Biosensor Strip for Rapid On-site Assessment of Levodopa Pharmacokinetics along with Motor Performance in Parkinson’s Disease

In-Vitro Analytical Performance Towards L-Dopa

The new L-Dopa biosensor strip has been evaluated for its stable fabrication and suitability of L-Dopa detection, (SI-4, Figure S1 and S2), where the analytical performance was assessed first under standard in vitro laboratory conditions. These initial studies included evaluation of the sensor sensitivity and selectivity, as well as assessments of the reproducibility and storage stability of the device, along with various control experiments. We initially examined the influence of the operating potential upon the sensitivity to 25 μM L-Dopa using applied potentials ranging from −0.1 to −0.5 V (Figure 2a). The gradual increase of the reduction current was observed upon decreasing the applied potential to −0.4 V, along with a slight decline afterward (due to oxygen interference). Consequently, 0.4 V was identified as the optimal potential for detecting the dopaquinone product of the enzymatic reaction. Subsequently, we evaluated the sensitivity of the biosensor by recording the current signals for increasing L-Dopa concentrations in 5 μM steps, over the pathophysiological range (of 0 to 30 μM) (Figure 2b). These well-defined current signals led to a highly linear calibration plot over the entire range, as shown in Figure 2c. The linear regression equation for L-Dopa sensing is expressed as follows: ΔI (nA)=21.5 [L-Dopa (μM)] +121.8, boasting a remarkable correlation coefficient of 0.99. Calculating the limit of detection based on the sensitivity from the calibration plot and the response of the blank (n=5), we found it to be 0.13 μM. The linear dynamic range encompasses the clinical range in blood. Thereafter, we examined the selectivity of the biosensor in the presence of a large excess of common potential interferences (uric acid (UA) and ascorbic acid (AA)) as well as potential tyrosinase substrates (tyrosine (Ty), acetaminophen (AP), and dopamine (DA)) (Figure 2d). These data indicate that the biosensor strip is highly selective towards L-Dopa, displaying negligible response towards these potential interferences compared to the target drug. Since carbidopa is a component of anti-Parkinsonian medication that is usually administered simultaneously with L-Dopa to prevent adverse effects and enhance bioavailability, we assessed its potential interference. Selectivity assessment against carbidopa was performed at the same ratio (4:1, 10 μM L-Dopa and 2.5 μM carbidopa), showing no significant interference in various scenarios (Figure 2e), indicating minimal interference (less than 5% at low concentration and 18% at the same concentration) of coexisting carbidopa molecules, as desired for accurate and reliable L-Dopa detection. Following these assessments, the reproducibility of the L-Dopa sensor was scrutinized by measuring 10 μM L-Dopa concentration on seven different sensors with three replicates (Figure 2f). The obtained relative standard deviation of 8.6% underscores its high fabrication reproducibility. Furthermore, the shelf-life of the L-Dopa sensors was investigated, demonstrating high storage stability upon storing the biosensor strip under refrigerated humid conditions (4°C), with retention of 88% of the initial response after 10 days (Figure 2g) Overall, the data in Figure 2 signifies the suitability of the L-Dopa sensor strip for practical clinical applications.

Development of the U-Slope for Calibration-Free L-Dopa Monitoring

Upon confirming the attractive analytical performance under the standard conditions, we introduced the U-slope concept for blood L-Dopa detection in a manner analogous to on-site insulin sensing.^[28] The U-Slope approach enables reliable calibration-free blood measurements, thus dramatically reducing the assay time, complexity, and costs towards routine decentralized POC testing. Exploiting these advantages, in this work, we established a U-slope using finger-pricked capillary blood samples based on a developed L-Dopa sensor. The U-slope for L-Dopa was traced by recording the chronoamperometric responses obtained at the blank and spiked L-Dopa concentration (5 μM) in capillary blood samples collected from numerous healthy volunteers and PwP. The detailed processes for U-slope preparation and obtaining the U-slope have been described and depicted in Figure S3. Briefly, a two-point chronoamperometric response was obtained for every capillary blood sample, and using the current values the individual slopes were traced. Upon assessing these individual slopes (n=50) statistically, we obtained the normal distribution pattern (with mean=0.133 μA/μM; standard deviation σ= 0.007 μA/μM), which has provided the value of the averaged slope of 0.133 μA/μM (Figure 2h, i). This resulting average slope ($\mathbf{K}$) allows the direct measurement of blood $\text{L-Dopa}$ concentrations by converting the recorded current sample signal using Eq. (1):

Where “$\left[\text{L-Dopa}\right]$” is concentrations of L-Dopa, “$\Delta$ Current” is the change in the current value from the blank, and “$\mathbf{K}$” is the obtained U-slope from the multiple blood samples.

