Dataset Information

A machine learning model for identifying patients at risk for wild-type transthyretin amyloid cardiomyopathy.

ABSTRACT: Transthyretin amyloid cardiomyopathy, an often unrecognized cause of heart failure, is now treatable with a transthyretin stabilizer. It is therefore important to identify at-risk patients who can undergo targeted testing for earlier diagnosis and treatment, prior to the development of irreversible heart failure. Here we show that a random forest machine learning model can identify potential wild-type transthyretin amyloid cardiomyopathy using medical claims data. We derive a machine learning model in 1071 cases and 1071 non-amyloid heart failure controls and validate the model in three nationally representative cohorts (9412 cases, 9412 matched controls), and a large, single-center electronic health record-based cohort (261 cases, 39393 controls). We show that the machine learning model performs well in identifying patients with cardiac amyloidosis in the derivation cohort and all four validation cohorts, thereby providing a systematic framework to increase the suspicion of transthyretin cardiac amyloidosis in patients with heart failure.

SUBMITTER: Huda A

PROVIDER: S-EPMC8113237 | biostudies-literature | 2021 May

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A machine learning model for identifying patients at risk for wild-type transthyretin amyloid cardiomyopathy.

Huda Ahsan A Castaño Adam A Niyogi Anindita A Schumacher Jennifer J Stewart Michelle M Bruno Marianna M Hu Mo M Ahmad Faraz S FS Deo Rahul C RC Shah Sanjiv J SJ

Nature communications 20210511 1

Transthyretin amyloid cardiomyopathy, an often unrecognized cause of heart failure, is now treatable with a transthyretin stabilizer. It is therefore important to identify at-risk patients who can undergo targeted testing for earlier diagnosis and treatment, prior to the development of irreversible heart failure. Here we show that a random forest machine learning model can identify potential wild-type transthyretin amyloid cardiomyopathy using medical claims data. We derive a machine learning mo ...[more]

PMID: 33976166

Similar Datasets

Project description:AimsThe focus on wild-type transthyretin amyloid cardiomyopathy (ATTRwt-CM) is increasing because of novel treatment options. There is currently no report on a large number of Japanese patients with ATTRwt-CM. The study aimed to examine the characteristics and prognosis of ATTRwt-CM in Japan.Methods and resultsConsecutive patients (78.5 ± 6.4 years old at diagnosis) with ATTRwt-CM diagnosed at Kumamoto University Hospital between December 2002 and December 2019 were retrospectively reviewed. Data, including demographic characteristics, co-morbidities, clinical manifestations at diagnosis, laboratory results, electrocardiographic and echocardiographic data, imaging and pathological findings, and treatment were obtained. Of 129 patients included in this study, 110 patients (85%) were male. The median period from initial symptom onset to diagnosis was 15.5 (2-75) months. Heart failure was the most common clinical manifestation leading to diagnosis (61%) and initial manifestations (49%). Of 106 patients, carpal tunnel syndrome was observed in 57 patients (54%), and the median period from initial symptom onset to diagnosis was 96 (48-120) months. Histopathological confirmation of transthyretin amyloid was achieved in 94 patients (73%), including 66 (51%) and 28 cases (22%) with endomyocardial and extracardiac biopsies. During the observation period (median 15.0 [inter-quartile range, 5.4-33.2] months after diagnosis), 34 patients (26%) died. Of these, 27 patients (79%) had cardiovascular deaths (heart failure, 25; sudden death, two). The median survival duration was 58.9 months and the 5 years' survival rate was 48%. According to a multivariate Cox hazard analysis, age [hazard ratio (HR), 1.14; 95% confidence interval (CI), 1.05-1.23, P = 0.002] and low serum sodium levels (HR, 0.89; 95% CI, 0.79-0.996; P = 0.04) contributed to all-cause mortality, and low serum sodium levels contributed to hospitalization for heart failure (HR, 0.86; 95% CI, 0.77-0.96; P = 0.005).ConclusionsClinical characteristics and prognosis of ATTRwt-CM patients in Japan were examined. Carpal tunnel syndrome can be considered an indication for diagnosis of ATTRwt-CM. Age and low serum sodium level were significant predictive factors of all survival outcomes. The clinical features of ATTRwt-CM should be recognized to provide appropriate treatment.

