Method Validation Results and Discussion

3.5.1. Selectivity

An absence of interfering signals at or near the retention times of the analytes of interest, diclazepam, delorazepam, lormetazepam, flubromazepam and pyrazolam, was demonstrated by this experiment, there were no viable peaks detected for the ten sources of blank blood tested. Figures 22 and 23 display example chromatograms with and without internal standard respectively, from one of the sources of blank blood tested; the analyte retention times have been added to show each analytes position and the lack of interference in the area of interest. The green and blue peaks are lorazepam-D4 and diazepam-D5 respectively.

68 Figure 22: Example chromatogram of method selectivity, with internal standard,

lack of interfering signal at the retention time of the analytes of interest.

Figure 23: Example of method selectivity, without internal standard, lack of interfering signal at the retention time of the analytes of interest.

69

3.5.2. Specificity

No interfering peaks were detected from any of the 31 drugs/metabolites at the retention times of interest of any of the analytes tested for in this method. Figure 24 shows the blank signal, which was obtained after the injection of solution 3 (see 3.4.2) which included cocaine, benzoylecognine, amphetamine, methamphetamine, MDMA, MDA, MDEA, morphine, codeine, dihydrocodeine and 6-MAM. All drugs within Table 20 gave no signal, similar to the example shown in Figure 24.

Figure 24: Example chromatogram of method specificity, from other

drugs/metabolites, shown by the lack of signal at the analytes of interest retention time.

3.5.3. Linearity

Linearity was tested with ten calibration curves which where ran over a ten-month period. The linearity data for all 5 analytes is displayed in Table 21. All analytes displayed good linearity with 1/x weighted linear regression over the calibration range 0.005 to 0.20 mg/L, with all curves r>0.99. All calibrators were within the ±20% accuracy criterion. 1/x weighting was chosen to minimise the distribution of

70 variance across the calibration range, larger deviations present in the higher concentrations will influence the curve more leading to impaired accuracy on the at the lower concentrations. To balance this out weighting is applied, when 1/x weighting is applied the slope more closely approximates the majority of calibrator points. Figure 25 gives example calibration curves for diclazepam, delorazepam, lormetazepam, flubromazepam and pyrazolam. The residuals from five calibration curves were plotted for each analyte (as per SWGTOX recommendation), see Figure 26. The residuals represent the difference in observed value from the predicted value. These values can be positive or negative depending on if the observed value is more or less than the predicted value. A residual plot with randomly distributed positive and negative values around the 0 axis with no defined trend demonstrates a linear model is appropriate. (Polettini, 2006) There is no convincing pattern in the residual plots for diclazepam, delorazepam, lormetazepam, and pyrazolam, demonstrating linearity is achieved. The range of variances increases with the concentration confirming a weighted calibration is required which was applied. (Polettini, 2006) Flubromazepam appears to be more positively distributed and therefore a quadratic fit may be more appropriate however the calibration curves produced have all been linear with a r>0.99. It is recognised that when calibration points are displaying accuracy within the acceptable limits then an alternative fit is not required and deviations can be overlooked. (Peters and Maurer, 2002)

Table 21: Blood method validation linearity data Analyte Mean r (n=10) SD %CV Calibrator % accuracy range Diclazepam 0.9994 0.0005 0.0539 93-108 Delorazepam 0.9992 0.0011 0.1130 84-114 Lormetazepam 0.9981 0.0029 0.2874 90-114 Pyrazolam 0.9963 0.0024 0.2362 81-112 Flubromazepam 0.9990 0.0007 0.0686 82-116

71 Figure 25: Example calibration curves for diclazepam, delorazepam,

72 Figure 26: Residual plots for diclazepam, delorazepam, lormetazepam,

flubromazepam and pyrazolam

The inter-day precision of the LLOQ and ULOQ was calculated in order to evaluate the low and high point of the calibration curve for all analytes. The inter- day %CV met the acceptance criterion of <20% for diclazepam (<1.5), delorazepam (<9.5), lormetazepam (11.2), flubromazepam (<11.6%) and pyrazolam (<17.2%) for both the LLOQ and ULOQ showing that both high and low points are precise for this method. Diclazepam LLOQ data demonstrates good precision giving a SD and %CV of 0 to 3 decimal places. The %bias was also calculated and demonstrated good accuracy for the LLOQ. Table 22 shows the LLOQ mean concentration, SD and inter-day precision and bias.

