A novel LC-MS/MS assay for low concentrations of creatinine in sweat and saliva to validate biosensors for continuous monitoring of renal function
Journal of Chromatography B, Volume 1252, 2025, 124444: Fig. 2. Creatinine concentrations at the start and at the end of the haemodialysis cycle, measured in plasma, sweat and saliva. All creatinine concentrations in sweat and saliva were significantly (***) lower than in plasma. *** = p < 0.001.
The goal of this study was to develop and validate a novel LC-MS/MS assay capable of accurately measuring low concentrations of creatinine in sweat and saliva. These non-invasively obtainable biofluids are explored as alternatives to blood for continuous monitoring of kidney function, particularly during haemodialysis.
By analyzing samples from forty dialysis patients, the study demonstrated strong correlations between creatinine levels in sweat, saliva, and plasma. The validated assay showed high accuracy and precision, with a low limit of quantitation. These findings support the potential use of sweat and saliva in wearable biosensor systems, offering a foundation for patient-friendly, non-invasive kidney monitoring technologies.
The original article
A novel LC-MS/MS assay for low concentrations of creatinine in sweat and saliva to validate biosensors for continuous monitoring of renal function
Sophie Adelaars, Chyara S.M. Lapré, Patricia Raaijmakers, Constantijn J.A.M. Konings, Massimo Mischi, R. Arthur Bouwman, Daan van de Kerkhof
Journal of Chromatography B, Volume 1252, 2025, 124444
https://doi.org/10.1016/j.jchromb.2024.124444
licensed under CC-BY 4.0
Selected sections from the article follow. Formats and hyperlinks were adapted from the original.
Accurately and timely monitoring of renal function is critical for managing patients at risk of acute kidney injury (AKI), a condition that affects approximately 12% of hospitalized patients [1] and up to 57% of ICU patients [2]. In about 20% of these cases, progression occurs towards higher stages of AKI, with significantly longer periods of hospitalization, increased mortality rates and more frequent need for invasive renal replacement therapy [1]. This progression could potentially be delayed when patients are proactively treated based on frequent or continuous monitoring of renal function related biomarkers, such as creatinine.
As renal failure can progress rapidly, frequent venepuncture for creatinine analysis is required. Continuous monitoring of creatinine concentration can possibly improve effective patient management by further minimizing diagnostic delay. Recent developments in microfluidic devices with biosensors have sparked interest in alternative biofluids, such as sweat and saliva, as non-invasive matrices for continuous creatinine monitoring [3]. The existing literature demonstrates moderate to high correlation coefficients between saliva and plasma creatinine concentrations, implying a potential utility of saliva in the diagnosis and monitoring of patients with AKI [4], [5], [6], [7]. However, the correlation between sweat and plasma creatinine concentrations remains poorly investigated. Contrarily to the relatively high urea concentrations in sweat, which supposedly contributes significantly to urea clearance in renal failure [8], [9], sweat creatinine concentration generally remains <15 µmol/L. Besides the fact that this complicates correlation studies, it also suggests that creatinine clearance via sweat is physiologically irrelevant in healthy conditions as well as in renal failure, as hypothesized by Al-Tamer et al. [16].
To enable accurate quantification of creatinine in these alternative biofluids, especially in sweat where concentrations are low, the development of a highly sensitive and specific analytical method is required. Conventional enzymatic assays, with limits of quantification (LoQ) around 10 µmol/L, lack the sensitivity needed for this purpose. Our developed LC-MS/MS assay, with its lower limit of quantification, aims to serve as a reference method to validate emerging biosensors for real-time, continuous renal function monitoring. These biosensors hold the potential to minimize diagnostic delays, offering more immediate insight into renal function status [10].
In this study, we aimed to develop and validate a novel LC-MS/MS assay that meets these sensitivity requirements with a significantly lower LoQ than conventional enzymatic methods. This work builds upon our previous research, in which we explored the concentrations of urea and creatinine in sweat and saliva from the same cohort of patients [11]. However, the enzymatic assay used in that study faced limitations; it was not sensitive enough to quantify creatinine in many samples due to the assay’s LoQ, and it required larger sweat volumes than practically collected. The current study seeks to address these limitations by presenting an LC-MS/MS method that requires smaller sample volumes while offering a lower LoQ. Additionally, the assay can be utilized as a reference method to validate the performance of emerging biosensors aimed at continuous monitoring of renal function. Sweat and saliva samples were collected from forty haemodialysis patients to explore the feasibility of using sweat and saliva for continuous, non-invasive monitoring of renal function.
2. Methods
2.1.3. LC-MS/MS
The analysis was performed using an Acquity Ultraperformance Liquid Chromatograph (UP-LC) from Waters (Milford, MA, USA). 5 µL of sample was injected onto a 4 mm × 3 mm strong cation exchange SecurityGuard column (Phenomenex, Macclesfield, UK). Every measurement started with 100% mobile phase A (water + 2 mmol/L ammoniumacetate + 0.1% FA), stepping up at 0.40 min to 100% mobile phase B (water + 50 mmol/L ammoniumacetate + 10% MeOH + 0.1% FA), holding for 0.20 min before returning to initial conditions until a total runtime of 1.10 min, with a flow rate of 0.6 mL/min and a column temperature of 30 °C. A typical chromatograms obtained for sweat samples is shown in Supplemental Fig. S1.
