Blood-based biomarkers for Alzheimer’s disease: influence of kidney function

doi:10.21203/rs.3.rs-8007954/v1

Background

Blood biomarkers may aid Alzheimer’s disease (AD) diagnosis; however, their reliability can be affected by systemic conditions such as kidney dysfunction (KD). We evaluated the impact of KD on plasma levels of amyloid-β (Aβ42/Aβ40), phosphorylated tau (p-Tau181, p-Tau217), and neurofilament light chain (NfL), and developed correction strategies.

Methods

We retrospectively selected 112 individuals (52 with AD, 21 with other neurodegenerative diseases, and 39 with non-neurodegenerative conditions) who had a previously documented abnormal renal function value. Participants were included based on the availability of: i)previously measured cerebrospinal fluid (CSF) Aβ42/Aβ40, p-Tau181, and NfL; ii) plasma samples for the measurement of Aβ42/Aβ40, p-Tau217, p-Tau181, and NfL; iii) paired serum samples collected at the same time to re-measure creatinine and ensure temporal correspondence between kidney function and biomarker assessments. Plasma biomarkers were quantified using the Lumipulse G1200 platform. Kidney dysfunction (KD) was defined as eGFR < 60 mL/min/1.73 m². Correction models were developed using linear regression and ROC curve analyses.

Results

KD was present in 51.8% of patients and associated with increased plasma levels of Aβ42, Aβ40, p-Tau181, p-Tau217, and NfL, not with the Aβ42/Aβ40 ratio. Creatinine-based correction improved the correlation between plasma and CSF NfL and enhanced the diagnostic performance of plasma p-Tau181 and p-Tau217 for AD. Corrected cut-offs aligned with those from KD-free cohorts.

Conclusions

KD alters plasma biomarker concentrations. Creatinine-based normalization in patients with KD may allow to rely on cut-offs generated on cohorts without KD.

Funding: Italian Ministry of Health; Italian Research Association on Alzheimer’s Disease (AirAlzh); Parkinson’s Foundation.

Alzheimer’s disease

plasma

biomarkers

kidney

renal dysfunction

p-Tau181

p-Tau217

NfL

Fluid biomarkers improve the accuracy in the diagnosis of Alzheimer's disease (AD).[1–3] The scalability, ease of collection, and patient acceptance of blood sampling make it a more viable option compared to lumbar puncture or nuclear medicine investigations.[4] Consequently, diagnostic algorithms are increasingly incorporating blood biomarkers into the diagnostic journey for suspected AD cases or when ruling out AD is necessary.[5] Whether blood biomarkers will serve as standalone indicators or as initial tests followed by further investigations remains a topic of debate, and both scenarios may be valid depending on individual cases.[6] In either situation, the reliability of blood biomarkers is of paramount importance.

Beyond diagnostic accuracy, which is notably high for certain biomarkers like tau phosphorylated at threonine 217 (p-Tau217),[7] and predictive values, which can vary significantly based on clinical context,[6] blood biomarkers must reliably reflect brain pathology without being influenced by external systemic factors. Research has shown that systemic variables, such as body mass index, can affect plasma neurofilament light chain (NfL) levels,[8] and certain comorbidities, like kidney dysfunction (KD), are associated with elevated concentrations of tau phosphorylated at threonine 181 (p-Tau181), p-Tau217, and NfL in plasma.[9–13] This effect is particularly pronounced for p-Tau181 and NfL, less so for p-Tau217, and could be clinically significant in severe KD cases where the need for a biomarker-based diagnosis may be less critical.

Given the potential for blood biomarkers to become widely used diagnostic tools, a thorough analysis of the impact of KD on their concentrations is essential. Moreover, strategies to adjust for these effects are necessary, but have not yet been adequately addressed.

For this reason, we conducted this study with the following aims:

To assess the impact of KD on plasma concentrations of Aβ42/Aβ40, p-Tau181, p-Tau217, and NfL measured using chemiluminescent immunoassay (CLEIA) on the fully automated Lumipulse G1200 platform (Fujirebio), which is a candidate for widespread use in measuring blood biomarkers, in a cohort of patients with AD and other neurological diseases, all with available CSF AT(N) profiles.

To develop and evaluate correction strategies to mitigate the impact of KD on plasma biomarker concentrations, with CSF as a reference.

Study population. We retrospectively analyzed plasma samples of 112 patients consecutively referred to the Section of Neurology, University Hospital of Perugia, Italy, between January 2017 and December 2023.

