Method for characterizing heroin characteristics based on UPLC-MS / MS principal component analysis method
The highly correlated heroin sample groups were screened through UPLC-MS/MS principal component analysis method and Pearson distance method. Combined with the PCA results, the problem of insufficient traceability of heroin in the existing technology was solved, and the accurate analysis and correlation judgment of heroin samples were achieved, and case investigation and intelligence construction were supported.
Patent Information
- Application Number
- CN202411986622.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-31
- Publication Date
- 2025-08-26
AI Technical Summary
In the prior art, there is insufficient method for tracing heroin in the traceability of heroin, and it is difficult to effectively determine whether heroin samples from different sources come from the same processing batch or manufacturer.
UPLC-MS/MS principal component analysis method was used, combined with mathematical and statistical analysis methods, and 15 neutral/acid characteristic target substances in illegal heroin products were detected. Sample correlation was calculated by Pearson distance method, sample groups with extremely strong correlation were screened, and characteristic analysis was performed based on PCA results.
Accurate analysis of heroin samples is achieved, the correlation between samples can be determined, whether the same batch of samples comes from the same upstream distributor or processing plant, providing clues for the investigation of major cases and supporting the construction of political and legal intelligence.
Smart Images
Figure SMS_2 
Figure SMS_4 
Figure SMS_5
Abstract
Description
Technical Field
[0001] The present invention relates to a method for characterizing characteristics of heroin based on UPLC-MS / MS principal component analysis, and belongs to the field of dangerous goods detection. Background Art
[0002] Heroin (3,6-diacetylmorphine or diamorphine) is a semi-synthetic derivative of morphine, first synthesized in 1874. In addition to its powerful analgesic effects, it has also been found to cause severe dependence and illegal use (Wolff 1956), which can easily cause public health problems.
[0003] Previously reported heroin tracing and correlation studies have used gas chromatography (GC) (Dufey et al., 2007), gas chromatography tandem mass spectrometry (GC / MS-MS) (Yan et al., 2013), ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UPLC-Q-TOF) (Liu et al., 2015), and ultra-high performance liquid chromatography-mass spectrometry (UPLC-MS / MS) (Debrus et al., 2011). Salinity and purity can also be determined, as can analysis of samples for major alkaloids, adulterants, trace impurities, and inorganic elements (Dams et al., 2001). Heroin samples from sites A and B have been characterized using a combination of statistics, figures, charts, and other measures (Casale et al., 2006; Esseiva et al., 2005). Liu et al. clearly explained the theoretical and parameter basis of this method in their heroin traceability studies using neutral / acidic impurities (Liu et al., 2014; Liu et al., 2015). Furthermore, after years of research and accumulation, the National Narcotics Laboratory has established industry standards for heroin source tracing and characterization analysis, playing a significant role in the national fight against drug abuse. Summary of the Invention
[0004] The technical problem to be solved by the present invention is to provide a method for characterizing the characteristics of heroin based on UPLC-MS / MS principal component analysis, so as to supplement the traceability of heroin in the existing technology.
[0005] The technical solution of the present invention is: a method for characterizing the characteristics of heroin based on UPLC-MS / MS principal component analysis. Based on UPLC-MS / MS, 15 neutral / acidic characteristic target substances in illegal heroin products are detected and analyzed, and mathematical and statistical analysis methods are combined to complete sample analysis. By combining case information, relevant subgroups and Principal Component Analysis (PCA) results, the characteristics of heroin samples from different sources are analyzed. PCA is initially used to group the samples, and then the Pearson distance method is used to calculate the correlation between each group of samples. R>98 is selected as the threshold to screen out sample groups with extremely strong correlation.
[0006] Chromatographic conditions described: ACQUITY BEH C18 columns were 100 mm × 2.1 mm, 1.7 μm, the column oven temperature was 35 °C, the eluents were A: acetonitrile and B: 10 mM ammonium formate aqueous solution, and the flow rate was maintained at 0.4 mL / min.