An Acquity Xevo TQ-S triple MS/MS from Waters (Milford, MA, USA) was used with MassLynx software from Waters (Milford, MA, USA) and results were analysed with TargetLynx software from Waters (Milford, MA, USA). Multiple Reaction Monitoring (MRM) in the positive electrospray ionization (ESI+) mode was used as MS acquisition mode. MS settings were set as follows: capillary voltage: 2.40 kV, cone voltage: 40 V, desolvation temperature: 200 °C, source temperature: 150 °C, desolvation gas flow: 550 L/h, nebulizer gas: 7.0 bar, collision gas: 0.15 mL/min, dwell time: 0.063 s. The specific transitions and collision voltage were monitored and optimized in Multiple Reaction Monitoring (MRM) (Supplemental Table S1).
3. Results
3.4. Creatinine concentrations and correlation
The results of the creatinine concentrations in plasma, sweat and saliva are summarized in Table 2 and visualized in Fig. 2. Median creatinine concentrations in plasma decreased significantly during haemodialysis, from 677 µmol/L to 232 µmol/L (p < 0.001). Similarly, the concentrations decreased significantly in saliva, from 59.2 µmol/L to 14.0 µmol/L (p < 0.001), and in sweat, from 41.2 µmol/L to 18.4 µmol/L (p < 0.001).
Journal of Chromatography B, Volume 1252, 2025, 124444: Fig. 2. Creatinine concentrations at the start and at the end of the haemodialysis cycle, measured in plasma, sweat and saliva. All creatinine concentrations in sweat and saliva were significantly (***) lower than in plasma. *** = p < 0.001.
The mean sweat-to-plasma ratio was 0.071 (95% Confidence Interval (CI): 0.005–0.138). This ratio appeared to follow a decreasing pattern with increasing plasma creatinine concentrations (Fig. 3A). The mean saliva-to-plasma ratio was 0.109 (95% CI: −0.066 to 0.283) and appeared constant across the entire range of measured plasma creatinine concentrations (Fig. 3B).
Journal of Chromatography B, Volume 1252, 2025, 124444: Fig. 3. (A) The mean sweat-to-plasma was 0.071 and appeared to follow a decreasing pattern with increasing plasma creatinine concentrations. (B) The mean saliva-to-plasma ratio was 0.109 and appears constant across the range of the measured plasma creatinine concentration.
The Spearman’s correlation coefficient between creatinine concentrations in plasma with sweat was 0.68 (95% confidence interval (CI): 0.49–0.81, p < 0.001), while the correlation between creatinine concentrations in plasma with saliva was 0.80 (95% CI: 0.71–0.85, p < 0.001) (Supplemental Fig. S6). The results of the Passing–Bablok regression analysis, assessing the agreement between plasma and sweat and saliva creatinine concentrations, are also visualized in Supplemental Fig. S6.
3.5. Sweat rate dependency
Multivariable linear regression analysis to predict the plasma creatinine concentration based on sweat creatinine concentration revealed a significant positive effect of collected sweat volume (p < 0.001) (Fig. 4). To enable visualization, sweat volumes were fixed at specific values (2, 5, 6, and 10 µmol/L), and corresponding regression lines were plotted as examples. Specifically, higher sweat volumes were associated with lower concentrations in sweat and higher predicted concentrations in plasma. However, there was no direct correlation between the concentrations in sweat and sweat volume (r: −0.26, p < 0.05, R2adj: 0.06); therefore, the concentrations in sweat were not corrected for the collected volume (Supplemental Fig. S7).
Journal of Chromatography B, Volume 1252, 2025, 124444: Fig. 4. Multivariable linear regression analysis revealed a significant effect of the collected sweat volume on the prediction of plasma creatinine concentration with sweat creatinine concentration. Regression lines at different sweat volumes are visualized in blue. Sweat volumes of 10, 40 and 70 µL were filled in as constants in the multivariable linear regression formula (y = ax1 + bx2 + c, where y = plasma creatinine concentrations, x1 = sweat creatinine concentration, x2 = sweat volume and a, b and c are constants determined by the model). To demonstrate, a sweat creatinine concentration of 40 µmol/L without taking volume into account would predict a plasma creatinine concentration of ±633 µmol/L (black square). With the volumes 10 µL, 40 µL and 70 µL, the plasma concentration would be predicted to be 528, 700 and 872 µmol/L (red squares), respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
4. Discussion
In summary, this study builds upon our previous research by employing a novel LC-MS/MS assay to quantify creatinine concentrations in sweat and saliva, enabling the measurement of lower analyte concentrations than conventional enzymatic methods. While our earlier work was the first to explore urea and creatinine in these biofluids within this patient cohort, this follow-up study provides a more comprehensive dataset that enhances our understanding of the correlations between these non-invasive biomarkers and plasma creatinine. The strong correlations observed support the potential of sweat and salivary creatinine as valid indicators of renal function. Moreover, these findings might pave the way for future applications in validating wearable biosensors for continuous monitoring of patients with AKI in low-resource settings, ultimately improving patient care through real-time assessment and timely intervention.