Project description:Wild-type transthyretin amyloid cardiomyopathy (ATTRwt CM) is a more common disease than previously thought. Awareness of ATTRwt CM and its diagnosis has been challenged by its unspecific and widely distributed clinical manifestations and traditionally invasive diagnostic tools. Recent advances in echocardiography and cardiac magnetic resonance (CMR), non-invasive diagnosis by bone scintigraphy, and the development of disease-modifying treatments have resulted in an increased interest, reflected in multiple publications especially during the last decade. To get an overview of the scientific knowledge and gaps related to patient entry, suspicion, diagnosis, and systematic screening of ATTRwt CM, we developed a framework to systematically map the available evidence of (i) when to suspect ATTRwt CM in a patient, (ii) how to diagnose the disease, and (iii) which at-risk populations to screen for ATTRwt CM. Articles published between 2010 and August 2021 containing part of or a full diagnostic pathway for ATTRwt CM were included. From these articles, data for patient entry, suspicion, diagnosis, and screening were extracted, as were key study design and results from the original studies referred to. A total of 50 articles met the inclusion criteria. Of these, five were position statements from academic societies, while one was a clinical guideline. Three articles discussed the importance of primary care providers in terms of patient entry, while the remaining articles had the cardiovascular setting as point of departure. The most frequently mentioned suspicion criteria were ventricular wall thickening (44/50), carpal tunnel syndrome (42/50), and late gadolinium enhancement on CMR (43/50). Diagnostic pathways varied slightly, but most included bone scintigraphy, exclusion of light-chain amyloidosis, and the possibility of doing a biopsy. Systematic screening was mentioned in 16 articles, 10 of which suggested specific at-risk populations for screening. The European Society of Cardiology recommends to screen patients with a wall thickness ≥12 mm and heart failure, aortic stenosis, or red flag symptoms, especially if they are >65 years. The underlying evidence was generally good for diagnosis, while significant gaps were identified for the relevance and mutual ranking of the different suspicion criteria and for systematic screening. Conclusively, patient entry was neglected in the reviewed literature. While multiple red flags were described, high-quality prospective studies designed to evaluate their suitability as suspicion criteria were lacking. An upcoming task lies in defining and evaluating at-risk populations for screening. All are steps needed to promote early detection and diagnosis of ATTRwt CM, a prerequisite for timely treatment.

Project description:AimsThis study was performed to investigate whether left atrial (LA) strain by echocardiography provides prognostic information in patients with wild-type transthyretin amyloid cardiomyopathy (ATTRwt-CM).Methods and resultsAmong 129 patients who were diagnosed with ATTRwt-CM at Kumamoto University Hospital from December 2002 to December 2019, 113 patients who had enough information for two-dimensional speckle tracking echocardiography were enrolled in this study. During a median follow-up of 668 days, 28 cardiovascular deaths occurred. Compared with patients in the non-event group, those in the cardiovascular death group were significantly older (81.5 ± 7.4 vs. 78.1 ± 6.1 years, P < 0.01), had a lower incidence of carpal tunnel syndrome (21% vs. 47%, P < 0.05), and had a higher high-sensitivity cardiac troponin T [0.085 (0.063-0.105) vs. 0.049 (0.036-0.079) ng/mL, P < 0.01] and B-type natriuretic peptide concentrations [419 (239-541) vs. 271 (155-462) pg/mL, P < 0.01] and lower estimated glomerular filtration rate (41.8 ± 15.4 vs. 53.4 ± 14.6 mL/min/1.73 m2 , P < 0.01). Electrocardiography showed higher rate of a V1-V3 QS pattern (52% vs. 24%, P < 0.01) and complete left bundle branch block (27% vs. 6%, P < 0.01), and echocardiography showed a significantly lower peak LA strain rate during the contraction phase (0.16 ± 0.13 vs. 0.28 ± 0.27 S-1 , P < 0.05), LA strain during the reservoir phase (LASr) (5.84 ± 2.41 vs. 8.22 ± 4.05%, P < 0.01), and peak LA strain rate during the reservoir phase (0.26 ± 0.09 vs. 0.33 ± 0.15 S-1 , P < 0.05) in the cardiovascular death group than in non-event group. By contrast, conventional echocardiographic findings were not significantly different between these two groups. After adjusting for conventional predictive factors of ATTRwt-CM (age, high-sensitivity cardiac troponin T and B-type natriuretic peptide concentrations, and estimated glomerular filtration rate), multivariable Cox proportional hazard analyses showed that LASr was significantly and independently associated with cardiovascular death in patients with ATTRwt-CM (odds ratio, 0.84; 95% confidence interval, 0.72-0.98; P < 0.05). After adjusting for age and echocardiographic findings associated with cardiovascular death (LA volume index and peak LA strain rate during the contraction phase), LASr was significantly and independently associated with cardiovascular death in patients with ATTRwt-CM (odds ratio, 0.83; 95% confidence interval, 0.70-0.98; P < 0.05). Receiver operating characteristic curve analysis showed that the area under the curve of LASr for cardiovascular death was 0.686 and that the best cut-off value of LASr was 6.69% (sensitivity, 62.4%; specificity, 64.3%). In the Kaplan-Meier analysis, patients with low LASr (<6.69%) had a significantly higher probability of total cardiovascular death (P < 0.05) and heart failure-related hospitalization (P < 0.05).ConclusionsLeft atrial strain during the reservoir phase provides significant prognostic value in patients with ATTRwt-CM even after adjusting for conventional predictive factors.

Dataset Information

A machine learning model for identifying patients at risk for wild-type transthyretin amyloid cardiomyopathy.

Publications

A machine learning model for identifying patients at risk for wild-type transthyretin amyloid cardiomyopathy.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets