by Christos Evangelou, MSc, PhD – Medical Writer and Editor
In a recent study, researchers at the University of Milano-Bicocca, University and Hospital Trust of Verona, and Spedali Civili Hospital of Brescia developed a new method for detecting renal amyloidosis based on Congo red fluorescence analysis of whole slide images (WSIs).
Compared to the current gold standard of Congo red birefringence, analysis of Congo red fluorescence on WSIs provided excellent interobserver agreement and high sensitivity and specificity in detecting renal amyloidosis.1 These findings suggest that Congo red fluorescence could facilitate digitization and computational analysis of renal amyloidosis, overcoming the limitations of Congo red birefringence.
“This study demonstrates that Congo red fluorescence is a reliable alternative to Congo red birefringence for detecting renal amyloidosis and provides an automated computational approach to enhance the diagnostic performance of the proposed method,” said lead investigator Vincenzo L’Imperio, MD, who is an assistant professor at the University of Milano-Bicocca and nephropathologist at the Fondazione IRCCS San Gerardo in Monza, Italy.
“We believe that Congo red fluorescence analysis of digitized WSIs coupled with computational pathology tools could advance research and diagnosis of renal amyloidosis,” he added.
Study Rationale: Improving Renal Amyloidosis Detection
Renal amyloidosis is caused by the disposition and accumulation of misfolded proteins, also referred to as amyloids, in the kidneys. If not cleared, renal amyloids disrupt the renal structure and function and cause progressive kidney damage that can lead to kidney failure.
In renal pathology, the detection of amyloid is the first step in the characterization of monoclonal gammopathies of renal significance, which may prompt the initiation of chemotherapy for lymphoma or myeloma in the absence of detectable neoplastic clones.
Although the rapid diagnosis of renal amyloidosis is critical, the current gold standard test using Congo red dye and polarized light microscopy has limitations that create bottlenecks for digitization and research.
“Congo red stain is still the gold standard for the detection of amyloid deposits, and its evaluation still requires tissue examination under polarized light using light microscopy,” Dr. L’Imperio explained. “This stain allows the identification of even small amounts of deposits if adequately performed and assessed by experienced pathologists, but the concurrent birefringence of adjacent structures, such as collagen, or non-standardized staining protocols, can lead to false negative or positive results.”
The Problem with Polarized Light
Renal amyloidosis remains challenging to diagnose, hampering efforts for intervention before end-stage renal disease. When pathologists suspect renal amyloidosis, they examine kidney biopsy samples under polarized light after staining them with Congo red dye, which binds to amyloid proteins. This produces an apple-green birefringence pattern that confirms amyloid buildup.
“Our center collects renal biopsies from different spokes on a national level, and the creation of such a network prompted the full conversion towards a renal digital pathology facility, converting light microscopy and immunofluorescence to the WSI format,” said Dr. L’Imperio.
However, polarized light microscopy cannot be easily digitized, hampering efforts to automate the analysis using computational tools. “We wanted to find an alternative approach to the problem,” he said.
In addition, Congo red staining followed by polarized light microscopy fails to differentiate amyloids from similarly birefringent collagen deposits, which are abundant in kidney capsule and scar tissue. Therefore, pathologists struggle to accurately score amyloid severity, particularly in the kidney interstitium. Therefore, misdiagnosis can occur, causing delayed treatment of this progressive disease.
Testing a ‘Brighter’ Approach
To overcome the limitations of polarized light microscopy in the detection of renal amyloids, researchers turned to an alternative technique: analyzing Congo red’s distinctive red fluorescence under Texas red-filtered light in digitized biopsy images. They termed this method ‘Congo red fluorescence on virtual slide’ or simply ‘CRFvs.’ To test the performance of their proposed method, they retrospectively retrieved 154 Congo red-stained renal biopsy slides from their repository and digitized them in dark field using both autofluorescence and Texas red/TRITC filters to obtain Congo red fluorescence WSIs.1 WSIs were scored and evaluated for positivity and distribution of the eventual deposits by two independent pathologists.
“We investigate whether Congo red fluorescence under Texas red/TRITC filter may have equivalent performances when digitized as compared to the ‘conventional’ Congo red birefringence and polarized light microscopy,” noted Dr. L’Imperio.
CRFvs showed near-perfect interobserver agreement with standard Congo red birefringence for detecting amyloidosis(k = 0.90 [95% CI = 0.81–0.98]).1 When consensus-based CRFvs evaluation was taken into account, concordance was even higher (k = 0.98, 95% CI = 0.93–1).
Compared with polarized light microscopy, CRFvs readily differentiated renal amyloids from collagen fibers, enabling more accurate scoring of amyloid severity and more sensitive quantification of interstitial involvement. Two pathologists who independently analyzed the virtual slides consistently achieved higher amyloid scores with CRFvs than with the conventional method.
“We confirmed that Congo red fluorescence had comparable performance with the ‘conventional’ birefringence under polarized light, demonstrating noninferiority in the setting of amyloid detection and unlocking the possibility to digitize this refractory sub-field of nephropathology,” said Prof. Fabio Pagni, one of the senior authors of the study.
Commenting on the reasons contributing to variability in the evaluation of amyloid involvement and extension using validated amyloid scores, he said: “The different nature of the two techniques — light versus dark field — tends to overestimate or underestimate specific compartments, such as the interstitium, and this might have contributed to variability between digital fluorescence and birefringence.”
Streamlining Amyloid Detection Through Computation
Although CRFvs reliably detected renal amyloids in WSIs, background fluorescence sometimes obscured smaller deposits. To address this limitation, the team developed an automated image analysis pipeline called ‘streamlined pipeline for amyloid detection through Congo red fluorescence digital analysis’ or ‘SPADA’ to subtract background autofluorescence and isolate fluorescent amyloid signals.
SPADA processed slides in just over 1 minute per slide, providing enhanced images for quick and accurate diagnosis. When pathologists re-evaluated the biopsies with SPADA, their amyloid scores matched perfectly with those obtained using polarized light.
“By subtracting the true amyloid signal from autofluorescence background, SPADA enabled quick and accurate characterization of even tiny amounts of amyloids,” noted Prof. Federico Alberici, another senior author of the study.
Automatic detection of amyloid deposits using the SPADA pipeline represents a key milestone in the digital transition of nephropathology. Further work is needed to improve the detection of even smaller amounts of amyloids, which would help in the characterization of challenging cases handled by pathologists without renal expertise.
Unlike polarized light imaging, CRFvs generates high-quality digital images amenable to computational analysis. The authors see this as an exciting opportunity to deploy AI for automated amyloid detection and typing. However, future studies are needed to extend the SPADA pipeline to other amyloid diseases.
“SPADA detections can be directly translated as regions of interest for the laser capture microdissection of areas to submit to mass spectrometry tools and to couple innovative in situ proteomics techniques, such as matrix-assisted laser desorption ionization mass spectrometry imaging, which are already demonstrating revolutionizing role in the second step of amyloid diagnosis — the typing of deposits,” said Prof. Giovanni Gambaro, another senior author of the study.
The SPADA pipeline is freely available and compatible with popular open-source software, such as ImageJ and QuPath.
- Cazzaniga G, Bolognesi MM, Stefania MD, et al. Congo red staining in digital pathology: the “SPADA” pipeline. Lab Invest. August 2023:100243. doi:10.1016/j.labinv.2023.100243