Clinical Studies in PwP: On-Site L-Dopa and PD Motor Score Assessments

The developed L-Dopa sensor strips, with a U-slope value of 0.133 μA/μM, were evaluated in a clinical pilot trial (NCT04735627) to assess their ability to track rapidly the drug pharmacokinetics in decentralized (i.e., on-site) settings. Before starting on-site clinical studies in PwP, we first validated the sensor’s performance in blood samples from various healthy participants (controls) using the sensor without the immobilized Tyr enzyme and with the use of a placebo (in time-dependent studies involving both healthy and PwP). The participants in these studies have been summarized in Table S1. As illustrated in Figure S4, in both cases (controls), no current changes and temporal profiles were observed, because the sensor was unable to catalyze L-Dopa in the absence of the Tyr enzyme (in the first control), while in the second (placebo) case, no L-Dopa was available to be catalyzed. Following these control experiments, we tested various capillary blood samples from PwP (n=14). Ten microliters finger-prick capillary blood samples were thus analyzed at 10 minute intervals during 90 minutes, including before and after the participants took a single oral dose of carbidopa/L-Dopa in a 1:4 ratio, in a dose equal to their home dose or 1.5 times their home dose (up to 4 tablets maximum) to ensure adequate clinical effect. As illustrated in Figure 3a, at each point, the capillary blood was diluted with PBS (pH 7.4; 1:1 v/v ratio), and the L-Dopa level was estimated on-site using chronoamperometry. Such dilution equilibrates the blood sample’s pH and ionic strengths and delays blood clotting over the electrode surface, towards minimizing the instant fouling of the electrode surface. Thereafter, the obtained chronoamperometric values were extrapolated to estimate the corresponding L-Dopa concentrations using U-slope constant ($\mathbf{K}$) at respective points to track the L-Dopa blood profile. We obtained distinct temporal L-Dopa profiles (Figure 3c; red) which reflect the personal L-Dopa absorption and assimilation after the pill intake. The temporal profile of the capillary L-Dopa shows a distinct trend, where the L-Dopa level remains low and stagnant (called baseline level) before the carbidopa/L-Dopa pill administration, followed by increasing L-Dopa levels over time (~30 mins) for three consecutive recordings after medication ingestion, and then a gradual decline to the baseline (or closer to the baseline). A representative set of these chronoamperometric current signals obtained from these finger-prick blood samples is shown in Figure 3b. Notice the rising (blue) and falling (brown) current signals after the oral dose (at 0 min). This pattern reflects the L-Dopa absorption and assimilation at the gastrointestinal tract (increasing the concentration), thereby subsequent utilization of L-Dopa by the neuronal cells leads to the depletion of its concentration in blood. Similar observations of the L-Dopa pharmacokinetic temporal profiles were reported in previous studies, corroborating our findings.^[8,29–31]

Adequate L-Dopa therapy temporarily improves Parkinsonian motor symptoms to some extent. Symptom severity is generally assessed by medical providers’ visual observation using a validated scoring system, where higher scores indicate worse motor symptoms.^[32] We serially evaluated motor performance in PwP every 10 minutes along with the capillary blood testing. A movement disorders neurologist (KL) performed an abbreviated version of the Movement Disorders Society Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) Part III,^[32] which included the items for muscular rigidity, bradykinesia/hypokinesia, and tremor; each item was scored with integers from 0–4, where higher is worse, indicating more severe symptoms. The scores for these items were totaled (minimum score=0, maximum score=56). The motor scores were then plotted against the time along with the capillary blood L-Dopa concentration temporal profile. As expected, the motor scores were found to be inversely related to L-Dopa levels throughout the study for all subjects where the higher the signal and L-Dopa level, the lower the motor score (motor performance improves) (Figure 3c; black). This not only corroborates the fact that the elevated level of L-Dopa leads to reduced motor symptoms but also validates the functionality of the developed L-Dopameter for L-Dopa quantification in capillary blood.^[31] The representative temporal motor score plots obtained from six representative participants are displayed in Figure 3c (black) in correlation with the L-Dopa current responses temporal profiles following the administration of the L-Dopa carbidopa/L-Dopa tablet (red). Notice that for several participants, the falling L-Dopa levels (at times >60 min) are not accompanied by a worsening of the score, indicating a longer-lasting clinical effect of the drug. This supports that the same L-Dopa dose can affect motor symptoms differently among different individuals, as expected in PD, which is heterogeneous. Identifying such personal variations in the drug’s therapeutic effect could be useful for tailoring treatment regimens.

Further, we attempted to investigate the exact concentration fluctuations of the L-Dopa based on the U-slope approach from capillary blood. Thus, the biosensor current signals were converted to the corresponding L-Dopa concentrations (Figure 3d) using Eq. (1) (U-slope constant $\mathbf{K}$). The range of the L-Dopa levels obtained from our U-slope approach corroborates the findings from various clinical and pharmacokinetic studies,^[33–35] which indicate that the developed sensors and the demonstrated U-slope are capable of determining L-Dopa levels in capillary blood. We observed in PwP that the peak L-Dopa levels showed subtle fluctuations ranging between 20–40 minutes, which are likely attributable to the pharmacokinetic variability among the individuals, the stage of PD, etc.^[36,37] Thereafter, we attempted to model the estimated changing L-Dopa concentrations (from the chronoamperometric response) in correlation with the motor scores. Here, the motor scores were inversely transformed and plotted with the obtained L-Dopa concentrations (Figure 3d; red vs blue). The temporal profiles of the inverse motor scores show a similar trend as the L-Dopa concentration temporal profiles, with a lag time of 10 minutes in almost all the cases, which could be attributed to the duration of time required for the L-Dopa to cross the blood–brain-barrier to affect motor symptoms. The correlation of the temporal profiles (L-Dopa concentrations and transformed motor scores) indicates the reliable, quick, onsite estimation of the L-Dopa level in the PwP, which not only mitigates the existing challenge to the L-Dopa monitoring but also helps in knowing the accurate level instantly during the clinical examination for better predication of dose optimization. Although this correlation in temporal trends validates the proper functionality of the developed L-Dopameter and strip-based method for L-Dopa monitoring, we explored to establish the correlation of L-Dopa levels with lab-based centralized HPLC techniques, as discussed in the following section.

Analytical Comparison of the L-Dopa Sensors with the HPLC Method

We also attempted to find the correlations of L-Dopa concentrations between the venous blood and capillary blood which were sampled at the same time from PwP. The collected venous blood samples were then preprocessed before sending for the evaluation with HPLC at centralized lab to determine the plasma L-Dopa concentrations in PwP (n=14) (Figure S5a). This process of HPLC evaluation at the centralized lab has been adopted to mimic the routine clinical assessment procedure. The obtained results were then plotted to trace the temporal profile based on venous L-Dopa concentrations (representative plots have been shown in Figure S5b). The L-Dopa temporal profiles obtained using the capillary blood samples and the disposable biosensor strips follow the expected temporal profiles, with similar patterns as in previously reported studies (involving venous blood based HPLC methods).^[8,29–31] The profiles observed with current HPLC assessment, show unusual inconsistent trends (with random irregular ups and downs), that are most likely due to the cumulative impact of errors associated with the sample handling, transport steps, and corresponding sample aging and delays. Correlating, these HPLC results with the devised inverse motor scores we found inconsistent trends in all the cases, compared to the good correlation to the motor performance obtained with the on-site L-Dopa sensor-based assessment. As is shown in Figure 3c, our on-the-spot tests display a good correlation between the trend in L-Dopa concentrations and the devised inverse of motor scores, offering a useful method for the POC therapeutic drug monitoring and assessment of the L-Dopa pharmacokinetics.