All patients underwent medical history, physical and neurological examination, a thorough neuropsychological evaluation, brain imaging (computed tomography or magnetic resonance imaging), and lumbar puncture for the measurement of CSF core AD biomarkers, namely Aβ42/40, p-Tau181 and total tau (t-Tau). Patients were defined as A + or A- and as T + or T- based on previously calculated internal cut-off values of CSF Aβ42/40 and p-Tau181, measured with the use of the Lumipulse® G600 (Fujirebio, Japan).[14]

The cohort included 52 patients with AD, defined by a CSF profile A+/T+, 21 patients with non-AD neurodegenerative diseases (ND), including Parkinson's disease (PD) (n = 6), progressive supranuclear palsy (PSP) (n = 6), frontotemporal dementia (FTD) (n = 5), corticobasal syndrome (CBS) (n = 2), motor neuron disease (MND) (n = 1), and Huntington’s disease (HD) (n = 1), with different CSF profiles (A+/T+, A+/T-, A-/T+, and A-/T-), and 39 individuals with non-neurodegenerative neurological diseases (OND), who underwent lumbar puncture as part of their diagnostic workup (Supplementary Table 1), with CSF profiles A+/T-, A-/T+, and A-/T- (Table 1).

Table 1
Demographical and clinical features
	AD	ND	OND
n	52	21	39
F/M	19/33	5/16	14/25
Age – yrs mean ± SD	76.2 ± 5.9	74.3 ± 6.4	74.5 ± 5.3
MMSE mean ± SD	20.7 ± 7	23.8 ± 3.7	25.9 ± 3.2
A-T- n (%)	0	17 (81)	23 (59)
A-T+ n (%)	0	3 (14.3)	12 (30.8)
A + T- n (%)	0	1 (4.7)	4 (10.2)
A + T+ n (%)	52 (100)	0	0
KD+ n (%)	29 (55.8)	7 (33.3)	22 (56.4)
Severe KD n (%)	1 (1.9)	0	4 (10.2)

Abbreviations. AD: Alzheimer’s disease. A+/- T+/-: combination of cerebrospinal fluid biomarkers with or without abnormal Aβ42/Aβ40 and with or without abnormal p-Tau181. KD+: individuals with kidney dysfunction, i.e. estimated glomerular filtration rate < 60 ml/min/1.73 m². MMSE: Mini Mental State Examination. ND: neurodegenerative diseases non-AD. OND: other neurological diseases. Severe KD: severe kidney dysfunction, i.e. estimated glomerular filtration rate < 30 ml/min/1.73 m².

Patients were selected for this study if they had one or more blood tests with an abnormal estimated glomerular filtration rate (eGFR) value within a year from lumbar puncture (eGFR < 60 mL/min/1.73 m²). Creatinine was re-measured on serum samples collected together with plasma samples, and eGFR was re-calculated.

AD patients (n = 52) were diagnosed according to a CSF biomarker profile A+/T+, independent of the clinical stage, in line with the 2018 National Institute of Aging–Alzheimer’s Association criteria.[15] PD, PSP, FTD, CBS, MND and HD patients (n = 21) were diagnosed according to the current diagnostic criteria.[16–21]

The main demographic and clinical features of each diagnostic group are summarized in Table 1. Plasma biomarkers concentrations in each diagnostic group are reported in Supplementary Table 2.

Sample collection. All patients underwent lumbar puncture and venipuncture at the Section of Neurology, University Hospital of Perugia, Italy, and CSF and plasma were collected, handled, and stored according to international guidelines.[22] Collection of human CSF, serum and plasma samples has been performed following international guidelines and the same standard operating procedures (SOPs) throughout the study.[22] Lumbar puncture was performed between 8:00 a.m. and 10:00 a.m. CSF was collected into sterile polypropylene tubes and centrifuged for 10 min at 2000 × g at room temperature. At the same time, plasma was collected into sterile polypropylene tubes containing EDTA as the anticoagulant and centrifuged for 10 min at 2000 × g at room temperature. Serum tubes have micro silica particles attached to the inside that activate coagulation when the tubes are gently inverted after collection, they were also centrifuged for 10 min at 2000 × g at room temperature. Once processed, CSF, plasma and serum samples were stored in 0.5 mL tubes (72.730.007, Sarstedt, Germany) and immediately frozen at -80°C pending analysis.

Biomarker analysis. All plasma AD biomarkers were measured with the use of the Lumipulse® G1200 (Fujirebio, Japan) at the Clinical Neurochemistry Laboratory, Section of Neurology, Department of Medicine and Surgery, University of Perugia, Italy. Different commercially available Lumipulse® kits to analyse CSF and plasma levels of biomarkers have been used: Lumipulse®G Aβ1–40 CSF, Lumipulse®G Aβ1–40 Plasma, Lumipulse®G Aβ1–42 CSF, Lumipulse®G Aβ1–42 Plasma, Lumipulse®G pTau181 CSF, Lumipulse®G pTau181 Plasma, Lumipulse®G pTau217 Plasma, Lumipulse®G NfL CSF, Lumipulse®G NfL Blood. Appropriate calibrators and controls included in each kit were run together with samples. On serum samples collected at the same moment of CSF and plasma, creatinine was measured through automated assay on Beckman Coulter at the Clinical Pathology and Hematology Laboratory, S. Maria della Misericordia Hospital, Perugia, Italy, and eGFR was re-calculated.

Statistical analysis. Statistical analysis has been performed using R Studio software version 4.3.3 (2024-02-29 ucrt). Linear regression analysis was applied to determine the influence of KD on the association between plasma and CSF NfL. However, given the skewness in the distributions of NfL, concentrations were Log10-transformed to meet the assumptions of linear regression. Indeed, as a sensitivity analysis, we determined that without such transformation the regression residuals were not normally distributed. Probability density fitting was used to compare the empirical distribution of data with a theoretical or estimated distribution. Probability density fitting for plasma and CSF NfL, and for the models obtained by the correction were performed by applying the “fitdist” function of the “fitdistrplus” R package. The goodness of fit was assessed by comparison of empirical and estimated histograms and cumulative density functions (CDF), quartile-quartile (Q-Q) plots, and probability-probability (P-P) plots. Pearson’s correlation coefficient was used to quantify the strength of linear associations.

Box plots were used to visualise the influence of KD in plasma p-Tau181, p-Tau217, Aβ42, Aβ40 and their ratios by using the "boxplot" function of the "ggplot" R package. A two-way analysis of variance test (two-way-ANOVA) was used to analyse the impact of KD on plasma biomarkers by using KD status and A/T profile as grouping variables. Then, an ANCOVA test was used to find out whether the contribution of eGFR could be used as normalising factor. ROC analysis was performed to test the goodness of the corrections on p-Tau181 and p-Tau217 with creatinine by using the R package “pROC”. The change in AUC was tested by the DeLong test. To evaluate the performance of the classification models, several parameters, i.e., area under the ROC curve (AUC), threshold, accuracy, sensitivity, specificity, positive predictive values (PPV), negative predictive values (NPV) and Youden index, were evaluated. Confidence intervals were calculated by using 2000 bootstrap replicates.

Standard Protocol Approvals, Registrations, and Patient Consents. The study was conducted in accordance with the Declaration of Helsinki and was approved by the local Ethics Committee (Comitato Etico Aziende Sanitarie Regione Umbria; approvals n. 19369/AV and 20942/21/OV). Written informed consent was obtained from all participants prior to inclusion in the study.

Data Availability. Anonymized data not published within this article will be made available by request from any qualified investigator.

Role of funders. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the research article.

Prevalence of KD based on serum creatinine levels at the time of CSF and plasma collection. Patients were included in this study based on evidence of KD from retrospective serum creatinine determinations conducted within one year prior to CSF and plasma sampling. However, upon evaluating serum creatinine levels from samples collected concurrently with CSF and plasma, we observed that 58/112 (51.8%) individuals had confirmed evidence of KD. Of these, 5/112 (4.5%) had severe KD, i.e. eGFR < 30 ml/min/1.73m² (Table 1).

Impact of renal function on the correlation between CSF and plasma NfL. To assess the degree of correlation between CSF and plasma NfL, linear regression analysis was applied. Although the Pearson’s correlation between raw CSF and plasma NfL concentrations was moderate to strong (Fig. 1 – A), the analysis of the residuals (Fig. 1 – B) prompted us to consider Log10-transformed data to meet the assumptions of linear regression (Fig. 1 – C). Indeed, applying linear regression on Log10-transformed data resulted in approximately normally distributed residuals (Fig. 1 – D), thus meeting the assumptions of linear regression.

Subsequently, eGFR, serum creatinine, plasma Aβ40, and age were incorporated with plasma NfL to predict CSF NfL concentration. The covariate showing the most significant contribution when added to plasma NfL was eGFR (p-value = 6.2*10^− 7) (Supplementary Table 3), and it was accordingly chosen as the primary correction factor. This first adjustment applied in Log10 space, referred to as “Model 1”, resulted in an improved correlation (from the original R = 0.69 to R = 0.8, as shown in Fig. 2– A) when considering back-transformed data. Indeed, after applying eGFR correction in Log10 space, back-transformed data met the assumptions of linear regression (Fig. 2– B).

Next, the impact of introducing a second covariate was evaluated. Only age showed significant association (p-value = 0.01) (Supplementary Table 4) and was therefore chosen as an additional adjustment factor to be included. This adjustment, referred to as “Model 2”, resulted in a marginally improved correlation (from the initial R = 0.8 to R = 0.83 in linear space, depicted in Fig. 2 – C). Indeed, after applying eGFR + age correction in Log10 space, back-transformed data met the assumptions of linear regression (Fig. 2 – D). The value of the coefficients and the standard error of the variables used in Model 2 are reported in Supplementary Table 5.

Impact of renal function on plasma AD biomarkers and biomarker ratios. Concerning all other plasma biomarkers of AD considered, through a two-way ANOVA analysis, it was revealed that KD exerts a pronounced effect on plasma Aβ42, Aβ40 (not shown), p-Tau181 and p-Tau217, but not on plasma Aβ42/Aβ40. Within each A/T group, plasma levels of Aβ42, Aβ40 (not shown), p-Tau181 and p-Tau217 in KD + subjects were higher than those in KD- patients (Fig. 3).

When considering biomarker ratios, the influence of KD on plasma levels of p-Tau181/Aβ42 (p-value KD = 0.172), p-Tau217/Aβ42 (p-value KD = 0.064), p-Tau181/Aβ40 (p-value KD = 0.211) and p-Tau217/Aβ40 (p-value KD = 0.065) was less evident (Table 2) (Fig. 4A – D). However, these ratios could not be considered as optimal compensation in the presence of KD since they were still influenced by eGFR. Indeed, p-Tau181/Aβ42 was not significantly affected by KD according to two-way ANOVA but was still affected by eGFR as indicated by ANCOVA (p-value eGFR = 0.03). In contrast, p-Tau217/Aβ42 was more effective (p-value 2-way ANOVA KD = 0.064, p-value ANCOVA eGFR = 0.07). Almost the same results have been obtained from the ratio between p-Tau181 and p-Tau217 with Aβ40 (p-value p-Tau181/Aβ40: 2-way ANOVA KD = 0.211, p-value ANCOVA eGFR = 0.02; p-value p-Tau217/Aβ40: 2-way ANOVA KD = 0.065, p-value ANCOVA eGFR = 0.07) and for the same principle, Aβ40 could mitigate the effect of KD (Table 2).

Table 2
The impact of kidney dysfunction (KD + and KD-) and of eGFR on plasma biomarker ratios.
Biomarkers ratio	p-value ANOVA (A/T and KD+/-)	p-value ANCOVA (A/T and eGFR)
p-Tau181/Aβ42	0.17	0.03
p-Tau217/Aβ42	0.06	0.07
p-Tau181/Aβ40	0.21	0.02
p-Tau217/Aβ40	0.07	0.07
p-Tau181/creatinine	0.17	0.18
p-Tau217/creatinine	0.39	0.84
Abbreviations. ANOVA: analysis of variance of biomarker ratios based on groups defined by the CSF A/T profile and by the KD + and KD- status. ANCOVA: analysis of covariance to test the interaction effects of A/T profiles variables on biomarker ratios, controlling for the effects of eGFR as a continuous covariable. KD+: individuals with kidney dysfunction, i.e. estimated glomerular filtration rate < 60 ml/min/1.73 m². KD-: individuals without kidney dysfunction, i.e. estimated glomerular filtration rate ≥ 60 ml/min/1.73 m².

Conversely, the impact of KD was significantly minimized by the p-Tau181/creatinine ratio (p-value 2-way-ANOVA KD = 0.17, p-value ANCOVA eGFR = 0.18) and the p-Tau217/creatinine ratio (p-value 2-way-ANOVA KD = 0.39, p-value ANCOVA eGFR = 0.84) (Table 2) (Fig. 4E – F). Following these results, and the well-known sex-dependance of serum creatinine values, it was decided to normalize p-Tau values with creatinine values adjusted to the mean values by sex.[23]

Impact of creatinine, Aβ40, and Aβ42 corrections on plasma p-Tau181 and p-Tau217 accuracy. A ROC analysis was performed to evaluate the accuracy of p-Tau181 and p-Tau217 using serum creatinine and sex correction (p-Tau217corr. and p-Tau18 corr.), ratio on Aβ40, and ratio on Aβ42 in predicting an A + T + CSF profile. The following parameters were evaluated: threshold, accuracy, sensitivity, specificity, positive predictive values (PPV), negative predictive values (NPV) and Youden. Cut-offs (threshold) are shown based on two levels of sensitivity and specificity restriction (95% and 90%) and the best accuracy obtained for each analyte (Table 3) (Supplementary Fig. 1). For p-Tau217corr. and p-Tau181corr., ROC curves and distributions are displayed in Fig. 5 and Fig. 6, respectively.

