The stakes are higher than many people realize. Cardiac amyloidosis comes in several forms, the most common being ATTR amyloidosis, in which the protein transthyretin becomes unstable and deposits in the heart, and AL amyloidosis, which arises from abnormal antibody-producing cells. For decades the condition was considered rare and essentially untreatable, but that picture has changed dramatically. Today there are approved disease-modifying therapies that can slow or halt the buildup of amyloid fibrils, and more drugs are in late-stage development. These treatments work best when started early, before irreversible damage stiffens the heart and traps patients in a downward spiral of heart failure. The bottleneck is no longer therapy but diagnosis: confirmatory tests exist, yet clinicians struggle to decide which of the many patients with suspicious symptoms should be sent for them.

The confirmatory test itself is remarkably good. Bone scintigraphy, a nuclear medicine scan in which a radioactive tracer binds to amyloid deposits and lights them up on imaging, can detect transthyretin cardiac amyloidosis with high accuracy and, when combined with a blood test for monoclonal protein, can often establish the diagnosis without an invasive biopsy. The problem is upstream. Because the disease mimics hypertensive heart disease, aortic stenosis, and hypertrophic cardiomyopathy, many patients never get referred for the scan at all. Studies have repeatedly shown that diagnostic delays of several years are common, and that a meaningful fraction of patients are only identified once the disease is advanced. The new study set out to attack exactly this referral gap using data that hospitals already collect every day.

The researchers, led by Clemens Spielvogel and Christian Nitsche of Medical University of Vienna together with collaborators at University Hospital Essen, assembled an unusually large dataset. They included 11,616 consecutive patients, meaning every single patient referred for bone scintigraphy at Vienna General Hospital between 2010 and 2023, regardless of why they were sent. This all-comer design matters because it reflects the messy reality of clinical practice rather than a curated population. From each patient’s electronic health record, the team extracted 50 routinely available parameters: demographics, laboratory values, echocardiographic measurements, diagnoses, medications, and referral details. Nothing exotic, no specialized biomarkers or genetic panels, just the digital breadcrumbs that any hospital system generates as a byproduct of ordinary care.

The ground truth for training was the scintigraphy result itself. Patients whose scans showed Perugini grade 2 or 3 uptake, meaning substantial tracer retention in the heart that is suggestive of amyloidosis, were labeled positive. Of the more than eleven thousand patients, 388, or about 3 percent, fell into this high-grade category, underscoring how needle-in-a-haystack the problem is. Patients referred before August 2020 formed the development cohort on which the machine learning model was trained, while those referred afterward served as an untouched internal validation set. A separate external validation followed at University Hospital Essen in Germany with 1,521 additional patients, testing whether the model’s logic transferred to a different country, a different scanner fleet, and a different patient mix.

The results were strikingly consistent. Amylo-Detect achieved an area under the receiver operating characteristic curve, or AUC, of 0.93 in the development cohort, 0.91 in internal validation, and 0.91 in external validation. For context, an AUC of 0.5 represents random guessing while 1.0 is perfect discrimination, and values above 0.9 are considered excellent for a diagnostic screening tool. Crucially, the model outperformed both an existing clinical scoring system and the judgment embedded in routine referral practice. The performance held up across patient subgroups, and even when crucial predictors were missing from the record, a common occurrence in real-world data, the model degraded gracefully rather than collapsing. That robustness is what separates a promising algorithm from something a hospital can actually deploy.

Perhaps the most sobering number in the study concerns the patients that clinical routine missed. Of the 388 patients with CA-suggestive scan results, 42, or 10.8 percent, had been flagged by no one; their scintigraphy had been ordered for other reasons, and their amyloidosis was an incidental discovery. Amylo-Detect caught 12 of those 42 overlooked cases, roughly 29 percent of the missed patients, using nothing but their existing health record data. In other words, the algorithm rescued nearly a third of the people who would otherwise have slipped through the cracks entirely, people whose disease might have progressed untreated for years. Each of those rescued diagnoses represents a patient who could now be considered for therapies that change the course of the disease.

The model’s value extended beyond diagnosis into prognosis. When the researchers examined long-term outcomes, Amylo-Detect’s risk score carried significant predictive weight for mortality and for hospitalization due to heart failure. This makes biological sense: the same features that hint at occult amyloidosis, such as unexplained ventricular wall thickening, characteristic laboratory patterns, and cardiac symptoms, also track with overall cardiovascular risk. A single number that simultaneously identifies patients needing confirmatory testing and stratifies their prognosis could help clinicians prioritize referrals, sending the highest-risk patients for scintigraphy first in systems where imaging capacity is limited and waiting lists are long.

