraperitoneally in an established three-dimensional collagen-based model in healthy control animals and two rat models of CKD and vascular calcification: (1) 5/6 nephrectomy + high phosphorus diet; and (2) adenine nephropathy. As internal controls, collagen gels without MSCs were transplanted in the same animals. After 4 and 8 weeks, MSCs were still detectable and proliferating in the collagen gels (fluorescence-activated cell sorting [FACS] analysis and confocal microscopy after fluorescence labeling of the cells). Aortas and MSC-containing collagen gels in CKD animals showed distinct similarities in calcification (micro-computed tomography [µCT], energy-dispersive X-ray [EDX] analysis, calcium content), induction of osteogenic markers, (ie, bone morphogenic protein 2 [BMP-2], Runt related transcription factor 2 [Runx2], alkaline phosphatase [ALP]), upregulation of the osteocytic marker sclerostin and extracellular matrix remodeling with increased expression of osteopontin, collagen I/III/IV, fibronectin, and laminin. Calcification, osteogenesis, and matrix remodeling were never observed in healthy control animals and non-MSC-containing collagen gels in all groups. Paul Karl Horan 26 (PKH-26)-labeled, 3G5-positive MSCs expressed Runx2 and sclerostin in CKD animals whereas PKH-26-negative migrated cells did not express osteogenic markers. In conclusion, heterotopically implanted MSCs undergo osteogenic differentiation in rat models of CKD-induced vascular calcification, supporting our hypothesis of MSCs as possible players in heterotopic calcification processes of CKD patients.