Table 3

Cut-off, accuracy, sensitivity, specificity, positive predictive values (PPV), negative predictive values (NPV) and Youden evaluated for plasma p-Tau217 and p-Tau181 corrected for creatinine and sex, p-Tau217/Aβ40, p-Tau181/Aβ40, p-Tau217/Aβ42, p-Tau181/Aβ42, and Aβ42/Aβ40 for the CSF A+/T + vs other comparison.
Plasma p-Tau217 corrected for creatinine and sex, AUC = 0.965
Scenario	Threshold	Accuracy	Sensitivity	Specificity	PPV	NPV	Youden
95% sens	0.186	89%	96%	82%	83%	96%	1.79
Best accuracy	0.229	91%	92%	89%	89%	93%	1.82
90% sens	0.236	90%	90%	89%	88%	91%	1.80
90% spec	0.246	88%	86%	89%	88%	88%	1.76
95% spec	0.337	85%	73%	96%	95%	80%	1.69
Plasma p-Tau181 corrected for creatinine and sex, AUC = 0.862
Scenario	Threshold	Accuracy	Sensitivity	Specificity	PPV	NPV	Youden
95% sens	1.483	77%	94%	62%	68%	93%	1.56
90% sens ssensens	1.581	78%	90%	67%	70%	89%	1.57
Best accuracy	1.950	81%	77%	85%	82%	81%	1.62
90% spec	2.351	74%	56%	90%	83%	70%	1.46
95% spec	2.945	63%	25%	95%	81%	59%	1.20
Plasma p-Tau217/Aβ40, AUC = 0.956
Scenario	Threshold	Accuracy	Sensitivity	Specificity	PPV	NPV	Youden
95% sens	65.79×10^− 4	90%	96%	84%	84%	96%	1.80
Best accuracy	65.79×10^− 4	90%	96%	84%	84%	96%	1.80
90% sens	79.11×10^− 4	89%	90%	88%	87%	91%	1.78
90% spec	89.93×10^− 4	89%	86%	91%	90%	88%	1.77
95% spec	145.19×10^− 4	82%	67%	96%	94%	76%	1.63
Plasma p-Tau181/Aβ40, AUC = 0.853
Scenario	Threshold	Accuracy	Sensitivity	Specificity	PPV	NPV	Youden
95% sens	50.43×10^− 3	67%	96%	45%	60%	93%	1.41
90% sens	58.00×10^− 3	73%	90%	65%	65%	88%	1.49
Best accuracy	66.72×10^− 3	82%	83%	82%	80%	84%	1.64
90% spec	82.52×10^− 3	76%	60%	90%	84%	72%	1.50
95% spec	97.54×10^− 3	70%	40%	95%	88%	65%	1.35
Plasma p-Tau217/Aβ42, AUC = 0.959
Scenario	Threshold	Accuracy	Sensitivity	Specificity	PPV	NPV	Youden
95% sens	81.82×10^− 3	91%	96%	86%	86%	96%	1.82
Best accuracy	81.82×10^− 3	91%	96%	86%	86%	96%	1.82
90% sens	96.39×10^− 3	88%	90%	86%	85%	91%	1.76
90% spec	109.26×10^− 3	89%	86%	91%	90%	88%	1.78
95% spec	181.19×10^− 3	81%	65%	96%	94%	75%	1.61
Plasma p-Tau181/Aβ42, AUC = 0.884
Scenario	Threshold	Accuracy	Sensitivity	Specificity	PPV	NPV	Youden
95% sens	0.0601	74%	96%	55%	65%	94%	1.51
90% sens	0.0709	79%	90%	68%	71%	89%	1.59
Best accuracy	0.0804	83%	83%	83%	81%	85%	1.66
90% spec	0.0974	79%	65%	90%	85%	75%	1.55
95% spec	0.1098	71%	42%	95%	88%	65%	1.37
Plasma Aβ42/Aβ40, AUC = 0.780
Scenario	Threshold	Accuracy	Sensitivity	Specificity	PPV	NPV	Youden
95% sens	0.097	65%	95%	35%	60%	86%	1.29
Best accuracy	0.094	69%	90%	47%	64%	81%	1.37
90% sens	0.084	71%	65%	78%	46%	68%	1.43
90% spec	0.078	65%	40%	91%	82%	60%	1.31
95% spec	0.073	57%	21%	95%	80%	54%	1.16
Abbreviations. AUC: area under the curve. CSF: cerebrospinal fluid. NPV: negative predictive value. PPV: positive predictive value. Sens.: sensitivity. Spec.: specificity.

