Pre-Amyloidosis Red-Flag Clinical Diagnoses in Light Chain (AL) Versus Age-Related Transthyretin (ATTRwt) Amyloidosis: Electronic Health Record–Based Descriptive Study

Introduction

Amyloidosis represents diseases caused by misfolded protein deposits with multiorgan involvement. The two main systemic types are light chain amyloidosis (AL), caused by a disordered plasma cell clone, and age-related amyloidosis, also known as wild-type transthyretin amyloidosis (ATTRwt), from misfolded transthyretin protein [1]. Diagnosis is often delayed due to slow, heterogeneous presentations, leading patients through multiple specialists. Older adults and Black individuals are disproportionately affected by underdiagnosis and delays [2-5]. Although Food and Drug Administration–approved therapies exist, neither condition is curable; early AL diagnosis improves outcomes, and timely ATTRwt recognition enhances treatment access [3,6]. Common AL amyloidosis “red flag” clinical diagnoses include dyspnea, fatigue, neuropathy, renal disease, heart failure, and edema, symptoms that are often nonspecific [2,7,8]. ATTRwt often presents with cardiac, musculoskeletal, and carpal tunnel syndrome symptoms [9]. We compared red-flag claims preceding diagnosis of AL or ATTRwt amyloidosis to describe differences in symptom recognition prior to diagnosis between the two diseases.

Methods

Study Design

Data were collected using the TriNetX real-time platform, which included deidentified patient-level electronic health record (EHR) data from 77 US health care organizations. The AL amyloidosis cohort included adults ≥18 years old with ≥2 International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes (E85.81, E85.4, E85.89, E85.9) [5] between January 1, 2017, and December 31, 2023, and treated with plasma cell–directed chemotherapy; those receiving tafamidis were excluded. The ATTRwt cohort included patients with ≥2 occurrences of E85.82 in the same period and tafamidis treatment. Those receiving chemotherapy were excluded. Prespecified red-flag diagnoses were identified using ICD-CM-9/10 codes (Table 1).

Ethical Considerations

This project does not meet the federal definition of human subjects’ research thus did not require institutional review board oversight or approval.

Results

We identified 12,090 patients with AL amyloidosis and 2614 with ATTRwt (Table S1 in Multimedia Appendix 1). Patients with AL were younger (mean age 68, SD 11.8 versus 77, SD 8.1 years; P<.001), while the ATTRwt cohort had a higher proportion of males (2149/2614, 82% vs 6355/12,090, 53%; z=2.85; P<.001). After propensity score matching by age, sex, race, and type 2 diabetes mellitus (Figure 1, Table 1), clonal plasma cell disorders were more common in AL: monoclonal gammopathy of undetermined significance (550/2503, 22% vs 220/2503, 9%; z=−12.93; P<.001) and smoldering multiple myeloma (720/2503, 29% vs 53/2503, 2%; z=−26.09, P<.001). Cardiac red-flag diagnoses were more common among patients with ATTRwt. Heart failure was present in 72% (1795/2503) of patients with ATTRwt versus 45% (1114/2503) of patients with AL (z=19.51; P<.001). Atrial fibrillation followed a similar pattern (1388/2503, 56% vs 920/2503, 37%; z=13.27; P<.001).

‎

Renal involvement was higher in AL. Chronic kidney disease was observed in 43% (1087/2503) of patients with AL versus 37% (914/2503) of patients with ATTRwt any time prior (z=−4.99, P<.001). Proteinuria (422/2503, 17% vs 161/2503, 6%; z=−11.50) and nephrotic syndrome (141/2503, 6% vs <10/2503, 1%; z=−10.94) were more common in AL (P<.001). Gastrointestinal symptoms were more frequent in AL, including dysphagia (460/2503, 18% vs 258/2503, 10%; z=−8.15), constipation (27%; n=687/2503 vs 18%; n=444/2503; z=−8.21), nausea/vomiting (508/2503, 20% vs 304/2503, 12%; z=−7.82), and diarrhea (485/2503, 19% vs 270/2503, 11%; z=−8.49; P<.001 for all; Figure 1). Neurologic red flags like unspecified polyneuropathies were also higher in AL (526/2503, 21% vs 421/2503, 17%; z=−3.79; P<.001). Musculoskeletal conditions were uncommon; arthropathy was slightly more common in ATTRwt but not significantly different. Spinal stenosis was higher in ATTRwt overall (493/2503, 20% vs 424/2503, 17%; z=2.52; P=.01). Various multisystemic manifestations were more frequent in AL, such as edema (853/2503, 34% AL vs 788/2503, 32%; z=−1.96) and dizziness/giddiness (693/2503, 28% vs 588/2503, 24%; z=−3.40; P=.001 for both); however, dyspnea was more common in ATTRwt (1460/2503, 58% vs 1212/2503, 48%; z=7.03; P<.001). General symptoms like malaise and fatigue were more common in patients with AL (988/2503, 40% and 757/2503, 30%; z=−6.85; P<.001). Finally, carpal tunnel syndrome was more prevalent in ATTRwt (708/2503, 28% vs 303/2503, 12%; z=14.26; P<.001). Overall patterns remained consistent before and after matching, with a few exceptions that were significant prior to matching but that lost significance upon matching (male erectile dysfunction – unspecified, pleural effusion, liver disorders, injury of muscle); in addition, two diagnoses became significant after matching (dizziness and glaucoma). Findings remained consistent at 6 months, 1 year, and 5 years prior to amyloidosis diagnosis (Table S2 in Multimedia Appendix 1).

Discussion

Our findings highlight distinct clinical patterns of diagnoses prior to a diagnosis of AL and ATTRwt. Patients with AL commonly presented with clonal diseases and renal, gastrointestinal, and neurologic manifestations, while those with ATTRwt predominantly presented with cardiac precursors like heart failure, atrial fibrillation, dyspnea, and carpal tunnel syndrome, suggesting these as key differences [1,7,9]. Despite diagnostic delays, many symptoms were identified and documented in EHRs prior to diagnosis, suggesting the potential for further work in using these data toward identifying patients with amyloidosis sooner. While our findings and prior research support this approach in both AL and ATTR [2,7,10], further work is needed to study whether EHR alerts can reduce diagnostic timelines, especially when compared to appropriate controls without amyloidosis. Strengths of our study include a large, diverse dataset and the use of propensity score matching to adjust for baseline differences. Our descriptive analysis is limited by a reliance on ICD codes alone, which may miss subclinical manifestations and depends on health care recognition of red-flag diagnoses. Additionally, the sensitivity and specificity of ICD codes for AL and ATTRwt are unknown. For example, 10% of those with ATTRwt had prior monoclonal gammopathy of undetermined significance recognition, when the actual prevalence of monoclonal gammopathy of undetermined significance in patients with ATTRwt is much higher. Additionally, given the need to include treatment for AL and ATTRwt after diagnosis, we likely missed a proportion of cases—for example, those with AL who were too sick to get chemotherapy and died before treatment initiation, and noncardiac ATTRwt where tafamidis is not approved for use. We treated all red flags equally, though some are likely more clinically indicative of amyloidosis than others. Next steps include integrating diagnostic patterns with laboratory tests, imaging, and clinical notes to develop EHR-based diagnostic tools and define time intervals between red-flag presentation and amyloidosis diagnosis.