Page 96 - AN-2-3

Advanced Neurology eGFR and serum neurofilament light chain

lowering medications are also considered to have variable (Q2/Q1: β = −4.8, P = 0.03; Q3/Q1: β = −7.28,
hyperlipidemia. The criteria for hypertension include a P = 0.01; Q4/Q1: β = −5.78, P = 0.01) (Table 3). Notably,
blood pressure reading of 140/90 mmHg or higher or a this relationship persisted even after adjusting for all
documented history of taking medication to control high covariates (Table 3). Using a generalized additive model
blood pressure . followed by a threshold effect analysis, we identified an
[16]
inflection point at an eGFR value of 59.9. Specifically, for
2.5. Statistical analysis each unit increase in eGFR <59.9, a corresponding decrease
To obtain a representative estimate of the US adult (β = −1.0, P < 0.01) of 1 pg/ml in sNfL levels was observed
population, sample weights provided by NHANES were (Table 4 and Figure 1). Given that CKD is typically defined
used for analysis. In the baseline characteristics, participants’ as an eGFR <60, we conducted interaction and subgroup
eGFR was divided into quartiles, and its association with analyses to explore the potential impact of CKD on the
all covariates was calculated. Mean ± standard error relationship between eGFR and sNfL levels. However, the
represented continuous variables, while numbers and results indicated no significant interaction between CKD
percentages were used for categorical variables. Weighted and eGFR (Table 5). However, we did find a significant
linear regression was used for univariate analysis to interaction between eGFR and hypertension (P = 0.01),
calculate the relationship between sNfL and eGFR, along which we included in our subgroup analysis (Table 5).
with other covariates. To further explore the relationship Specifically, this study suggests that hypertension may
between eGFR and sNfL, three linear regression models modulate the effect of eGFR on sNfL levels.
were used. Model 1 adjusted for demographic variables
(age, sex, education, and marital status); Model 2 adjusted 4. Discussion
for alcohol consumption, smoking, CKD, HF, MI, The present study aimed to investigate the correlation
metabolic syndrome (MetS), and diabetes mellitus (DM) between eGFR and sNfL in the adult population using data
based on Model 1; and Model 3 adjusted for hypertension, from the NHANES database for the years 2013 – 2014.
hyperlipidemia, and stroke based on Model 2. Sensitivity Our analysis revealed a negative correlation between eGFR
analysis involved dividing eGFR into quartiles to assess the and sNfL after adjusting for demographic and clinical
robustness of the relationship with SNL. A two-stage linear covariates. These findings align with previous research
regression model was utilized to identify any potential that has reported an association between decreased kidney
inflection points in the curve and explore the potential function and increased neuronal injury, manifested as
correlation between eGFR and sNfL using a generalized elevated levels of sNfL . It is worth noting that the previous
[17]
additive model. Statistical analysis was conducted using research focused on sNfL levels in elderly individuals with
Empower Stats and R, with significance set at P < 0.05. atrial fibrillation. In contrast, our study aimed to broaden

3. Results the scope to encompass a wider population.
Interestingly, we identified an inflection point at an
Our study involved a sample size of 2071 participants, and eGFR of 59.9. When eGFR reached 59.9 or higher, each
Table 1 provides the baseline characteristics. From the unit increase in eGFR corresponded to a concomitant
table, it is evident that age (P < 0.01), gender (P = 0.03), and decrease of 1pg/mL in sNfL (P = 0.01). Conversely, when
marital status (P < 0.01) exhibited significant correlations eGFR was below 59.9, each unit increase in eGFR was
with eGFR. In addition, several comorbidities such as associated with a decline of 0.1 pg/mL in sNfL (β = −1.0,
CKD (P < 0.01), hyperlipidemia (P < 0.01), hypertension
(P < 0.01), and diabetes (P < 0.01) were also associated P < 0.01). This inflection point is particularly noteworthy
with eGFR (Table 1). Subsequently, univariate analyses because CKD is defined as an eGFR <60. We considered
were performed to explore the association between sNfL the presence of CKD in our analysis, suggesting that
the observed inflection point may reflect a different
levels, eGFR, and other covariates. The results indicated mechanism than the one underlying the pathophysiology
that age (β = 0.38, P < 0.01) and all comorbidities were
positively correlated with sNfL levels (Table 2), while of CKD. One potential explanation for this observation is
eGFR exhibited a significantly negative correlation with that an eGFR <60 may represent a threshold at which the
sNfL levels (β = −0.31, P < 0.01) (Table 2). After adjusting decline in kidney function becomes significant enough to
for demographic and clinical covariates in a multivariate induce neurotoxicity and elevated sNfL levels.
model, our findings revealed that eGFR remained negatively Furthermore, our subgroup analysis revealed a significant
associated with sNfL levels (P < 0.01). This association interaction between hypertension and eGFR (P = 0.01).
remained valid for eGFR when considered as both a A previous meta-analysis reported that prehypertension or
continuous variable (β = −0.2, P < 0.01) and a categorical hypertension can serve as predictors of decreased eGFR .
[18]

Volume 2 Issue 3 (2023) 3 https://doi.org/10.36922/an.1394