Table 22: Blood method validation LLOQ (0.005 mg/L) inter-day precision and bias data

Diclazepam Delorazepam Lormetazepam Flubromazepam Pyrazolam Mean

concentration 0.005 0.005 0.005 0.005 0.005

SD 0.000 0.000 0.001 0.001 0.001

%CV 0.0 9.4 11 12 17

73 Table 23 shows the data used to calculate the ULOQ precision and bias. The precision and bias for the ULOQ for all analytes is particularly good at <5% and <2%, respectively.

Table 23: Blood method validation ULOQ (0.20 mg/L) inter-day precision and bias data

Diclazepam Delorazepam Lormetazepam Flubromazepam Pyrazolam Mean concentration 0.197 0.198 0.199 0.199 0.200 SD 0.003 0.004 0.006 0.004 0.010 %CV 1.4 2.1 3.2 2.1 5.0 %Bias 3 1 1 1 2 3.5.4. Limit of detection

The limit of detection (LOD) was determined by evaluating the signal-to-noise ratios for the analytes of interest. LOD is an approximation of the lowest concentration detectable by the method and can change depending on the set of experimental data used, laboratory conditions and how clean the instrument is. Since this is just an approximation of detectability the S/N ratio is not used to determine LOQ. S/N uses the baseline as a blank in which to compare to the signal generated by the analyte, a blank without the analyte cannot be used accurately as the noise is dependent on the signal magnitude. (Desimoni and Brunetti, 2015) A signal-to-noise ratio ≥3 at the correct retention is considered to be a detectable peak. Using the baseline means instrumental noise is considered but also can widely vary between injections. The S/N ratio alone is not the only aspect considered when using this technique for LOD determination; the peaks must also be visually inspected to ensure good Gaussian peak shapes. Diclazepam, delorazepam, lormetazepam, flubromazepam and pyrazolam all had signal-to-noise ratios >3 for all the concentrations tested therefore the LOD for all five drugs was determined as 0.002 mg/L (2 ng/mL). Flubromazepam gave the lowest signal-to-noise ratio across the three concentrations. Table 24 displays the signal-to-noise ratio results for all analytes at the three concentrations tested.

74 Table 24: Blood method - Limit of detection (LOD) data

Analyte Signal-to-noise ratio

LOD 3 LOD 2.5 LOD 2

QT QL QT QL QT QL Diclazepam 15.2 26.5 15.2 71.5 13.5 44.5 Delorazepam 53.5 33.5 53.5 4.5 17.5 8.5 Lormetazepam 26.2 7.0 26.2 9.8 12.8 6.5 Pyrazolam 12.0 15.0 12.0 11.0 6.5 12.5 Flubromazepam 6.0 16.0 6.0 24.5 4.5 14.5

3.5.5. Precision and bias

The results for the intra- and inter-day precision and bias data are shown in Tables 25 and 26. The precision evaluated by the %CV was within the acceptance criterion of <20% for diclazepam, delorazepam, lormetazepam, pyrazolam and flubromazepam for all three QC concentrations for both intra- and inter-day batches. The bias evaluated by %bias was within the acceptance criterion of ±20% (accuracy of 80-120%) for diclazepam, delorazepam, lormetazepam, pyrazolam and flubromazepam for all three concentrations for both intra- and inter-day batches.

Table 25: Blood method validation intra-day precision and bias Analyte

Diclazepam Delorazepam Lormetazepam Flubromazepam Pyrazolam

QC 1 0.015 mg/L Mean (n = 5) 0.016 0.016 0.015 0.016 0.014 (n = 4) SD 0.001 0.001 0.002 0.000 0.002 %CV 3.1 9.3 11 2.5 13 %Bias 10 4 -1 5 -4 QC 2 0.042 mg/L Mean (n = 5) 0.043 0.041 0.043 0.043 0.050 SD 0.002 0.002 0.002 0.002 0.001 %CV 3.7 4.7 4.3 4.4 3.1 %Bias 2 -3 3 3 18 QC 3 0.150 mg/L Mean (n = 5) 0.152 0.146 0.151 0.160 0.179 (n = 4) SD 0.005 0.002 0.007 0.003 0.022 %CV 3.0 1.7 4.7 2.1 12 %Bias 2 --3 0 7 19