Renal function correction formulas for plasma NfL, p-Tau181 and p-Tau217. The ready-to-apply corrections are reported below:

$$\:{CSF\:NfL}_{Model\:1}={eGFR}^{0.9}*\:{plasma\:NfL}^{0.77}*\:{10}^{0.54}$$

1

$$\:{CSF\:NfL}_{Model\:2}={eGFR}^{0.94}*\:{plasma\:NfL}^{0.72}*\:{age}^{1.45}*\:{10}^{-2.16}$$

2

$$\:{Plasma\:pTau}_{corr}=s*0.9*\:\frac{pTau}{Cr}+0.7*\:\left(1-s\right)*\:\frac{pTau}{Cr}$$

3

The first Eq. (1) represents the result of the log-linear regression of CSF NfL vs. NfL and eGFR, referred to as “Model 1”. The second Eq. (2) represents the log-linear regression of CSF NfL vs. NfL, eGFR and age, referred to as “Model 2”. The third Eq. (3) represents the correction for plasma p-Tau, both p-Tau181 and p-Tau217 (s = 1 for males, s = 0 for females, Cr = serum creatinine concentration).

In this study, we evaluated the impact of KD on plasma concentrations of AD biomarkers (Aβ42, Aβ40, p-Tau181, p-Tau217, and NfL), with the aim of proposing correction models to reduce this effect. Our results confirm that KD increases plasma levels of these biomarkers, leading to higher concentrations in individuals with impaired renal function. For the first time, we also propose practical correction strategies to account for this effect.

The effect on Aβ42 can be compensated by using the Aβ42/Aβ40 ratio. We therefore tested whether Aβ40 could also normalize the concentrations of other plasma biomarkers. For p-Tau181 and p-Tau217, the ratio with Aβ40 reduced the impact of KD, although less effectively than for Aβ42. For NfL, plasma Aβ40 had little influence in normalizing the KD effect. Similar findings were obtained when using Aβ42 as a normalizer for p-Tau181, p-Tau217, and NfL.

Aβ40 and Aβ42 share a similar metabolic pathway, systemic catabolism, and are likely filtered by the kidneys in the same way.[24] Aβ40 can be considered a reliable proxy for total Aβ levels, and Aβ42/Aβ40 ratio adjusts for inter-individual differences in Aβ production.[25] This relationship does not necessarily extend to other biomarkers with different molecular weights, which may undergo different renal processing. Thus, while Aβ40 is suitable to normalize Aβ42, its use for other plasma biomarkers is questionable.

For plasma p-Tau181 and p-Tau217, the best normalization for renal function was obtained by using the ratio with creatinine. After correcting for creatinine and sex, no differences were found in plasma p-Tau181 and p-Tau217 between AD patients with and without KD. Based on these results, we developed a formula to adjust plasma p-Tau levels for KD when present.

Compared to raw plasma values, creatinine-corrected p-Tau181 and p-Tau217 showed slightly higher accuracy in distinguishing AD from non-AD cases, although this difference was not statistically significant. Previous studies have shown that KD affects p-Tau217 levels, but the effect is smaller than the impact of being A+.[26] KD may be relevant for p-Tau217 only in severe cases. In routine clinical practice, p-Tau217 correction for renal function is usually unnecessary. However, creatinine adjustment allowed us to obtain cut-off values consistent with those reported by Arranz et al.[26], and Bellomo et al.,[27] in independent cohorts with only small numbers of KD patients. This suggests that correcting for renal function may enable the use of established cut-offs from populations without KD.

For NfL, the variables most strongly influencing the correlation between plasma and CSF concentrations were eGFR and age. This led to a formula (Model 2) that improved the plasma–CSF correlation. Adjusting plasma NfL for age and eGFR may therefore increase the reliability of peripheral NfL measurements.

A strength of our study is the use of an automated immunoassay platform, which ensures highly reproducible and standardized results, minimizes pre-analytical and analytical variability, and is cost-effective and scalable.[28] A limitation is the current unavailability of the p-Tau217 assay in CSF, which prevents assessing its plasma–CSF correlation and potential adjustment for other variables. Furthermore, although we accounted for renal function, other systemic factors such as body mass index and liver function were not considered and should be explored in future studies.

Our findings may have clinically relevant implications. Given the increasing use of plasma biomarkers in AD diagnosis and in clinical trials, systematic correction for renal function should be considered, particularly in older populations where KD is prevalent, in order to rely with highest confidence on cut-offs generated on cohorts unaffected by KD. Moreover, validation in independent larger cohorts and prospective studies will be crucial to confirm the generalizability of these correction strategies.

Acknowledgements

We express our sincere gratitude to Fujirebio Europe for the invaluable support in providing the research kits free of charge for our scientific investigation.

