GC-EI-MS Canonical Discriminant Analysis Achieves 100% Classification Rate for Pentylone vs. 2,3-Pentylone Isomer Differentiation
In a direct head-to-head study, Liliedahl and Davidson (2021) applied canonical discriminant analysis (CDA) to GC-EI-MS data for the constitutional isomer set comprising eutylone, dibutylone, pentylone, and 2,3-pentylone. Using 15 ions and peak apex data, the CDA leave-one-out cross-validation (LOOCV) classification rate reached 100.0% for this isomer set, meaning all four compounds — including pentylone and its 2,3-isomer — were correctly identified in every cross-validation iteration [1]. Even when the number of ions used to construct the CDA model was reduced, classification rates remained at 98.1% or higher (99.4% when lowest-concentration samples were excluded) [1]. This demonstrates that the 2,3-pentylone isomer generates a quantitatively distinct and statistically separable GC-EI-MS fragmentation pattern from pentylone, despite their visually similar mass spectra.
| Evidence Dimension | GC-EI-MS multivariate classification accuracy (CDA LOOCV) |
|---|---|
| Target Compound Data | 100.0% correct classification as 2,3-pentylone (15-ion CDA model, peak apex data); ≥98.1% under ion-reduction strategies |
| Comparator Or Baseline | Pentylone (3,4-isomer): 100.0% correct classification within the same 4-compound constitutional isomer CDA model; eutylone and dibutylone also classified at 100.0% |
| Quantified Difference | 100.0% LOOCV for all four constitutional isomers including pentylone/2,3-pentylone pair; reduced-model classification 99.4% after low-concentration sample exclusion |
| Conditions | GC-EI-MS (70 eV), Agilent 7890B GC with 5977A MSD, HP-5MS UI column (30 m × 0.25 mm × 0.25 μm), CDA performed via SPSS, 15 most abundant ions used; sample concentrations: 0.01–0.50 mg/mL in methanol |
Why This Matters
Forensic laboratories can achieve definitive, statistically validated differentiation between pentylone and its 2,3-isomer using standard GC-EI-MS instrumentation combined with CDA, eliminating reliance on subjective visual spectral comparison for isomer identification in legal casework.
- [1] Liliedahl RE, Davidson JT. The differentiation of synthetic cathinone isomers using GC-EI-MS and multivariate analysis. Forensic Chemistry. 2021;26:100349. (Abstract and Results: LOOCV classification rates of 100.0% for constitutional isomer set; 98.1% and 99.4% under reduction strategies.) View Source