75 Table 26: Blood Validation Method Inter-day precision and bias

Analyte

Diclazepam Delorazepam Lormetazepam Flubromazepam Pyrazolam

QC 1 _{0.015 mg/L} Mean (n = 5) _{0.015 0.014 0.014 0.016 0.014} Low _{0.012 0.008 0.010 0.013 0.007} High _{0.017 0.017 0.017 0.019 0.020} SD _{0.001 0.002 0.002 0.001 0.003} %CV _{7.9 17 13 8.4 20} %Bias _{2 -6 -5 8 -6} QC 2 _{0.042 mg/L} Mean (n = 5) _{0.044 0.041 0.041 0.048 0.046} Low _{0.038 0.030 0.031 0.037 0.039} High _{0.052 0.047 0.047 0.057 0.056} SD _{0.004 0.005 0.005 0.005 0.007} %CV _{8.8 12 11 11 14} %Bias _{5 -2 -2 13 9} QC 3 _{0.150 mg/L} Mean (n = 5) _{0.159 0.152 0.159 0.177 0.179} Low _{0.135 0.114 0.113 0.139 0.115} High _{0.180 0.173 0.191 0.216 0.221} SD _{0.011 0.016 0.022 0.021 0.034} %CV _{7.1 11 14 12 19} %Bias _{6 1 6 18 19}

3.5.6. Matrix effects, recovery and process efficiency

The matrix effects for diclazepam, delorazepam, lormetazepam, pyrazolam and flubromazepam are shown for the low concentration (0.010 mg/L) in Table 27 and the high concentration (0.17 mg/L) in Table 28.

Table 27: Blood method validation - ME results at 0.010 mg/L %ME (n=2) 0.010 mg/L

Diclazepam Delorazepam Lormetazepam Pyrazolam Flubromazepam

QT QT QT QT QT Overall Mean %ME (n=10) -14 -19 -21 -10 -14 Overall ME %CV 7.5 7.6 7.4 8.5 8.8

The mean %ME for diclazepam, delorazepam, lormetazepam, pyrazolam and flubromazepam were -14, -19, -21, -10 and -14% respectively at the low concentration tested. All mean results were within the ±25% criterion for %ME and <15% for precision.

76 Table 28: Blood method validation - ME results at 0.170 mg/L

%ME (n=2) 0.170 mg/L

Diclazepam Delorazepam Lormetazepam Pyrazolam Flubromazepam

QT QT QT QT QT Overall Mean %ME (n=10) 16 14 12 9 21 Overall ME %CV 1.2 2.0 2.0 2.6 1.8

The mean %ME for diclazepam, delorazepam, lormetazepam, flubromazepam and pyrazolam were 16, 14, 12, 21 and 9% respectively at the high concentration tested. All mean results were within the ±25% criterion and the precision well under the <15% criteria.

Table 29: Blood method validation - ME internal standard results

Diazepam-D5 %ME (n=2) (0.010 mg/L) (0.170 mg/L) Overall Mean %ME (n=10) -6 6 Overall ME %CV 3.5 2.4

The %ME for the internal standard is shown in Table 29; there is no significant ME evident for the internal standard.

The %RE and %PE for diclazepam, delorazepam, lormetazepam, pyrazolam and flubromazepam are shown for the low concentration (0.010 mg/L) in Table 30 and the high concentration (0.17 mg/L) in Table 31.

77 Table 30: Blood method validation - RE and PE results at 0.010 mg/L

Drug QT (0.010 mg/L) %RE mean (%CV) (n=10) %PE mean (%CV) (n=10) Diclazepam 57 (10) 49 (7.7) Delorazepam 64 (9.0) 52 (9.2) Lormetazepam 42 (9.9) 33 (10) Flubromazepam 31 (18) 26 (16) Pyrazolam 28 (18) 25 (17)

Table 31: Blood method validation - RE and PE results 0.17 mg/L Drug QT (0.017 mg/L) %RE mean (%CV) (n=10) %PE mean (%CV) (n=10) Diclazepam 64 (6.7) 75 (7.4) Delorazepam 63 (6.8) 73 (7.9) Lormetazepam 45 (7.6) 50 (9.0) Flubromazepam 43 (12) 52 (13) Pyrazolam 27 (7.6) 30 (9.5)

