Method validation. Linearity and limit of detection Linearity was evaluated by preparing calibration lines (n = 3) on three consecutive days. The calibration ranges are shown in Table 3. All calibration lines were fitted to a 1/x2 weighted linear regression model. The limits of detection (LOD) and limits of quantification (LOQ) were calculated as LOD = 3 × Sa/b, LOQ = 10 × Sa/b, where Sa is the standard deviation of the y-intercept, b is the slope of the calibration curve25.
Method validation. Linearity The linearity within the calibration range was investigated with calibration lines (n=3 in the first batch, n=1 in the second and third batch). Calibration samples were prepared by spiking calibrations standards (C0 - C9) into brain lysate samples (1mg protein/mL, 100 μL). All the calibration lines were fitted to a non-weighted linear regression model. Precision The within-run and between-run precisions samples were unspiked brain lysates (n=5) and spiked brain lysates at C8 level (n=5) over three separate batches. Within-run precision was calculated as the RSD of the peak area ratio of the precision samples from day 1 (n=5). Between-run precision was calculated as the RSD of the peak area ratio of the precision samples from all three batches (n=15). Recovery and matrix effect Recovery and matrix effect were evaluated by spiking IS solutions [low level(C3), medium level(C5) and high level(C8)] into brain lysate samples (n=5). Recovery was the ratio of ISTD peak areas acquired from brain lysate samples with IS solutions spiked-before and spiked-after extraction. Matrix effect was the ratio of ISTD peak areas acquired from brain lysate samples with IS solutions spiked-after extraction and academic IS standard samples.
Method validation. The optimized method was validated based on guidelines from FDA, EMA and ICH for bioanalytical method validation. Validation including linearity, precision, recovery and matrix effect was performed. The linearity and precision of analytes from each class are summarized in Table 3, recovery and matrix effect were summarized in Figure 2. Linearity The square of correlation coefficient (R2) values are all between 0.994 and 0.999 (significance F <0.05), and for at least 75% of the back calculated concentrations of analytes were within ±15% (20% for LLOQ) of the nominal value, which indicated that the linearity is good for all the analytes in the calibration range. Unexpectedly, different from pre-test results, 18:1-OH-PE-N-20:4 and p18:0-OH-PE-N-20:4 were not detected at endogenous levels, and peaks were only seen at higher than C6 levels, which might be caused by deteriorated sensitivity of the instrument. Precision The within-run and between-run precisions samples were unspiked brain lysates (endogenous level, n=5) and spiked brain lysates (C8 level, n=5) over three separate batches. For most of the analytes, within-run and between-run precisions were lower than 15%, indicating good repeatability (Table 3). However, 2-AG showed a within-run variability of 21.8% at endogenous level, and very high between-run variability. Multiple reasons including high background at the transition of 2-AG and low-signal of the internal standard 2-AG-d8 may have caused the high variability. With these method, 2-AG can only be reported for samples that can be measured in one single batch and the data should be treated with caution. Recovery and matrix effect were assessed at three different levels [low-level (C3), medium- level (C5), and high-level (C8), n=5 at each level]. For all the analytes except 2-AG, the recoveries ranged from 71.7% to 100.2%. Matrix effects ranged from 78.1% to 204.2% (Figure 2). For each analyte, the effect of recovery and matrix effect was consistent at three different levels. The exception was 2-AG-d8, which showed extremely low recovery at low and medium level, which may be the reason that caused the high variability. Minor ion suppression was observed for the NAEs, while significant ion enhancement was observed for FFAs, NAPEs and pNAPEs. The mechanism of ion suppression can be explained by the competition from the co-eluting matrix for the limited charges on the sprayed droplets formed by ESI20,21. Besides, co-eluting matrix may increase the visco...
Method validation. A procedure blank was analyzed periodically and it was prepared using the same reagent and procedures for the samples. In addition, to assess efficiency of recovery procedure, two of the 14 samples were spiked with 20(mg/L) PAHs standard. Since the LoD calculation performed in the lab was insufficient, the LoD was determined based on an interpolated method using the calibration curves provided by the laboratory. Percent recovery calculations = {(Spike result/expected result) * (spike volume/original volume)} *100
Method validation. Feasibility studies have proven a [+] method acceptable for the determination of drug levels in [+] matrices. Each component of the study drug is assayed independently. The method will be validated (GLP) in matrices corresponding to samples specified by the clinical study protocol for pharmacokinetic analysis [+].
Method validation. All methods used to test / inspect the components involved in manufacturing / packaging of the Product or testing / inspection of the Product shall be validated according to cGMP requirements and ____________ procedures.
Method validation. All methods used to test / inspect the material shall be validated according to cGMP requirements and KING’s procedures.
Method validation. Heading 4.1.7
Method validation. The validation of the method was discussed in detail in our previous study (Xxxx et al., 2020). In that study, the model presented here was applied to a relatively large data set of human eCAP growth function recordings. This data set consisted of 4982 eCAPs from 111 CI recipients who received a HiRes90K device (Advanced Bionics, Valencia, CA), either with a 1J or a Mid-Scala electrode array. The eCAPs were recorded measured on eight odd electrode contacts with stimulus levels from 50 to 500 current unit. We have validated both steps of the method. First, we validated step one, namely the estimation of the URh, by comparing the resulting eCAPps obtained with our estimated URh (Dong et al., 2020) to the eCAPps obtained with URgp (Xxxxxxx et al., 1992) in step two. Based on the goodness of fit measure (NRMSE, the normalized root mean square error provided in MATLAB), the eCAPs achieved with URh were better than those achieved with URgp (Xxxx et al., 2020). The URh reduced the fitting error for all eCAPs by approximately 18%. This result supported our assumption that the UR of human AN fibers differs from the URgp (Xxxxxxx et al., 1992). The assumption that the UR is constant may be contested, as it can hypothetically vary across subjects, electrodes and/or current levels. However, the assumption of UR constancy is necessary, because a fixed UR is needed to optimize the derivation of CDLD in step two. As such, the UR is used solely as a necessary intermediate step to extract a valid CDLD from the eCAP. While a fixed UR is necessary and sufficient for our goal, our deconvolution model can nonetheless be used to investigate whether the UR differs across subjects or different stimulus conditions by running the deconvolution model for each condition separately. However, to more conclusively resolve such questions, direct recordings of the URh are necessary. Second, we validated the extraction of the CDLD with the fixed UR by evaluating the goodness of fit of the predicted eCAPs. In general, 93.6% of the recorded eCAPs were predicted accurately, with a >0.9 goodness of fit (NRMSE). Thus, these CDLDs provided a good picture of the temporal firing properties of the AN fibers in eCAPs. Importantly, realistic CDLDs were obtained that lacked any negative phases without any post-processing. The remaining 322 eCAPs had deviant waveforms, with relatively small N1 peaks and large P1 peaks; thus, they could not be predicted well with our model (NRMSE <0.9). This may have be...