The team has made the tool available as a web application, allowing physicians to enter patient parameters and receive a risk estimate, and inviting further evaluation by other centers. That openness is notable in a field where many algorithms remain locked inside proprietary systems. Still, the authors are careful about what the model is and is not. Amylo-Detect does not diagnose amyloidosis; it identifies patients at risk who should be referred for confirmatory scintigraphy and monoclonal protein testing. And because it was developed and validated retrospectively, on patients who already had scans, its true impact on outcomes remains to be proven. The decisive test will be prospective: deploying the tool in live clinical workflows and measuring whether it actually shortens diagnostic delays and gets more patients onto effective treatment earlier.

Even with those caveats, the study offers a template for how machine learning could chip away at diagnostic inertia across medicine. Rather than chasing exotic new biomarkers, Amylo-Detect demonstrates that the data hospitals already possess, when combined thoughtfully with a well-defined clinical question and rigorous external validation, can surface hidden disease in populations that are being seen every day but not being seen clearly. Cardiac amyloidosis is far more common than the textbooks long assumed, and experts believe a large reservoir of undiagnosed patients remains in every health system. If tools like this one can pull even a fraction of those patients into the light, the payoff in prevented heart failure and extended lives could be substantial, and the same playbook could be applied to other stealth conditions waiting in the margins of the medical record.

Subject of Research: Machine learning screening for cardiac amyloidosis risk using electronic health record data

Article Title: Screening for patients at risk for cardiac amyloidosis via electronic health records: A multicenter machine learning development and validation study

Article References: Spielvogel, C. P., Kersting, D., Haberl, D., Autherith, M., Hauptmann, L., Yu, J., Hennenberg, J., Kluge, K., Moon, K., Settelmeier, S., Ning, J., Kumpf, K., Köfler, M., Hofer, F., Mascherbauer, K., Mascherbauer, J., Kammerlander, A., Traub-Weidinger, T., Kasprian, G., … Nitsche, C. (2026). Screening for patients at risk for cardiac amyloidosis via electronic health records: A multicenter machine learning development and validation study. PLOS Digital Health, 5(10), e0001637. https://doi.org/10.1371/journal.pdig.0001637

Image Credits: AI Generated

DOI: 10.1371/journal.pdig.0001637

Keywords: cardiac amyloidosis, machine learning, electronic health records, bone scintigraphy, ATTR amyloidosis, diagnostic delay, risk prediction, external validation, heart failure, nuclear medicine, clinical decision support, PLOS Digital Health

Mass spectrometry-based amyloid typing has earned a reputation as the gold standard for identifying the amyloidogenic protein in tissue deposits. By shredding laser-microdissected amyloid plaques into peptides and reading their mass spectra, laboratories can distinguish among more than thirty amyloid types with a specificity that immunohistochemistry and immunofluorescence simply cannot match. Yet the technique has remained confined largely to a handful of reference laboratories because it depends on nanoflow liquid chromatography, in which solvents are pushed through capillaries with inner diameters measured in tens of micrometers. Nanoflow systems deliver exquisite sensitivity, but they come with steep up-front costs, demanding maintenance, long column equilibration times, and a level of operational complexity that deters routine clinical adoption.

The Stanford team, led by Morgan W. Mann and Fangjun Chen with senior authors Megan L. Troxell and Ruben Y. Luo, set out to re-engineer the workflow so that it could run on the high-flow liquid chromatography systems already humming away in hospital laboratories. Their solution rests on two complementary innovations: a sample preparation method called suspension trapping, known commercially as S-Trap, and an acquisition strategy known as data-independent acquisition, or DIA. Together, these techniques allow the analysis of microdissected amyloid plaques using conventional-flow chromatography coupled to tandem mass spectrometry, without sacrificing the diagnostic accuracy that clinicians rely upon.

Suspension trapping is the quieter revolution of the two. In traditional proteomics sample preparation, proteins extracted from tiny tissue fragments must be digested into peptides through a series of solution exchanges, buffer removals, and cleanups that consume time and lose precious material at every step. The S-Trap approach instead traps proteins on a quartz filter matrix as they precipitate out of an acidic methanol-containing buffer. Digestion enzymes then pass through the trap, generating peptides that are eluted in a single, efficient step. The method is fast, robust, and minimizes sample loss, which matters enormously when the starting material consists of a few thousand cells harvested by laser capture microdissection from a formalin-fixed paraffin-embedded tissue section.