Table 32 Blood method validation- RE and PE internal standard data Drug %RE mean

(%CV) (n=10) %PE mean (%CV) (n=10) Diazepam-D5 17 (3.1) 16 (2.8)

The RE and PE results for diclazepam and delorazepam are acceptable and around 50% or above for both concentrations tested. Lormetazepam was below 50% for RE and PE but over 40% for RE for both concentrations, this is considered suboptimal but should not have a deleterious effect. Flubromazepam has sufficient PE and above 40% RE at the high concentration tested but under 40% for RE and PE at the low concentration tested, therefore the accuracy at the lower end of the calibration curve may not be as good as at the high end; however

78 the LLOQ shows good precision and bias (see Table 22) and the analyte is detectable down to 0.002 mg/L. The RE and PE for pyrazolam at both concentrations is poor at 30% and under however the LLOQ shows good precision and bias (see Table 22) and the analyte is detectable down to 0.002 mg/L. This suggests an issue with this extraction for pyrazolam. One possibility is that pyrazolam is less soluble in methanol than it is in acetonitrile (pyrazolam is purchased in acetonitrile) and therefore does not fully dissolve at the reconstitution step or there is a stability issue with pyrazolam in the combined methanolic working solution. The ME for all analytes was acceptable and although there was low RE and PE for pyrazolam and flubromazepam the precision and bias were acceptable and an acceptable LOD down to 0.002 mg/L was achieved for both.

The mean PE and RE for the internal standard was suboptimal as it is low at 16% and 17%, see Table 32 however it is consistent across all ten sources and the peak areas are sufficiently strong at 1,000,000 cps or above.

3.5.7. Carryover

Carryover is assessed by the absence or presence of peaks in the injections of drug-free reconstitution solution directly following injections of a high concentration of the analytes of interest. There were no peaks in the drug-free reconstitution solution injection that followed three injections of 0.50 mg/L of diclazepam, delorazepam, lormetazepam, pyrazolam and flubromazepam, see Figure 27.

79 Figure 27: Example chromatogram of the absence of peaks after injections of 0.50

mg/L, demonstrating the lack of carryover

3.5.8. Autosampler stability

The results from this experiment were plotted on graphs to evaluate autosampler stability. Figures 28 and 29 shows all analytes for the low (0.015 mg/L) and the high (0.15 mg/L) concentration, respectively. The response did not fall below 20% of the time zero response (t0) within 48 hours in a temperature controlled room (16°C - 24°C) and all analytes were considered stable under these conditions.

Lormetazepam appeared to be the least stable and showed a downward trend for both concentrations compared to the other analytes however the responses for all time points were within 20% of t0.

80 Figure 28: Autosampler stability for all blood method analytes at a concentration of

0.015 mg/L over 48 hours at 16-24 °C

Figure 29: Autosampler stability for all blood method analytes at a concentration of 0.15 mg/L over 48 hours at 16-24 °C.

81 3.6. Conclusion

This method is suitable for the quantitative analysis of blood samples for diclazepam and its metabolites (delorazepam and lormetazepam), flubromazepam and pyrazolam. The method spans a concentration range (0.005-0.20 mg/L) for these drugs in blood with acceptable linearity. The method is specific and selective, and demonstrated no carryover at the concentration tested. Bias and precision were acceptable for all analytes. Matrix effects were found to be acceptable for all analytes and within the ±25% criterion for %ME and <15% for precision. Recovery and process efficiency was suitable for diclazepam and delorazepam, it was suboptimal for lormetazepam and flubromazepam however the bias, precision and sensitivity were good. Pyrazolams recovery and process efficiency was particularly low and this should be considered when analysing real samples in particular decomposed samples as they may give a lower concentration and samples which are positive at the lower end of the concentration range may give a negative result. All analytes were stable in the autosampler stability up to 48 hours in a laboratory with the temperature range 16- 24 °C. A temperature-controlled autosampler would allow more control over the analyte stabilty however it has been demonstrated that analysis within 48 hours should not be an issue. As further work, freeze and thaw stability should be investigated for these analytes, as the long-term stabilty in blood is still unknown.

82

4. Development and validation of a qualitative LC-MS/MS method for the