[0007] The mass spectrometry conditions are as follows: electrospray positive ion source (ESI+), multiple reaction monitoring (MRM) mode; capillary voltage 0.50 kV; cone voltage 20 V; desolvation temperature 500°C; desolvation gas flow rate 1000 liters / hour; cone gas flow rate 100 liters / hour; collision gas: argon, flow rate 0.12 ml / minute; acquisition time 0-9.49 minutes.
[0008] The Pearson distance method: When the number of targets detected in the sample to be tested is greater than or equal to 8, the peak areas of the 15 targets are recorded. The peak area of the detected target is the peak area of its semi-quantitative ion pair, and the peak area of the undetected target is recorded as 200. The peak area of each target substance in the sample to be tested is divided by the sum of the peak areas of the 15 target substances in the sample to obtain the normalized peak area Si of each target substance. The correlation coefficient r between the two samples is:
[0009]
[0010] Where r is the correlation coefficient between the two samples, SXi and SYi are the normalized peak areas of each target in samples X and Y, respectively. If the correlation coefficient is greater than 0.98 and the correlation between samples X and Y is judged to be extremely strong, it is highly likely that the heroin samples in both cases originated from the same processing batch.
[0011] The present invention has the following beneficial effects: Correlation analysis of heroin samples seized in heroin drug cases was conducted in conjunction with standard methods. Using ultra-high performance liquid chromatography-tandem triple quadrupole mass spectrometry and modeling for precise analysis, correlations between samples can be determined, ultimately confirming whether the same batch of samples originated from the same upstream distributor or processing plant. Furthermore, comparative attribution of cases seized from different units, at different times, and at different locations can be performed, providing clues for major case investigations and offering technical support for the development of political and legal intelligence. DETAILED DESCRIPTION
[0012] 1.1 Reagents and Instruments
[0013] Acetonitrile, methanol, and ammonium formate (chromatographic grade, >99% purity) were purchased from Merck (Darmstadt, Germany). Analytical-grade dichloromethane, petroleum ether, and concentrated sulfuric acid (>98% purity) were purchased from Tianjin Damao Chemical Reagent Factory (Tianjin, China), Chengdu Kelong Chemical Reagent Factory (Sichuan, China), and Chongqing Chuandong Chemical (Chongqing, China), respectively. Deionized water was purified using a Milli-Q water purification system (Shanghai, China).
[0014] The UPLC-MS / MS instrument with an electrospray ionization (ESI) source was purchased from Waters Corporation (Aqcuity / Xevo TQ-S, USA); the analytical balance was purchased from Mettler-Toledo GmbH (XSR205DU, USA); the ultrasonic cleaning machine was manufactured by Kunshan Ultrasonic Instrument Co., Ltd. (KQ-100DV, China); and the vortex analyzer was purchased from Scientific Industries, Inc. (G560E, USA).
[0015] 1.2 Sampling
[0016] Thirty-eight heroin samples were obtained from drug-related cases seized by public security organs in various cities in Guizhou Province. P1-P2 were from city A, P3-P27, P32-P33 from city B, P28-P31 from city C, and P34-P38 from city D. Quality control (QC) samples of heroin were obtained from the National Narcotics Laboratory of the Ministry of Public Security (Beijing, China).
[0017] 1.3 Analysis methods
[0018] 1.3.1 Sample preparation
[0019] Referring to the industry standard JD / Y JY03.01-2017 (JD / Y JY03.01-2017), the sample was ground into powder before collecting the spectrum. Weigh 100 mg of heroin sample (the allowed range is 100±10 mg), place it in a centrifuge tube, and sonicate with 2 mL of dilute sulfuric acid solution (0.25 mol / L). Add 2.5 mL of a mixture of petroleum ether and dichloromethane (v / v=3:2) to dissolve the sample, shake it horizontally for 30 minutes, and centrifuge it for 5 minutes. Then remove 1 mL of the organic layer and evaporate it naturally, redissolve it in methanol, and filter it through a 0.22 μm microporous filter membrane before UPLC-MS / MS analysis. QC and blank sample processing are as described above. When applying the above method in a blank experiment, it should be confirmed that the target substance to be measured is not interfered with.
[0020] 1.3.2 Liquid chromatography-mass spectrometry
[0021] Chromatographic conditions: BEH C18 columns (100 mm × 2.1 mm, 1.7 μm) were used, the column oven temperature was 35°C, and the eluents were A: acetonitrile and B: 10 mM ammonium formate in water. The elution program is shown in Table 1. The injection volume was 2 μL, and the flow rate was maintained at 0.4 mL / min.
[0022] Table 1. LC-MS / MS chromatographic conditions
[0023]
[0024] Mass spectrometry conditions included an electrospray ionization source (ESI+) in multiple reaction monitoring (MRM) mode; capillary voltage 0.50 kV; cone voltage 20 V; desolvation temperature 500°C; desolvation gas flow rate 1000 L / h; cone gas flow rate 100 L / h; collision gas: argon at a flow rate of 0.12 mL / min; and acquisition time 0–9.49 minutes. A total of 15 characteristic impurities were selected as analytical targets. The specific mass spectrometric parameters and reference retention times for each target are shown in Table 2.
[0025] Table 2 Formulas, semi-quantitative ion pairs, collision energies, reference abundance ratios, and reference retention times for the 15 targets
[0026]
[0027]
[0028] “*” indicates a semi-quantitative ion pair; the name of each target in the table is cited from Liu et al. (Liu et al., 2015).
[0029] 1.3.3 Methodological stability assessment
[0030] Quality control samples were pretreated according to standard methods. Following the study by Duan et al. (Duan et al. 2022), interday stability tests were performed six times on each of the three heroin samples on the same day. Interday stability tests were also performed once on each of the three heroin samples over four days. The relative standard deviations (RSD%) of the interday and intraday stability values were calculated.
[0031] 1.4 Results Analysis
[0032] 1.4.1 Qualitative analysis
[0033] The chromatogram and mass spectrum of the test sample were compared with those of the QC sample. If the following two conditions were met, the corresponding target substance was determined to have been detected: (1) The relative error (RE%) of the tR of the target substance in the test sample and the QC sample was less than 2%. (2) Two parent / daughter ion pairs of the target substance were detected in the test sample (Table 2), and the peak area ratios of the qualifier ion pair and the semi-quantitative ion pair did not exceed the specified RE% limit compared with the corresponding peak area ratios in the QC sample. Referring to the industry standard JD / Y JY03.01-2017 (JD / Y JY03.01-2017), the maximum allowable relative error of the peak area ratio is defined as: ≥50%, ±20%, 20%–50%, ±25%, 10%–20%, ±30%; ≤10%, ±50%.
[0034] 1.4.2 Pearson Correlation Analysis
[0035] When the number of targets detected in the sample to be tested is greater than or equal to 8, the peak areas of 15 targets are recorded (JD / YJY03.01-2017). Among them, the peak area of the detected target is the peak area of its semi-quantitative ion pair, and the peak area of the undetected target is recorded as 200. With reference to the research of Lociro (Lociro et al., 2008), Durourdy (Dujourdy et al., 2008) and standard method JD / Y JY03.01-2017 (JD / Y JY03.01-2017), the peak areas of 15 target substances are normalized, that is, the peak area of each target substance in the sample to be tested is divided by the sum of the peak areas of the 15 target substances in the sample to obtain the normalized peak area Si of each target substance. The correlation coefficient r between two samples is:
[0036]
[0037] Where r is the correlation coefficient between the two samples, and SXi and SYi are the normalized peak areas of each target in samples X and Y, respectively. If the correlation coefficient is greater than 0.98 and the correlation between samples X and Y is judged to be extremely strong, it is highly likely that the heroin samples in both cases originated from the same processing batch.