Study Funding

L. Gaetani received support from AirAlzh (Associazione Italiana Ricerca Alzheimer Onlus). G. Bellomo was supported by the Postdoctoral Fellowship for Basic Scientists grant of the Parkinson’s Foundation (award ID: PF-PRF-934916). A. Tozzi received support by grants from the Italian Ministry of University and Research (PRIN 2022 grant 2022CAKAHL; PRIN 2022 Next Generation EU-PNRR-M4C2 grant P2022374Y9).

Disclosures

L. Gaetani has participated in advisory boards for and received writing or speaker honoraria and travel grants from Almirall, Biogen, Eisai, Euroimmun, Fujirebio, Lilly, Merck, Mylan, Novartis, Roche, Sanofi, Siemens Healthineers, and Teva. G. Bellomo received honoraria from Fujirebio and completed paid consultancies for Parkinson’s Foundation. He received travel/educational grants from Fujirebio and Alzheimer’s Association. D. Chiasserini has nothing to disclose. L. Parnetti served as Member of Advisory Boards for Fujirebio, IBL, Roche, and Merck.

Beach TG, Monsell SE, Phillips LE, et al. Accuracy of the clinical diagnosis of Alzheimer disease at National Institute on Aging Alzheimer Disease Centers, 2005-2010. J Neuropathol Exp Neurol. 2012;71:266–73. doi: 10.1097/NEN.0b013e31824b211b
Mattsson-Carlgren N, Grinberg LT, Boxer A, et al. Cerebrospinal Fluid Biomarkers in Autopsy-Confirmed Alzheimer Disease and Frontotemporal Lobar Degeneration. Neurology. 2022;98:e1137–50. doi: 10.1212/WNL.0000000000200040
Teunissen CE, Verberk IMW, Thijssen EH, et al. Blood-based biomarkers for Alzheimer’s disease: towards clinical implementation. Lancet Neurol. 2022;21:66–77. doi: 10.1016/S1474-4422(21)00361-6
Mielke MM, Anderson M, Ashford JW, et al. Considerations for widespread implementation of blood-based biomarkers of Alzheimer’s disease. Alzheimer’s & dementia : the journal of the Alzheimer’s Association. 2024;20:8209–15. doi: 10.1002/alz.14150
Mielke MM, Anderson M, Ashford JW, et al. Recommendations for clinical implementation of blood-based biomarkers for Alzheimer’s disease. Alzheimers Dement. 2024;20:8216–24. doi: 10.1002/alz.14184
Schindler SE, Galasko D, Pereira AC, et al. Acceptable performance of blood biomarker tests of amyloid pathology - recommendations from the Global CEO Initiative on Alzheimer’s Disease. Nat Rev Neurol. 2024;20:426–39. doi: 10.1038/s41582-024-00977-5
Anastasi F, Fernández-Lebrero A, Ashton NJ, et al. A head-to-head comparison of plasma biomarkers to detect Alzheimer’s disease in a memory clinic. Alzheimer’s & dementia : the journal of the Alzheimer’s Association. 2025;21:e14609. doi: 10.1002/alz.14609
Manouchehrinia A, Piehl F, Hillert J, et al. Confounding effect of blood volume and body mass index on blood neurofilament light chain levels. Annals of clinical and translational neurology. 2020;7:139–43. doi: 10.1002/acn3.50972
Arslan B, Brum WS, Pola I, et al. The impact of kidney function on Alzheimer’s disease blood biomarkers: implications for predicting amyloid-β positivity. Alzheimer’s Research & Therapy. 2025;17:48. doi: 10.1186/s13195-025-01692-z
Dittrich A, Ashton NJ, Zetterberg H, et al. Association of Chronic Kidney Disease With Plasma NfL and Other Biomarkers of Neurodegeneration. Neurology. 2023;101. doi: 10.1212/WNL.0000000000207419
Akamine S, Marutani N, Kanayama D, et al. Renal function is associated with blood neurofilament light chain level in older adults. Scientific Reports. 2020;10:20350. doi: 10.1038/s41598-020-76990-7
Mielke MM, Dage JL, Frank RD, et al. Performance of plasma phosphorylated tau 181 and 217 in the community. Nat Med. 2022;28:1398–405. doi: 10.1038/s41591-022-01822-2
Janelidze S, Barthélemy NR, He Y, et al. Mitigating the Associations of Kidney Dysfunction With Blood Biomarkers of Alzheimer Disease by Using Phosphorylated Tau to Total Tau Ratios. JAMA Neurol. 2023;80:516–22. doi: 10.1001/jamaneurol.2023.0199