Data-independent acquisition addresses the other half of the problem. In classical data-dependent acquisition, the mass spectrometer selects the most intense ions in each scan for fragmentation, a stochastic process that can miss low-abundance peptides and produces data that are tedious to align across runs. DIA, by contrast, fragments all ions within successive wide mass windows, systematically recording fragment spectra for essentially everything in the sample. The resulting data are reconstructed computationally against spectral libraries, yielding consistent and quantitative peptide identifications. Because DIA tolerates the higher flow rates, wider chromatographic peaks, and greater sample loads of high-flow chromatography, it made the leap away from nanoflow feasible rather than merely aspirational.

To translate raw proteomic data into a diagnosis, the researchers also developed a custom selection heuristic that sifts through the hundreds of proteins detected in each plaque and identifies the most likely amyloidogenic culprit. Amyloid deposits are not pure; they sweep up serum proteins, immunoglobulin fragments, and extracellular matrix components as they form. Distinguishing the genuine amyloidogenic protein from innocent bystanders requires weighing total protein abundance, the presence of type-specific signature peptides, and the biological plausibility of each candidate. The heuristic automates this judgment, reducing an expert curation task that could take hours into a reproducible computational step suitable for a busy clinical workflow.

The performance data are striking. Working with laser capture microdissected plaques from 47 patient samples, 10 from cardiac biopsies and 37 from renal biopsies, the team split the material into a training set of 25 samples and an independent testing set of 22. The high-flow LC-DIA-MS/MS method identified the amyloidogenic protein in approximately 96 percent of plaques in both sets, a success rate comparable to established nanoflow approaches. The two inaccuracies observed across the cohort both traced to difficulties in identifying immunoglobulin lambda proteins, once in a specimen obtained after treatment had altered the composition of the deposits and once in a plaque containing two distinct amyloid types simultaneously. These edge cases, the authors note, define the boundaries of the method rather than undermining its general utility.

Reducing the sample preparation workflow to under eight hours carries practical significance beyond convenience. Amyloidosis is increasingly recognized as an urgent diagnosis; cardiac involvement can progress rapidly, and new therapies such as transthyretin stabilizers and silencing agents work best when started early. Every hour shaved from the analytical pipeline shortens the time between biopsy and answer. Equally important, the shift to high-flow chromatography means the method can run on instruments of the kind already deployed for clinical chemistry assays, drug monitoring, and newborn screening in hospital laboratories. That compatibility could transform amyloid typing from a rarefied referral test into an accessible diagnostic performed locally, at lower cost and with faster turnaround.

The study, published in Clinical Proteomics, was conducted on remnant patient specimens under approved institutional review board protocols and funded through the Stanford University Department of Pathology’s Test Development Program, underscoring its explicitly clinical orientation. The authors acknowledge one commercial entanglement: a collaborative research relationship between one investigator and Thermo Fisher Scientific, which provided technical support. While the method will still require validation in other laboratories and across broader panels of amyloid types, the combination of suspension trapping, data-independent acquisition, and high-flow chromatography represents a credible path toward democratizing molecular amyloid typing, offering patients everywhere the prospect of a faster, more precise answer to the question their treatment ultimately depends on.

Subject of Research: A high-flow liquid chromatography-tandem mass spectrometry method for typing amyloidosis in clinical laboratories

Article Title: Suspension trapping and data-independent acquisition enable high-flow liquid chromatography-tandem mass spectrometry-based amyloidosis typing in clinical laboratories

Article References: Mann, M. W., Chen, F., Zhu, C., Lu, C., Liang, B., Kambham, N., Troxell, M. L., & Luo, R. Y. (2026). Suspension trapping and data-independent acquisition enable high-flow liquid chromatography-tandem mass spectrometry-based amyloidosis typing in clinical laboratories. Clinical Proteomics. https://doi.org/10.1186/s12014-026-09628-x

Image Credits: AI Generated

DOI: 10.1186/s12014-026-09628-x

Keywords: amyloidosis, mass spectrometry, proteomics, data-independent acquisition, suspension trapping, liquid chromatography, clinical diagnostics, laser capture microdissection, AL amyloidosis, ATTR amyloidosis, tandem mass spectrometry, clinical proteomics