Bornhorst, Joshua A.

This cross-sectional study evaluates the performance of the US Food and Drug Administration–cleared phosphorylated tau 217 (p-tau217) to β-amyloid 42 ratio cut points used to detect amyloid pathology associated with Alzheimer disease.

Abstract Background Artificial intelligence (AI) augmented laboratory tests can improve quality, efficiency, cost-effectiveness, and staff satisfaction. However, the clinical success of these tests can introduce unforeseen challenges for model retraining. This study describes the discovery of an “AI label feedback loop” in a clinically implemented AI-augmented kidney stone composition test. Methods An AI-augmented kidney stone composition test (V1) has been previously deployed for clinical kidney stone characterization. After several years of clinical use, a retrained model (V2) was developed using 6 times more data. Model performance of both V1 and V2 were evaluated across 3 datasets: a recent production validation (hold-out) set (mostly AI-influenced labels), the original V1 validation set (pre-AI, entirely human-labeled), and a subset of recent cases with exclusively human-generated or human-corrected labels. Results V2 demonstrated a 10% lower concordance rate than V1 when evaluated on the recent production hold-out set, despite a much larger training dataset. Performance between V1 and V2 was similar when applied to the pre-AI validation set. Notably, V2 outperformed V1 on the recent subset of cases with human-only or human-corrected labels, particularly for less-common stone types. These findings revealed an AI label feedback loop, confounding retraining and evaluation. Conclusions The integration of AI into clinical practice can potentially influence reported test results, complicating the development and evaluation of future models. To mitigate AI label feedback loops, ongoing human annotation and careful validation set construction are essential. These strategies can ensure reliable performance assessment and support the safe evolution of clinical AI systems.

Abstract INTRODUCTION Plasma phosphorylated tau at threonine 217 (p‐tau217) has emerged as one of the most promising blood‐based biomarkers for Alzheimer's disease (AD). While rare, heterophilic antibodies (HAb) are a persistent potential confounding factor in immunoassays. METHODS Potential HAb interference in the Lumipulse G pTau217 assay in samples exhibiting concentrations > 10 pg/mL were investigated. Samples were subjected to HAb blocking reagent (HBT), and in some cases, polyethylene glycol (PEG) precipitation and serial dilutions. RESULTS In 14 of the 15 suspected HAb cases, HBT treatment greatly reduced measured p‐tau217 concentrations. and in one of three specimens selected for further investigation, reduced Aβ42 concentrations PEG precipitation in the three selected samples also reduced p‐tau217 concentrations, while serial dilution yielded mixed effects. Control specimens were unaffected by sample treatment or dilutions. DISCUSSION HAb may result in falsely elevated p‐tau217 and p‐tau217/Aβ42 ratio. HAb interference should be considered in cases of unusually high p‐tau217 concentrations prior to clinical interpretation. Highlights Potential heterophile antibody (HAb) interference in the Lumipulse p‐tau217 immunoassay was investigated. Investigation was performed using HAb blocking reagent, and in some cases also using PEG precipitation . Patients with p‐tau217 concentrations > 10 pg/mL exhibited positive HAb interference in 14 of 15 cases. HAb interference can lead to falsely positive p‐tau217/Aβ42 ratios.