Bellomo G, Indaco A, Chiasserini D, et al. Machine Learning Driven Profiling of Cerebrospinal Fluid Core Biomarkers in Alzheimer’s Disease and Other Neurological Disorders. Frontiers in neuroscience. 2021;15:647783. doi: 10.3389/fnins.2021.647783
Jack CR, Bennett DA, Blennow K, et al. NIA-AA Research Framework: Toward a biological definition of Alzheimer’s disease. Alzheimers Dement. 2018;14:535–62. doi: 10.1016/j.jalz.2018.02.018
Carvalho M de, Dengler R, Eisen A, et al. Electrodiagnostic criteria for diagnosis of ALS. Clinical neurophysiology : official journal of the International Federation of Clinical Neurophysiology. 2008;119:497–503. doi: 10.1016/j.clinph.2007.09.143
Gorno-Tempini ML, Hillis AE, Weintraub S, et al. Classification of primary progressive aphasia and its variants. Neurology. 2011;76:1006–14. doi: 10.1212/WNL.0b013e31821103e6
Rascovsky K, Hodges JR, Knopman D, et al. Sensitivity of revised diagnostic criteria for the behavioural variant of frontotemporal dementia. Brain. 2011;134:2456–77. doi: 10.1093/brain/awr179
Postuma RB, Berg D, Stern M, et al. MDS clinical diagnostic criteria for Parkinson’s disease. Movement Disorders. 2015.
Ross CA, Reilmann R, Cardoso F, et al. Movement Disorder Society Task Force Viewpoint: Huntington’s Disease Diagnostic Categories. Movement Disorders Clinical Practice. 2019;6:541–6. doi: 10.1002/mdc3.12808
Armstrong MJ, Litvan I, Lang AE, et al. Criteria for the diagnosis of corticobasal degeneration. Neurology. 2013;80:496–503. doi: 10.1212/WNL.0b013e31827f0fd1
Teunissen CE, Petzold A, Bennett JL, et al. A consensus protocol for the standardization of cerebrospinal fluid collection and biobanking. Neurology. 2009;73:1914–22. doi: 10.1212/WNL.0b013e3181c47cc2
Pottel H, Vrydags N, Mahieu B, et al. Establishing age/sex related serum creatinine reference intervals from hospital laboratory data based on different statistical methods. Clinica chimica acta; international journal of clinical chemistry. 2008;396:49–55. doi: 10.1016/j.cca.2008.06.017
Ghiso J, Shayo M, Calero M, et al. Systemic Catabolism of Alzheimer’s Aβ40 and Aβ42. Journal of Biological Chemistry. 2004;279:45897–908. doi: 10.1074/jbc.M407668200
Janelidze S, Zetterberg H, Mattsson N, et al. CSF Aβ42/Aβ40 and Aβ42/Aβ38 ratios: better diagnostic markers of Alzheimer disease. Annals of clinical and translational neurology. 2016;3:154–65. doi: 10.1002/acn3.274
Arranz J, Zhu N, Rubio-Guerra S, et al. Diagnostic performance of plasma pTau217, pTau181, Aβ1-42 and Aβ1-40 in the LUMIPULSE automated platform for the detection of Alzheimer disease. Alzheimer’s Research & Therapy. 2024;16:139. doi: 10.1186/s13195-024-01513-9
Bellomo G, Bayoumy S, Megaro A, et al. Fully automated measurement of plasma Aβ42/40 and p-tau181: Analytical robustness and concordance with cerebrospinal fluid profile along the Alzheimer’s disease continuum in two independent cohorts. Alzheimers Dement. 2024;20:2453–68. doi: 10.1002/alz.13687
Teunissen CE, Kimble L, Bayoumy S, et al. Methods to Discover and Validate Biofluid-Based Biomarkers in Neurodegenerative Dementias. Mol Cell Proteomics. 2023;22:100629. doi: 10.1016/j.mcpro.2023.100629

Competing interest reported. L. Gaetani has participated in advisory boards for and received writing or speaker honoraria and travel grants from Almirall, Biogen, Eisai, Euroimmun, Fujirebio, Lilly, Merck, Mylan, Novartis, Roche, Sanofi, Siemens Healthineers, and Teva. G. Bellomo received honoraria from Fujirebio and completed paid consultancies for Parkinson’s Foundation. He received travel/educational grants from Fujirebio and Alzheimer’s Association. D. Chiasserini has nothing to disclose. L. Parnetti served as Member of Advisory Boards for Fujirebio, IBL, Roche, and Merck.

SupplementaryMaterial.docx

Blood-based biomarkers for Alzheimer’s disease: influence of kidney function

Status:

Version 1

Abstract

Figures

Introduction

Methods

Results

Discussion

Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1