[0038] 1.5 Data Analysis
[0039] Data processing was performed using Microsoft Excel 2021, and one-way analysis of variance (ANOVA) was performed using IBM SPSS Statistical Software 27.0. To assess the similarity of sample composition, the principal component analysis (PCA) method was drawn using the free online platform Majorbio Cloud Platform (https: / / cloud.majorbio.com / page / tools / ).
[0040] 2. Results and Discussion
[0041] 2.1 Methodological stability assessment
[0042] To verify the stability of the experimental method, an inter-day stability test was conducted, using the RSDs of 15 target impurities to test the stability of the method. The results showed that the RSD range of the intra-day data was 0.2%–5.1%, and the RSD range of the inter-day data was 1.5%–7.4%.
[0043] 2.2 Qualitative analysis
[0044] In the qualitative analysis of 38 heroin experimental samples, the corresponding target peaks did not appear in the blank test, demonstrating the effectiveness of parallel analysis of the test samples. The retention time percent relative error (tR) between the QC samples and the authentic test samples is shown in Table 3. For all test and QC samples, the percent relative error (tR) for the 15 acidic / neutral targets was less than 2%. Furthermore, as shown in Table 4, the peak area ratios of the two parent / subion pairs of the target in the test samples, the percent relative error (RE) for the qualifier and semi-quantitative ion pairs in the QC samples to the corresponding peak area ratios were within the range of -10.83–13.74, all below the maximum allowable relative error.
[0045] Table 3-1. Relative error (RE%) of retention times of seven target substances in quality control samples and real samples to be tested
[0046]
[0047]
[0048]
[0049] Table 3-2. Relative error (RE%) of retention times of eight target substances in quality control samples and real samples to be tested
[0050]
[0051]
[0052] “—” indicates no peak; when the RE% value is 0, it is indicated by “ / ”; the name of each target in the table is cited from Liu et al. (Liu et al., 2015).
[0053] Table 4-1. Relative error (RE%) of the peak area ratios of the seven target substances in the quality control samples and the test samples
[0054]
[0055]
[0056]
[0057] Table 4-2. Relative error (RE%) of peak area ratios of eight target substances in quality control samples and test samples
[0058]
[0059]
[0060]
[0061] 2.3 Main groups of correlations by PCA
[0062] Principal component analysis (PCA) is a data mining technique in multivariate statistics. Its advantage is that it eliminates correlations between evaluation indicators, ensuring objectivity (Duan et al. 2022; Lever et al. 2017). PCA was used to reduce the order of the parameters of 15 target impurity variables in the samples, extracting two principal components, PC1 and PC2, with a cumulative contribution rate of 91.86%. PCA scatter plot analysis showed good separation between heroin samples seized in different case types. The sample data was initially divided into three regions, with the distance between each point representing the distance between samples. Samples in the same region on the plane appeared similar. Samples from cities A (P1-P2), B (P3-P27, P32-P33), and D (P34-P38) were well separated in the different regions. This may indicate that the sources or processing methods of the heroin samples seized in A, B, and D may differ. Using similar PCA methods and tools, Duan determined that the sources or processing of heroin samples seized in Dehong and Lincang may be different (Duan et al., 2022). Some samples seized in City A (P1-P2), City C (P28-P31), and City D (P37-P38) were distributed in the same area, and the distances between samples were relatively close. This suggests that samples from different cases may have similar sources. However, samples may be located in the same area of PCA but not actually related, mainly because the two data obtained by PCA downscaling only partially represent the information in the 15 data (Lever et al., 2017). Therefore, principal component analysis can be used as a preliminary estimate of sample correlation, but it cannot be used as a final basis for determining the correlation of heroin samples.