Abstract Background Cognitive impairment (CI) has diverse neurological etiologies, including Alzheimer’s Disease (AD), non-degenerative (ND) conditions, and cerebrospinal fluid (CSF) dynamic disorders, such as normal pressure hydrocephalus (NPH). The etiology of CI may be different to diagnose due to the overlap of non-specific symptoms. Various CSF biomarkers, such as amyloid beta (Aβ) 42, Aβ40, total Tau (t-Tau), and phosphorylated Tau 181 (p-Tau181), have been evaluated for their diagnostic utility in identifying the underlying cause for CI. Previous studies have suggested an isolated abnormal Aβ42 levels may be indicative of NPH in the appropriate clinical context. This study aims to further explore the clinical utility of AD-related CSF biomarkers to distinguish NPH from AD and ND causes of CI. Methods 240 patients were chart reviewed and categorized as either AD (n = 124), ND (n = 68), or NPH (n = 47) based on the etiologies of CI. Concentrations of Aβ40, Aβ42, t-Tau, p-Tau181 were measured in CSF (Cobas e601, Roche). Receiver operating characteristic (ROC) curves were plotted and optimal Youden cutoffs and areas under curve (AUCs) were calculated for each biomarker, and for selected ratios of biomarker combinations, for their ability to differentiate NPH from either AD or ND. Additionally, qualitative (abnormal/normal) result patterns for Aβ42, t-Tau, p-Tau181, and p-Tau181/Aβ42 ratio were reviewed for each patient, using established cut-offs for identifying AD (Aβ42≤1026 pg/mL, t-Tau>238 pg/mL, p-Tau181>21.7 pg/mL, p-Tau181/Aβ42>0.023). Results p-Tau181/Aβ42 ratio demonstrated one of the highest AUCs (0.931) to separate NPH from AD (sensitivity=84.8%, specificity= 91.9%), and Aβ42 showed the highest AUC (0.856) to separate NPH from ND (sensitivity=71.6%, specificity=89.7%). When using previously established AD cut-offs, the combination of an abnormally low Aβ42 along with a normal p-Tau181/Aβ42 ratio yielded a sensitivity of 40.4% and a specificity of 96.4% in distinguishing NPH from other CI etiologies in this cohort. In contrast, abnormal results for both Aβ42 and p-Tau181/Aβ42 ratio had a sensitivity of 89.5% and a specificity of 87.0% in identifying AD, while normal results for both biomarkers had a sensitivity of 91.2% and a specificity of 90.1% in identifying ND. Incorporating the qualitative results for t-Tau and p-Tau181 into the evaluation did not enhance the diagnostic performance for distinguishing NPH, AD or ND in this cohort. Conclusion ROC analysis indicated that p-Tau181/Aβ42 and Aβ42 may help differentiate NPH from AD and ND, respectively. However, the combination of an abnormally low Aβ42 and a normal p-Tau181/Aβ42 ratio, using current AD cut-offs, demonstrated a lower sensitivity but high specificity in distinguishing NPH from the other CI etiologies in this cohort, suggesting this result pattern may be used for NPH rule-in in patients with CI.

Abstract Peanut allergy is a leading cause of food-induced allergy (prevalence of ∼1%). The diagnosis relies on evaluating clinical symptoms consistent with peanut exposure, supported by the detection of total peanut specific IgE (sIgE) and peanut component IgE antibodies (Ara h1, Ara h2, Ara h3, Ara h6, Ara h8, Ara h9 and Bet v2) through serum or skin testing. However, sensitization patterns across different ages and regions in the United States remain largely unknown. In this study, we investigated regional and demographic patterns of peanut component sensitization in a large U.S. population tested for suspected peanut allergy. Retrospective data were collected from 13,702 individuals tested at Mayo Clinic Laboratories. sIgEs were measured using the ImmunoCAP system (Thermo Fisher Scientific). Samples with detectable total peanut sIgE (≥0.1 kU/L) were reflexed to testing for seven peanut component IgEs. Statistical analyses including linear regression and correlation matrix analysis were conducted using R Studio. Regional differences were assessed using care provider zip code information. Among the study population, 92.0% (12,604 of 13,702) had detectable total peanut sIgE, with the Western U.S. region exhibiting the highest detectable rate (95.6%), and the Southeast the lowest (85.6%) (p < 0.001). On average, 3.38 different component IgEs were detectable per person with a detectable total peanut sIgE; 7.1% (892) had no detectable component IgEs, possibly due to antibodies against other peanut components. Correlation analysis revealed strong associations between Ara h2 and Ara h6 (r = 0.94), with moderate correlations observed between Ara h2, Ara h3, and Ara h6 to Ara h1 (r ranging from 0.72 to 0.77). Age-stratified analysis indicated higher total peanut sIgE detectable rates in children and adolescents (95.2%, ≤18 years) compared to adults (76.5%, >18 years). Interestingly, adults were more likely than children to have negative results for all component IgEs. Sensitization rates to different peanut components varied, with Ara h2 (77.5%) and Ara h6 (73.2%) being the most frequently detected, followed by Ara h1 (56%), and Ara h3 (47.3%). Lower detectable rates were observed for Ara h8, Ara h9, and Bet v2 (34.8%, 26.1% and 22.7%, respectively). Regional analysis showed consistently high detectable rates for Ara h2 and Ara h6 (>70%) across most areas (p < 0.001) with slightly lower rates in the Southwest (<63%). Bet v2, despite its overall low prevalence (<24%), had a relatively higher detectable rate in the Southwest (29%). This study provides valuable insights into potential demographic and geographic factors influencing peanut allergy sensitization. Our study demonstrates that peanut sensitization patterns vary by age and geographic location. Children and adolescents exhibit higher detectable rates for total peanut sIgE and component IgEs compared to adults, with Ara h2 and Ara h6 as the most frequently detected and highly correlated components.

Abstract Background Alzheimer’s disease (AD) cerebrospinal fluid (CSF) biomarkers, including the Aß42/40 and p-tau181/Aß42 ratios, are used in clinical practice to detect the presence of amyloid pathology for the diagnosis of AD. However, in some instances, discrepancies between Aß42/40 and p-tau181/Aß42 are observed. Such discrepancies may introduce uncertainties in the clinical management of patients presenting with cognitive decline. Identifying the underlying causes of these discrepancies and determining which biomarker abnormalities better predict disease progression is crucial to enhancing confidence in the use of these CSF biomarkers for the diagnosis of AD. This study aims to investigate the discordance between the Aß42/40 and p-tau181/Aß42 ratios, evaluate the frequency of these discrepancies, and identify potential causes. Methods Data from 958 individuals who had both CSF Aß42/40 and p-tau181/Aß42 tested at Mayo Clinic Laboratories between 2022 and 2024 and where the tests were ordered in CSF collected during the same lumbar puncture procedure were included. Cases with an inadequate sample collection, which could cause a false decrease in Aß42 concentrations, were excluded from the analysis. Aß42/40 was measured using the Fujirebio LUMIPULSE assays and reported as negative, likely positive, or positive. For this analysis, likely positive and positive results were combined and considered as positive (<0.072). P-tau181/Aß42 was measured using the Roche Elecsys assays and above 0.028 was considered positive. Statistical analyses were conducted using R Studio. Results Aß42/40 was positive in 69.1% of cases (n=661), while p-tau181/Aß42 was positive in 62.1% (n=594). There were 8.0% discordant cases (n=77). Most frequently, the discordant cases had an abnormal Aß42/40 with a normal p-tau181/Aß42 (Aß42/40+/p-tau181/Aß42-; n=72, 94%). Compared to the concordant Aß42/40+/p-tau181/Aß42+ group, the discordant Aß42/40+/p-tau181/Aß42- group showed significantly higher Aß42 concentrations (p <0.001, median [pg/mL]: 878.5 vs 587.0, respectively); whereas p-tau181 concentrations were significantly lower (p <0.001, median [pg/mL]: 19.3 vs 32.1, respectively). Among the Aß42/40+/p-tau181/Aß42- group, 30 cases exhibited abnormally low Aß42 and normal p-tau181, possibly reflecting early-stage AD. There were 41 cases with a normal Aß42 concentration despite an abnormal Aß42/40, possibly reflecting an overall higher production of Aß peptides resulting in an increased Aß40 peptide (median [pg/mL]: 12627 compared to 9487 in the low Aß42 and normal p-tau181 cases (n=30); p <0.001). Of these 41, 19 showed normal Aß42 but elevated p-tau181 concentrations where the use of Aß42/40 compensated for the overall higher production of Aß peptides, resulting in an abnormal ratio. In contrast, the p-tau181/Aß42 ratio was not able to normalize for these differences, leading to a normal ratio. Conclusion While overall concordance was 92.0%, our findings suggest that discrepancies between CSF Aß42/40 and p-Tau181/Aß42 are more often due to the presence of an abnormal Aß42/40 and a normal p-Tau181/Aß42. Two potential explanations for these discrepancies are early-stage AD or higher production of Aß peptides in some individuals. In the latter, differences in amyloid production would be normalized by ratios of the same biomarker (i.e. Aß42/40) but not by ratios of different biomarkers (i.e. p-Tau181/Aß42).