[0063] 2.4 Pearson Correlation Analysis
[0064] Organic impurity analysis is one of the earliest and most widely used analytical methods, providing both qualitative and quantitative results and enabling comparative analysis. Chemical composition varies with sample origin (Wang et al., 2006). In a cohort of 38 heroin samples, the number of target compounds detected was greater than eight in each sample. When the correlation coefficient r > 0.98, the correlation between two samples is considered extremely strong (JD / Y JY03.01-2017). Eleven highly correlated sample groups were identified, 10 of which contained 2-5 samples, and one contained 7 samples. Samples seized in City A (P1-P2) and City C (P28-P31) were highly correlated with each other. Samples P15, P18, P23, and P26 were all located in the same region in the PCA analysis and were highly correlated with other heroin samples in City B, where case information indicates these samples were all seized. However, no correlation was observed between samples from different cities. It is inferred that the heroin samples seized in City A, City B, and City C may have different sources or processing methods. Only P37 and P38 showed correlation between the heroin samples seized in City D. The results indicate that the correlated samples came from the same region and the same batch of seized samples, demonstrating the reliability of our correlation analysis method.
[0065] 3 Conclusion
[0066] This study used UPLC-MS / MS to establish a Pearson correlation analysis, analyzing the peak areas of 15 neutral / acidic impurities in heroin. A correlation coefficient of r > 0.98 was used as the correlation threshold for heroin samples. Characteristic analysis of 38 heroin samples seized from different cases identified 11 groups of highly correlated samples, each containing 2-7 samples. Assuming that the heroin samples in the same group share the same raw materials and production process, the similarity between samples from the same batch is extremely high. Furthermore, combining case information, relevant subgroups, and PCA results, it is speculated that the heroin samples seized in cities A, B, and C may have different sources or processing methods.
Claims
1. A method for characterizing heroin characteristics based on UPLC-MS / MS principal component analysis, characterized in that: Fifteen neutral / acidic target substances in illicit heroin products were detected and analyzed using UPLC-MS / MS. Sample analysis was completed using a combination of mathematical and statistical methods. Characteristic analysis of heroin samples from different sources was performed by combining case information, relevant subgroups, and Principal Component Analysis (PCA) results. PCA was initially used to group the samples, and the Pearson distance method was then used to calculate the correlations between each group of samples. An R value greater than 98 was selected as the threshold to identify groups of samples with strong correlations.
2. The method for characterizing heroin characteristics based on UPLC-MS / MS principal component analysis according to claim 1, characterized in that: Chromatographic conditions described: ACQUITY BEH C18 columns were 100 mm × 2.1 mm, 1.7 μm, the column oven temperature was 35 °C, the eluents were A: acetonitrile and B: 10 mM ammonium formate aqueous solution, and the flow rate was maintained at 0.4 mL / min.
3. The method for characterizing heroin characteristics based on UPLC-MS / MS principal component analysis according to claim 1, characterized in that: The mass spectrometry conditions are as follows: electrospray positive ion source (ESI+), multiple reaction monitoring (MRM) mode; capillary voltage 0.50 kV; cone voltage 20 V; desolvation temperature 500°C; desolvation gas flow rate 1000 liters / hour; cone gas flow rate 100 liters / hour; collision gas: argon, flow rate 0.12 ml / minute; acquisition time 0-9.49 minutes.
4. The method for characterizing heroin characteristics based on UPLC-MS / MS principal component analysis according to claim 1, characterized in that: The Pearson distance method: When the number of targets detected in the sample to be tested is greater than or equal to 8, the peak areas of the 15 targets are recorded. The peak area of the detected target is the peak area of its semi-quantitative ion pair, and the peak area of the undetected target is recorded as 200. The peak area of each target substance in the sample to be tested is divided by the sum of the peak areas of the 15 target substances in the sample to obtain the normalized peak area Si of each target substance. The correlation coefficient r between the two samples is: Where r is the correlation coefficient between the two samples, SXi and SYi are the normalized peak areas of each target in samples X and Y, respectively. If the correlation coefficient is greater than 0.98 and the correlation between samples X and Y is judged to be extremely strong, it is highly likely that the heroin samples in both cases originated from the same processing batch.