Abstract Aims Blood biomarkers (BBMs) are promising minimally invasive tools for the evaluation of Alzheimer’s disease (AD). However, recent studies have demonstrated that comorbidities, such as severe chronic kidney disease (CKD), may elevate circulating concentrations of various BBMs including amyloid-ß (Aß) and phosphorylated tau proteins (p-tau). Previous work has found that the effects of CKD on BBM plasma concentrations could be partially mitigated through use of biomarker ratios (e.g., the proportion of p-tau217 to non-phosphorylated tau217 (np-tau217)). In this work, we quantitatively evaluate how different combinations of BBM ratios and single biomarkers are affected in the context of decreased kidney function. Methods Participants (n=83, >57 years of age) with CKD stages 1-2 (eGFR 60-98, n = 50), and stages 3-4 (eGFR 25-59, n = 33) and without amyloid pathology (A-) were included. eGFR (mL/min/1.73 m2) was calculated using the 2021 CKD-EPI equation based on serum-paired creatinine measurements. Plasma p-tau217, np-tau217, p-tau181, np-tau181, Aß42, and Aß40 were measured by mass spectrometry (MS) (C2N Diagnostics). %p-tau217 [(p-tau217/np-tau217) x 100], %p-tau181 [(p-tau181/np-tau181) x 100], Aß42/Aß40 ratio, and p-tau217/Aß42 ratio were calculated. Log-Log (power function) best-fit regression analysis was used to quantitatively predict the relationship of BBM changes as a function of eGFR. Results Median concentrations of p-tau217, np-tau217, p-tau181, np-tau181, Aß42, and Aß40 were all significantly elevated in individuals with eGFR <60 mL/min/1.73 m2 compared to those with eGFR > or =60 (all p < 0.05). There were no differences in the %p-tau217 and Aß42/Aß40 median values between these two groups (eGFR <60 versus > or =60) (p > 0.05); whereas the proportions %p-tau181 and p-tau217/Aß42 median values were higher in the eGFR > or =60 group versus the <60 subset. Observed best-fit analyte regression equations of eGFR versus BBM concentrations predict that an eGFR decrease from 60 to 45 results in p-tau217 and p-tau181 increases of 52% (95% confidence interval (CI): 30–78%) and 40% (95% CI: 29–53%), respectively. A 16% (95% CI: 12–21%), and a 19% (95% CI: 16–23%) increase was predicted in response to the same eGFR decline for Aß42 and Aß40. Furthermore, increases of 19% (95% CI: 2–38%) for %p-tau217, 6% (95% CI: -0.02–13%) for %p-tau181 and 31% (95% CI: 12–53%) for p-tau217/Aß42 were predicted as eGFR declined from 60 to 45, while a minor decrease of 2.3% (95% CI: -5.2–0.6%) was predicted for Aß42/Aß40. Conclusions: Reduction in eGFR is associated with significant elevations in plasma concentrations of p-tau217, p-tau181, Aß42, and Aß40, which may affect the interpretations of these biomarkers. However, the use of %p-tau217, %p-tau181, or p-tau217/Aß42 could partially attenuate this effect. Although the Aß42/Aß40 ratio is not considered an effective diagnostic BBM for AD, this ratio was relatively unaffected by CKD. For example, in comparison to p-tau217 alone, using %p-tau217 or p-tau217/Aß42, the relative increase in analyte values associated with decreased eGFR was reduced by 63% and 40%, respectively. These findings provide a better understanding of the effect of reduced eGFR on AD BBM interpretation in the presence of CKD.

Abstract INTRODUCTION When cerebrospinal fluid (CSF) and positron emission tomography (PET) measurements for amyloid‐beta‐peptide (Aβ) related pathology are discordant, therapeutic decision‐making becomes uncertain. METHODS Using data from patients with mild cognitive impairment ( n = 541) from the Alzheimer's Disease Neuroimaging Initiative, we examined baseline characteristics and longitudinal clinical outcomes in persons grouped according to normal/abnormal Aβ via concurrent CSF and PET determinations using standard cutpoints. RESULTS Discordant groups for brain Aβ status (CSF+/PET− and CSF−/PET+) each represented about 5% of the mild cognitive impairment (MCI) population. Longitudinally, neither discordant group declined more than the CSF−/PET− group on either a memory measure or the Clinical Dementia Rating Sum of Boxes scores over a median 4 years of observation, while the CSF+/PET+ group exhibited worsening on both measures. DISCUSSION In contrast to the clinical decline observed in the CSF+/PET+ group, persons with MCI and CSF+/PET− or CSF−/PET+ brain amyloid patterns did not exhibit incipient decline. Highlights Discrepant abnormal cerebrospinal fluid (CSF) and positron emission tomography (PET) brain amyloid indicators are uncommon in mild cognitive impairment (MCI). CSF‐PET discrepant persons with MCI tend to have less abnormal values initially. CSF‐PET discrepant persons with MCI have a benign prognosis at 4 years.